Merge pull request #1095 from adamhathcock/copilot/add-github-action-nuget-release

Add GitHub Actions workflow for automated NuGet releases with multi-platform builds

.config/dotnet-tools.json

@@ -3,11 +3,11 @@
   "isRoot": true,
   "tools": {
     "csharpier": {
-      "version": "1.2.1",
+      "version": "1.2.4",
       "commands": [
         "csharpier"
       ],
       "rollForward": false
     }
   }
-}
+}

.github/COPILOT_AGENT_README.md

@@ -1,15 +0,0 @@
-# Copilot Coding Agent Configuration
-
-This repository includes a minimal opt-in configuration and CI workflow to allow the GitHub Copilot coding agent to open and validate PRs.
-
-- .copilot-agent.yml: opt-in config for automated agents
-- .github/agents/copilot-agent.yml: detailed agent policy configuration
-- .github/workflows/dotnetcore.yml: CI runs on PRs touching the solution, source, or tests to validate changes
-- AGENTS.md: general instructions for Copilot coding agent with project-specific guidelines
-
-Maintainers can adjust the allowed paths or disable the agent by editing or removing .copilot-agent.yml.
-
-Notes:
-- The agent can create, modify, and delete files within the allowed paths (src, tests, README.md, AGENTS.md)
-- All changes require review before merge
-- If build/test paths are different, update the workflow accordingly; this workflow targets SharpCompress.sln and the SharpCompress.Test test project.

.github/prompts/plan-async.prompt.md

@@ -0,0 +1,25 @@
+# Plan: Implement Missing Async Functionality in SharpCompress
+
+SharpCompress has async support for low-level stream operations and Reader/Writer APIs, but critical entry points (Archive.Open, factory methods, initialization) remain synchronous. This plan adds async overloads for all user-facing I/O operations and fixes existing async bugs, enabling full end-to-end async workflows.
+
+## Steps
+
+1. **Add async factory methods** to [ArchiveFactory.cs](src/SharpCompress/Factories/ArchiveFactory.cs), [ReaderFactory.cs](src/SharpCompress/Factories/ReaderFactory.cs), and [WriterFactory.cs](src/SharpCompress/Factories/WriterFactory.cs) with `OpenAsync` and `CreateAsync` overloads accepting `CancellationToken`
+
+2. **Implement async Open methods** on concrete archive types ([ZipArchive.cs](src/SharpCompress/Archives/Zip/ZipArchive.cs), [TarArchive.cs](src/SharpCompress/Archives/Tar/TarArchive.cs), [RarArchive.cs](src/SharpCompress/Archives/Rar/RarArchive.cs), [GZipArchive.cs](src/SharpCompress/Archives/GZip/GZipArchive.cs), [SevenZipArchive.cs](src/SharpCompress/Archives/SevenZip/SevenZipArchive.cs)) and reader types ([ZipReader.cs](src/SharpCompress/Readers/Zip/ZipReader.cs), [TarReader.cs](src/SharpCompress/Readers/Tar/TarReader.cs), etc.)
+
+3. **Convert archive initialization logic to async** including header reading, volume loading, and format signature detection across archive constructors and internal initialization methods
+
+4. **Fix LZMA decoder async bugs** in [LzmaStream.cs](src/SharpCompress/Compressors/LZMA/LzmaStream.cs), [Decoder.cs](src/SharpCompress/Compressors/LZMA/Decoder.cs), and [OutWindow.cs](src/SharpCompress/Compressors/LZMA/OutWindow.cs) to enable true async 7Zip support and remove `NonDisposingStream` workaround
+
+5. **Complete Rar async implementation** by converting `UnpackV2017` methods to async in [UnpackV2017.cs](src/SharpCompress/Compressors/Rar/UnpackV2017.cs) and updating Rar20 decompression
+
+6. **Add comprehensive async tests** covering all new async entry points, cancellation scenarios, and concurrent operations across all archive formats in test files
+
+## Further Considerations
+
+1. **Breaking changes** - Should new async methods be added alongside existing sync methods (non-breaking), or should sync methods eventually be deprecated? Recommend additive approach for backward compatibility.
+
+2. **Performance impact** - Header parsing for formats like Zip/Tar is often small; consider whether truly async parsing adds value vs sync parsing wrapped in Task, or make it conditional based on stream type (network vs file).
+
+3. **7Zip complexity** - The LZMA async bug fix (Step 4) may be challenging due to state management in the decoder; consider whether to scope it separately or implement a simpler workaround that maintains correctness.

.github/prompts/plan-for-next.prompt.md

@@ -0,0 +1,123 @@
+# Plan: Modernize SharpCompress Public API
+
+Based on comprehensive analysis, the API has several inconsistencies around factory patterns, async support, format capabilities, and options classes. Most improvements can be done incrementally without breaking changes.
+
+## Steps
+
+1. **Standardize factory patterns** by deprecating format-specific static `Open` methods in [Archives/Zip/ZipArchive.cs](src/SharpCompress/Archives/Zip/ZipArchive.cs), [Archives/Tar/TarArchive.cs](src/SharpCompress/Archives/Tar/TarArchive.cs), etc. in favor of centralized [Factories/ArchiveFactory.cs](src/SharpCompress/Factories/ArchiveFactory.cs)
+
+2. **Complete async implementation** in [Writers/Zip/ZipWriter.cs](src/SharpCompress/Writers/Zip/ZipWriter.cs) and other writers that currently use sync-over-async, implementing true async I/O throughout the writer hierarchy
+
+3. **Unify options classes** by making [Common/ExtractionOptions.cs](src/SharpCompress/Common/ExtractionOptions.cs) inherit from `OptionsBase` and adding progress reporting to extraction methods consistently
+
+4. **Clarify GZip semantics** in [Archives/GZip/GZipArchive.cs](src/SharpCompress/Archives/GZip/GZipArchive.cs) by adding XML documentation explaining single-entry limitation and relationship to GZip compression used in Tar.gz
+
+## Further Considerations
+
+1. **Breaking changes roadmap** - Should we plan a major version (2.0) to remove deprecated factory methods, clean up `ArchiveType` enum (remove Arc/Arj or add full support), and consolidate naming patterns?
+
+2. **Progress reporting consistency** - Should `IProgress<ArchiveExtractionProgress<IEntry>>` be added to all extraction extension methods or consolidated into options classes?
+
+## Detailed Analysis
+
+### Factory Pattern Issues
+
+Three different factory patterns exist with overlapping functionality:
+
+1. **Static Factories**: ArchiveFactory, ReaderFactory, WriterFactory
+2. **Instance Factories**: IArchiveFactory, IReaderFactory, IWriterFactory
+3. **Format-specific static methods**: Each archive class has static `Open` methods
+
+**Example confusion:**
+```csharp
+// Three ways to open a Zip archive - which is recommended?
+var archive1 = ArchiveFactory.Open("file.zip");
+var archive2 = ZipArchive.Open("file.zip");
+var archive3 = ArchiveFactory.AutoFactory.Open(fileInfo, options);
+```
+
+### Async Support Gaps
+
+Base `IWriter` interface has async methods, but writer implementations provide minimal async support. Most writers just call synchronous methods:
+
+```csharp
+public virtual async Task WriteAsync(...)
+{
+    // Default implementation calls synchronous version
+    Write(filename, source, modificationTime);
+    await Task.CompletedTask.ConfigureAwait(false);
+}
+```
+
+Real async implementations only in:
+- `TarWriter` - Proper async implementation
+- Most other writers use sync-over-async
+
+### GZip Archive Special Case
+
+GZip is treated as both a compression format and an archive format, but only supports single-entry archives:
+
+```csharp
+protected override GZipArchiveEntry CreateEntryInternal(...)
+{
+    if (Entries.Any())
+    {
+        throw new InvalidFormatException("Only one entry is allowed in a GZip Archive");
+    }
+    // ...
+}
+```
+
+### Options Class Hierarchy
+
+```
+OptionsBase (LeaveStreamOpen, ArchiveEncoding)
+  ├─ ReaderOptions (LookForHeader, Password, DisableCheckIncomplete, BufferSize, ExtensionHint, Progress)
+  ├─ WriterOptions (CompressionType, CompressionLevel, Progress)
+  │   ├─ ZipWriterOptions (ArchiveComment, UseZip64)
+  │   ├─ TarWriterOptions (FinalizeArchiveOnClose, HeaderFormat)
+  │   └─ GZipWriterOptions (no additional properties)
+  └─ ExtractionOptions (standalone - Overwrite, ExtractFullPath, PreserveFileTime, PreserveAttributes)
+```
+
+**Issues:**
+- `ExtractionOptions` doesn't inherit from `OptionsBase` - no encoding support during extraction
+- Progress reporting inconsistency between readers and extraction
+- Obsolete properties (`ChecksumIsValid`, `Version`) with unclear migration path
+
+### Implementation Priorities
+
+**High Priority (Non-Breaking):**
+1. Add API usage guide (Archive vs Reader, factory recommendations, async best practices)
+2. Fix progress reporting consistency
+3. Complete async implementation in writers
+
+**Medium Priority (Next Major Version):**
+1. Unify factory pattern - deprecate format-specific static `Open` methods
+2. Clean up options classes - make `ExtractionOptions` inherit from `OptionsBase`
+3. Clarify archive types - remove Arc/Arj from `ArchiveType` enum or add full support
+4. Standardize naming across archive types
+
+**Low Priority:**
+1. Add BZip2 archive support similar to GZipArchive
+2. Complete obsolete property cleanup with migration guide
+
+### Backward Compatibility Strategy
+
+**Safe (Non-Breaking) Changes:**
+- Add new methods to interfaces (use default implementations)
+- Add new options properties (with defaults)
+- Add new factory methods
+- Improve async implementations
+- Add progress reporting support
+
+**Breaking Changes to Avoid:**
+- ❌ Removing format-specific `Open` methods (deprecate instead)
+- ❌ Changing `LeaveStreamOpen` default (currently `true`)
+- ❌ Removing obsolete properties before major version bump
+- ❌ Changing return types or signatures of existing methods
+
+**Deprecation Pattern:**
+- Use `[Obsolete]` for one major version
+- Use `[EditorBrowsable(EditorBrowsableState.Never)]` in next major version
+- Remove in following major version

.github/workflows/NUGET_RELEASE.md

@@ -0,0 +1,155 @@
+# NuGet Release Workflow
+
+This document describes the automated NuGet release workflow for SharpCompress.
+
+## Overview
+
+The `nuget-release.yml` workflow automatically builds, tests, and publishes SharpCompress packages to NuGet.org when:
+- Changes are pushed to the `master` or `release` branch
+- A version tag (format: `MAJOR.MINOR.PATCH`) is pushed
+
+The workflow runs on both Windows and Ubuntu, but only the Windows build publishes to NuGet.
+
+## How It Works
+
+### Version Determination
+
+The workflow automatically determines the version based on whether the commit is tagged using C# code in the build project:
+
+1. **Tagged Release (Stable)**:
+   - If the current commit has a version tag (e.g., `0.42.1`)
+   - Uses the tag as the version number
+   - Published as a stable release
+
+2. **Untagged Release (Prerelease)**:
+   - If the current commit is NOT tagged
+   - Creates a prerelease version based on the next minor version
+   - Format: `{NEXT_MINOR_VERSION}-beta.{COMMIT_COUNT}`
+   - Example: `0.43.0-beta.123` (if last tag is 0.42.x)
+   - Published as a prerelease to NuGet.org (Windows build only)
+
+### Workflow Steps
+
+The workflow runs on a matrix of operating systems (Windows and Ubuntu):
+
+1. **Checkout**: Fetches the repository with full history for version detection
+2. **Setup .NET**: Installs .NET 10.0
+3. **Determine Version**: Runs `determine-version` build target to check for tags and determine version
+4. **Update Version**: Runs `update-version` build target to update the version in the project file
+5. **Build and Test**: Runs the full build and test suite on both platforms
+6. **Upload Artifacts**: Uploads the generated `.nupkg` files as workflow artifacts (separate for each OS)
+7. **Push to NuGet**: (Windows only) Runs `push-to-nuget` build target to publish the package to NuGet.org using the API key
+
+All version detection, file updates, and publishing logic is implemented in C# in the `build/Program.cs` file using build targets.
+
+## Setup Requirements
+
+### 1. NuGet API Key Secret
+
+The workflow requires a `NUGET_API_KEY` secret to be configured in the repository settings:
+
+1. Go to https://www.nuget.org/account/apikeys
+2. Create a new API key with "Push" permission for the SharpCompress package
+3. In GitHub, go to: **Settings** → **Secrets and variables** → **Actions**
+4. Create a new secret named `NUGET_API_KEY` with the API key value
+
+### 2. Branch Protection (Recommended)
+
+Consider enabling branch protection rules for the `release` branch to ensure:
+- Code reviews are required before merging
+- Status checks pass before merging
+- Only authorized users can push to the branch
+
+## Usage
+
+### Creating a Stable Release
+
+There are two ways to trigger a stable release:
+
+**Method 1: Push tag to trigger workflow**
+1. Ensure all changes are committed on the `master` or `release` branch
+2. Create and push a version tag:
+   ```bash
+   git checkout master  # or release
+   git tag 0.43.0
+   git push origin 0.43.0
+   ```
+3. The workflow will automatically trigger, build, test, and publish `SharpCompress 0.43.0` to NuGet.org (Windows build)
+
+**Method 2: Tag after pushing to branch**
+1. Ensure all changes are merged and pushed to the `master` or `release` branch
+2. Create and push a version tag on the already-pushed commit:
+   ```bash
+   git checkout master  # or release
+   git tag 0.43.0
+   git push origin 0.43.0
+   ```
+3. The workflow will automatically trigger, build, test, and publish `SharpCompress 0.43.0` to NuGet.org (Windows build)
+
+### Creating a Prerelease
+
+1. Push changes to the `master` or `release` branch without tagging:
+   ```bash
+   git checkout master  # or release
+   git push origin master  # or release
+   ```
+2. The workflow will automatically:
+   - Build and test the project on both Windows and Ubuntu
+   - Publish a prerelease version like `0.43.0-beta.456` to NuGet.org (Windows build)
+
+## Troubleshooting
+
+### Workflow Fails to Push to NuGet
+
+- **Check the API Key**: Ensure `NUGET_API_KEY` is set correctly in repository secrets
+- **Check API Key Permissions**: Verify the API key has "Push" permission for SharpCompress
+- **Check API Key Expiration**: NuGet API keys may expire; create a new one if needed
+
+### Version Conflict
+
+If you see "Package already exists" errors:
+- The workflow uses `--skip-duplicate` flag to handle this gracefully
+- If you need to republish the same version, delete it from NuGet.org first (if allowed)
+
+### Build or Test Failures
+
+- The workflow will not push to NuGet if build or tests fail
+- Check the workflow logs in GitHub Actions for details
+- Fix the issues and push again
+
+## Manual Package Creation
+
+If you need to create a package manually without publishing:
+
+```bash
+dotnet run --project build/build.csproj -- publish
+```
+
+The package will be created in the `artifacts/` directory.
+
+## Build Targets
+
+The workflow uses the following C# build targets defined in `build/Program.cs`:
+
+- **determine-version**: Detects version from git tags and outputs VERSION and PRERELEASE variables
+- **update-version**: Updates VersionPrefix, AssemblyVersion, and FileVersion in the project file
+- **push-to-nuget**: Pushes the generated NuGet packages to NuGet.org (requires NUGET_API_KEY)
+
+These targets can be run manually for testing:
+
+```bash
+# Determine the version
+dotnet run --project build/build.csproj -- determine-version
+
+# Update version in project file
+VERSION=0.43.0 dotnet run --project build/build.csproj -- update-version
+
+# Push to NuGet (requires NUGET_API_KEY environment variable)
+NUGET_API_KEY=your-key dotnet run --project build/build.csproj -- push-to-nuget
+```
+
+## Related Files
+
+- `.github/workflows/nuget-release.yml` - The workflow definition
+- `build/Program.cs` - Build script with version detection and publishing logic
+- `src/SharpCompress/SharpCompress.csproj` - Project file with version information

.github/workflows/TESTING.md

@@ -0,0 +1,120 @@
+# Testing Guide for NuGet Release Workflow
+
+This document describes how to test the NuGet release workflow.
+
+## Testing Strategy
+
+Since this workflow publishes to NuGet.org and requires repository secrets, testing should be done carefully. The workflow runs on both Windows and Ubuntu, but only the Windows build publishes to NuGet.
+
+## Pre-Testing Checklist
+
+- [x] Workflow YAML syntax validated
+- [x] Version determination logic tested locally
+- [x] Version update logic tested locally
+- [x] Build script works (`dotnet run --project build/build.csproj`)
+
+## Manual Testing Steps
+
+### 1. Test Prerelease Publishing (Recommended First Test)
+
+This tests the workflow on untagged commits to the master or release branch.
+
+**Steps:**
+1. Ensure `NUGET_API_KEY` secret is configured in repository settings
+2. Create a test commit on the `master` or `release` branch (e.g., update a comment or README)
+3. Push to the `master` or `release` branch
+4. Monitor the GitHub Actions workflow at: https://github.com/adamhathcock/sharpcompress/actions
+5. Verify:
+   - Workflow triggers and runs successfully on both Windows and Ubuntu
+   - Version is determined correctly (e.g., `0.43.0-beta.XXX` if last tag is 0.42.x)
+   - Build and tests pass on both platforms
+   - Package artifacts are uploaded for both platforms
+   - Package is pushed to NuGet.org as prerelease (Windows build only)
+
+**Expected Outcome:**
+- A new prerelease package appears on NuGet.org: https://www.nuget.org/packages/SharpCompress/
+- Package version follows pattern: `{NEXT_MINOR_VERSION}-beta.{COMMIT_COUNT}`
+
+### 2. Test Tagged Release Publishing
+
+This tests the workflow when a version tag is pushed.
+
+**Steps:**
+1. Prepare the `master` or `release` branch with all desired changes
+2. Create a version tag (must be a pure semantic version like `MAJOR.MINOR.PATCH`):
+   ```bash
+   git checkout master  # or release
+   git tag 0.42.2
+   git push origin 0.42.2
+   ```
+3. Monitor the GitHub Actions workflow
+4. Verify:
+   - Workflow triggers and runs successfully on both Windows and Ubuntu
+   - Version is determined as the tag (e.g., `0.42.2`)
+   - Build and tests pass on both platforms
+   - Package artifacts are uploaded for both platforms
+   - Package is pushed to NuGet.org as stable release (Windows build only)
+
+**Expected Outcome:**
+- A new stable release package appears on NuGet.org
+- Package version matches the tag
+
+### 3. Test Duplicate Package Handling
+
+This tests the `--skip-duplicate` flag behavior.
+
+**Steps:**
+1. Push to the `release` branch without making changes
+2. Monitor the workflow
+3. Verify:
+   - Workflow runs but NuGet push is skipped with "duplicate" message
+   - No errors occur
+
+### 4. Test Build Failure Handling
+
+This tests that failed builds don't publish packages.
+
+**Steps:**
+1. Introduce a breaking change in a test or code
+2. Push to the `release` branch
+3. Verify:
+   - Workflow runs and detects the failure
+   - Build or test step fails
+   - NuGet push step is skipped
+   - No package is published
+
+## Verification
+
+After each test, verify:
+
+1. **GitHub Actions Logs**: Check the workflow logs for any errors or warnings
+2. **NuGet.org**: Verify the package appears with correct version and metadata
+3. **Artifacts**: Download and inspect the uploaded artifacts
+
+## Rollback/Cleanup
+
+If testing produces unwanted packages:
+
+1. **Prerelease packages**: Can be unlisted on NuGet.org (Settings → Unlist)
+2. **Stable packages**: Cannot be deleted, only unlisted (use test versions)
+3. **Tags**: Can be deleted with:
+   ```bash
+   git tag -d 0.42.2
+   git push origin :refs/tags/0.42.2
+   ```
+
+## Known Limitations
+
+- NuGet.org does not allow re-uploading the same version
+- Deleted packages on NuGet.org reserve the version number
+- The workflow requires the `NUGET_API_KEY` secret to be set
+
+## Success Criteria
+
+The workflow is considered successful if:
+
+- ✅ Prerelease versions are published correctly with beta suffix
+- ✅ Tagged versions are published as stable releases
+- ✅ Build and test failures prevent publishing
+- ✅ Duplicate packages are handled gracefully
+- ✅ Workflow logs are clear and informative

.github/workflows/dotnetcore.yml

@@ -17,9 +17,9 @@ jobs:
     - uses: actions/checkout@v6
     - uses: actions/setup-dotnet@v5
       with:
-        dotnet-version: 8.0.x
+        dotnet-version: 10.0.x
     - run: dotnet run --project build/build.csproj
-    - uses: actions/upload-artifact@v5
+    - uses: actions/upload-artifact@v6
       with:
         name: ${{ matrix.os }}-sharpcompress.nupkg
         path: artifacts/*

.github/workflows/nuget-release.yml

@@ -0,0 +1,57 @@
+name: NuGet Release
+
+on:
+  push:
+    branches:
+      - 'master'
+      - 'release'
+    tags:
+      - '[0-9]+.[0-9]+.[0-9]+'
+
+permissions:
+  contents: read
+
+jobs:
+  build-and-publish:
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [windows-latest, ubuntu-latest]
+    
+    steps:
+    - uses: actions/checkout@v6
+      with:
+        fetch-depth: 0  # Fetch all history for versioning
+    
+    - uses: actions/setup-dotnet@v5
+      with:
+        dotnet-version: 10.0.x
+    
+    # Determine version using C# build target
+    - name: Determine Version
+      id: version
+      run: dotnet run --project build/build.csproj -- determine-version
+    
+    # Update version in project file using C# build target
+    - name: Update Version in Project
+      run: dotnet run --project build/build.csproj -- update-version
+      env:
+        VERSION: ${{ steps.version.outputs.version }}
+    
+    # Build and test
+    - name: Build and Test
+      run: dotnet run --project build/build.csproj
+    
+    # Upload artifacts for verification
+    - name: Upload NuGet Package
+      uses: actions/upload-artifact@v6
+      with:
+        name: ${{ matrix.os }}-nuget-package
+        path: artifacts/*.nupkg
+    
+    # Push to NuGet.org using C# build target (Windows only)
+    - name: Push to NuGet
+      if: success() && matrix.os == 'windows-latest'
+      run: dotnet run --project build/build.csproj -- push-to-nuget
+      env:
+        NUGET_API_KEY: ${{ secrets.NUGET_API_KEY }}

.gitignore

@@ -15,8 +15,8 @@ tests/TestArchives/*/Scratch
 tests/TestArchives/*/Scratch2
 .vs
 tools
-.vscode
 .idea/
+artifacts/
 
 .DS_Store
 *.snupkg

.vscode/extensions.json

@@ -0,0 +1,9 @@
+{
+  "recommendations": [
+    "ms-dotnettools.csdevkit",
+    "ms-dotnettools.csharp",
+    "ms-dotnettools.vscode-dotnet-runtime",
+    "csharpier.csharpier-vscode",
+    "formulahendry.dotnet-test-explorer"
+  ]
+}

.vscode/launch.json

@@ -0,0 +1,97 @@
+{
+  "version": "0.2.0",
+  "configurations": [
+    {
+      "name": "Debug Tests (net10.0)",
+      "type": "coreclr",
+      "request": "launch",
+      "preLaunchTask": "build",
+      "program": "dotnet",
+      "args": [
+        "test",
+        "${workspaceFolder}/tests/SharpCompress.Test/SharpCompress.Test.csproj",
+        "-f",
+        "net10.0",
+        "--no-build",
+        "--verbosity=normal"
+      ],
+      "cwd": "${workspaceFolder}",
+      "console": "internalConsole",
+      "stopAtEntry": false
+    },
+    {
+      "name": "Debug Specific Test (net10.0)",
+      "type": "coreclr",
+      "request": "launch",
+      "preLaunchTask": "build",
+      "program": "dotnet",
+      "args": [
+        "test",
+        "${workspaceFolder}/tests/SharpCompress.Test/SharpCompress.Test.csproj",
+        "-f",
+        "net10.0",
+        "--no-build",
+        "--filter",
+        "FullyQualifiedName~${input:testName}"
+      ],
+      "cwd": "${workspaceFolder}",
+      "console": "internalConsole",
+      "stopAtEntry": false
+    },
+    {
+      "name": "Debug Performance Tests",
+      "type": "coreclr",
+      "request": "launch",
+      "preLaunchTask": "build",
+      "program": "dotnet",
+      "args": [
+        "run",
+        "--project",
+        "${workspaceFolder}/tests/SharpCompress.Performance/SharpCompress.Performance.csproj",
+        "--no-build"
+      ],
+      "cwd": "${workspaceFolder}",
+      "console": "internalConsole",
+      "stopAtEntry": false
+    },
+    {
+      "name": "Debug Build Script",
+      "type": "coreclr",
+      "request": "launch",
+      "program": "dotnet",
+      "args": [
+        "run",
+        "--project",
+        "${workspaceFolder}/build/build.csproj",
+        "--",
+        "${input:buildTarget}"
+      ],
+      "cwd": "${workspaceFolder}",
+      "console": "internalConsole",
+      "stopAtEntry": false
+    }
+  ],
+  "inputs": [
+    {
+      "id": "testName",
+      "type": "promptString",
+      "description": "Enter test name or pattern (e.g., TestMethodName or ClassName)",
+      "default": ""
+    },
+    {
+      "id": "buildTarget",
+      "type": "pickString",
+      "description": "Select build target",
+      "options": [
+        "clean",
+        "restore",
+        "build",
+        "test",
+        "format",
+        "publish",
+        "default"
+      ],
+      "default": "build"
+    }
+  ]
+}

.vscode/settings.json

@@ -0,0 +1,29 @@
+{
+  "dotnet.defaultSolution": "SharpCompress.sln",
+  "files.exclude": {
+    "**/bin": true,
+    "**/obj": true
+  },
+  "files.watcherExclude": {
+    "**/bin/**": true,
+    "**/obj/**": true,
+    "**/artifacts/**": true
+  },
+  "search.exclude": {
+    "**/bin": true,
+    "**/obj": true,
+    "**/artifacts": true
+  },
+  "editor.formatOnSave": false,
+  "[csharp]": {
+    "editor.defaultFormatter": "csharpier.csharpier-vscode",
+    "editor.formatOnSave": true,
+    "editor.codeActionsOnSave": {
+      "source.fixAll": "explicit"
+    }
+  },
+  "csharpier.enableDebugLogs": false,
+  "omnisharp.enableRoslynAnalyzers": true,
+  "omnisharp.enableEditorConfigSupport": true,
+  "dotnet-test-explorer.testProjectPath": "tests/**/*.csproj"
+}

.vscode/tasks.json

@@ -0,0 +1,178 @@
+{
+  "version": "2.0.0",
+  "tasks": [
+    {
+      "label": "build",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "build",
+        "${workspaceFolder}/SharpCompress.sln",
+        "/property:GenerateFullPaths=true",
+        "/consoleloggerparameters:NoSummary;ForceNoAlign"
+      ],
+      "problemMatcher": "$msCompile",
+      "group": {
+        "kind": "build",
+        "isDefault": true
+      }
+    },
+    {
+      "label": "build-release",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "build",
+        "${workspaceFolder}/SharpCompress.sln",
+        "-c",
+        "Release",
+        "/property:GenerateFullPaths=true",
+        "/consoleloggerparameters:NoSummary;ForceNoAlign"
+      ],
+      "problemMatcher": "$msCompile",
+      "group": "build"
+    },
+    {
+      "label": "build-library",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "build",
+        "${workspaceFolder}/src/SharpCompress/SharpCompress.csproj",
+        "/property:GenerateFullPaths=true",
+        "/consoleloggerparameters:NoSummary;ForceNoAlign"
+      ],
+      "problemMatcher": "$msCompile",
+      "group": "build"
+    },
+    {
+      "label": "restore",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "restore",
+        "${workspaceFolder}/SharpCompress.sln"
+      ],
+      "problemMatcher": "$msCompile"
+    },
+    {
+      "label": "clean",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "clean",
+        "${workspaceFolder}/SharpCompress.sln"
+      ],
+      "problemMatcher": "$msCompile"
+    },
+    {
+      "label": "test",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "test",
+        "${workspaceFolder}/tests/SharpCompress.Test/SharpCompress.Test.csproj",
+        "--no-build",
+        "--verbosity=normal"
+      ],
+      "problemMatcher": "$msCompile",
+      "group": {
+        "kind": "test",
+        "isDefault": true
+      },
+      "dependsOn": "build"
+    },
+    {
+      "label": "test-net10",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "test",
+        "${workspaceFolder}/tests/SharpCompress.Test/SharpCompress.Test.csproj",
+        "-f",
+        "net10.0",
+        "--no-build",
+        "--verbosity=normal"
+      ],
+      "problemMatcher": "$msCompile",
+      "group": "test",
+      "dependsOn": "build"
+    },
+    {
+      "label": "test-net48",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "test",
+        "${workspaceFolder}/tests/SharpCompress.Test/SharpCompress.Test.csproj",
+        "-f",
+        "net48",
+        "--no-build",
+        "--verbosity=normal"
+      ],
+      "problemMatcher": "$msCompile",
+      "group": "test",
+      "dependsOn": "build"
+    },
+    {
+      "label": "format",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "csharpier",
+        "."
+      ],
+      "problemMatcher": []
+    },
+    {
+      "label": "format-check",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "csharpier",
+        "check",
+        "."
+      ],
+      "problemMatcher": []
+    },
+    {
+      "label": "run-build-script",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "run",
+        "--project",
+        "${workspaceFolder}/build/build.csproj"
+      ],
+      "problemMatcher": "$msCompile"
+    },
+    {
+      "label": "pack",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "pack",
+        "${workspaceFolder}/src/SharpCompress/SharpCompress.csproj",
+        "-c",
+        "Release",
+        "-o",
+        "${workspaceFolder}/artifacts/"
+      ],
+      "problemMatcher": "$msCompile",
+      "dependsOn": "build-release"
+    },
+    {
+      "label": "performance-tests",
+      "command": "dotnet",
+      "type": "process",
+      "args": [
+        "run",
+        "--project",
+        "${workspaceFolder}/tests/SharpCompress.Performance/SharpCompress.Performance.csproj",
+        "-c",
+        "Release"
+      ],
+      "problemMatcher": "$msCompile"
+    }
+  ]
+}

AGENTS.md

@@ -28,14 +28,38 @@ SharpCompress is a pure C# compression library supporting multiple archive forma
 
 ## Code Formatting
 
+**Copilot agents: You MUST run the `format` task after making code changes to ensure consistency.**
+
 - Use CSharpier for code formatting to ensure consistent style across the project
 - CSharpier is configured as a local tool in `.config/dotnet-tools.json`
-- Restore tools with: `dotnet tool restore`
-- Format files from the project root with: `dotnet csharpier .`
-- **Run `dotnet csharpier .` from the project root after making code changes before committing**
-- Configure your IDE to format on save using CSharpier for the best experience
+
+### Commands
+
+1. **Restore tools** (first time only):
+   ```bash
+   dotnet tool restore
+   ```
+
+2. **Check if files are formatted correctly** (doesn't modify files):
+   ```bash
+   dotnet csharpier check .
+   ```
+   - Exit code 0: All files are properly formatted
+   - Exit code 1: Some files need formatting (will show which files and differences)
+
+3. **Format files** (modifies files):
+   ```bash
+   dotnet csharpier format .
+   ```
+   - Formats all files in the project to match CSharpier style
+   - Run from project root directory
+
+4. **Configure your IDE** to format on save using CSharpier for the best experience
+
+### Additional Notes
 - The project also uses `.editorconfig` for editor settings (indentation, encoding, etc.)
 - Let CSharpier handle code style while `.editorconfig` handles editor behavior
+- Always run `dotnet csharpier check .` before committing to verify formatting
 
 ## Project Setup and Structure
 
@@ -49,6 +73,30 @@ SharpCompress is a pure C# compression library supporting multiple archive forma
 - Use `dotnet test` to run tests
 - Solution file: `SharpCompress.sln`
 
+### Directory Structure
+```
+src/SharpCompress/
+  ├── Archives/        # IArchive implementations (Zip, Tar, Rar, 7Zip, GZip)
+  ├── Readers/         # IReader implementations (forward-only)
+  ├── Writers/         # IWriter implementations (forward-only)
+  ├── Compressors/     # Low-level compression streams (BZip2, Deflate, LZMA, etc.)
+  ├── Factories/       # Format detection and factory pattern
+  ├── Common/          # Shared types (ArchiveType, Entry, Options)
+  ├── Crypto/          # Encryption implementations
+  └── IO/              # Stream utilities and wrappers
+
+tests/SharpCompress.Test/
+  ├── Zip/, Tar/, Rar/, SevenZip/, GZip/, BZip2/  # Format-specific tests
+  ├── TestBase.cs      # Base test class with helper methods
+  └── TestArchives/    # Test data (not checked into main test project)
+```
+
+### Factory Pattern
+All format types implement factory interfaces (`IArchiveFactory`, `IReaderFactory`, `IWriterFactory`) for auto-detection:
+- `ReaderFactory.Open()` - Auto-detects format by probing stream
+- `WriterFactory.Open()` - Creates writer for specified `ArchiveType`
+- Factories located in: `src/SharpCompress/Factories/`
+
 ## Nullable Reference Types
 
 - Declare variables non-nullable, and check for `null` at entry points.
@@ -116,3 +164,18 @@ SharpCompress supports multiple archive and compression formats:
 - Use test archives from `tests/TestArchives` directory for consistency.
 - Test stream disposal and `LeaveStreamOpen` behavior.
 - Test edge cases: empty archives, large files, corrupted archives, encrypted archives.
+
+### Test Organization
+- Base class: `TestBase` - Provides `TEST_ARCHIVES_PATH`, `SCRATCH_FILES_PATH`, temp directory management
+- Framework: xUnit with AwesomeAssertions
+- Test archives: `tests/TestArchives/` - Use existing archives, don't create new ones unnecessarily
+- Match naming style of nearby test files
+
+## Common Pitfalls
+
+1. **Don't mix Archive and Reader APIs** - Archive needs seekable stream, Reader doesn't
+2. **Solid archives (Rar, 7Zip)** - Use `ExtractAllEntries()` for best performance, not individual entry extraction
+3. **Stream disposal** - Always set `LeaveStreamOpen` explicitly when needed (default is to close)
+4. **Tar + non-seekable stream** - Must provide file size or it will throw
+5. **Multi-framework differences** - Some features differ between .NET Framework and modern .NET (e.g., Mono.Posix)
+6. **Format detection** - Use `ReaderFactory.Open()` for auto-detection, test with actual archive files

Directory.Packages.props

@@ -1,20 +1,19 @@
 <Project>
   <ItemGroup>
-    <PackageVersion Include="Bullseye" Version="6.0.0" />
+    <PackageVersion Include="Bullseye" Version="6.1.0" />
     <PackageVersion Include="AwesomeAssertions" Version="9.3.0" />
     <PackageVersion Include="Glob" Version="1.1.9" />
-    <PackageVersion Include="JetBrains.Profiler.SelfApi" Version="2.5.14" />
-    <PackageVersion Include="Microsoft.Bcl.AsyncInterfaces" Version="8.0.0" />
-    <PackageVersion Include="Microsoft.NET.Test.Sdk" Version="18.0.0" />
+    <PackageVersion Include="JetBrains.Profiler.SelfApi" Version="2.5.15" />
+    <PackageVersion Include="Microsoft.Bcl.AsyncInterfaces" Version="10.0.0" />
+    <PackageVersion Include="Microsoft.NET.Test.Sdk" Version="18.0.1" />
     <PackageVersion Include="Mono.Posix.NETStandard" Version="1.0.0" />
-    <PackageVersion Include="SimpleExec" Version="12.0.0" />
+    <PackageVersion Include="SimpleExec" Version="12.1.0" />
+    <PackageVersion Include="System.Text.Encoding.CodePages" Version="10.0.0" />
     <PackageVersion Include="System.Buffers" Version="4.6.1" />
     <PackageVersion Include="System.Memory" Version="4.6.3" />
-    <PackageVersion Include="System.Text.Encoding.CodePages" Version="8.0.0" />
     <PackageVersion Include="xunit" Version="2.9.3" />
     <PackageVersion Include="xunit.runner.visualstudio" Version="3.1.5" />
-    <PackageVersion Include="ZstdSharp.Port" Version="0.8.6" />
-    <PackageVersion Include="Microsoft.NET.ILLink.Tasks" Version="8.0.21" />
+    <PackageVersion Include="Microsoft.NET.ILLink.Tasks" Version="10.0.0" />
     <PackageVersion Include="Microsoft.SourceLink.GitHub" Version="8.0.0" />
     <PackageVersion Include="Microsoft.NETFramework.ReferenceAssemblies" Version="1.0.3" />
   </ItemGroup>

FORMATS.md

@@ -22,11 +22,28 @@
 | 7Zip (4)               | LZMA, LZMA2, BZip2, PPMd, BCJ, BCJ2, Deflate      | Decompress          | SevenZipArchive | N/A        | N/A           |
 
 1. SOLID Rars are only supported in the RarReader API.
-2. Zip format supports pkware and WinzipAES encryption. However, encrypted LZMA is not supported. Zip64 reading/writing is supported but only with seekable streams as the Zip spec doesn't support Zip64 data in post data descriptors. Deflate64 is only supported for reading.
+2. Zip format supports pkware and WinzipAES encryption. However, encrypted LZMA is not supported. Zip64 reading/writing is supported but only with seekable streams as the Zip spec doesn't support Zip64 data in post data descriptors. Deflate64 is only supported for reading. See [Zip Format Notes](#zip-format-notes) for details on multi-volume archives and streaming behavior.
 3. The Tar format requires a file size in the header. If no size is specified to the TarWriter and the stream is not seekable, then an exception will be thrown.
-4. The 7Zip format doesn't allow for reading as a forward-only stream so 7Zip is only supported through the Archive API
+4. The 7Zip format doesn't allow for reading as a forward-only stream so 7Zip is only supported through the Archive API. See [7Zip Format Notes](#7zip-format-notes) for details on async extraction behavior.
 5. LZip has no support for extra data like the file name or timestamp. There is a default filename used when looking at the entry Key on the archive.
 
+### Zip Format Notes
+
+- Multi-volume/split ZIP archives require ZipArchive (seekable streams) as ZipReader cannot seek across volume files.
+- ZipReader processes entries from LocalEntry headers (which include directory entries ending with `/`) and intentionally skips DirectoryEntry headers from the central directory, as they are redundant in streaming mode - all entry data comes from LocalEntry headers which ZipReader has already processed.
+
+### 7Zip Format Notes
+
+- **Async Extraction Performance**: When using async extraction methods (e.g., `ExtractAllEntries()` with `MoveToNextEntryAsync()`), each file creates its own decompression stream to avoid state corruption in the LZMA decoder. This is less efficient than synchronous extraction, which can reuse a single decompression stream for multiple files in the same folder.
+  
+  **Performance Impact**: For archives with many small files in the same compression folder, async extraction will be slower than synchronous extraction because it must:
+  1. Create a new LZMA decoder for each file
+  2. Skip through the decompressed data to reach each file's starting position
+  
+  **Recommendation**: For best performance with 7Zip archives, use synchronous extraction methods (`MoveToNextEntry()` and `WriteEntryToDirectory()`) when possible. Use async methods only when you need to avoid blocking the thread (e.g., in UI applications or async-only contexts).
+  
+  **Technical Details**: 7Zip archives group files into "folders" (compression units), where all files in a folder share one continuous LZMA-compressed stream. The LZMA decoder maintains internal state (dictionary window, decoder positions) that assumes sequential, non-interruptible processing. Async operations can yield control during awaits, which would corrupt this shared state. To avoid this, async extraction creates a fresh decoder stream for each file.
+
 ## Compression Streams
 
 For those who want to directly compress/decompress bits. The single file formats are represented here as well. However, BZip2, LZip and XZ have no metadata (GZip has a little) so using them without something like a Tar file makes little sense.

README.md

@@ -1,6 +1,6 @@
 # SharpCompress
 
-SharpCompress is a compression library in pure C# for .NET Framework 4.62, .NET Standard 2.1, .NET 6.0 and NET 8.0 that can unrar, un7zip, unzip, untar unbzip2, ungzip, unlzip, unzstd, unarc and unarj with forward-only reading and file random access APIs. Write support for zip/tar/bzip2/gzip/lzip are implemented.
+SharpCompress is a compression library in pure C# for .NET Framework 4.8, .NET 8.0 and .NET 10.0 that can unrar, un7zip, unzip, untar unbzip2, ungzip, unlzip, unzstd, unarc and unarj with forward-only reading and file random access APIs. Write support for zip/tar/bzip2/gzip/lzip are implemented.
 
 The major feature is support for non-seekable streams so large files can be processed on the fly (i.e. download stream).
 

USAGE.md

@@ -87,20 +87,17 @@ memoryStream.Position = 0;
 ### Extract all files from a rar file to a directory using RarArchive
 
 Note: Extracting a solid rar or 7z file needs to be done in sequential order to get acceptable decompression speed.
-It is explicitly recommended to use `ExtractAllEntries` when extracting an entire `IArchive` instead of iterating over all its `Entries`.
-Alternatively, use `IArchive.WriteToDirectory`.
+`ExtractAllEntries` is primarily intended for solid archives (like solid Rar) or 7Zip archives, where sequential extraction provides the best performance. For general/simple extraction with any supported archive type, use `archive.WriteToDirectory()` instead.
 
 ```C#
 using (var archive = RarArchive.Open("Test.rar"))
 {
-    using (var reader = archive.ExtractAllEntries())
+    // Simple extraction with RarArchive; this WriteToDirectory pattern works for all archive types
+    archive.WriteToDirectory(@"D:\temp", new ExtractionOptions()
     {
-        reader.WriteAllToDirectory(@"D:\temp", new ExtractionOptions()
-        {
-            ExtractFullPath = true,
-            Overwrite = true
-        });
-    }
+        ExtractFullPath = true,
+        Overwrite = true
+    });
 }
 ```
 
@@ -116,6 +113,41 @@ using (var archive = RarArchive.Open("Test.rar"))
 }
 ```
 
+### Extract solid Rar or 7Zip archives with manual progress reporting
+
+`ExtractAllEntries` only works for solid archives (Rar) or 7Zip archives. For optimal performance with these archive types, use this method:
+
+```C#
+using (var archive = RarArchive.Open("archive.rar")) // Must be solid Rar or 7Zip
+{
+    if (archive.IsSolid || archive.Type == ArchiveType.SevenZip)
+    {
+        // Calculate total size for progress reporting
+        double totalSize = archive.Entries.Where(e => !e.IsDirectory).Sum(e => e.Size);
+        long completed = 0;
+
+        using (var reader = archive.ExtractAllEntries())
+        {
+            while (reader.MoveToNextEntry())
+            {
+                if (!reader.Entry.IsDirectory)
+                {
+                    reader.WriteEntryToDirectory(@"D:\output", new ExtractionOptions()
+                    {
+                        ExtractFullPath = true,
+                        Overwrite = true
+                    });
+
+                    completed += reader.Entry.Size;
+                    double progress = completed / totalSize;
+                    Console.WriteLine($"Progress: {progress:P}");
+                }
+            }
+        }
+    }
+}
+```
+
 ### Use ReaderFactory to autodetect archive type and Open the entry stream
 
 ```C#
@@ -298,14 +330,12 @@ using (var writer = WriterFactory.Open(stream, ArchiveType.Zip, CompressionType.
 ```C#
 using (var archive = ZipArchive.Open("archive.zip"))
 {
-    using (var reader = archive.ExtractAllEntries())
-    {
-        await reader.WriteAllToDirectoryAsync(
-            @"C:\output",
-            new ExtractionOptions() { ExtractFullPath = true, Overwrite = true },
-            cancellationToken
-        );
-    }
+    // Simple async extraction - works for all archive types
+    await archive.WriteToDirectoryAsync(
+        @"C:\output",
+        new ExtractionOptions() { ExtractFullPath = true, Overwrite = true },
+        cancellationToken
+    );
 }
 ```
 

build/Program.cs

@@ -1,7 +1,10 @@
 using System;
 using System.Collections.Generic;
 using System.IO;
+using System.Linq;
 using System.Runtime.InteropServices;
+using System.Text.RegularExpressions;
+using System.Threading.Tasks;
 using GlobExpressions;
 using static Bullseye.Targets;
 using static SimpleExec.Command;
@@ -11,7 +14,11 @@ const string Restore = "restore";
 const string Build = "build";
 const string Test = "test";
 const string Format = "format";
+const string CheckFormat = "check-format";
 const string Publish = "publish";
+const string DetermineVersion = "determine-version";
+const string UpdateVersion = "update-version";
+const string PushToNuGet = "push-to-nuget";
 
 Target(
     Clean,
@@ -42,12 +49,20 @@ Target(
 Target(
     Format,
     () =>
+    {
+        Run("dotnet", "tool restore");
+        Run("dotnet", "csharpier format .");
+    }
+);
+Target(
+    CheckFormat,
+    () =>
     {
         Run("dotnet", "tool restore");
         Run("dotnet", "csharpier check .");
     }
 );
-Target(Restore, [Format], () => Run("dotnet", "restore"));
+Target(Restore, [CheckFormat], () => Run("dotnet", "restore"));
 
 Target(
     Build,
@@ -61,7 +76,7 @@ Target(
 Target(
     Test,
     [Build],
-    ["net8.0", "net48"],
+    ["net10.0", "net48"],
     framework =>
     {
         IEnumerable<string> GetFiles(string d)
@@ -90,6 +105,164 @@ Target(
     }
 );
 
+Target(
+    DetermineVersion,
+    async () =>
+    {
+        var (version, isPrerelease) = await GetVersion();
+        Console.WriteLine($"VERSION={version}");
+        Console.WriteLine($"PRERELEASE={isPrerelease.ToString().ToLower()}");
+
+        // Write to environment file for GitHub Actions
+        var githubOutput = Environment.GetEnvironmentVariable("GITHUB_OUTPUT");
+        if (!string.IsNullOrEmpty(githubOutput))
+        {
+            File.AppendAllText(githubOutput, $"version={version}\n");
+            File.AppendAllText(githubOutput, $"prerelease={isPrerelease.ToString().ToLower()}\n");
+        }
+    }
+);
+
+Target(
+    UpdateVersion,
+    async () =>
+    {
+        var version = Environment.GetEnvironmentVariable("VERSION");
+        if (string.IsNullOrEmpty(version))
+        {
+            var (detectedVersion, _) = await GetVersion();
+            version = detectedVersion;
+        }
+
+        Console.WriteLine($"Updating project file with version: {version}");
+
+        var projectPath = "src/SharpCompress/SharpCompress.csproj";
+        var content = File.ReadAllText(projectPath);
+
+        // Get base version (without prerelease suffix)
+        var baseVersion = version.Split('-')[0];
+
+        // Update VersionPrefix
+        content = Regex.Replace(
+            content,
+            @"<VersionPrefix>[^<]*</VersionPrefix>",
+            $"<VersionPrefix>{version}</VersionPrefix>"
+        );
+
+        // Update AssemblyVersion
+        content = Regex.Replace(
+            content,
+            @"<AssemblyVersion>[^<]*</AssemblyVersion>",
+            $"<AssemblyVersion>{baseVersion}</AssemblyVersion>"
+        );
+
+        // Update FileVersion
+        content = Regex.Replace(
+            content,
+            @"<FileVersion>[^<]*</FileVersion>",
+            $"<FileVersion>{baseVersion}</FileVersion>"
+        );
+
+        File.WriteAllText(projectPath, content);
+        Console.WriteLine($"Updated VersionPrefix to: {version}");
+        Console.WriteLine($"Updated AssemblyVersion and FileVersion to: {baseVersion}");
+    }
+);
+
+Target(
+    PushToNuGet,
+    () =>
+    {
+        var apiKey = Environment.GetEnvironmentVariable("NUGET_API_KEY");
+        if (string.IsNullOrEmpty(apiKey))
+        {
+            Console.WriteLine(
+                "NUGET_API_KEY environment variable is not set. Skipping NuGet push."
+            );
+            return;
+        }
+
+        var packages = Directory.GetFiles("artifacts", "*.nupkg");
+        if (packages.Length == 0)
+        {
+            Console.WriteLine("No packages found in artifacts directory.");
+            return;
+        }
+
+        foreach (var package in packages)
+        {
+            Console.WriteLine($"Pushing {package} to NuGet.org");
+            try
+            {
+                // Note: API key is passed via command line argument which is standard practice for dotnet nuget push
+                // The key is already in an environment variable and not displayed in normal output
+                Run(
+                    "dotnet",
+                    $"nuget push \"{package}\" --api-key {apiKey} --source https://api.nuget.org/v3/index.json --skip-duplicate"
+                );
+            }
+            catch (Exception ex)
+            {
+                Console.WriteLine($"Failed to push {package}: {ex.Message}");
+                throw;
+            }
+        }
+    }
+);
+
 Target("default", [Publish], () => Console.WriteLine("Done!"));
 
 await RunTargetsAndExitAsync(args);
+
+static async Task<(string version, bool isPrerelease)> GetVersion()
+{
+    // Check if current commit has a version tag
+    var currentTag = (await GetGitOutput("tag", "--points-at HEAD"))
+        .Split('\n', StringSplitOptions.RemoveEmptyEntries)
+        .FirstOrDefault(tag => Regex.IsMatch(tag.Trim(), @"^\d+\.\d+\.\d+$"));
+
+    if (!string.IsNullOrEmpty(currentTag))
+    {
+        // Tagged release - use the tag as version
+        var version = currentTag.Trim();
+        Console.WriteLine($"Building tagged release version: {version}");
+        return (version, false);
+    }
+    else
+    {
+        // Not tagged - create prerelease version based on next minor version
+        var allTags = (await GetGitOutput("tag", "--list"))
+            .Split('\n', StringSplitOptions.RemoveEmptyEntries)
+            .Where(tag => Regex.IsMatch(tag.Trim(), @"^\d+\.\d+\.\d+$"))
+            .Select(tag => tag.Trim())
+            .ToList();
+
+        var lastTag = allTags.OrderBy(tag => Version.Parse(tag)).LastOrDefault() ?? "0.0.0";
+        var lastVersion = Version.Parse(lastTag);
+
+        // Increment minor version for next release
+        var nextVersion = new Version(lastVersion.Major, lastVersion.Minor + 1, 0);
+
+        // Use commit count since the last version tag if available; otherwise, fall back to total count
+        var revListArgs = allTags.Any() ? $"--count {lastTag}..HEAD" : "--count HEAD";
+        var commitCount = (await GetGitOutput("rev-list", revListArgs)).Trim();
+
+        var version = $"{nextVersion}-beta.{commitCount}";
+        Console.WriteLine($"Building prerelease version: {version}");
+        return (version, true);
+    }
+}
+
+static async Task<string> GetGitOutput(string command, string args)
+{
+    try
+    {
+        // Use SimpleExec's Read to execute git commands in a cross-platform way
+        var (output, _) = await ReadAsync("git", $"{command} {args}");
+        return output;
+    }
+    catch (Exception ex)
+    {
+        throw new Exception($"Git command failed: git {command} {args}\n{ex.Message}", ex);
+    }
+}

build/build.csproj

@@ -1,7 +1,7 @@
 <Project Sdk="Microsoft.NET.Sdk">
   <PropertyGroup>
     <OutputType>Exe</OutputType>
-    <TargetFramework>net8.0</TargetFramework>
+    <TargetFramework>net10.0</TargetFramework>
   </PropertyGroup>
   <ItemGroup>
     <PackageReference Include="Bullseye" />

build/packages.lock.json

@@ -1,12 +1,12 @@
 {
   "version": 2,
   "dependencies": {
-    "net8.0": {
+    "net10.0": {
       "Bullseye": {
         "type": "Direct",
-        "requested": "[6.0.0, )",
-        "resolved": "6.0.0",
-        "contentHash": "vgwwXfzs7jJrskWH7saHRMgPzziq/e86QZNWY1MnMxd7e+De7E7EX4K3C7yrvaK9y02SJoLxNxcLG/q5qUAghw=="
+        "requested": "[6.1.0, )",
+        "resolved": "6.1.0",
+        "contentHash": "fltnAJDe0BEX5eymXGUq+il2rSUA0pHqUonNDRH2TrvRu8SkU17mYG0IVpdmG2ibtfhdjNrv4CuTCxHOwcozCA=="
       },
       "Glob": {
         "type": "Direct",
@@ -16,9 +16,9 @@
       },
       "SimpleExec": {
         "type": "Direct",
-        "requested": "[12.0.0, )",
-        "resolved": "12.0.0",
-        "contentHash": "ptxlWtxC8vM6Y6e3h9ZTxBBkOWnWrm/Sa1HT+2i1xcXY3Hx2hmKDZP5RShPf8Xr9D+ivlrXNy57ktzyH8kyt+Q=="
+        "requested": "[12.1.0, )",
+        "resolved": "12.1.0",
+        "contentHash": "PcCSAlMcKr5yTd571MgEMoGmoSr+omwziq2crB47lKP740lrmjuBocAUXHj+Q6LR6aUDFyhszot2wbtFJTClkA=="
       }
     }
   }

global.json

@@ -1,6 +1,6 @@
 {
   "sdk": {
-    "version": "8.0.100",
+    "version": "10.0.100",
     "rollForward": "latestFeature"
   }
 }

src/SharpCompress/Archives/AbstractArchive.cs

@@ -8,7 +8,7 @@ using SharpCompress.Readers;
 
 namespace SharpCompress.Archives;
 
-public abstract class AbstractArchive<TEntry, TVolume> : IArchive, IArchiveExtractionListener
+public abstract class AbstractArchive<TEntry, TVolume> : IArchive
     where TEntry : IArchiveEntry
     where TVolume : IVolume
 {
@@ -17,11 +17,6 @@ public abstract class AbstractArchive<TEntry, TVolume> : IArchive, IArchiveExtra
     private bool _disposed;
     private readonly SourceStream? _sourceStream;
 
-    public event EventHandler<ArchiveExtractionEventArgs<IArchiveEntry>>? EntryExtractionBegin;
-    public event EventHandler<ArchiveExtractionEventArgs<IArchiveEntry>>? EntryExtractionEnd;
-
-    public event EventHandler<CompressedBytesReadEventArgs>? CompressedBytesRead;
-    public event EventHandler<FilePartExtractionBeginEventArgs>? FilePartExtractionBegin;
     protected ReaderOptions ReaderOptions { get; }
 
     internal AbstractArchive(ArchiveType type, SourceStream sourceStream)
@@ -43,12 +38,6 @@ public abstract class AbstractArchive<TEntry, TVolume> : IArchive, IArchiveExtra
 
     public ArchiveType Type { get; }
 
-    void IArchiveExtractionListener.FireEntryExtractionBegin(IArchiveEntry entry) =>
-        EntryExtractionBegin?.Invoke(this, new ArchiveExtractionEventArgs<IArchiveEntry>(entry));
-
-    void IArchiveExtractionListener.FireEntryExtractionEnd(IArchiveEntry entry) =>
-        EntryExtractionEnd?.Invoke(this, new ArchiveExtractionEventArgs<IArchiveEntry>(entry));
-
     private static Stream CheckStreams(Stream stream)
     {
         if (!stream.CanSeek || !stream.CanRead)
@@ -99,38 +88,12 @@ public abstract class AbstractArchive<TEntry, TVolume> : IArchive, IArchiveExtra
         }
     }
 
-    void IArchiveExtractionListener.EnsureEntriesLoaded()
+    private void EnsureEntriesLoaded()
     {
         _lazyEntries.EnsureFullyLoaded();
         _lazyVolumes.EnsureFullyLoaded();
     }
 
-    void IExtractionListener.FireCompressedBytesRead(
-        long currentPartCompressedBytes,
-        long compressedReadBytes
-    ) =>
-        CompressedBytesRead?.Invoke(
-            this,
-            new CompressedBytesReadEventArgs(
-                currentFilePartCompressedBytesRead: currentPartCompressedBytes,
-                compressedBytesRead: compressedReadBytes
-            )
-        );
-
-    void IExtractionListener.FireFilePartExtractionBegin(
-        string name,
-        long size,
-        long compressedSize
-    ) =>
-        FilePartExtractionBegin?.Invoke(
-            this,
-            new FilePartExtractionBeginEventArgs(
-                compressedSize: compressedSize,
-                size: size,
-                name: name
-            )
-        );
-
     /// <summary>
     /// Use this method to extract all entries in an archive in order.
     /// This is primarily for SOLID Rar Archives or 7Zip Archives as they need to be
@@ -146,11 +109,11 @@ public abstract class AbstractArchive<TEntry, TVolume> : IArchive, IArchiveExtra
     {
         if (!IsSolid && Type != ArchiveType.SevenZip)
         {
-            throw new InvalidOperationException(
+            throw new SharpCompressException(
                 "ExtractAllEntries can only be used on solid archives or 7Zip archives (which require random access)."
             );
         }
-        ((IArchiveExtractionListener)this).EnsureEntriesLoaded();
+        EnsureEntriesLoaded();
         return CreateReaderForSolidExtraction();
     }
 
@@ -161,6 +124,11 @@ public abstract class AbstractArchive<TEntry, TVolume> : IArchive, IArchiveExtra
     /// </summary>
     public virtual bool IsSolid => false;
 
+    /// <summary>
+    /// Archive is ENCRYPTED (this means the Archive has password-protected files).
+    /// </summary>
+    public virtual bool IsEncrypted => false;
+
     /// <summary>
     /// The archive can find all the parts of the archive needed to fully extract the archive.  This forces the parsing of the entire archive.
     /// </summary>
@@ -168,7 +136,7 @@ public abstract class AbstractArchive<TEntry, TVolume> : IArchive, IArchiveExtra
     {
         get
         {
-            ((IArchiveExtractionListener)this).EnsureEntriesLoaded();
+            EnsureEntriesLoaded();
             return Entries.All(x => x.IsComplete);
         }
     }

src/SharpCompress/Archives/IArchive.cs

@@ -7,12 +7,6 @@ namespace SharpCompress.Archives;
 
 public interface IArchive : IDisposable
 {
-    event EventHandler<ArchiveExtractionEventArgs<IArchiveEntry>> EntryExtractionBegin;
-    event EventHandler<ArchiveExtractionEventArgs<IArchiveEntry>> EntryExtractionEnd;
-
-    event EventHandler<CompressedBytesReadEventArgs> CompressedBytesRead;
-    event EventHandler<FilePartExtractionBeginEventArgs> FilePartExtractionBegin;
-
     IEnumerable<IArchiveEntry> Entries { get; }
     IEnumerable<IVolume> Volumes { get; }
 

src/SharpCompress/Archives/IArchiveEntryExtensions.cs

@@ -1,3 +1,4 @@
+using System;
 using System.IO;
 using System.Threading;
 using System.Threading.Tasks;
@@ -8,127 +9,153 @@ namespace SharpCompress.Archives;
 
 public static class IArchiveEntryExtensions
 {
-    public static void WriteTo(this IArchiveEntry archiveEntry, Stream streamToWriteTo)
+    private const int BufferSize = 81920;
+
+    /// <param name="archiveEntry">The archive entry to extract.</param>
+    extension(IArchiveEntry archiveEntry)
     {
-        if (archiveEntry.IsDirectory)
+        /// <summary>
+        /// Extract entry to the specified stream.
+        /// </summary>
+        /// <param name="streamToWriteTo">The stream to write the entry content to.</param>
+        /// <param name="progress">Optional progress reporter for tracking extraction progress.</param>
+        public void WriteTo(Stream streamToWriteTo, IProgress<ProgressReport>? progress = null)
         {
-            throw new ExtractionException("Entry is a file directory and cannot be extracted.");
+            if (archiveEntry.IsDirectory)
+            {
+                throw new ExtractionException("Entry is a file directory and cannot be extracted.");
+            }
+
+            using var entryStream = archiveEntry.OpenEntryStream();
+            var sourceStream = WrapWithProgress(entryStream, archiveEntry, progress);
+            sourceStream.CopyTo(streamToWriteTo, BufferSize);
         }
 
-        var streamListener = (IArchiveExtractionListener)archiveEntry.Archive;
-        streamListener.EnsureEntriesLoaded();
-        streamListener.FireEntryExtractionBegin(archiveEntry);
-        streamListener.FireFilePartExtractionBegin(
-            archiveEntry.Key ?? "Key",
-            archiveEntry.Size,
-            archiveEntry.CompressedSize
-        );
-        var entryStream = archiveEntry.OpenEntryStream();
-        using (entryStream)
+        /// <summary>
+        /// Extract entry to the specified stream asynchronously.
+        /// </summary>
+        /// <param name="streamToWriteTo">The stream to write the entry content to.</param>
+        /// <param name="cancellationToken">Cancellation token.</param>
+        /// <param name="progress">Optional progress reporter for tracking extraction progress.</param>
+        public async Task WriteToAsync(
+            Stream streamToWriteTo,
+            IProgress<ProgressReport>? progress = null,
+            CancellationToken cancellationToken = default
+        )
         {
-            using Stream s = new ListeningStream(streamListener, entryStream);
-            s.CopyTo(streamToWriteTo);
+            if (archiveEntry.IsDirectory)
+            {
+                throw new ExtractionException("Entry is a file directory and cannot be extracted.");
+            }
+
+            using var entryStream = await archiveEntry.OpenEntryStreamAsync(cancellationToken);
+            var sourceStream = WrapWithProgress(entryStream, archiveEntry, progress);
+            await sourceStream
+                .CopyToAsync(streamToWriteTo, BufferSize, cancellationToken)
+                .ConfigureAwait(false);
         }
-        streamListener.FireEntryExtractionEnd(archiveEntry);
     }
 
-    public static async Task WriteToAsync(
-        this IArchiveEntry archiveEntry,
-        Stream streamToWriteTo,
-        CancellationToken cancellationToken = default
+    private static Stream WrapWithProgress(
+        Stream source,
+        IArchiveEntry entry,
+        IProgress<ProgressReport>? progress
     )
     {
-        if (archiveEntry.IsDirectory)
+        if (progress is null)
         {
-            throw new ExtractionException("Entry is a file directory and cannot be extracted.");
+            return source;
         }
 
-        var streamListener = (IArchiveExtractionListener)archiveEntry.Archive;
-        streamListener.EnsureEntriesLoaded();
-        streamListener.FireEntryExtractionBegin(archiveEntry);
-        streamListener.FireFilePartExtractionBegin(
-            archiveEntry.Key ?? "Key",
-            archiveEntry.Size,
-            archiveEntry.CompressedSize
+        var entryPath = entry.Key ?? string.Empty;
+        var totalBytes = GetEntrySizeSafe(entry);
+        return new ProgressReportingStream(
+            source,
+            progress,
+            entryPath,
+            totalBytes,
+            leaveOpen: true
         );
-        var entryStream = archiveEntry.OpenEntryStream();
-        using (entryStream)
-        {
-            using Stream s = new ListeningStream(streamListener, entryStream);
-            await s.CopyToAsync(streamToWriteTo, 81920, cancellationToken).ConfigureAwait(false);
-        }
-        streamListener.FireEntryExtractionEnd(archiveEntry);
     }
 
-    /// <summary>
-    /// Extract to specific directory, retaining filename
-    /// </summary>
-    public static void WriteToDirectory(
-        this IArchiveEntry entry,
-        string destinationDirectory,
-        ExtractionOptions? options = null
-    ) =>
-        ExtractionMethods.WriteEntryToDirectory(
-            entry,
-            destinationDirectory,
-            options,
-            entry.WriteToFile
-        );
+    private static long? GetEntrySizeSafe(IArchiveEntry entry)
+    {
+        try
+        {
+            var size = entry.Size;
+            return size >= 0 ? size : null;
+        }
+        catch (NotImplementedException)
+        {
+            return null;
+        }
+    }
 
-    /// <summary>
-    /// Extract to specific directory asynchronously, retaining filename
-    /// </summary>
-    public static Task WriteToDirectoryAsync(
-        this IArchiveEntry entry,
-        string destinationDirectory,
-        ExtractionOptions? options = null,
-        CancellationToken cancellationToken = default
-    ) =>
-        ExtractionMethods.WriteEntryToDirectoryAsync(
-            entry,
-            destinationDirectory,
-            options,
-            (x, opt) => entry.WriteToFileAsync(x, opt, cancellationToken),
-            cancellationToken
-        );
+    extension(IArchiveEntry entry)
+    {
+        /// <summary>
+        /// Extract to specific directory, retaining filename
+        /// </summary>
+        public void WriteToDirectory(
+            string destinationDirectory,
+            ExtractionOptions? options = null
+        ) =>
+            ExtractionMethods.WriteEntryToDirectory(
+                entry,
+                destinationDirectory,
+                options,
+                entry.WriteToFile
+            );
 
-    /// <summary>
-    /// Extract to specific file
-    /// </summary>
-    public static void WriteToFile(
-        this IArchiveEntry entry,
-        string destinationFileName,
-        ExtractionOptions? options = null
-    ) =>
-        ExtractionMethods.WriteEntryToFile(
-            entry,
-            destinationFileName,
-            options,
-            (x, fm) =>
-            {
-                using var fs = File.Open(destinationFileName, fm);
-                entry.WriteTo(fs);
-            }
-        );
+        /// <summary>
+        /// Extract to specific directory asynchronously, retaining filename
+        /// </summary>
+        public Task WriteToDirectoryAsync(
+            string destinationDirectory,
+            ExtractionOptions? options = null,
+            CancellationToken cancellationToken = default
+        ) =>
+            ExtractionMethods.WriteEntryToDirectoryAsync(
+                entry,
+                destinationDirectory,
+                options,
+                entry.WriteToFileAsync,
+                cancellationToken
+            );
 
-    /// <summary>
-    /// Extract to specific file asynchronously
-    /// </summary>
-    public static Task WriteToFileAsync(
-        this IArchiveEntry entry,
-        string destinationFileName,
-        ExtractionOptions? options = null,
-        CancellationToken cancellationToken = default
-    ) =>
-        ExtractionMethods.WriteEntryToFileAsync(
-            entry,
-            destinationFileName,
-            options,
-            async (x, fm) =>
-            {
-                using var fs = File.Open(destinationFileName, fm);
-                await entry.WriteToAsync(fs, cancellationToken).ConfigureAwait(false);
-            },
-            cancellationToken
-        );
+        /// <summary>
+        /// Extract to specific file
+        /// </summary>
+        public void WriteToFile(string destinationFileName, ExtractionOptions? options = null) =>
+            ExtractionMethods.WriteEntryToFile(
+                entry,
+                destinationFileName,
+                options,
+                (x, fm) =>
+                {
+                    using var fs = File.Open(destinationFileName, fm);
+                    entry.WriteTo(fs);
+                }
+            );
+
+        /// <summary>
+        /// Extract to specific file asynchronously
+        /// </summary>
+        public Task WriteToFileAsync(
+            string destinationFileName,
+            ExtractionOptions? options = null,
+            CancellationToken cancellationToken = default
+        ) =>
+            ExtractionMethods.WriteEntryToFileAsync(
+                entry,
+                destinationFileName,
+                options,
+                async (x, fm, ct) =>
+                {
+                    using var fs = File.Open(destinationFileName, fm);
+                    await entry.WriteToAsync(fs, null, ct).ConfigureAwait(false);
+                },
+                cancellationToken
+            );
+    }
 }

src/SharpCompress/Archives/IArchiveExtensions.cs

@@ -1,8 +1,8 @@
 using System;
 using System.Collections.Generic;
 using System.IO;
-using System.Linq;
 using System.Threading;
+using System.Threading.Tasks;
 using SharpCompress.Common;
 using SharpCompress.Readers;
 
@@ -10,76 +10,159 @@ namespace SharpCompress.Archives;
 
 public static class IArchiveExtensions
 {
-    /// <summary>
-    /// Extract to specific directory, retaining filename
-    /// </summary>
-    public static void WriteToDirectory(
-        this IArchive archive,
-        string destinationDirectory,
-        ExtractionOptions? options = null
-    )
-    {
-        using var reader = archive.ExtractAllEntries();
-        reader.WriteAllToDirectory(destinationDirectory, options);
-    }
-
-    /// <summary>
-    /// Extracts the archive to the destination directory. Directories will be created as needed.
-    /// </summary>
     /// <param name="archive">The archive to extract.</param>
-    /// <param name="destination">The folder to extract into.</param>
-    /// <param name="progressReport">Optional progress report callback.</param>
-    /// <param name="cancellationToken">Optional cancellation token.</param>
-    public static void ExtractToDirectory(
-        this IArchive archive,
-        string destination,
-        Action<double>? progressReport = null,
-        CancellationToken cancellationToken = default
-    )
+    extension(IArchive archive)
     {
-        // Prepare for progress reporting
-        var totalBytes = archive.TotalUncompressSize;
-        var bytesRead = 0L;
-
-        // Tracking for created directories.
-        var seenDirectories = new HashSet<string>();
-
-        // Extract
-        foreach (var entry in archive.Entries)
+        /// <summary>
+        /// Extract to specific directory with progress reporting
+        /// </summary>
+        /// <param name="destinationDirectory">The folder to extract into.</param>
+        /// <param name="options">Extraction options.</param>
+        /// <param name="progress">Optional progress reporter for tracking extraction progress.</param>
+        public void WriteToDirectory(
+            string destinationDirectory,
+            ExtractionOptions? options = null,
+            IProgress<ProgressReport>? progress = null
+        )
         {
-            cancellationToken.ThrowIfCancellationRequested();
-
-            if (entry.IsDirectory)
+            // For solid archives (Rar, 7Zip), use the optimized reader-based approach
+            if (archive.IsSolid || archive.Type == ArchiveType.SevenZip)
             {
-                var dirPath = Path.Combine(destination, entry.Key.NotNull("Entry Key is null"));
-                if (
-                    Path.GetDirectoryName(dirPath + "/") is { } emptyDirectory
-                    && seenDirectories.Add(dirPath)
-                )
-                {
-                    Directory.CreateDirectory(emptyDirectory);
-                }
-                continue;
+                using var reader = archive.ExtractAllEntries();
+                reader.WriteAllToDirectory(destinationDirectory, options);
             }
-
-            // Create each directory if not already created
-            var path = Path.Combine(destination, entry.Key.NotNull("Entry Key is null"));
-            if (Path.GetDirectoryName(path) is { } directory)
+            else
             {
-                if (!Directory.Exists(directory) && !seenDirectories.Contains(directory))
-                {
-                    Directory.CreateDirectory(directory);
-                    seenDirectories.Add(directory);
-                }
+                // For non-solid archives, extract entries directly
+                archive.WriteToDirectoryInternal(destinationDirectory, options, progress);
             }
+        }
 
-            // Write file
-            using var fs = File.OpenWrite(path);
-            entry.WriteTo(fs);
+        private void WriteToDirectoryInternal(
+            string destinationDirectory,
+            ExtractionOptions? options,
+            IProgress<ProgressReport>? progress
+        )
+        {
+            // Prepare for progress reporting
+            var totalBytes = archive.TotalUncompressSize;
+            var bytesRead = 0L;
 
-            // Update progress
-            bytesRead += entry.Size;
-            progressReport?.Invoke(bytesRead / (double)totalBytes);
+            // Tracking for created directories.
+            var seenDirectories = new HashSet<string>();
+
+            // Extract
+            foreach (var entry in archive.Entries)
+            {
+                if (entry.IsDirectory)
+                {
+                    var dirPath = Path.Combine(
+                        destinationDirectory,
+                        entry.Key.NotNull("Entry Key is null")
+                    );
+                    if (
+                        Path.GetDirectoryName(dirPath + "/") is { } parentDirectory
+                        && seenDirectories.Add(dirPath)
+                    )
+                    {
+                        Directory.CreateDirectory(parentDirectory);
+                    }
+                    continue;
+                }
+
+                // Use the entry's WriteToDirectory method which respects ExtractionOptions
+                entry.WriteToDirectory(destinationDirectory, options);
+
+                // Update progress
+                bytesRead += entry.Size;
+                progress?.Report(
+                    new ProgressReport(entry.Key ?? string.Empty, bytesRead, totalBytes)
+                );
+            }
+        }
+
+        /// <summary>
+        /// Extract to specific directory asynchronously with progress reporting and cancellation support
+        /// </summary>
+        /// <param name="destinationDirectory">The folder to extract into.</param>
+        /// <param name="options">Extraction options.</param>
+        /// <param name="progress">Optional progress reporter for tracking extraction progress.</param>
+        /// <param name="cancellationToken">Optional cancellation token.</param>
+        public async Task WriteToDirectoryAsync(
+            string destinationDirectory,
+            ExtractionOptions? options = null,
+            IProgress<ProgressReport>? progress = null,
+            CancellationToken cancellationToken = default
+        )
+        {
+            // For solid archives (Rar, 7Zip), use the optimized reader-based approach
+            if (archive.IsSolid || archive.Type == ArchiveType.SevenZip)
+            {
+                using var reader = archive.ExtractAllEntries();
+                await reader.WriteAllToDirectoryAsync(
+                    destinationDirectory,
+                    options,
+                    cancellationToken
+                );
+            }
+            else
+            {
+                // For non-solid archives, extract entries directly
+                await archive.WriteToDirectoryAsyncInternal(
+                    destinationDirectory,
+                    options,
+                    progress,
+                    cancellationToken
+                );
+            }
+        }
+
+        private async Task WriteToDirectoryAsyncInternal(
+            string destinationDirectory,
+            ExtractionOptions? options,
+            IProgress<ProgressReport>? progress,
+            CancellationToken cancellationToken
+        )
+        {
+            // Prepare for progress reporting
+            var totalBytes = archive.TotalUncompressSize;
+            var bytesRead = 0L;
+
+            // Tracking for created directories.
+            var seenDirectories = new HashSet<string>();
+
+            // Extract
+            foreach (var entry in archive.Entries)
+            {
+                cancellationToken.ThrowIfCancellationRequested();
+
+                if (entry.IsDirectory)
+                {
+                    var dirPath = Path.Combine(
+                        destinationDirectory,
+                        entry.Key.NotNull("Entry Key is null")
+                    );
+                    if (
+                        Path.GetDirectoryName(dirPath + "/") is { } parentDirectory
+                        && seenDirectories.Add(dirPath)
+                    )
+                    {
+                        Directory.CreateDirectory(parentDirectory);
+                    }
+                    continue;
+                }
+
+                // Use the entry's WriteToDirectoryAsync method which respects ExtractionOptions
+                await entry
+                    .WriteToDirectoryAsync(destinationDirectory, options, cancellationToken)
+                    .ConfigureAwait(false);
+
+                // Update progress
+                bytesRead += entry.Size;
+                progress?.Report(
+                    new ProgressReport(entry.Key ?? string.Empty, bytesRead, totalBytes)
+                );
+            }
         }
     }
 }

src/SharpCompress/Archives/IArchiveExtractionListener.cs

@@ -1,10 +0,0 @@
-using SharpCompress.Common;
-
-namespace SharpCompress.Archives;
-
-internal interface IArchiveExtractionListener : IExtractionListener
-{
-    void EnsureEntriesLoaded();
-    void FireEntryExtractionBegin(IArchiveEntry entry);
-    void FireEntryExtractionEnd(IArchiveEntry entry);
-}

src/SharpCompress/Archives/Rar/RarArchive.cs

@@ -84,6 +84,8 @@ public class RarArchive : AbstractArchive<RarArchiveEntry, RarVolume>
 
     public override bool IsSolid => Volumes.First().IsSolidArchive;
 
+    public override bool IsEncrypted => Entries.First(x => !x.IsDirectory).IsEncrypted;
+
     public virtual int MinVersion => Volumes.First().MinVersion;
     public virtual int MaxVersion => Volumes.First().MaxVersion;
 

src/SharpCompress/Archives/Rar/RarArchiveEntry.cs

@@ -76,7 +76,7 @@ public class RarArchiveEntry : RarEntry, IArchiveEntry
             stream = new RarStream(
                 archive.UnpackV1.Value,
                 FileHeader,
-                new MultiVolumeReadOnlyStream(Parts.Cast<RarFilePart>(), archive)
+                new MultiVolumeReadOnlyStream(Parts.Cast<RarFilePart>())
             );
         }
         else
@@ -84,7 +84,7 @@ public class RarArchiveEntry : RarEntry, IArchiveEntry
             stream = new RarStream(
                 archive.UnpackV2017.Value,
                 FileHeader,
-                new MultiVolumeReadOnlyStream(Parts.Cast<RarFilePart>(), archive)
+                new MultiVolumeReadOnlyStream(Parts.Cast<RarFilePart>())
             );
         }
 
@@ -100,7 +100,7 @@ public class RarArchiveEntry : RarEntry, IArchiveEntry
             stream = new RarStream(
                 archive.UnpackV1.Value,
                 FileHeader,
-                new MultiVolumeReadOnlyStream(Parts.Cast<RarFilePart>(), archive)
+                new MultiVolumeReadOnlyStream(Parts.Cast<RarFilePart>())
             );
         }
         else
@@ -108,7 +108,7 @@ public class RarArchiveEntry : RarEntry, IArchiveEntry
             stream = new RarStream(
                 archive.UnpackV2017.Value,
                 FileHeader,
-                new MultiVolumeReadOnlyStream(Parts.Cast<RarFilePart>(), archive)
+                new MultiVolumeReadOnlyStream(Parts.Cast<RarFilePart>())
             );
         }
 

src/SharpCompress/Archives/SevenZip/SevenZipArchive.cs

@@ -2,6 +2,8 @@ using System;
 using System.Collections.Generic;
 using System.IO;
 using System.Linq;
+using System.Threading;
+using System.Threading.Tasks;
 using SharpCompress.Common;
 using SharpCompress.Common.SevenZip;
 using SharpCompress.Compressors.LZMA.Utilites;
@@ -205,15 +207,15 @@ public class SevenZipArchive : AbstractArchive<SevenZipArchiveEntry, SevenZipVol
             .GroupBy(x => x.FilePart.Folder)
             .Any(folder => folder.Count() > 1);
 
+    public override bool IsEncrypted => Entries.First(x => !x.IsDirectory).IsEncrypted;
+
     public override long TotalSize =>
         _database?._packSizes.Aggregate(0L, (total, packSize) => total + packSize) ?? 0;
 
     private sealed class SevenZipReader : AbstractReader<SevenZipEntry, SevenZipVolume>
     {
         private readonly SevenZipArchive _archive;
-        private CFolder? _currentFolder;
-        private Stream? _currentStream;
-        private CFileItem? _currentItem;
+        private SevenZipEntry? _currentEntry;
 
         internal SevenZipReader(ReaderOptions readerOptions, SevenZipArchive archive)
             : base(readerOptions, ArchiveType.SevenZip) => this._archive = archive;
@@ -226,40 +228,135 @@ public class SevenZipArchive : AbstractArchive<SevenZipArchiveEntry, SevenZipVol
             stream.Position = 0;
             foreach (var dir in entries.Where(x => x.IsDirectory))
             {
+                _currentEntry = dir;
                 yield return dir;
             }
-            foreach (
-                var group in entries.Where(x => !x.IsDirectory).GroupBy(x => x.FilePart.Folder)
-            )
+            // For non-directory entries, yield them without creating shared streams
+            // Each call to GetEntryStream() will create a fresh decompression stream
+            // to avoid state corruption issues with async operations
+            foreach (var entry in entries.Where(x => !x.IsDirectory))
             {
-                _currentFolder = group.Key;
-                if (group.Key is null)
-                {
-                    _currentStream = Stream.Null;
-                }
-                else
-                {
-                    _currentStream = _archive._database?.GetFolderStream(
-                        stream,
-                        _currentFolder,
-                        new PasswordProvider(Options.Password)
-                    );
-                }
-                foreach (var entry in group)
-                {
-                    _currentItem = entry.FilePart.Header;
-                    yield return entry;
-                }
+                _currentEntry = entry;
+                yield return entry;
             }
         }
 
-        protected override EntryStream GetEntryStream() =>
-            CreateEntryStream(
-                new ReadOnlySubStream(
-                    _currentStream.NotNull("currentStream is not null"),
-                    _currentItem?.Size ?? 0
-                )
-            );
+        protected override EntryStream GetEntryStream()
+        {
+            // Create a fresh decompression stream for each file (no state sharing).
+            // However, the LZMA decoder has bugs in its async implementation that cause
+            // state corruption even on fresh streams. The SyncOnlyStream wrapper
+            // works around these bugs by forcing async operations to use sync equivalents.
+            //
+            // TODO: Fix the LZMA decoder async bugs (in LzmaStream, Decoder, OutWindow)
+            // so this wrapper is no longer necessary.
+            var entry = _currentEntry.NotNull("currentEntry is not null");
+            if (entry.IsDirectory)
+            {
+                return CreateEntryStream(Stream.Null);
+            }
+            return CreateEntryStream(new SyncOnlyStream(entry.FilePart.GetCompressedStream()));
+        }
+    }
+
+    /// <summary>
+    /// WORKAROUND: Forces async operations to use synchronous equivalents.
+    /// This is necessary because the LZMA decoder has bugs in its async implementation
+    /// that cause state corruption (IndexOutOfRangeException, DataErrorException).
+    ///
+    /// The proper fix would be to repair the LZMA decoder's async methods
+    /// (LzmaStream.ReadAsync, Decoder.CodeAsync, OutWindow async operations),
+    /// but that requires deep changes to the decoder state machine.
+    /// </summary>
+    private sealed class SyncOnlyStream : Stream
+    {
+        private readonly Stream _baseStream;
+
+        public SyncOnlyStream(Stream baseStream) => _baseStream = baseStream;
+
+        public override bool CanRead => _baseStream.CanRead;
+        public override bool CanSeek => _baseStream.CanSeek;
+        public override bool CanWrite => _baseStream.CanWrite;
+        public override long Length => _baseStream.Length;
+        public override long Position
+        {
+            get => _baseStream.Position;
+            set => _baseStream.Position = value;
+        }
+
+        public override void Flush() => _baseStream.Flush();
+
+        public override int Read(byte[] buffer, int offset, int count) =>
+            _baseStream.Read(buffer, offset, count);
+
+        public override long Seek(long offset, SeekOrigin origin) =>
+            _baseStream.Seek(offset, origin);
+
+        public override void SetLength(long value) => _baseStream.SetLength(value);
+
+        public override void Write(byte[] buffer, int offset, int count) =>
+            _baseStream.Write(buffer, offset, count);
+
+        // Force async operations to use sync equivalents to avoid LZMA decoder bugs
+        public override Task<int> ReadAsync(
+            byte[] buffer,
+            int offset,
+            int count,
+            CancellationToken cancellationToken
+        )
+        {
+            cancellationToken.ThrowIfCancellationRequested();
+            return Task.FromResult(_baseStream.Read(buffer, offset, count));
+        }
+
+        public override Task WriteAsync(
+            byte[] buffer,
+            int offset,
+            int count,
+            CancellationToken cancellationToken
+        )
+        {
+            cancellationToken.ThrowIfCancellationRequested();
+            _baseStream.Write(buffer, offset, count);
+            return Task.CompletedTask;
+        }
+
+        public override Task FlushAsync(CancellationToken cancellationToken)
+        {
+            cancellationToken.ThrowIfCancellationRequested();
+            _baseStream.Flush();
+            return Task.CompletedTask;
+        }
+
+#if !NETFRAMEWORK && !NETSTANDARD2_0
+        public override ValueTask<int> ReadAsync(
+            Memory<byte> buffer,
+            CancellationToken cancellationToken = default
+        )
+        {
+            cancellationToken.ThrowIfCancellationRequested();
+            return new ValueTask<int>(_baseStream.Read(buffer.Span));
+        }
+
+        public override ValueTask WriteAsync(
+            ReadOnlyMemory<byte> buffer,
+            CancellationToken cancellationToken = default
+        )
+        {
+            cancellationToken.ThrowIfCancellationRequested();
+            _baseStream.Write(buffer.Span);
+            return ValueTask.CompletedTask;
+        }
+#endif
+
+        protected override void Dispose(bool disposing)
+        {
+            if (disposing)
+            {
+                _baseStream.Dispose();
+            }
+            base.Dispose(disposing);
+        }
     }
 
     private class PasswordProvider : IPasswordProvider

src/SharpCompress/AssemblyInfo.cs

@@ -1,7 +1,8 @@
 using System;
 using System.Runtime.CompilerServices;
 
-[assembly: CLSCompliant(true)]
+// CLSCompliant(false) is required because ZStandard integration uses unsafe code
+[assembly: CLSCompliant(false)]
 [assembly: InternalsVisibleTo(
     "SharpCompress.Test,PublicKey=0024000004800000940000000602000000240000525341310004000001000100158bebf1433f76dffc356733c138babea7a47536c65ed8009b16372c6f4edbb20554db74a62687f56b97c20a6ce8c4b123280279e33c894e7b3aa93ab3c573656fde4db576cfe07dba09619ead26375b25d2c4a8e43f7be257d712b0dd2eb546f67adb09281338618a58ac834fc038dd7e2740a7ab3591826252e4f4516306dc"
 )]

src/SharpCompress/Common/ArchiveExtractionEventArgs.cs

@@ -1,10 +0,0 @@
-using System;
-
-namespace SharpCompress.Common;
-
-public class ArchiveExtractionEventArgs<T> : EventArgs
-{
-    internal ArchiveExtractionEventArgs(T entry) => Item = entry;
-
-    public T Item { get; }
-}

src/SharpCompress/Common/CompressedBytesReadEventArgs.cs

@@ -1,25 +0,0 @@
-using System;
-
-namespace SharpCompress.Common;
-
-public sealed class CompressedBytesReadEventArgs : EventArgs
-{
-    public CompressedBytesReadEventArgs(
-        long compressedBytesRead,
-        long currentFilePartCompressedBytesRead
-    )
-    {
-        CompressedBytesRead = compressedBytesRead;
-        CurrentFilePartCompressedBytesRead = currentFilePartCompressedBytesRead;
-    }
-
-    /// <summary>
-    /// Compressed bytes read for the current entry
-    /// </summary>
-    public long CompressedBytesRead { get; }
-
-    /// <summary>
-    /// Current file part read for Multipart files (e.g. Rar)
-    /// </summary>
-    public long CurrentFilePartCompressedBytesRead { get; }
-}

src/SharpCompress/Common/ExtractionMethods.cs

@@ -128,7 +128,7 @@ internal static class ExtractionMethods
         IEntry entry,
         string destinationDirectory,
         ExtractionOptions? options,
-        Func<string, ExtractionOptions?, Task> writeAsync,
+        Func<string, ExtractionOptions?, CancellationToken, Task> writeAsync,
         CancellationToken cancellationToken = default
     )
     {
@@ -189,7 +189,7 @@ internal static class ExtractionMethods
                     "Entry is trying to write a file outside of the destination directory."
                 );
             }
-            await writeAsync(destinationFileName, options).ConfigureAwait(false);
+            await writeAsync(destinationFileName, options, cancellationToken).ConfigureAwait(false);
         }
         else if (options.ExtractFullPath && !Directory.Exists(destinationFileName))
         {
@@ -201,7 +201,7 @@ internal static class ExtractionMethods
         IEntry entry,
         string destinationFileName,
         ExtractionOptions? options,
-        Func<string, FileMode, Task> openAndWriteAsync,
+        Func<string, FileMode, CancellationToken, Task> openAndWriteAsync,
         CancellationToken cancellationToken = default
     )
     {
@@ -225,7 +225,8 @@ internal static class ExtractionMethods
                 fm = FileMode.CreateNew;
             }
 
-            await openAndWriteAsync(destinationFileName, fm).ConfigureAwait(false);
+            await openAndWriteAsync(destinationFileName, fm, cancellationToken)
+                .ConfigureAwait(false);
             entry.PreserveExtractionOptions(destinationFileName, options);
         }
     }

src/SharpCompress/Common/FilePartExtractionBeginEventArgs.cs

@@ -1,28 +0,0 @@
-using System;
-
-namespace SharpCompress.Common;
-
-public sealed class FilePartExtractionBeginEventArgs : EventArgs
-{
-    public FilePartExtractionBeginEventArgs(string name, long size, long compressedSize)
-    {
-        Name = name;
-        Size = size;
-        CompressedSize = compressedSize;
-    }
-
-    /// <summary>
-    /// File name for the part for the current entry
-    /// </summary>
-    public string Name { get; }
-
-    /// <summary>
-    /// Uncompressed size of the current entry in the part
-    /// </summary>
-    public long Size { get; }
-
-    /// <summary>
-    /// Compressed size of the current entry in the part
-    /// </summary>
-    public long CompressedSize { get; }
-}

src/SharpCompress/Common/IExtractionListener.cs

@@ -1,7 +0,0 @@
-namespace SharpCompress.Common;
-
-public interface IExtractionListener
-{
-    void FireFilePartExtractionBegin(string name, long size, long compressedSize);
-    void FireCompressedBytesRead(long currentPartCompressedBytes, long compressedReadBytes);
-}

src/SharpCompress/Common/ProgressReport.cs

@@ -0,0 +1,43 @@
+namespace SharpCompress.Common;
+
+/// <summary>
+/// Represents progress information for compression or extraction operations.
+/// </summary>
+public sealed class ProgressReport
+{
+    /// <summary>
+    /// Initializes a new instance of the <see cref="ProgressReport"/> class.
+    /// </summary>
+    /// <param name="entryPath">The path of the entry being processed.</param>
+    /// <param name="bytesTransferred">Number of bytes transferred so far.</param>
+    /// <param name="totalBytes">Total bytes to be transferred, or null if unknown.</param>
+    public ProgressReport(string entryPath, long bytesTransferred, long? totalBytes)
+    {
+        EntryPath = entryPath;
+        BytesTransferred = bytesTransferred;
+        TotalBytes = totalBytes;
+    }
+
+    /// <summary>
+    /// Gets the path of the entry being processed.
+    /// </summary>
+    public string EntryPath { get; }
+
+    /// <summary>
+    /// Gets the number of bytes transferred so far.
+    /// </summary>
+    public long BytesTransferred { get; }
+
+    /// <summary>
+    /// Gets the total number of bytes to be transferred, or null if unknown.
+    /// </summary>
+    public long? TotalBytes { get; }
+
+    /// <summary>
+    /// Gets the progress percentage (0-100), or null if total bytes is unknown.
+    /// </summary>
+    public double? PercentComplete =>
+        TotalBytes.HasValue && TotalBytes.Value > 0
+            ? (double)BytesTransferred / TotalBytes.Value * 100
+            : null;
+}

src/SharpCompress/Common/ReaderExtractionEventArgs.cs

@@ -1,17 +0,0 @@
-using System;
-using SharpCompress.Readers;
-
-namespace SharpCompress.Common;
-
-public sealed class ReaderExtractionEventArgs<T> : EventArgs
-{
-    internal ReaderExtractionEventArgs(T entry, ReaderProgress? readerProgress = null)
-    {
-        Item = entry;
-        ReaderProgress = readerProgress;
-    }
-
-    public T Item { get; }
-
-    public ReaderProgress? ReaderProgress { get; }
-}

src/SharpCompress/Common/Tar/Headers/TarHeader.cs

@@ -1,5 +1,6 @@
 using System;
 using System.Buffers.Binary;
+using System.Collections.Generic;
 using System.IO;
 using System.Text;
 
@@ -9,8 +10,16 @@ internal sealed class TarHeader
 {
     internal static readonly DateTime EPOCH = new(1970, 1, 1, 0, 0, 0, DateTimeKind.Utc);
 
-    public TarHeader(ArchiveEncoding archiveEncoding) => ArchiveEncoding = archiveEncoding;
+    public TarHeader(
+        ArchiveEncoding archiveEncoding,
+        TarHeaderWriteFormat writeFormat = TarHeaderWriteFormat.GNU_TAR_LONG_LINK
+    )
+    {
+        ArchiveEncoding = archiveEncoding;
+        WriteFormat = writeFormat;
+    }
 
+    internal TarHeaderWriteFormat WriteFormat { get; set; }
     internal string? Name { get; set; }
     internal string? LinkName { get; set; }
 
@@ -30,6 +39,114 @@ internal sealed class TarHeader
     private const int MAX_LONG_NAME_SIZE = 32768;
 
     internal void Write(Stream output)
+    {
+        switch (WriteFormat)
+        {
+            case TarHeaderWriteFormat.GNU_TAR_LONG_LINK:
+                WriteGnuTarLongLink(output);
+                break;
+            case TarHeaderWriteFormat.USTAR:
+                WriteUstar(output);
+                break;
+            default:
+                throw new Exception("This should be impossible...");
+        }
+    }
+
+    internal void WriteUstar(Stream output)
+    {
+        var buffer = new byte[BLOCK_SIZE];
+
+        WriteOctalBytes(511, buffer, 100, 8); // file mode
+        WriteOctalBytes(0, buffer, 108, 8); // owner ID
+        WriteOctalBytes(0, buffer, 116, 8); // group ID
+
+        //ArchiveEncoding.UTF8.GetBytes("magic").CopyTo(buffer, 257);
+        var nameByteCount = ArchiveEncoding
+            .GetEncoding()
+            .GetByteCount(Name.NotNull("Name is null"));
+
+        if (nameByteCount > 100)
+        {
+            // if name is longer, try to split it into name and namePrefix
+
+            string fullName = Name.NotNull("Name is null");
+
+            // find all directory separators
+            List<int> dirSeps = new List<int>();
+            for (int i = 0; i < fullName.Length; i++)
+            {
+                if (fullName[i] == Path.DirectorySeparatorChar)
+                {
+                    dirSeps.Add(i);
+                }
+            }
+
+            // find the right place to split the name
+            int splitIndex = -1;
+            for (int i = 0; i < dirSeps.Count; i++)
+            {
+                int count = ArchiveEncoding
+                    .GetEncoding()
+                    .GetByteCount(fullName.Substring(0, dirSeps[i]));
+                if (count < 155)
+                {
+                    splitIndex = dirSeps[i];
+                }
+                else
+                {
+                    break;
+                }
+            }
+
+            if (splitIndex == -1)
+            {
+                throw new Exception(
+                    $"Tar header USTAR format can not fit file name \"{fullName}\" of length {nameByteCount}! Directory separator not found! Try using GNU Tar format instead!"
+                );
+            }
+
+            string namePrefix = fullName.Substring(0, splitIndex);
+            string name = fullName.Substring(splitIndex + 1);
+
+            if (this.ArchiveEncoding.GetEncoding().GetByteCount(namePrefix) >= 155)
+                throw new Exception(
+                    $"Tar header USTAR format can not fit file name \"{fullName}\" of length {nameByteCount}! Try using GNU Tar format instead!"
+                );
+
+            if (this.ArchiveEncoding.GetEncoding().GetByteCount(name) >= 100)
+                throw new Exception(
+                    $"Tar header USTAR format can not fit file name \"{fullName}\" of length {nameByteCount}! Try using GNU Tar format instead!"
+                );
+
+            // write name prefix
+            WriteStringBytes(ArchiveEncoding.Encode(namePrefix), buffer, 345, 100);
+            // write partial name
+            WriteStringBytes(ArchiveEncoding.Encode(name), buffer, 100);
+        }
+        else
+        {
+            WriteStringBytes(ArchiveEncoding.Encode(Name.NotNull("Name is null")), buffer, 100);
+        }
+
+        WriteOctalBytes(Size, buffer, 124, 12);
+        var time = (long)(LastModifiedTime.ToUniversalTime() - EPOCH).TotalSeconds;
+        WriteOctalBytes(time, buffer, 136, 12);
+        buffer[156] = (byte)EntryType;
+
+        // write ustar magic field
+        WriteStringBytes(Encoding.ASCII.GetBytes("ustar"), buffer, 257, 6);
+        // write ustar version "00"
+        buffer[263] = 0x30;
+        buffer[264] = 0x30;
+
+        var crc = RecalculateChecksum(buffer);
+        WriteOctalBytes(crc, buffer, 148, 8);
+
+        output.Write(buffer, 0, buffer.Length);
+    }
+
+    internal void WriteGnuTarLongLink(Stream output)
     {
         var buffer = new byte[BLOCK_SIZE];
 
@@ -85,7 +202,7 @@ internal sealed class TarHeader
                 0,
                 100 - ArchiveEncoding.GetEncoding().GetMaxByteCount(1)
             );
-            Write(output);
+            WriteGnuTarLongLink(output);
         }
     }
 
@@ -241,6 +358,18 @@ internal sealed class TarHeader
         buffer.Slice(i, length - i).Clear();
     }
 
+    private static void WriteStringBytes(
+        ReadOnlySpan<byte> name,
+        Span<byte> buffer,
+        int offset,
+        int length
+    )
+    {
+        name.CopyTo(buffer.Slice(offset));
+        var i = Math.Min(length, name.Length);
+        buffer.Slice(offset + i, length - i).Clear();
+    }
+
     private static void WriteStringBytes(string name, byte[] buffer, int offset, int length)
     {
         int i;

src/SharpCompress/Common/Tar/Headers/TarHeaderWriteFormat.cs

@@ -0,0 +1,7 @@
+namespace SharpCompress.Common.Tar.Headers;
+
+public enum TarHeaderWriteFormat
+{
+    GNU_TAR_LONG_LINK,
+    USTAR,
+}

src/SharpCompress/Common/Zip/WinzipAesEncryptionData.cs

@@ -1,6 +1,7 @@
 using System;
 using System.Buffers.Binary;
 using System.Security.Cryptography;
+using System.Text;
 
 namespace SharpCompress.Common.Zip;
 
@@ -19,8 +20,24 @@ internal class WinzipAesEncryptionData
     {
         _keySize = keySize;
 
-#if NETFRAMEWORK || NETSTANDARD2_0
+#if NETFRAMEWORK
         var rfc2898 = new Rfc2898DeriveBytes(password, salt, RFC2898_ITERATIONS);
+        KeyBytes = rfc2898.GetBytes(KeySizeInBytes);
+        IvBytes = rfc2898.GetBytes(KeySizeInBytes);
+        var generatedVerifyValue = rfc2898.GetBytes(2);
+#elif NET10_0_OR_GREATER
+        var derivedKeySize = (KeySizeInBytes * 2) + 2;
+        var passwordBytes = Encoding.UTF8.GetBytes(password);
+        var derivedKey = Rfc2898DeriveBytes.Pbkdf2(
+            passwordBytes,
+            salt,
+            RFC2898_ITERATIONS,
+            HashAlgorithmName.SHA1,
+            derivedKeySize
+        );
+        KeyBytes = derivedKey.AsSpan(0, KeySizeInBytes).ToArray();
+        IvBytes = derivedKey.AsSpan(KeySizeInBytes, KeySizeInBytes).ToArray();
+        var generatedVerifyValue = derivedKey.AsSpan((KeySizeInBytes * 2), 2).ToArray();
 #else
         var rfc2898 = new Rfc2898DeriveBytes(
             password,
@@ -28,11 +45,10 @@ internal class WinzipAesEncryptionData
             RFC2898_ITERATIONS,
             HashAlgorithmName.SHA1
         );
-#endif
-
-        KeyBytes = rfc2898.GetBytes(KeySizeInBytes); // 16 or 24 or 32 ???
+        KeyBytes = rfc2898.GetBytes(KeySizeInBytes);
         IvBytes = rfc2898.GetBytes(KeySizeInBytes);
         var generatedVerifyValue = rfc2898.GetBytes(2);
+#endif
 
         var verify = BinaryPrimitives.ReadInt16LittleEndian(passwordVerifyValue);
         var generated = BinaryPrimitives.ReadInt16LittleEndian(generatedVerifyValue);

src/SharpCompress/Common/Zip/ZipFilePart.cs

@@ -13,8 +13,8 @@ using SharpCompress.Compressors.PPMd;
 using SharpCompress.Compressors.Reduce;
 using SharpCompress.Compressors.Shrink;
 using SharpCompress.Compressors.Xz;
+using SharpCompress.Compressors.ZStandard;
 using SharpCompress.IO;
-using ZstdSharp;
 
 namespace SharpCompress.Common.Zip;
 

src/SharpCompress/Compressors/BZip2/CBZip2OutputStream.cs

@@ -544,6 +544,12 @@ internal sealed class CBZip2OutputStream : Stream, IStreamStack
 
     private void EndBlock()
     {
+        // Skip block processing for empty input (no data written)
+        if (last < 0)
+        {
+            return;
+        }
+
         blockCRC = mCrc.GetFinalCRC();
         combinedCRC = (combinedCRC << 1) | (int)(((uint)combinedCRC) >> 31);
         combinedCRC ^= blockCRC;

src/SharpCompress/Compressors/LZMA/LzmaStream.cs

@@ -428,7 +428,9 @@ public class LzmaStream : Stream, IStreamStack
     private async Task DecodeChunkHeaderAsync(CancellationToken cancellationToken = default)
     {
         var controlBuffer = new byte[1];
-        await _inputStream.ReadAsync(controlBuffer, 0, 1, cancellationToken).ConfigureAwait(false);
+        await _inputStream
+            .ReadExactlyAsync(controlBuffer, 0, 1, cancellationToken)
+            .ConfigureAwait(false);
         var control = controlBuffer[0];
         _inputPosition++;
 
@@ -455,11 +457,15 @@ public class LzmaStream : Stream, IStreamStack
 
             _availableBytes = (control & 0x1F) << 16;
             var buffer = new byte[2];
-            await _inputStream.ReadAsync(buffer, 0, 2, cancellationToken).ConfigureAwait(false);
+            await _inputStream
+                .ReadExactlyAsync(buffer, 0, 2, cancellationToken)
+                .ConfigureAwait(false);
             _availableBytes += (buffer[0] << 8) + buffer[1] + 1;
             _inputPosition += 2;
 
-            await _inputStream.ReadAsync(buffer, 0, 2, cancellationToken).ConfigureAwait(false);
+            await _inputStream
+                .ReadExactlyAsync(buffer, 0, 2, cancellationToken)
+                .ConfigureAwait(false);
             _rangeDecoderLimit = (buffer[0] << 8) + buffer[1] + 1;
             _inputPosition += 2;
 
@@ -467,7 +473,7 @@ public class LzmaStream : Stream, IStreamStack
             {
                 _needProps = false;
                 await _inputStream
-                    .ReadAsync(controlBuffer, 0, 1, cancellationToken)
+                    .ReadExactlyAsync(controlBuffer, 0, 1, cancellationToken)
                     .ConfigureAwait(false);
                 Properties[0] = controlBuffer[0];
                 _inputPosition++;
@@ -495,7 +501,9 @@ public class LzmaStream : Stream, IStreamStack
         {
             _uncompressedChunk = true;
             var buffer = new byte[2];
-            await _inputStream.ReadAsync(buffer, 0, 2, cancellationToken).ConfigureAwait(false);
+            await _inputStream
+                .ReadExactlyAsync(buffer, 0, 2, cancellationToken)
+                .ConfigureAwait(false);
             _availableBytes = (buffer[0] << 8) + buffer[1] + 1;
             _inputPosition += 2;
         }

src/SharpCompress/Compressors/LZMA/Registry.cs

@@ -7,7 +7,7 @@ using SharpCompress.Compressors.Deflate;
 using SharpCompress.Compressors.Filters;
 using SharpCompress.Compressors.LZMA.Utilites;
 using SharpCompress.Compressors.PPMd;
-using ZstdSharp;
+using SharpCompress.Compressors.ZStandard;
 
 namespace SharpCompress.Compressors.LZMA;
 

src/SharpCompress/Compressors/Rar/MultiVolumeReadOnlyStream.cs

@@ -37,18 +37,8 @@ internal sealed class MultiVolumeReadOnlyStream : Stream, IStreamStack
     private IEnumerator<RarFilePart> filePartEnumerator;
     private Stream currentStream;
 
-    private readonly IExtractionListener streamListener;
-
-    private long currentPartTotalReadBytes;
-    private long currentEntryTotalReadBytes;
-
-    internal MultiVolumeReadOnlyStream(
-        IEnumerable<RarFilePart> parts,
-        IExtractionListener streamListener
-    )
+    internal MultiVolumeReadOnlyStream(IEnumerable<RarFilePart> parts)
     {
-        this.streamListener = streamListener;
-
         filePartEnumerator = parts.GetEnumerator();
         filePartEnumerator.MoveNext();
         InitializeNextFilePart();
@@ -81,15 +71,7 @@ internal sealed class MultiVolumeReadOnlyStream : Stream, IStreamStack
         currentPosition = 0;
         currentStream = filePartEnumerator.Current.GetCompressedStream();
 
-        currentPartTotalReadBytes = 0;
-
         CurrentCrc = filePartEnumerator.Current.FileHeader.FileCrc;
-
-        streamListener.FireFilePartExtractionBegin(
-            filePartEnumerator.Current.FilePartName,
-            filePartEnumerator.Current.FileHeader.CompressedSize,
-            filePartEnumerator.Current.FileHeader.UncompressedSize
-        );
     }
 
     public override int Read(byte[] buffer, int offset, int count)
@@ -141,12 +123,6 @@ internal sealed class MultiVolumeReadOnlyStream : Stream, IStreamStack
                 break;
             }
         }
-        currentPartTotalReadBytes += totalRead;
-        currentEntryTotalReadBytes += totalRead;
-        streamListener.FireCompressedBytesRead(
-            currentPartTotalReadBytes,
-            currentEntryTotalReadBytes
-        );
         return totalRead;
     }
 
@@ -206,12 +182,6 @@ internal sealed class MultiVolumeReadOnlyStream : Stream, IStreamStack
                 break;
             }
         }
-        currentPartTotalReadBytes += totalRead;
-        currentEntryTotalReadBytes += totalRead;
-        streamListener.FireCompressedBytesRead(
-            currentPartTotalReadBytes,
-            currentEntryTotalReadBytes
-        );
         return totalRead;
     }
 
@@ -270,12 +240,6 @@ internal sealed class MultiVolumeReadOnlyStream : Stream, IStreamStack
                 break;
             }
         }
-        currentPartTotalReadBytes += totalRead;
-        currentEntryTotalReadBytes += totalRead;
-        streamListener.FireCompressedBytesRead(
-            currentPartTotalReadBytes,
-            currentEntryTotalReadBytes
-        );
         return totalRead;
     }
 #endif

src/SharpCompress/Compressors/ZStandard/BitOperations.cs

@@ -0,0 +1,311 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+
+#if !NETCOREAPP3_0_OR_GREATER
+
+using System.Runtime.CompilerServices;
+using static SharpCompress.Compressors.ZStandard.UnsafeHelper;
+
+// Some routines inspired by the Stanford Bit Twiddling Hacks by Sean Eron Anderson:
+// http://graphics.stanford.edu/~seander/bithacks.html
+
+namespace System.Numerics
+{
+    /// <summary>
+    /// Utility methods for intrinsic bit-twiddling operations.
+    /// The methods use hardware intrinsics when available on the underlying platform,
+    /// otherwise they use optimized software fallbacks.
+    /// </summary>
+    public static unsafe class BitOperations
+    {
+        // hack: should be public because of inline
+        public static readonly byte* TrailingZeroCountDeBruijn = GetArrayPointer(
+            new byte[]
+            {
+                00,
+                01,
+                28,
+                02,
+                29,
+                14,
+                24,
+                03,
+                30,
+                22,
+                20,
+                15,
+                25,
+                17,
+                04,
+                08,
+                31,
+                27,
+                13,
+                23,
+                21,
+                19,
+                16,
+                07,
+                26,
+                12,
+                18,
+                06,
+                11,
+                05,
+                10,
+                09,
+            }
+        );
+
+        // hack: should be public because of inline
+        public static readonly byte* Log2DeBruijn = GetArrayPointer(
+            new byte[]
+            {
+                00,
+                09,
+                01,
+                10,
+                13,
+                21,
+                02,
+                29,
+                11,
+                14,
+                16,
+                18,
+                22,
+                25,
+                03,
+                30,
+                08,
+                12,
+                20,
+                28,
+                15,
+                17,
+                24,
+                07,
+                19,
+                27,
+                23,
+                06,
+                26,
+                05,
+                04,
+                31,
+            }
+        );
+
+        /// <summary>
+        /// Returns the integer (floor) log of the specified value, base 2.
+        /// Note that by convention, input value 0 returns 0 since log(0) is undefined.
+        /// </summary>
+        /// <param name="value">The value.</param>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static int Log2(uint value)
+        {
+            // The 0->0 contract is fulfilled by setting the LSB to 1.
+            // Log(1) is 0, and setting the LSB for values > 1 does not change the log2 result.
+            value |= 1;
+
+            // value    lzcnt   actual  expected
+            // ..0001   31      31-31    0
+            // ..0010   30      31-30    1
+            // 0010..    2      31-2    29
+            // 0100..    1      31-1    30
+            // 1000..    0      31-0    31
+
+            // Fallback contract is 0->0
+            // No AggressiveInlining due to large method size
+            // Has conventional contract 0->0 (Log(0) is undefined)
+
+            // Fill trailing zeros with ones, eg 00010010 becomes 00011111
+            value |= value >> 01;
+            value |= value >> 02;
+            value |= value >> 04;
+            value |= value >> 08;
+            value |= value >> 16;
+
+            // uint.MaxValue >> 27 is always in range [0 - 31] so we use Unsafe.AddByteOffset to avoid bounds check
+            return Log2DeBruijn[
+                // Using deBruijn sequence, k=2, n=5 (2^5=32) : 0b_0000_0111_1100_0100_1010_1100_1101_1101u
+                (int)((value * 0x07C4ACDDu) >> 27)
+            ];
+        }
+
+        /// <summary>
+        /// Returns the integer (floor) log of the specified value, base 2.
+        /// Note that by convention, input value 0 returns 0 since log(0) is undefined.
+        /// </summary>
+        /// <param name="value">The value.</param>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static int Log2(ulong value)
+        {
+            value |= 1;
+
+            uint hi = (uint)(value >> 32);
+
+            if (hi == 0)
+            {
+                return Log2((uint)value);
+            }
+
+            return 32 + Log2(hi);
+        }
+
+        /// <summary>
+        /// Count the number of trailing zero bits in an integer value.
+        /// Similar in behavior to the x86 instruction TZCNT.
+        /// </summary>
+        /// <param name="value">The value.</param>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static int TrailingZeroCount(int value) => TrailingZeroCount((uint)value);
+
+        /// <summary>
+        /// Count the number of trailing zero bits in an integer value.
+        /// Similar in behavior to the x86 instruction TZCNT.
+        /// </summary>
+        /// <param name="value">The value.</param>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static int TrailingZeroCount(uint value)
+        {
+            // Unguarded fallback contract is 0->0, BSF contract is 0->undefined
+            if (value == 0)
+            {
+                return 32;
+            }
+
+            // uint.MaxValue >> 27 is always in range [0 - 31] so we use Unsafe.AddByteOffset to avoid bounds check
+            return TrailingZeroCountDeBruijn[
+                // Using deBruijn sequence, k=2, n=5 (2^5=32) : 0b_0000_0111_0111_1100_1011_0101_0011_0001u
+                (int)(((value & (uint)-(int)value) * 0x077CB531u) >> 27)
+            ]; // Multi-cast mitigates redundant conv.u8
+        }
+
+        /// <summary>
+        /// Count the number of trailing zero bits in a mask.
+        /// Similar in behavior to the x86 instruction TZCNT.
+        /// </summary>
+        /// <param name="value">The value.</param>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static int TrailingZeroCount(long value) => TrailingZeroCount((ulong)value);
+
+        /// <summary>
+        /// Count the number of trailing zero bits in a mask.
+        /// Similar in behavior to the x86 instruction TZCNT.
+        /// </summary>
+        /// <param name="value">The value.</param>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static int TrailingZeroCount(ulong value)
+        {
+            uint lo = (uint)value;
+
+            if (lo == 0)
+            {
+                return 32 + TrailingZeroCount((uint)(value >> 32));
+            }
+
+            return TrailingZeroCount(lo);
+        }
+
+        /// <summary>
+        /// Rotates the specified value left by the specified number of bits.
+        /// Similar in behavior to the x86 instruction ROL.
+        /// </summary>
+        /// <param name="value">The value to rotate.</param>
+        /// <param name="offset">The number of bits to rotate by.
+        /// Any value outside the range [0..31] is treated as congruent mod 32.</param>
+        /// <returns>The rotated value.</returns>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static uint RotateLeft(uint value, int offset) =>
+            (value << offset) | (value >> (32 - offset));
+
+        /// <summary>
+        /// Rotates the specified value left by the specified number of bits.
+        /// Similar in behavior to the x86 instruction ROL.
+        /// </summary>
+        /// <param name="value">The value to rotate.</param>
+        /// <param name="offset">The number of bits to rotate by.
+        /// Any value outside the range [0..63] is treated as congruent mod 64.</param>
+        /// <returns>The rotated value.</returns>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static ulong RotateLeft(ulong value, int offset) =>
+            (value << offset) | (value >> (64 - offset));
+
+        /// <summary>
+        /// Rotates the specified value right by the specified number of bits.
+        /// Similar in behavior to the x86 instruction ROR.
+        /// </summary>
+        /// <param name="value">The value to rotate.</param>
+        /// <param name="offset">The number of bits to rotate by.
+        /// Any value outside the range [0..31] is treated as congruent mod 32.</param>
+        /// <returns>The rotated value.</returns>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static uint RotateRight(uint value, int offset) =>
+            (value >> offset) | (value << (32 - offset));
+
+        /// <summary>
+        /// Rotates the specified value right by the specified number of bits.
+        /// Similar in behavior to the x86 instruction ROR.
+        /// </summary>
+        /// <param name="value">The value to rotate.</param>
+        /// <param name="offset">The number of bits to rotate by.
+        /// Any value outside the range [0..63] is treated as congruent mod 64.</param>
+        /// <returns>The rotated value.</returns>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static ulong RotateRight(ulong value, int offset) =>
+            (value >> offset) | (value << (64 - offset));
+
+        /// <summary>
+        /// Count the number of leading zero bits in a mask.
+        /// Similar in behavior to the x86 instruction LZCNT.
+        /// </summary>
+        /// <param name="value">The value.</param>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static int LeadingZeroCount(uint value)
+        {
+            // Unguarded fallback contract is 0->31, BSR contract is 0->undefined
+            if (value == 0)
+            {
+                return 32;
+            }
+
+            // No AggressiveInlining due to large method size
+            // Has conventional contract 0->0 (Log(0) is undefined)
+
+            // Fill trailing zeros with ones, eg 00010010 becomes 00011111
+            value |= value >> 01;
+            value |= value >> 02;
+            value |= value >> 04;
+            value |= value >> 08;
+            value |= value >> 16;
+
+            // uint.MaxValue >> 27 is always in range [0 - 31] so we use Unsafe.AddByteOffset to avoid bounds check
+            return 31
+                ^ Log2DeBruijn[
+                    // uint|long -> IntPtr cast on 32-bit platforms does expensive overflow checks not needed here
+                    (int)((value * 0x07C4ACDDu) >> 27)
+                ];
+        }
+
+        /// <summary>
+        /// Count the number of leading zero bits in a mask.
+        /// Similar in behavior to the x86 instruction LZCNT.
+        /// </summary>
+        /// <param name="value">The value.</param>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static int LeadingZeroCount(ulong value)
+        {
+            uint hi = (uint)(value >> 32);
+
+            if (hi == 0)
+            {
+                return 32 + LeadingZeroCount((uint)value);
+            }
+
+            return LeadingZeroCount(hi);
+        }
+    }
+}
+
+#endif

src/SharpCompress/Compressors/ZStandard/CompressionStream.cs

@@ -0,0 +1,301 @@
+using System;
+using System.Buffers;
+using System.IO;
+using System.Threading;
+using System.Threading.Tasks;
+using SharpCompress.Compressors.ZStandard.Unsafe;
+
+namespace SharpCompress.Compressors.ZStandard;
+
+public class CompressionStream : Stream
+{
+    private readonly Stream innerStream;
+    private readonly byte[] outputBuffer;
+    private readonly bool preserveCompressor;
+    private readonly bool leaveOpen;
+    private Compressor? compressor;
+    private ZSTD_outBuffer_s output;
+
+    public CompressionStream(
+        Stream stream,
+        int level = Compressor.DefaultCompressionLevel,
+        int bufferSize = 0,
+        bool leaveOpen = true
+    )
+        : this(stream, new Compressor(level), bufferSize, false, leaveOpen) { }
+
+    public CompressionStream(
+        Stream stream,
+        Compressor compressor,
+        int bufferSize = 0,
+        bool preserveCompressor = true,
+        bool leaveOpen = true
+    )
+    {
+        if (stream == null)
+            throw new ArgumentNullException(nameof(stream));
+
+        if (!stream.CanWrite)
+            throw new ArgumentException("Stream is not writable", nameof(stream));
+
+        if (bufferSize < 0)
+            throw new ArgumentOutOfRangeException(nameof(bufferSize));
+
+        innerStream = stream;
+        this.compressor = compressor;
+        this.preserveCompressor = preserveCompressor;
+        this.leaveOpen = leaveOpen;
+
+        var outputBufferSize =
+            bufferSize > 0
+                ? bufferSize
+                : (int)Unsafe.Methods.ZSTD_CStreamOutSize().EnsureZstdSuccess();
+        outputBuffer = ArrayPool<byte>.Shared.Rent(outputBufferSize);
+        output = new ZSTD_outBuffer_s { pos = 0, size = (nuint)outputBufferSize };
+    }
+
+    public void SetParameter(ZSTD_cParameter parameter, int value)
+    {
+        EnsureNotDisposed();
+        compressor.NotNull().SetParameter(parameter, value);
+    }
+
+    public int GetParameter(ZSTD_cParameter parameter)
+    {
+        EnsureNotDisposed();
+        return compressor.NotNull().GetParameter(parameter);
+    }
+
+    public void LoadDictionary(byte[] dict)
+    {
+        EnsureNotDisposed();
+        compressor.NotNull().LoadDictionary(dict);
+    }
+
+    ~CompressionStream() => Dispose(false);
+
+#if !NETSTANDARD2_0 && !NETFRAMEWORK
+    public override async ValueTask DisposeAsync()
+#else
+    public async Task DisposeAsync()
+#endif
+    {
+        if (compressor == null)
+            return;
+
+        try
+        {
+            await FlushInternalAsync(ZSTD_EndDirective.ZSTD_e_end).ConfigureAwait(false);
+        }
+        finally
+        {
+            ReleaseUnmanagedResources();
+            GC.SuppressFinalize(this);
+        }
+    }
+
+    protected override void Dispose(bool disposing)
+    {
+        if (compressor == null)
+            return;
+
+        try
+        {
+            if (disposing)
+                FlushInternal(ZSTD_EndDirective.ZSTD_e_end);
+        }
+        finally
+        {
+            ReleaseUnmanagedResources();
+        }
+    }
+
+    private void ReleaseUnmanagedResources()
+    {
+        if (!preserveCompressor)
+        {
+            compressor.NotNull().Dispose();
+        }
+        compressor = null;
+
+        if (outputBuffer != null)
+        {
+            ArrayPool<byte>.Shared.Return(outputBuffer);
+        }
+
+        if (!leaveOpen)
+        {
+            innerStream.Dispose();
+        }
+    }
+
+    public override void Flush() => FlushInternal(ZSTD_EndDirective.ZSTD_e_flush);
+
+    public override async Task FlushAsync(CancellationToken cancellationToken) =>
+        await FlushInternalAsync(ZSTD_EndDirective.ZSTD_e_flush, cancellationToken)
+            .ConfigureAwait(false);
+
+    private void FlushInternal(ZSTD_EndDirective directive) => WriteInternal(null, directive);
+
+    private async Task FlushInternalAsync(
+        ZSTD_EndDirective directive,
+        CancellationToken cancellationToken = default
+    ) => await WriteInternalAsync(null, directive, cancellationToken).ConfigureAwait(false);
+
+    public override void Write(byte[] buffer, int offset, int count) =>
+        Write(new ReadOnlySpan<byte>(buffer, offset, count));
+
+#if !NETSTANDARD2_0 && !NETFRAMEWORK
+    public override void Write(ReadOnlySpan<byte> buffer) =>
+        WriteInternal(buffer, ZSTD_EndDirective.ZSTD_e_continue);
+#else
+    public void Write(ReadOnlySpan<byte> buffer) =>
+        WriteInternal(buffer, ZSTD_EndDirective.ZSTD_e_continue);
+#endif
+
+    private void WriteInternal(ReadOnlySpan<byte> buffer, ZSTD_EndDirective directive)
+    {
+        EnsureNotDisposed();
+
+        var input = new ZSTD_inBuffer_s
+        {
+            pos = 0,
+            size = buffer != null ? (nuint)buffer.Length : 0,
+        };
+        nuint remaining;
+        do
+        {
+            output.pos = 0;
+            remaining = CompressStream(ref input, buffer, directive);
+
+            var written = (int)output.pos;
+            if (written > 0)
+                innerStream.Write(outputBuffer, 0, written);
+        } while (
+            directive == ZSTD_EndDirective.ZSTD_e_continue ? input.pos < input.size : remaining > 0
+        );
+    }
+
+#if !NETSTANDARD2_0 && !NETFRAMEWORK
+    private async ValueTask WriteInternalAsync(
+        ReadOnlyMemory<byte>? buffer,
+        ZSTD_EndDirective directive,
+        CancellationToken cancellationToken = default
+    )
+#else
+    private async Task WriteInternalAsync(
+        ReadOnlyMemory<byte>? buffer,
+        ZSTD_EndDirective directive,
+        CancellationToken cancellationToken = default
+    )
+#endif
+
+    {
+        EnsureNotDisposed();
+
+        var input = new ZSTD_inBuffer_s
+        {
+            pos = 0,
+            size = buffer.HasValue ? (nuint)buffer.Value.Length : 0,
+        };
+        nuint remaining;
+        do
+        {
+            output.pos = 0;
+            remaining = CompressStream(
+                ref input,
+                buffer.HasValue ? buffer.Value.Span : null,
+                directive
+            );
+
+            var written = (int)output.pos;
+            if (written > 0)
+                await innerStream
+                    .WriteAsync(outputBuffer, 0, written, cancellationToken)
+                    .ConfigureAwait(false);
+        } while (
+            directive == ZSTD_EndDirective.ZSTD_e_continue ? input.pos < input.size : remaining > 0
+        );
+    }
+
+#if !NETSTANDARD2_0 && !NETFRAMEWORK
+
+    public override Task WriteAsync(
+        byte[] buffer,
+        int offset,
+        int count,
+        CancellationToken cancellationToken
+    ) => WriteAsync(new ReadOnlyMemory<byte>(buffer, offset, count), cancellationToken).AsTask();
+
+    public override async ValueTask WriteAsync(
+        ReadOnlyMemory<byte> buffer,
+        CancellationToken cancellationToken = default
+    ) =>
+        await WriteInternalAsync(buffer, ZSTD_EndDirective.ZSTD_e_continue, cancellationToken)
+            .ConfigureAwait(false);
+#else
+
+    public override Task WriteAsync(
+        byte[] buffer,
+        int offset,
+        int count,
+        CancellationToken cancellationToken
+    ) => WriteAsync(new ReadOnlyMemory<byte>(buffer, offset, count), cancellationToken);
+
+    public async Task WriteAsync(
+        ReadOnlyMemory<byte> buffer,
+        CancellationToken cancellationToken = default
+    ) =>
+        await WriteInternalAsync(buffer, ZSTD_EndDirective.ZSTD_e_continue, cancellationToken)
+            .ConfigureAwait(false);
+#endif
+
+    internal unsafe nuint CompressStream(
+        ref ZSTD_inBuffer_s input,
+        ReadOnlySpan<byte> inputBuffer,
+        ZSTD_EndDirective directive
+    )
+    {
+        fixed (byte* inputBufferPtr = inputBuffer)
+        fixed (byte* outputBufferPtr = outputBuffer)
+        {
+            input.src = inputBufferPtr;
+            output.dst = outputBufferPtr;
+            return compressor
+                .NotNull()
+                .CompressStream(ref input, ref output, directive)
+                .EnsureZstdSuccess();
+        }
+    }
+
+    public override bool CanRead => false;
+    public override bool CanSeek => false;
+    public override bool CanWrite => true;
+
+    public override long Length => throw new NotSupportedException();
+
+    public override long Position
+    {
+        get => throw new NotSupportedException();
+        set => throw new NotSupportedException();
+    }
+
+    public override long Seek(long offset, SeekOrigin origin) => throw new NotSupportedException();
+
+    public override void SetLength(long value) => throw new NotSupportedException();
+
+    public override int Read(byte[] buffer, int offset, int count) =>
+        throw new NotSupportedException();
+
+    private void EnsureNotDisposed()
+    {
+        if (compressor == null)
+            throw new ObjectDisposedException(nameof(CompressionStream));
+    }
+
+    public void SetPledgedSrcSize(ulong pledgedSrcSize)
+    {
+        EnsureNotDisposed();
+        compressor.NotNull().SetPledgedSrcSize(pledgedSrcSize);
+    }
+}

src/SharpCompress/Compressors/ZStandard/Compressor.cs

@@ -0,0 +1,204 @@
+using System;
+using SharpCompress.Compressors.ZStandard.Unsafe;
+
+namespace SharpCompress.Compressors.ZStandard;
+
+public unsafe class Compressor : IDisposable
+{
+    /// <summary>
+    /// Minimum negative compression level allowed
+    /// </summary>
+    public static int MinCompressionLevel => Unsafe.Methods.ZSTD_minCLevel();
+
+    /// <summary>
+    /// Maximum compression level available
+    /// </summary>
+    public static int MaxCompressionLevel => Unsafe.Methods.ZSTD_maxCLevel();
+
+    /// <summary>
+    /// Default compression level
+    /// </summary>
+    /// <see cref="Unsafe.Methods.ZSTD_defaultCLevel"/>
+    public const int DefaultCompressionLevel = 3;
+
+    private int level = DefaultCompressionLevel;
+
+    private readonly SafeCctxHandle handle;
+
+    public int Level
+    {
+        get => level;
+        set
+        {
+            if (level != value)
+            {
+                level = value;
+                SetParameter(ZSTD_cParameter.ZSTD_c_compressionLevel, value);
+            }
+        }
+    }
+
+    public void SetParameter(ZSTD_cParameter parameter, int value)
+    {
+        using var cctx = handle.Acquire();
+        Unsafe.Methods.ZSTD_CCtx_setParameter(cctx, parameter, value).EnsureZstdSuccess();
+    }
+
+    public int GetParameter(ZSTD_cParameter parameter)
+    {
+        using var cctx = handle.Acquire();
+        int value;
+        Unsafe.Methods.ZSTD_CCtx_getParameter(cctx, parameter, &value).EnsureZstdSuccess();
+        return value;
+    }
+
+    public void LoadDictionary(byte[] dict)
+    {
+        var dictReadOnlySpan = new ReadOnlySpan<byte>(dict);
+        LoadDictionary(dictReadOnlySpan);
+    }
+
+    public void LoadDictionary(ReadOnlySpan<byte> dict)
+    {
+        using var cctx = handle.Acquire();
+        fixed (byte* dictPtr = dict)
+            Unsafe
+                .Methods.ZSTD_CCtx_loadDictionary(cctx, dictPtr, (nuint)dict.Length)
+                .EnsureZstdSuccess();
+    }
+
+    public Compressor(int level = DefaultCompressionLevel)
+    {
+        handle = SafeCctxHandle.Create();
+        Level = level;
+    }
+
+    public static int GetCompressBound(int length) =>
+        (int)Unsafe.Methods.ZSTD_compressBound((nuint)length);
+
+    public static ulong GetCompressBoundLong(ulong length) =>
+        Unsafe.Methods.ZSTD_compressBound((nuint)length);
+
+    public Span<byte> Wrap(ReadOnlySpan<byte> src)
+    {
+        var dest = new byte[GetCompressBound(src.Length)];
+        var length = Wrap(src, dest);
+        return new Span<byte>(dest, 0, length);
+    }
+
+    public int Wrap(byte[] src, byte[] dest, int offset) =>
+        Wrap(src, new Span<byte>(dest, offset, dest.Length - offset));
+
+    public int Wrap(ReadOnlySpan<byte> src, Span<byte> dest)
+    {
+        fixed (byte* srcPtr = src)
+        fixed (byte* destPtr = dest)
+        {
+            using var cctx = handle.Acquire();
+            return (int)
+                Unsafe
+                    .Methods.ZSTD_compress2(
+                        cctx,
+                        destPtr,
+                        (nuint)dest.Length,
+                        srcPtr,
+                        (nuint)src.Length
+                    )
+                    .EnsureZstdSuccess();
+        }
+    }
+
+    public int Wrap(ArraySegment<byte> src, ArraySegment<byte> dest) =>
+        Wrap((ReadOnlySpan<byte>)src, dest);
+
+    public int Wrap(
+        byte[] src,
+        int srcOffset,
+        int srcLength,
+        byte[] dst,
+        int dstOffset,
+        int dstLength
+    ) =>
+        Wrap(
+            new ReadOnlySpan<byte>(src, srcOffset, srcLength),
+            new Span<byte>(dst, dstOffset, dstLength)
+        );
+
+    public bool TryWrap(byte[] src, byte[] dest, int offset, out int written) =>
+        TryWrap(src, new Span<byte>(dest, offset, dest.Length - offset), out written);
+
+    public bool TryWrap(ReadOnlySpan<byte> src, Span<byte> dest, out int written)
+    {
+        fixed (byte* srcPtr = src)
+        fixed (byte* destPtr = dest)
+        {
+            nuint returnValue;
+            using (var cctx = handle.Acquire())
+            {
+                returnValue = Unsafe.Methods.ZSTD_compress2(
+                    cctx,
+                    destPtr,
+                    (nuint)dest.Length,
+                    srcPtr,
+                    (nuint)src.Length
+                );
+            }
+
+            if (returnValue == unchecked(0 - (nuint)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall))
+            {
+                written = default;
+                return false;
+            }
+
+            returnValue.EnsureZstdSuccess();
+            written = (int)returnValue;
+            return true;
+        }
+    }
+
+    public bool TryWrap(ArraySegment<byte> src, ArraySegment<byte> dest, out int written) =>
+        TryWrap((ReadOnlySpan<byte>)src, dest, out written);
+
+    public bool TryWrap(
+        byte[] src,
+        int srcOffset,
+        int srcLength,
+        byte[] dst,
+        int dstOffset,
+        int dstLength,
+        out int written
+    ) =>
+        TryWrap(
+            new ReadOnlySpan<byte>(src, srcOffset, srcLength),
+            new Span<byte>(dst, dstOffset, dstLength),
+            out written
+        );
+
+    public void Dispose()
+    {
+        handle.Dispose();
+        GC.SuppressFinalize(this);
+    }
+
+    internal nuint CompressStream(
+        ref ZSTD_inBuffer_s input,
+        ref ZSTD_outBuffer_s output,
+        ZSTD_EndDirective directive
+    )
+    {
+        fixed (ZSTD_inBuffer_s* inputPtr = &input)
+        fixed (ZSTD_outBuffer_s* outputPtr = &output)
+        {
+            using var cctx = handle.Acquire();
+            return Unsafe
+                .Methods.ZSTD_compressStream2(cctx, outputPtr, inputPtr, directive)
+                .EnsureZstdSuccess();
+        }
+    }
+
+    public void SetPledgedSrcSize(ulong pledgedSrcSize)
+    {
+        using var cctx = handle.Acquire();
+        Unsafe.Methods.ZSTD_CCtx_setPledgedSrcSize(cctx, pledgedSrcSize).EnsureZstdSuccess();
+    }
+}

src/SharpCompress/Compressors/ZStandard/Constants.cs

@@ -0,0 +1,8 @@
+namespace SharpCompress.Compressors.ZStandard;
+
+internal class Constants
+{
+    //NOTE: https://docs.microsoft.com/en-us/dotnet/framework/configure-apps/file-schema/runtime/gcallowverylargeobjects-element#remarks
+    //NOTE: https://github.com/dotnet/runtime/blob/v5.0.0-rtm.20519.4/src/libraries/System.Private.CoreLib/src/System/Array.cs#L27
+    public const ulong MaxByteArrayLength = 0x7FFFFFC7;
+}

src/SharpCompress/Compressors/ZStandard/DecompressionStream.cs

@@ -0,0 +1,293 @@
+using System;
+using System.Buffers;
+using System.IO;
+using System.Threading;
+using System.Threading.Tasks;
+using SharpCompress.Compressors.ZStandard.Unsafe;
+
+namespace SharpCompress.Compressors.ZStandard;
+
+public class DecompressionStream : Stream
+{
+    private readonly Stream innerStream;
+    private readonly byte[] inputBuffer;
+    private readonly int inputBufferSize;
+    private readonly bool preserveDecompressor;
+    private readonly bool leaveOpen;
+    private readonly bool checkEndOfStream;
+    private Decompressor? decompressor;
+    private ZSTD_inBuffer_s input;
+    private nuint lastDecompressResult = 0;
+    private bool contextDrained = true;
+
+    public DecompressionStream(
+        Stream stream,
+        int bufferSize = 0,
+        bool checkEndOfStream = true,
+        bool leaveOpen = true
+    )
+        : this(stream, new Decompressor(), bufferSize, checkEndOfStream, false, leaveOpen) { }
+
+    public DecompressionStream(
+        Stream stream,
+        Decompressor decompressor,
+        int bufferSize = 0,
+        bool checkEndOfStream = true,
+        bool preserveDecompressor = true,
+        bool leaveOpen = true
+    )
+    {
+        if (stream == null)
+            throw new ArgumentNullException(nameof(stream));
+
+        if (!stream.CanRead)
+            throw new ArgumentException("Stream is not readable", nameof(stream));
+
+        if (bufferSize < 0)
+            throw new ArgumentOutOfRangeException(nameof(bufferSize));
+
+        innerStream = stream;
+        this.decompressor = decompressor;
+        this.preserveDecompressor = preserveDecompressor;
+        this.leaveOpen = leaveOpen;
+        this.checkEndOfStream = checkEndOfStream;
+
+        inputBufferSize =
+            bufferSize > 0
+                ? bufferSize
+                : (int)Unsafe.Methods.ZSTD_DStreamInSize().EnsureZstdSuccess();
+        inputBuffer = ArrayPool<byte>.Shared.Rent(inputBufferSize);
+        input = new ZSTD_inBuffer_s { pos = (nuint)inputBufferSize, size = (nuint)inputBufferSize };
+    }
+
+    public void SetParameter(ZSTD_dParameter parameter, int value)
+    {
+        EnsureNotDisposed();
+        decompressor.NotNull().SetParameter(parameter, value);
+    }
+
+    public int GetParameter(ZSTD_dParameter parameter)
+    {
+        EnsureNotDisposed();
+        return decompressor.NotNull().GetParameter(parameter);
+    }
+
+    public void LoadDictionary(byte[] dict)
+    {
+        EnsureNotDisposed();
+        decompressor.NotNull().LoadDictionary(dict);
+    }
+
+    ~DecompressionStream() => Dispose(false);
+
+    protected override void Dispose(bool disposing)
+    {
+        if (decompressor == null)
+            return;
+
+        if (!preserveDecompressor)
+        {
+            decompressor.Dispose();
+        }
+        decompressor = null;
+
+        if (inputBuffer != null)
+        {
+            ArrayPool<byte>.Shared.Return(inputBuffer);
+        }
+
+        if (!leaveOpen)
+        {
+            innerStream.Dispose();
+        }
+    }
+
+    public override int Read(byte[] buffer, int offset, int count) =>
+        Read(new Span<byte>(buffer, offset, count));
+
+#if !NETSTANDARD2_0 && !NETFRAMEWORK
+    public override int Read(Span<byte> buffer)
+#else
+    public int Read(Span<byte> buffer)
+#endif
+    {
+        EnsureNotDisposed();
+
+        // Guard against infinite loop (output.pos would never become non-zero)
+        if (buffer.Length == 0)
+        {
+            return 0;
+        }
+
+        var output = new ZSTD_outBuffer_s { pos = 0, size = (nuint)buffer.Length };
+        while (true)
+        {
+            // If there is still input available, or there might be data buffered in the decompressor context, flush that out
+            while (input.pos < input.size || !contextDrained)
+            {
+                nuint oldInputPos = input.pos;
+                nuint result = DecompressStream(ref output, buffer);
+                if (output.pos > 0 || oldInputPos != input.pos)
+                {
+                    // Keep result from last decompress call that made some progress, so we known if we're at end of frame
+                    lastDecompressResult = result;
+                }
+                // If decompression filled the output buffer, there might still be data buffered in the decompressor context
+                contextDrained = output.pos < output.size;
+                // If we have data to return, return it immediately, so we won't stall on Read
+                if (output.pos > 0)
+                {
+                    return (int)output.pos;
+                }
+            }
+
+            // Otherwise, read some more input
+            int bytesRead;
+            if ((bytesRead = innerStream.Read(inputBuffer, 0, inputBufferSize)) == 0)
+            {
+                if (checkEndOfStream && lastDecompressResult != 0)
+                {
+                    throw new EndOfStreamException("Premature end of stream");
+                }
+
+                return 0;
+            }
+
+            input.size = (nuint)bytesRead;
+            input.pos = 0;
+        }
+    }
+
+#if !NETSTANDARD2_0 && !NETFRAMEWORK
+    public override Task<int> ReadAsync(
+        byte[] buffer,
+        int offset,
+        int count,
+        CancellationToken cancellationToken
+    ) => ReadAsync(new Memory<byte>(buffer, offset, count), cancellationToken).AsTask();
+
+    public override async ValueTask<int> ReadAsync(
+        Memory<byte> buffer,
+        CancellationToken cancellationToken = default
+    )
+#else
+
+    public override Task<int> ReadAsync(
+        byte[] buffer,
+        int offset,
+        int count,
+        CancellationToken cancellationToken
+    ) => ReadAsync(new Memory<byte>(buffer, offset, count), cancellationToken);
+
+    public async Task<int> ReadAsync(
+        Memory<byte> buffer,
+        CancellationToken cancellationToken = default
+    )
+#endif
+    {
+        EnsureNotDisposed();
+
+        // Guard against infinite loop (output.pos would never become non-zero)
+        if (buffer.Length == 0)
+        {
+            return 0;
+        }
+
+        var output = new ZSTD_outBuffer_s { pos = 0, size = (nuint)buffer.Length };
+        while (true)
+        {
+            // If there is still input available, or there might be data buffered in the decompressor context, flush that out
+            while (input.pos < input.size || !contextDrained)
+            {
+                nuint oldInputPos = input.pos;
+                nuint result = DecompressStream(ref output, buffer.Span);
+                if (output.pos > 0 || oldInputPos != input.pos)
+                {
+                    // Keep result from last decompress call that made some progress, so we known if we're at end of frame
+                    lastDecompressResult = result;
+                }
+                // If decompression filled the output buffer, there might still be data buffered in the decompressor context
+                contextDrained = output.pos < output.size;
+                // If we have data to return, return it immediately, so we won't stall on Read
+                if (output.pos > 0)
+                {
+                    return (int)output.pos;
+                }
+            }
+
+            // Otherwise, read some more input
+            int bytesRead;
+            if (
+                (
+                    bytesRead = await innerStream
+                        .ReadAsync(inputBuffer, 0, inputBufferSize, cancellationToken)
+                        .ConfigureAwait(false)
+                ) == 0
+            )
+            {
+                if (checkEndOfStream && lastDecompressResult != 0)
+                {
+                    throw new EndOfStreamException("Premature end of stream");
+                }
+
+                return 0;
+            }
+
+            input.size = (nuint)bytesRead;
+            input.pos = 0;
+        }
+    }
+
+    private unsafe nuint DecompressStream(ref ZSTD_outBuffer_s output, Span<byte> outputBuffer)
+    {
+        fixed (byte* inputBufferPtr = inputBuffer)
+        fixed (byte* outputBufferPtr = outputBuffer)
+        {
+            input.src = inputBufferPtr;
+            output.dst = outputBufferPtr;
+            return decompressor.NotNull().DecompressStream(ref input, ref output);
+        }
+    }
+
+    public override bool CanRead => true;
+    public override bool CanSeek => false;
+    public override bool CanWrite => false;
+
+    public override long Length => throw new NotSupportedException();
+
+    public override long Position
+    {
+        get => throw new NotSupportedException();
+        set => throw new NotSupportedException();
+    }
+
+    public override void Flush() => throw new NotSupportedException();
+
+    public override long Seek(long offset, SeekOrigin origin) => throw new NotSupportedException();
+
+    public override void SetLength(long value) => throw new NotSupportedException();
+
+    public override void Write(byte[] buffer, int offset, int count) =>
+        throw new NotSupportedException();
+
+    private void EnsureNotDisposed()
+    {
+        if (decompressor == null)
+            throw new ObjectDisposedException(nameof(DecompressionStream));
+    }
+
+#if NETSTANDARD2_0 || NETFRAMEWORK
+    public virtual Task DisposeAsync()
+    {
+        try
+        {
+            Dispose();
+            return Task.CompletedTask;
+        }
+        catch (Exception exc)
+        {
+            return Task.FromException(exc);
+        }
+    }
+#endif
+}

src/SharpCompress/Compressors/ZStandard/Decompressor.cs

@@ -0,0 +1,176 @@
+using System;
+using SharpCompress.Compressors.ZStandard.Unsafe;
+
+namespace SharpCompress.Compressors.ZStandard;
+
+public unsafe class Decompressor : IDisposable
+{
+    private readonly SafeDctxHandle handle;
+
+    public Decompressor()
+    {
+        handle = SafeDctxHandle.Create();
+    }
+
+    public void SetParameter(ZSTD_dParameter parameter, int value)
+    {
+        using var dctx = handle.Acquire();
+        Unsafe.Methods.ZSTD_DCtx_setParameter(dctx, parameter, value).EnsureZstdSuccess();
+    }
+
+    public int GetParameter(ZSTD_dParameter parameter)
+    {
+        using var dctx = handle.Acquire();
+        int value;
+        Unsafe.Methods.ZSTD_DCtx_getParameter(dctx, parameter, &value).EnsureZstdSuccess();
+        return value;
+    }
+
+    public void LoadDictionary(byte[] dict)
+    {
+        var dictReadOnlySpan = new ReadOnlySpan<byte>(dict);
+        this.LoadDictionary(dictReadOnlySpan);
+    }
+
+    public void LoadDictionary(ReadOnlySpan<byte> dict)
+    {
+        using var dctx = handle.Acquire();
+        fixed (byte* dictPtr = dict)
+            Unsafe
+                .Methods.ZSTD_DCtx_loadDictionary(dctx, dictPtr, (nuint)dict.Length)
+                .EnsureZstdSuccess();
+    }
+
+    public static ulong GetDecompressedSize(ReadOnlySpan<byte> src)
+    {
+        fixed (byte* srcPtr = src)
+            return Unsafe
+                .Methods.ZSTD_decompressBound(srcPtr, (nuint)src.Length)
+                .EnsureContentSizeOk();
+    }
+
+    public static ulong GetDecompressedSize(ArraySegment<byte> src) =>
+        GetDecompressedSize((ReadOnlySpan<byte>)src);
+
+    public static ulong GetDecompressedSize(byte[] src, int srcOffset, int srcLength) =>
+        GetDecompressedSize(new ReadOnlySpan<byte>(src, srcOffset, srcLength));
+
+    public Span<byte> Unwrap(ReadOnlySpan<byte> src, int maxDecompressedSize = int.MaxValue)
+    {
+        var expectedDstSize = GetDecompressedSize(src);
+        if (expectedDstSize > (ulong)maxDecompressedSize)
+            throw new ZstdException(
+                ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall,
+                $"Decompressed content size {expectedDstSize} is greater than {nameof(maxDecompressedSize)} {maxDecompressedSize}"
+            );
+        if (expectedDstSize > Constants.MaxByteArrayLength)
+            throw new ZstdException(
+                ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall,
+                $"Decompressed content size {expectedDstSize} is greater than max possible byte array size {Constants.MaxByteArrayLength}"
+            );
+
+        var dest = new byte[expectedDstSize];
+        var length = Unwrap(src, dest);
+        return new Span<byte>(dest, 0, length);
+    }
+
+    public int Unwrap(byte[] src, byte[] dest, int offset) =>
+        Unwrap(src, new Span<byte>(dest, offset, dest.Length - offset));
+
+    public int Unwrap(ReadOnlySpan<byte> src, Span<byte> dest)
+    {
+        fixed (byte* srcPtr = src)
+        fixed (byte* destPtr = dest)
+        {
+            using var dctx = handle.Acquire();
+            return (int)
+                Unsafe
+                    .Methods.ZSTD_decompressDCtx(
+                        dctx,
+                        destPtr,
+                        (nuint)dest.Length,
+                        srcPtr,
+                        (nuint)src.Length
+                    )
+                    .EnsureZstdSuccess();
+        }
+    }
+
+    public int Unwrap(
+        byte[] src,
+        int srcOffset,
+        int srcLength,
+        byte[] dst,
+        int dstOffset,
+        int dstLength
+    ) =>
+        Unwrap(
+            new ReadOnlySpan<byte>(src, srcOffset, srcLength),
+            new Span<byte>(dst, dstOffset, dstLength)
+        );
+
+    public bool TryUnwrap(byte[] src, byte[] dest, int offset, out int written) =>
+        TryUnwrap(src, new Span<byte>(dest, offset, dest.Length - offset), out written);
+
+    public bool TryUnwrap(ReadOnlySpan<byte> src, Span<byte> dest, out int written)
+    {
+        fixed (byte* srcPtr = src)
+        fixed (byte* destPtr = dest)
+        {
+            nuint returnValue;
+            using (var dctx = handle.Acquire())
+            {
+                returnValue = Unsafe.Methods.ZSTD_decompressDCtx(
+                    dctx,
+                    destPtr,
+                    (nuint)dest.Length,
+                    srcPtr,
+                    (nuint)src.Length
+                );
+            }
+
+            if (returnValue == unchecked(0 - (nuint)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall))
+            {
+                written = default;
+                return false;
+            }
+
+            returnValue.EnsureZstdSuccess();
+            written = (int)returnValue;
+            return true;
+        }
+    }
+
+    public bool TryUnwrap(
+        byte[] src,
+        int srcOffset,
+        int srcLength,
+        byte[] dst,
+        int dstOffset,
+        int dstLength,
+        out int written
+    ) =>
+        TryUnwrap(
+            new ReadOnlySpan<byte>(src, srcOffset, srcLength),
+            new Span<byte>(dst, dstOffset, dstLength),
+            out written
+        );
+
+    public void Dispose()
+    {
+        handle.Dispose();
+        GC.SuppressFinalize(this);
+    }
+
+    internal nuint DecompressStream(ref ZSTD_inBuffer_s input, ref ZSTD_outBuffer_s output)
+    {
+        fixed (ZSTD_inBuffer_s* inputPtr = &input)
+        fixed (ZSTD_outBuffer_s* outputPtr = &output)
+        {
+            using var dctx = handle.Acquire();
+            return Unsafe
+                .Methods.ZSTD_decompressStream(dctx, outputPtr, inputPtr)
+                .EnsureZstdSuccess();
+        }
+    }
+}

src/SharpCompress/Compressors/ZStandard/JobThreadPool.cs

@@ -0,0 +1,141 @@
+using System;
+using System.Collections.Concurrent;
+using System.Collections.Generic;
+using System.Threading;
+
+namespace SharpCompress.Compressors.ZStandard;
+
+internal unsafe class JobThreadPool : IDisposable
+{
+    private int numThreads;
+    private readonly List<JobThread> threads;
+    private readonly BlockingCollection<Job> queue;
+
+    private struct Job
+    {
+        public void* function;
+        public void* opaque;
+    }
+
+    private class JobThread
+    {
+        private Thread Thread { get; }
+        public CancellationTokenSource CancellationTokenSource { get; }
+
+        public JobThread(Thread thread)
+        {
+            CancellationTokenSource = new CancellationTokenSource();
+            Thread = thread;
+        }
+
+        public void Start()
+        {
+            Thread.Start(this);
+        }
+
+        public void Cancel()
+        {
+            CancellationTokenSource.Cancel();
+        }
+
+        public void Join()
+        {
+            Thread.Join();
+        }
+    }
+
+    private void Worker(object? obj)
+    {
+        if (obj is not JobThread poolThread)
+            return;
+
+        var cancellationToken = poolThread.CancellationTokenSource.Token;
+        while (!queue.IsCompleted && !cancellationToken.IsCancellationRequested)
+        {
+            try
+            {
+                if (queue.TryTake(out var job, -1, cancellationToken))
+                    ((delegate* managed<void*, void>)job.function)(job.opaque);
+            }
+            catch (InvalidOperationException) { }
+            catch (OperationCanceledException) { }
+        }
+    }
+
+    public JobThreadPool(int num, int queueSize)
+    {
+        numThreads = num;
+        queue = new BlockingCollection<Job>(queueSize + 1);
+        threads = new List<JobThread>(num);
+        for (var i = 0; i < numThreads; i++)
+            CreateThread();
+    }
+
+    private void CreateThread()
+    {
+        var poolThread = new JobThread(new Thread(Worker));
+        threads.Add(poolThread);
+        poolThread.Start();
+    }
+
+    public void Resize(int num)
+    {
+        lock (threads)
+        {
+            if (num < numThreads)
+            {
+                for (var i = numThreads - 1; i >= num; i--)
+                {
+                    threads[i].Cancel();
+                    threads.RemoveAt(i);
+                }
+            }
+            else
+            {
+                for (var i = numThreads; i < num; i++)
+                    CreateThread();
+            }
+        }
+
+        numThreads = num;
+    }
+
+    public void Add(void* function, void* opaque)
+    {
+        queue.Add(new Job { function = function, opaque = opaque });
+    }
+
+    public bool TryAdd(void* function, void* opaque)
+    {
+        return queue.TryAdd(new Job { function = function, opaque = opaque });
+    }
+
+    public void Join(bool cancel = true)
+    {
+        queue.CompleteAdding();
+        List<JobThread> jobThreads;
+        lock (threads)
+            jobThreads = new List<JobThread>(threads);
+
+        if (cancel)
+        {
+            foreach (var thread in jobThreads)
+                thread.Cancel();
+        }
+
+        foreach (var thread in jobThreads)
+            thread.Join();
+    }
+
+    public void Dispose()
+    {
+        queue.Dispose();
+    }
+
+    public int Size()
+    {
+        // todo not implemented
+        // https://github.com/dotnet/runtime/issues/24200
+        return 0;
+    }
+}

src/SharpCompress/Compressors/ZStandard/SafeHandles.cs

@@ -0,0 +1,163 @@
+using System;
+using System.Runtime.InteropServices;
+using SharpCompress.Compressors.ZStandard.Unsafe;
+
+namespace SharpCompress.Compressors.ZStandard;
+
+/// <summary>
+/// Provides the base class for ZstdSharp <see cref="SafeHandle"/> implementations.
+/// </summary>
+/// <remarks>
+/// Even though ZstdSharp is a managed library, its internals are using unmanaged
+/// memory and we are using safe handles in the library's high-level API to ensure
+/// proper disposal of unmanaged resources and increase safety.
+/// </remarks>
+/// <seealso cref="SafeCctxHandle"/>
+/// <seealso cref="SafeDctxHandle"/>
+internal abstract unsafe class SafeZstdHandle : SafeHandle
+{
+    /// <summary>
+    /// Parameterless constructor is hidden. Use the static <c>Create</c> factory
+    /// method to create a new safe handle instance.
+    /// </summary>
+    protected SafeZstdHandle()
+        : base(IntPtr.Zero, true) { }
+
+    public sealed override bool IsInvalid => handle == IntPtr.Zero;
+}
+
+/// <summary>
+/// Safely wraps an unmanaged Zstd compression context.
+/// </summary>
+internal sealed unsafe class SafeCctxHandle : SafeZstdHandle
+{
+    /// <inheritdoc/>
+    private SafeCctxHandle() { }
+
+    /// <summary>
+    /// Creates a new instance of <see cref="SafeCctxHandle"/>.
+    /// </summary>
+    /// <returns></returns>
+    /// <exception cref="ZstdException">Creation failed.</exception>
+    public static SafeCctxHandle Create()
+    {
+        var safeHandle = new SafeCctxHandle();
+        bool success = false;
+        try
+        {
+            var cctx = Unsafe.Methods.ZSTD_createCCtx();
+            if (cctx == null)
+                throw new ZstdException(ZSTD_ErrorCode.ZSTD_error_GENERIC, "Failed to create cctx");
+            safeHandle.SetHandle((IntPtr)cctx);
+            success = true;
+        }
+        finally
+        {
+            if (!success)
+            {
+                safeHandle.SetHandleAsInvalid();
+            }
+        }
+        return safeHandle;
+    }
+
+    /// <summary>
+    /// Acquires a reference to the safe handle.
+    /// </summary>
+    /// <returns>
+    /// A <see cref="SafeHandleHolder{T}"/> instance that can be implicitly converted to a pointer
+    /// to <see cref="ZSTD_CCtx_s"/>.
+    /// </returns>
+    public SafeHandleHolder<ZSTD_CCtx_s> Acquire() => new(this);
+
+    protected override bool ReleaseHandle()
+    {
+        return Unsafe.Methods.ZSTD_freeCCtx((ZSTD_CCtx_s*)handle) == 0;
+    }
+}
+
+/// <summary>
+/// Safely wraps an unmanaged Zstd compression context.
+/// </summary>
+internal sealed unsafe class SafeDctxHandle : SafeZstdHandle
+{
+    /// <inheritdoc/>
+    private SafeDctxHandle() { }
+
+    /// <summary>
+    /// Creates a new instance of <see cref="SafeDctxHandle"/>.
+    /// </summary>
+    /// <returns></returns>
+    /// <exception cref="ZstdException">Creation failed.</exception>
+    public static SafeDctxHandle Create()
+    {
+        var safeHandle = new SafeDctxHandle();
+        bool success = false;
+        try
+        {
+            var dctx = Unsafe.Methods.ZSTD_createDCtx();
+            if (dctx == null)
+                throw new ZstdException(ZSTD_ErrorCode.ZSTD_error_GENERIC, "Failed to create dctx");
+            safeHandle.SetHandle((IntPtr)dctx);
+            success = true;
+        }
+        finally
+        {
+            if (!success)
+            {
+                safeHandle.SetHandleAsInvalid();
+            }
+        }
+        return safeHandle;
+    }
+
+    /// <summary>
+    /// Acquires a reference to the safe handle.
+    /// </summary>
+    /// <returns>
+    /// A <see cref="SafeHandleHolder{T}"/> instance that can be implicitly converted to a pointer
+    /// to <see cref="ZSTD_DCtx_s"/>.
+    /// </returns>
+    public SafeHandleHolder<ZSTD_DCtx_s> Acquire() => new(this);
+
+    protected override bool ReleaseHandle()
+    {
+        return Unsafe.Methods.ZSTD_freeDCtx((ZSTD_DCtx_s*)handle) == 0;
+    }
+}
+
+/// <summary>
+/// Provides a convenient interface to safely acquire pointers of a specific type
+/// from a <see cref="SafeHandle"/>, by utilizing <see langword="using"/> blocks.
+/// </summary>
+/// <typeparam name="T">The type of pointers to return.</typeparam>
+/// <remarks>
+/// Safe handle holders can be <see cref="Dispose"/>d to decrement the safe handle's
+/// reference count, and can be implicitly converted to pointers to <see cref="T"/>.
+/// </remarks>
+internal unsafe ref struct SafeHandleHolder<T>
+    where T : unmanaged
+{
+    private readonly SafeHandle _handle;
+
+    private bool _refAdded;
+
+    public SafeHandleHolder(SafeHandle safeHandle)
+    {
+        _handle = safeHandle;
+        _refAdded = false;
+        safeHandle.DangerousAddRef(ref _refAdded);
+    }
+
+    public static implicit operator T*(SafeHandleHolder<T> holder) =>
+        (T*)holder._handle.DangerousGetHandle();
+
+    public void Dispose()
+    {
+        if (_refAdded)
+        {
+            _handle.DangerousRelease();
+            _refAdded = false;
+        }
+    }
+}

src/SharpCompress/Compressors/ZStandard/SynchronizationWrapper.cs

@@ -0,0 +1,22 @@
+using System.Threading;
+
+namespace SharpCompress.Compressors.ZStandard;
+
+internal static unsafe class SynchronizationWrapper
+{
+    private static object UnwrapObject(void** obj) => UnmanagedObject.Unwrap<object>(*obj);
+
+    public static void Init(void** obj) => *obj = UnmanagedObject.Wrap(new object());
+
+    public static void Free(void** obj) => UnmanagedObject.Free(*obj);
+
+    public static void Enter(void** obj) => Monitor.Enter(UnwrapObject(obj));
+
+    public static void Exit(void** obj) => Monitor.Exit(UnwrapObject(obj));
+
+    public static void Pulse(void** obj) => Monitor.Pulse(UnwrapObject(obj));
+
+    public static void PulseAll(void** obj) => Monitor.PulseAll(UnwrapObject(obj));
+
+    public static void Wait(void** mutex) => Monitor.Wait(UnwrapObject(mutex));
+}

src/SharpCompress/Compressors/ZStandard/ThrowHelper.cs

@@ -0,0 +1,48 @@
+using SharpCompress.Compressors.ZStandard.Unsafe;
+
+namespace SharpCompress.Compressors.ZStandard;
+
+public static unsafe class ThrowHelper
+{
+    private const ulong ZSTD_CONTENTSIZE_UNKNOWN = unchecked(0UL - 1);
+    private const ulong ZSTD_CONTENTSIZE_ERROR = unchecked(0UL - 2);
+
+    public static nuint EnsureZstdSuccess(this nuint returnValue)
+    {
+        if (Unsafe.Methods.ZSTD_isError(returnValue))
+            ThrowException(returnValue, Unsafe.Methods.ZSTD_getErrorName(returnValue));
+
+        return returnValue;
+    }
+
+    public static nuint EnsureZdictSuccess(this nuint returnValue)
+    {
+        if (Unsafe.Methods.ZDICT_isError(returnValue))
+            ThrowException(returnValue, Unsafe.Methods.ZDICT_getErrorName(returnValue));
+
+        return returnValue;
+    }
+
+    public static ulong EnsureContentSizeOk(this ulong returnValue)
+    {
+        if (returnValue == ZSTD_CONTENTSIZE_UNKNOWN)
+            throw new ZstdException(
+                ZSTD_ErrorCode.ZSTD_error_GENERIC,
+                "Decompressed content size is not specified"
+            );
+
+        if (returnValue == ZSTD_CONTENTSIZE_ERROR)
+            throw new ZstdException(
+                ZSTD_ErrorCode.ZSTD_error_GENERIC,
+                "Decompressed content size cannot be determined (e.g. invalid magic number, srcSize too small)"
+            );
+
+        return returnValue;
+    }
+
+    private static void ThrowException(nuint returnValue, string message)
+    {
+        var code = 0 - returnValue;
+        throw new ZstdException((ZSTD_ErrorCode)code, message);
+    }
+}

src/SharpCompress/Compressors/ZStandard/UnmanagedObject.cs

@@ -0,0 +1,18 @@
+using System;
+using System.Runtime.InteropServices;
+
+namespace SharpCompress.Compressors.ZStandard;
+
+/*
+ * Wrap object to void* to make it unmanaged
+ */
+internal static unsafe class UnmanagedObject
+{
+    public static void* Wrap(object obj) => (void*)GCHandle.ToIntPtr(GCHandle.Alloc(obj));
+
+    private static GCHandle UnwrapGcHandle(void* value) => GCHandle.FromIntPtr((IntPtr)value);
+
+    public static T Unwrap<T>(void* value) => (T)UnwrapGcHandle(value).Target!;
+
+    public static void Free(void* value) => UnwrapGcHandle(value).Free();
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/Allocations.cs

@@ -0,0 +1,52 @@
+using System.Runtime.CompilerServices;
+using static SharpCompress.Compressors.ZStandard.UnsafeHelper;
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    /* custom memory allocation functions */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void* ZSTD_customMalloc(nuint size, ZSTD_customMem customMem)
+    {
+        if (customMem.customAlloc != null)
+            return ((delegate* managed<void*, nuint, void*>)customMem.customAlloc)(
+                customMem.opaque,
+                size
+            );
+        return malloc(size);
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void* ZSTD_customCalloc(nuint size, ZSTD_customMem customMem)
+    {
+        if (customMem.customAlloc != null)
+        {
+            /* calloc implemented as malloc+memset;
+             * not as efficient as calloc, but next best guess for custom malloc */
+            void* ptr = ((delegate* managed<void*, nuint, void*>)customMem.customAlloc)(
+                customMem.opaque,
+                size
+            );
+            memset(ptr, 0, (uint)size);
+            return ptr;
+        }
+
+        return calloc(1, size);
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void ZSTD_customFree(void* ptr, ZSTD_customMem customMem)
+    {
+        if (ptr != null)
+        {
+            if (customMem.customFree != null)
+                ((delegate* managed<void*, void*, void>)customMem.customFree)(
+                    customMem.opaque,
+                    ptr
+                );
+            else
+                free(ptr);
+        }
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/BIT_CStream_t.cs

@@ -0,0 +1,14 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/* bitStream can mix input from multiple sources.
+ * A critical property of these streams is that they encode and decode in **reverse** direction.
+ * So the first bit sequence you add will be the last to be read, like a LIFO stack.
+ */
+public unsafe struct BIT_CStream_t
+{
+    public nuint bitContainer;
+    public uint bitPos;
+    public sbyte* startPtr;
+    public sbyte* ptr;
+    public sbyte* endPtr;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/BIT_DStream_status.cs

@@ -0,0 +1,16 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public enum BIT_DStream_status
+{
+    /* fully refilled */
+    BIT_DStream_unfinished = 0,
+
+    /* still some bits left in bitstream */
+    BIT_DStream_endOfBuffer = 1,
+
+    /* bitstream entirely consumed, bit-exact */
+    BIT_DStream_completed = 2,
+
+    /* user requested more bits than present in bitstream */
+    BIT_DStream_overflow = 3,
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/BIT_DStream_t.cs

@@ -0,0 +1,13 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/*-********************************************
+ *  bitStream decoding API (read backward)
+ **********************************************/
+public unsafe struct BIT_DStream_t
+{
+    public nuint bitContainer;
+    public uint bitsConsumed;
+    public sbyte* ptr;
+    public sbyte* start;
+    public sbyte* limitPtr;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/Bits.cs

@@ -0,0 +1,60 @@
+using System;
+using System.Numerics;
+using System.Runtime.CompilerServices;
+using static SharpCompress.Compressors.ZStandard.UnsafeHelper;
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint ZSTD_countTrailingZeros32(uint val)
+    {
+        assert(val != 0);
+        return (uint)BitOperations.TrailingZeroCount(val);
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint ZSTD_countLeadingZeros32(uint val)
+    {
+        assert(val != 0);
+        return (uint)BitOperations.LeadingZeroCount(val);
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint ZSTD_countTrailingZeros64(ulong val)
+    {
+        assert(val != 0);
+        return (uint)BitOperations.TrailingZeroCount(val);
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint ZSTD_countLeadingZeros64(ulong val)
+    {
+        assert(val != 0);
+        return (uint)BitOperations.LeadingZeroCount(val);
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint ZSTD_NbCommonBytes(nuint val)
+    {
+        assert(val != 0);
+        if (BitConverter.IsLittleEndian)
+        {
+            return MEM_64bits
+                ? (uint)BitOperations.TrailingZeroCount(val) >> 3
+                : (uint)BitOperations.TrailingZeroCount((uint)val) >> 3;
+        }
+
+        return MEM_64bits
+            ? (uint)BitOperations.LeadingZeroCount(val) >> 3
+            : (uint)BitOperations.LeadingZeroCount((uint)val) >> 3;
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint ZSTD_highbit32(uint val)
+    {
+        assert(val != 0);
+        return (uint)BitOperations.Log2(val);
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/Bitstream.cs

@@ -0,0 +1,739 @@
+using System;
+using System.Runtime.CompilerServices;
+using System.Runtime.InteropServices;
+using static SharpCompress.Compressors.ZStandard.UnsafeHelper;
+#if NETCOREAPP3_0_OR_GREATER
+using System.Runtime.Intrinsics.X86;
+#endif
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+#if NET7_0_OR_GREATER
+    private static ReadOnlySpan<uint> Span_BIT_mask =>
+        new uint[32]
+        {
+            0,
+            1,
+            3,
+            7,
+            0xF,
+            0x1F,
+            0x3F,
+            0x7F,
+            0xFF,
+            0x1FF,
+            0x3FF,
+            0x7FF,
+            0xFFF,
+            0x1FFF,
+            0x3FFF,
+            0x7FFF,
+            0xFFFF,
+            0x1FFFF,
+            0x3FFFF,
+            0x7FFFF,
+            0xFFFFF,
+            0x1FFFFF,
+            0x3FFFFF,
+            0x7FFFFF,
+            0xFFFFFF,
+            0x1FFFFFF,
+            0x3FFFFFF,
+            0x7FFFFFF,
+            0xFFFFFFF,
+            0x1FFFFFFF,
+            0x3FFFFFFF,
+            0x7FFFFFFF,
+        };
+    private static uint* BIT_mask =>
+        (uint*)
+            System.Runtime.CompilerServices.Unsafe.AsPointer(
+                ref MemoryMarshal.GetReference(Span_BIT_mask)
+            );
+#else
+
+    private static readonly uint* BIT_mask = GetArrayPointer(
+        new uint[32]
+        {
+            0,
+            1,
+            3,
+            7,
+            0xF,
+            0x1F,
+            0x3F,
+            0x7F,
+            0xFF,
+            0x1FF,
+            0x3FF,
+            0x7FF,
+            0xFFF,
+            0x1FFF,
+            0x3FFF,
+            0x7FFF,
+            0xFFFF,
+            0x1FFFF,
+            0x3FFFF,
+            0x7FFFF,
+            0xFFFFF,
+            0x1FFFFF,
+            0x3FFFFF,
+            0x7FFFFF,
+            0xFFFFFF,
+            0x1FFFFFF,
+            0x3FFFFFF,
+            0x7FFFFFF,
+            0xFFFFFFF,
+            0x1FFFFFFF,
+            0x3FFFFFFF,
+            0x7FFFFFFF,
+        }
+    );
+#endif
+    /*-**************************************************************
+     *  bitStream encoding
+     ****************************************************************/
+    /*! BIT_initCStream() :
+     *  `dstCapacity` must be > sizeof(size_t)
+     *  @return : 0 if success,
+     *            otherwise an error code (can be tested using ERR_isError()) */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_initCStream(ref BIT_CStream_t bitC, void* startPtr, nuint dstCapacity)
+    {
+        bitC.bitContainer = 0;
+        bitC.bitPos = 0;
+        bitC.startPtr = (sbyte*)startPtr;
+        bitC.ptr = bitC.startPtr;
+        bitC.endPtr = bitC.startPtr + dstCapacity - sizeof(nuint);
+        if (dstCapacity <= (nuint)sizeof(nuint))
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall));
+        return 0;
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_getLowerBits(nuint bitContainer, uint nbBits)
+    {
+        assert(nbBits < sizeof(uint) * 32 / sizeof(uint));
+#if NETCOREAPP3_1_OR_GREATER
+        if (Bmi2.X64.IsSupported)
+        {
+            return (nuint)Bmi2.X64.ZeroHighBits(bitContainer, nbBits);
+        }
+
+        if (Bmi2.IsSupported)
+        {
+            return Bmi2.ZeroHighBits((uint)bitContainer, nbBits);
+        }
+#endif
+
+        return bitContainer & BIT_mask[nbBits];
+    }
+
+    /*! BIT_addBits() :
+     *  can add up to 31 bits into `bitC`.
+     *  Note : does not check for register overflow ! */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void BIT_addBits(
+        ref nuint bitC_bitContainer,
+        ref uint bitC_bitPos,
+        nuint value,
+        uint nbBits
+    )
+    {
+        assert(nbBits < sizeof(uint) * 32 / sizeof(uint));
+        assert(nbBits + bitC_bitPos < (uint)(sizeof(nuint) * 8));
+        bitC_bitContainer |= BIT_getLowerBits(value, nbBits) << (int)bitC_bitPos;
+        bitC_bitPos += nbBits;
+    }
+
+    /*! BIT_addBitsFast() :
+     *  works only if `value` is _clean_,
+     *  meaning all high bits above nbBits are 0 */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void BIT_addBitsFast(
+        ref nuint bitC_bitContainer,
+        ref uint bitC_bitPos,
+        nuint value,
+        uint nbBits
+    )
+    {
+        assert(value >> (int)nbBits == 0);
+        assert(nbBits + bitC_bitPos < (uint)(sizeof(nuint) * 8));
+        bitC_bitContainer |= value << (int)bitC_bitPos;
+        bitC_bitPos += nbBits;
+    }
+
+    /*! BIT_flushBitsFast() :
+     *  assumption : bitContainer has not overflowed
+     *  unsafe version; does not check buffer overflow */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void BIT_flushBitsFast(
+        ref nuint bitC_bitContainer,
+        ref uint bitC_bitPos,
+        ref sbyte* bitC_ptr,
+        sbyte* bitC_endPtr
+    )
+    {
+        nuint nbBytes = bitC_bitPos >> 3;
+        assert(bitC_bitPos < (uint)(sizeof(nuint) * 8));
+        assert(bitC_ptr <= bitC_endPtr);
+        MEM_writeLEST(bitC_ptr, bitC_bitContainer);
+        bitC_ptr += nbBytes;
+        bitC_bitPos &= 7;
+        bitC_bitContainer >>= (int)(nbBytes * 8);
+    }
+
+    /*! BIT_flushBits() :
+     *  assumption : bitContainer has not overflowed
+     *  safe version; check for buffer overflow, and prevents it.
+     *  note : does not signal buffer overflow.
+     *  overflow will be revealed later on using BIT_closeCStream() */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void BIT_flushBits(
+        ref nuint bitC_bitContainer,
+        ref uint bitC_bitPos,
+        ref sbyte* bitC_ptr,
+        sbyte* bitC_endPtr
+    )
+    {
+        nuint nbBytes = bitC_bitPos >> 3;
+        assert(bitC_bitPos < (uint)(sizeof(nuint) * 8));
+        assert(bitC_ptr <= bitC_endPtr);
+        MEM_writeLEST(bitC_ptr, bitC_bitContainer);
+        bitC_ptr += nbBytes;
+        if (bitC_ptr > bitC_endPtr)
+            bitC_ptr = bitC_endPtr;
+        bitC_bitPos &= 7;
+        bitC_bitContainer >>= (int)(nbBytes * 8);
+    }
+
+    /*! BIT_closeCStream() :
+     *  @return : size of CStream, in bytes,
+     *            or 0 if it could not fit into dstBuffer */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_closeCStream(
+        ref nuint bitC_bitContainer,
+        ref uint bitC_bitPos,
+        sbyte* bitC_ptr,
+        sbyte* bitC_endPtr,
+        sbyte* bitC_startPtr
+    )
+    {
+        BIT_addBitsFast(ref bitC_bitContainer, ref bitC_bitPos, 1, 1);
+        BIT_flushBits(ref bitC_bitContainer, ref bitC_bitPos, ref bitC_ptr, bitC_endPtr);
+        if (bitC_ptr >= bitC_endPtr)
+            return 0;
+        return (nuint)(bitC_ptr - bitC_startPtr) + (nuint)(bitC_bitPos > 0 ? 1 : 0);
+    }
+
+    /*-********************************************************
+     *  bitStream decoding
+     **********************************************************/
+    /*! BIT_initDStream() :
+     *  Initialize a BIT_DStream_t.
+     * `bitD` : a pointer to an already allocated BIT_DStream_t structure.
+     * `srcSize` must be the *exact* size of the bitStream, in bytes.
+     * @return : size of stream (== srcSize), or an errorCode if a problem is detected
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_initDStream(BIT_DStream_t* bitD, void* srcBuffer, nuint srcSize)
+    {
+        if (srcSize < 1)
+        {
+            *bitD = new BIT_DStream_t();
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_srcSize_wrong));
+        }
+
+        bitD->start = (sbyte*)srcBuffer;
+        bitD->limitPtr = bitD->start + sizeof(nuint);
+        if (srcSize >= (nuint)sizeof(nuint))
+        {
+            bitD->ptr = (sbyte*)srcBuffer + srcSize - sizeof(nuint);
+            bitD->bitContainer = MEM_readLEST(bitD->ptr);
+            {
+                byte lastByte = ((byte*)srcBuffer)[srcSize - 1];
+                bitD->bitsConsumed = lastByte != 0 ? 8 - ZSTD_highbit32(lastByte) : 0;
+                if (lastByte == 0)
+                    return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+            }
+        }
+        else
+        {
+            bitD->ptr = bitD->start;
+            bitD->bitContainer = *(byte*)bitD->start;
+            switch (srcSize)
+            {
+                case 7:
+                    bitD->bitContainer += (nuint)((byte*)srcBuffer)[6] << sizeof(nuint) * 8 - 16;
+                    goto case 6;
+                case 6:
+                    bitD->bitContainer += (nuint)((byte*)srcBuffer)[5] << sizeof(nuint) * 8 - 24;
+                    goto case 5;
+                case 5:
+                    bitD->bitContainer += (nuint)((byte*)srcBuffer)[4] << sizeof(nuint) * 8 - 32;
+                    goto case 4;
+                case 4:
+                    bitD->bitContainer += (nuint)((byte*)srcBuffer)[3] << 24;
+                    goto case 3;
+                case 3:
+                    bitD->bitContainer += (nuint)((byte*)srcBuffer)[2] << 16;
+                    goto case 2;
+                case 2:
+                    bitD->bitContainer += (nuint)((byte*)srcBuffer)[1] << 8;
+                    goto default;
+                default:
+                    break;
+            }
+
+            {
+                byte lastByte = ((byte*)srcBuffer)[srcSize - 1];
+                bitD->bitsConsumed = lastByte != 0 ? 8 - ZSTD_highbit32(lastByte) : 0;
+                if (lastByte == 0)
+                    return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+            }
+
+            bitD->bitsConsumed += (uint)((nuint)sizeof(nuint) - srcSize) * 8;
+        }
+
+        return srcSize;
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_getUpperBits(nuint bitContainer, uint start)
+    {
+        return bitContainer >> (int)start;
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_getMiddleBits(nuint bitContainer, uint start, uint nbBits)
+    {
+        uint regMask = (uint)(sizeof(nuint) * 8 - 1);
+        assert(nbBits < sizeof(uint) * 32 / sizeof(uint));
+#if NETCOREAPP3_1_OR_GREATER
+        if (Bmi2.X64.IsSupported)
+        {
+            return (nuint)Bmi2.X64.ZeroHighBits(bitContainer >> (int)(start & regMask), nbBits);
+        }
+
+        if (Bmi2.IsSupported)
+        {
+            return Bmi2.ZeroHighBits((uint)(bitContainer >> (int)(start & regMask)), nbBits);
+        }
+#endif
+
+        return (nuint)(bitContainer >> (int)(start & regMask) & ((ulong)1 << (int)nbBits) - 1);
+    }
+
+    /*! BIT_lookBits() :
+     *  Provides next n bits from local register.
+     *  local register is not modified.
+     *  On 32-bits, maxNbBits==24.
+     *  On 64-bits, maxNbBits==56.
+     * @return : value extracted */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_lookBits(BIT_DStream_t* bitD, uint nbBits)
+    {
+        return BIT_getMiddleBits(
+            bitD->bitContainer,
+            (uint)(sizeof(nuint) * 8) - bitD->bitsConsumed - nbBits,
+            nbBits
+        );
+    }
+
+    /*! BIT_lookBitsFast() :
+     *  unsafe version; only works if nbBits >= 1 */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_lookBitsFast(BIT_DStream_t* bitD, uint nbBits)
+    {
+        uint regMask = (uint)(sizeof(nuint) * 8 - 1);
+        assert(nbBits >= 1);
+        return bitD->bitContainer
+            << (int)(bitD->bitsConsumed & regMask)
+            >> (int)(regMask + 1 - nbBits & regMask);
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void BIT_skipBits(BIT_DStream_t* bitD, uint nbBits)
+    {
+        bitD->bitsConsumed += nbBits;
+    }
+
+    /*! BIT_readBits() :
+     *  Read (consume) next n bits from local register and update.
+     *  Pay attention to not read more than nbBits contained into local register.
+     * @return : extracted value. */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_readBits(BIT_DStream_t* bitD, uint nbBits)
+    {
+        nuint value = BIT_lookBits(bitD, nbBits);
+        BIT_skipBits(bitD, nbBits);
+        return value;
+    }
+
+    /*! BIT_readBitsFast() :
+     *  unsafe version; only works if nbBits >= 1 */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_readBitsFast(BIT_DStream_t* bitD, uint nbBits)
+    {
+        nuint value = BIT_lookBitsFast(bitD, nbBits);
+        assert(nbBits >= 1);
+        BIT_skipBits(bitD, nbBits);
+        return value;
+    }
+
+    /*! BIT_reloadDStream_internal() :
+     *  Simple variant of BIT_reloadDStream(), with two conditions:
+     *  1. bitstream is valid : bitsConsumed <= sizeof(bitD->bitContainer)*8
+     *  2. look window is valid after shifted down : bitD->ptr >= bitD->start
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static BIT_DStream_status BIT_reloadDStream_internal(BIT_DStream_t* bitD)
+    {
+        assert(bitD->bitsConsumed <= (uint)(sizeof(nuint) * 8));
+        bitD->ptr -= bitD->bitsConsumed >> 3;
+        assert(bitD->ptr >= bitD->start);
+        bitD->bitsConsumed &= 7;
+        bitD->bitContainer = MEM_readLEST(bitD->ptr);
+        return BIT_DStream_status.BIT_DStream_unfinished;
+    }
+
+    /*! BIT_reloadDStreamFast() :
+     *  Similar to BIT_reloadDStream(), but with two differences:
+     *  1. bitsConsumed <= sizeof(bitD->bitContainer)*8 must hold!
+     *  2. Returns BIT_DStream_overflow when bitD->ptr < bitD->limitPtr, at this
+     *     point you must use BIT_reloadDStream() to reload.
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static BIT_DStream_status BIT_reloadDStreamFast(BIT_DStream_t* bitD)
+    {
+        if (bitD->ptr < bitD->limitPtr)
+            return BIT_DStream_status.BIT_DStream_overflow;
+        return BIT_reloadDStream_internal(bitD);
+    }
+
+#if NET7_0_OR_GREATER
+    private static ReadOnlySpan<byte> Span_static_zeroFilled =>
+        new byte[] { 0, 0, 0, 0, 0, 0, 0, 0 };
+    private static nuint* static_zeroFilled =>
+        (nuint*)
+            System.Runtime.CompilerServices.Unsafe.AsPointer(
+                ref MemoryMarshal.GetReference(Span_static_zeroFilled)
+            );
+#else
+
+    private static readonly nuint* static_zeroFilled = (nuint*)GetArrayPointer(
+        new byte[] { 0, 0, 0, 0, 0, 0, 0, 0 }
+    );
+#endif
+    /*! BIT_reloadDStream() :
+     *  Refill `bitD` from buffer previously set in BIT_initDStream() .
+     *  This function is safe, it guarantees it will not never beyond src buffer.
+     * @return : status of `BIT_DStream_t` internal register.
+     *           when status == BIT_DStream_unfinished, internal register is filled with at least 25 or 57 bits */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
+    {
+        if (bitD->bitsConsumed > (uint)(sizeof(nuint) * 8))
+        {
+            bitD->ptr = (sbyte*)&static_zeroFilled[0];
+            return BIT_DStream_status.BIT_DStream_overflow;
+        }
+
+        assert(bitD->ptr >= bitD->start);
+        if (bitD->ptr >= bitD->limitPtr)
+        {
+            return BIT_reloadDStream_internal(bitD);
+        }
+
+        if (bitD->ptr == bitD->start)
+        {
+            if (bitD->bitsConsumed < (uint)(sizeof(nuint) * 8))
+                return BIT_DStream_status.BIT_DStream_endOfBuffer;
+            return BIT_DStream_status.BIT_DStream_completed;
+        }
+
+        {
+            uint nbBytes = bitD->bitsConsumed >> 3;
+            BIT_DStream_status result = BIT_DStream_status.BIT_DStream_unfinished;
+            if (bitD->ptr - nbBytes < bitD->start)
+            {
+                nbBytes = (uint)(bitD->ptr - bitD->start);
+                result = BIT_DStream_status.BIT_DStream_endOfBuffer;
+            }
+
+            bitD->ptr -= nbBytes;
+            bitD->bitsConsumed -= nbBytes * 8;
+            bitD->bitContainer = MEM_readLEST(bitD->ptr);
+            return result;
+        }
+    }
+
+    /*! BIT_endOfDStream() :
+     * @return : 1 if DStream has _exactly_ reached its end (all bits consumed).
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint BIT_endOfDStream(BIT_DStream_t* DStream)
+    {
+        return DStream->ptr == DStream->start && DStream->bitsConsumed == (uint)(sizeof(nuint) * 8)
+            ? 1U
+            : 0U;
+    }
+
+    /*-********************************************************
+     *  bitStream decoding
+     **********************************************************/
+    /*! BIT_initDStream() :
+     *  Initialize a BIT_DStream_t.
+     * `bitD` : a pointer to an already allocated BIT_DStream_t structure.
+     * `srcSize` must be the *exact* size of the bitStream, in bytes.
+     * @return : size of stream (== srcSize), or an errorCode if a problem is detected
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_initDStream(ref BIT_DStream_t bitD, void* srcBuffer, nuint srcSize)
+    {
+        if (srcSize < 1)
+        {
+            bitD = new BIT_DStream_t();
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_srcSize_wrong));
+        }
+
+        bitD.start = (sbyte*)srcBuffer;
+        bitD.limitPtr = bitD.start + sizeof(nuint);
+        if (srcSize >= (nuint)sizeof(nuint))
+        {
+            bitD.ptr = (sbyte*)srcBuffer + srcSize - sizeof(nuint);
+            bitD.bitContainer = MEM_readLEST(bitD.ptr);
+            {
+                byte lastByte = ((byte*)srcBuffer)[srcSize - 1];
+                bitD.bitsConsumed = lastByte != 0 ? 8 - ZSTD_highbit32(lastByte) : 0;
+                if (lastByte == 0)
+                    return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+            }
+        }
+        else
+        {
+            bitD.ptr = bitD.start;
+            bitD.bitContainer = *(byte*)bitD.start;
+            switch (srcSize)
+            {
+                case 7:
+                    bitD.bitContainer += (nuint)((byte*)srcBuffer)[6] << sizeof(nuint) * 8 - 16;
+                    goto case 6;
+                case 6:
+                    bitD.bitContainer += (nuint)((byte*)srcBuffer)[5] << sizeof(nuint) * 8 - 24;
+                    goto case 5;
+                case 5:
+                    bitD.bitContainer += (nuint)((byte*)srcBuffer)[4] << sizeof(nuint) * 8 - 32;
+                    goto case 4;
+                case 4:
+                    bitD.bitContainer += (nuint)((byte*)srcBuffer)[3] << 24;
+                    goto case 3;
+                case 3:
+                    bitD.bitContainer += (nuint)((byte*)srcBuffer)[2] << 16;
+                    goto case 2;
+                case 2:
+                    bitD.bitContainer += (nuint)((byte*)srcBuffer)[1] << 8;
+                    goto default;
+                default:
+                    break;
+            }
+
+            {
+                byte lastByte = ((byte*)srcBuffer)[srcSize - 1];
+                bitD.bitsConsumed = lastByte != 0 ? 8 - ZSTD_highbit32(lastByte) : 0;
+                if (lastByte == 0)
+                    return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+            }
+
+            bitD.bitsConsumed += (uint)((nuint)sizeof(nuint) - srcSize) * 8;
+        }
+
+        return srcSize;
+    }
+
+    /*! BIT_lookBits() :
+     *  Provides next n bits from local register.
+     *  local register is not modified.
+     *  On 32-bits, maxNbBits==24.
+     *  On 64-bits, maxNbBits==56.
+     * @return : value extracted */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_lookBits(nuint bitD_bitContainer, uint bitD_bitsConsumed, uint nbBits)
+    {
+        return BIT_getMiddleBits(
+            bitD_bitContainer,
+            (uint)(sizeof(nuint) * 8) - bitD_bitsConsumed - nbBits,
+            nbBits
+        );
+    }
+
+    /*! BIT_lookBitsFast() :
+     *  unsafe version; only works if nbBits >= 1 */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_lookBitsFast(
+        nuint bitD_bitContainer,
+        uint bitD_bitsConsumed,
+        uint nbBits
+    )
+    {
+        uint regMask = (uint)(sizeof(nuint) * 8 - 1);
+        assert(nbBits >= 1);
+        return bitD_bitContainer
+            << (int)(bitD_bitsConsumed & regMask)
+            >> (int)(regMask + 1 - nbBits & regMask);
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void BIT_skipBits(ref uint bitD_bitsConsumed, uint nbBits)
+    {
+        bitD_bitsConsumed += nbBits;
+    }
+
+    /*! BIT_readBits() :
+     *  Read (consume) next n bits from local register and update.
+     *  Pay attention to not read more than nbBits contained into local register.
+     * @return : extracted value. */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_readBits(
+        nuint bitD_bitContainer,
+        ref uint bitD_bitsConsumed,
+        uint nbBits
+    )
+    {
+        nuint value = BIT_lookBits(bitD_bitContainer, bitD_bitsConsumed, nbBits);
+        BIT_skipBits(ref bitD_bitsConsumed, nbBits);
+        return value;
+    }
+
+    /*! BIT_readBitsFast() :
+     *  unsafe version; only works if nbBits >= 1 */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint BIT_readBitsFast(
+        nuint bitD_bitContainer,
+        ref uint bitD_bitsConsumed,
+        uint nbBits
+    )
+    {
+        nuint value = BIT_lookBitsFast(bitD_bitContainer, bitD_bitsConsumed, nbBits);
+        assert(nbBits >= 1);
+        BIT_skipBits(ref bitD_bitsConsumed, nbBits);
+        return value;
+    }
+
+    /*! BIT_reloadDStreamFast() :
+     *  Similar to BIT_reloadDStream(), but with two differences:
+     *  1. bitsConsumed <= sizeof(bitD->bitContainer)*8 must hold!
+     *  2. Returns BIT_DStream_overflow when bitD->ptr < bitD->limitPtr, at this
+     *     point you must use BIT_reloadDStream() to reload.
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static BIT_DStream_status BIT_reloadDStreamFast(
+        ref nuint bitD_bitContainer,
+        ref uint bitD_bitsConsumed,
+        ref sbyte* bitD_ptr,
+        sbyte* bitD_start,
+        sbyte* bitD_limitPtr
+    )
+    {
+        if (bitD_ptr < bitD_limitPtr)
+            return BIT_DStream_status.BIT_DStream_overflow;
+        return BIT_reloadDStream_internal(
+            ref bitD_bitContainer,
+            ref bitD_bitsConsumed,
+            ref bitD_ptr,
+            bitD_start
+        );
+    }
+
+    /*! BIT_reloadDStream() :
+     *  Refill `bitD` from buffer previously set in BIT_initDStream() .
+     *  This function is safe, it guarantees it will not never beyond src buffer.
+     * @return : status of `BIT_DStream_t` internal register.
+     *           when status == BIT_DStream_unfinished, internal register is filled with at least 25 or 57 bits */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static BIT_DStream_status BIT_reloadDStream(
+        ref nuint bitD_bitContainer,
+        ref uint bitD_bitsConsumed,
+        ref sbyte* bitD_ptr,
+        sbyte* bitD_start,
+        sbyte* bitD_limitPtr
+    )
+    {
+        if (bitD_bitsConsumed > (uint)(sizeof(nuint) * 8))
+        {
+            bitD_ptr = (sbyte*)&static_zeroFilled[0];
+            return BIT_DStream_status.BIT_DStream_overflow;
+        }
+
+        assert(bitD_ptr >= bitD_start);
+        if (bitD_ptr >= bitD_limitPtr)
+        {
+            return BIT_reloadDStream_internal(
+                ref bitD_bitContainer,
+                ref bitD_bitsConsumed,
+                ref bitD_ptr,
+                bitD_start
+            );
+        }
+
+        if (bitD_ptr == bitD_start)
+        {
+            if (bitD_bitsConsumed < (uint)(sizeof(nuint) * 8))
+                return BIT_DStream_status.BIT_DStream_endOfBuffer;
+            return BIT_DStream_status.BIT_DStream_completed;
+        }
+
+        {
+            uint nbBytes = bitD_bitsConsumed >> 3;
+            BIT_DStream_status result = BIT_DStream_status.BIT_DStream_unfinished;
+            if (bitD_ptr - nbBytes < bitD_start)
+            {
+                nbBytes = (uint)(bitD_ptr - bitD_start);
+                result = BIT_DStream_status.BIT_DStream_endOfBuffer;
+            }
+
+            bitD_ptr -= nbBytes;
+            bitD_bitsConsumed -= nbBytes * 8;
+            bitD_bitContainer = MEM_readLEST(bitD_ptr);
+            return result;
+        }
+    }
+
+    /*! BIT_reloadDStream_internal() :
+     *  Simple variant of BIT_reloadDStream(), with two conditions:
+     *  1. bitstream is valid : bitsConsumed <= sizeof(bitD->bitContainer)*8
+     *  2. look window is valid after shifted down : bitD->ptr >= bitD->start
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static BIT_DStream_status BIT_reloadDStream_internal(
+        ref nuint bitD_bitContainer,
+        ref uint bitD_bitsConsumed,
+        ref sbyte* bitD_ptr,
+        sbyte* bitD_start
+    )
+    {
+        assert(bitD_bitsConsumed <= (uint)(sizeof(nuint) * 8));
+        bitD_ptr -= bitD_bitsConsumed >> 3;
+        assert(bitD_ptr >= bitD_start);
+        bitD_bitsConsumed &= 7;
+        bitD_bitContainer = MEM_readLEST(bitD_ptr);
+        return BIT_DStream_status.BIT_DStream_unfinished;
+    }
+
+    /*! BIT_endOfDStream() :
+     * @return : 1 if DStream has _exactly_ reached its end (all bits consumed).
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint BIT_endOfDStream(
+        uint DStream_bitsConsumed,
+        sbyte* DStream_ptr,
+        sbyte* DStream_start
+    )
+    {
+        return DStream_ptr == DStream_start && DStream_bitsConsumed == (uint)(sizeof(nuint) * 8)
+            ? 1U
+            : 0U;
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/BlockSummary.cs

@@ -0,0 +1,8 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public struct BlockSummary
+{
+    public nuint nbSequences;
+    public nuint blockSize;
+    public nuint litSize;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/COVER_best_s.cs

@@ -0,0 +1,20 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/**
+ * COVER_best_t is used for two purposes:
+ * 1. Synchronizing threads.
+ * 2. Saving the best parameters and dictionary.
+ *
+ * All of the methods except COVER_best_init() are thread safe if zstd is
+ * compiled with multithreaded support.
+ */
+public unsafe struct COVER_best_s
+{
+    public void* mutex;
+    public void* cond;
+    public nuint liveJobs;
+    public void* dict;
+    public nuint dictSize;
+    public ZDICT_cover_params_t parameters;
+    public nuint compressedSize;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/COVER_ctx_t.cs

@@ -0,0 +1,19 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/*-*************************************
+ * Context
+ ***************************************/
+public unsafe struct COVER_ctx_t
+{
+    public byte* samples;
+    public nuint* offsets;
+    public nuint* samplesSizes;
+    public nuint nbSamples;
+    public nuint nbTrainSamples;
+    public nuint nbTestSamples;
+    public uint* suffix;
+    public nuint suffixSize;
+    public uint* freqs;
+    public uint* dmerAt;
+    public uint d;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/COVER_dictSelection.cs

@@ -0,0 +1,11 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/**
+ * Struct used for the dictionary selection function.
+ */
+public unsafe struct COVER_dictSelection
+{
+    public byte* dictContent;
+    public nuint dictSize;
+    public nuint totalCompressedSize;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/COVER_epoch_info_t.cs

@@ -0,0 +1,10 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/**
+ *Number of epochs and size of each epoch.
+ */
+public struct COVER_epoch_info_t
+{
+    public uint num;
+    public uint size;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/COVER_map_pair_t_s.cs

@@ -0,0 +1,7 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public struct COVER_map_pair_t_s
+{
+    public uint key;
+    public uint value;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/COVER_map_s.cs

@@ -0,0 +1,9 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public unsafe struct COVER_map_s
+{
+    public COVER_map_pair_t_s* data;
+    public uint sizeLog;
+    public uint size;
+    public uint sizeMask;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/COVER_segment_t.cs

@@ -0,0 +1,11 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/**
+ * A segment is a range in the source as well as the score of the segment.
+ */
+public struct COVER_segment_t
+{
+    public uint begin;
+    public uint end;
+    public uint score;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/COVER_tryParameters_data_s.cs

@@ -0,0 +1,12 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/**
+ * Parameters for COVER_tryParameters().
+ */
+public unsafe struct COVER_tryParameters_data_s
+{
+    public COVER_ctx_t* ctx;
+    public COVER_best_s* best;
+    public nuint dictBufferCapacity;
+    public ZDICT_cover_params_t parameters;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/Clevels.cs

@@ -0,0 +1,849 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    private static readonly ZSTD_compressionParameters[][] ZSTD_defaultCParameters =
+        new ZSTD_compressionParameters[4][]
+        {
+            new ZSTD_compressionParameters[23]
+            {
+                new ZSTD_compressionParameters(
+                    windowLog: 19,
+                    chainLog: 12,
+                    hashLog: 13,
+                    searchLog: 1,
+                    minMatch: 6,
+                    targetLength: 1,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 19,
+                    chainLog: 13,
+                    hashLog: 14,
+                    searchLog: 1,
+                    minMatch: 7,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 20,
+                    chainLog: 15,
+                    hashLog: 16,
+                    searchLog: 1,
+                    minMatch: 6,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 21,
+                    chainLog: 16,
+                    hashLog: 17,
+                    searchLog: 1,
+                    minMatch: 5,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_dfast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 21,
+                    chainLog: 18,
+                    hashLog: 18,
+                    searchLog: 1,
+                    minMatch: 5,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_dfast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 21,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 3,
+                    minMatch: 5,
+                    targetLength: 2,
+                    strategy: ZSTD_strategy.ZSTD_greedy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 21,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 3,
+                    minMatch: 5,
+                    targetLength: 4,
+                    strategy: ZSTD_strategy.ZSTD_lazy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 21,
+                    chainLog: 19,
+                    hashLog: 20,
+                    searchLog: 4,
+                    minMatch: 5,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 21,
+                    chainLog: 19,
+                    hashLog: 20,
+                    searchLog: 4,
+                    minMatch: 5,
+                    targetLength: 16,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 22,
+                    chainLog: 20,
+                    hashLog: 21,
+                    searchLog: 4,
+                    minMatch: 5,
+                    targetLength: 16,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 22,
+                    chainLog: 21,
+                    hashLog: 22,
+                    searchLog: 5,
+                    minMatch: 5,
+                    targetLength: 16,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 22,
+                    chainLog: 21,
+                    hashLog: 22,
+                    searchLog: 6,
+                    minMatch: 5,
+                    targetLength: 16,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 22,
+                    chainLog: 22,
+                    hashLog: 23,
+                    searchLog: 6,
+                    minMatch: 5,
+                    targetLength: 32,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 22,
+                    chainLog: 22,
+                    hashLog: 22,
+                    searchLog: 4,
+                    minMatch: 5,
+                    targetLength: 32,
+                    strategy: ZSTD_strategy.ZSTD_btlazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 22,
+                    chainLog: 22,
+                    hashLog: 23,
+                    searchLog: 5,
+                    minMatch: 5,
+                    targetLength: 32,
+                    strategy: ZSTD_strategy.ZSTD_btlazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 22,
+                    chainLog: 23,
+                    hashLog: 23,
+                    searchLog: 6,
+                    minMatch: 5,
+                    targetLength: 32,
+                    strategy: ZSTD_strategy.ZSTD_btlazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 22,
+                    chainLog: 22,
+                    hashLog: 22,
+                    searchLog: 5,
+                    minMatch: 5,
+                    targetLength: 48,
+                    strategy: ZSTD_strategy.ZSTD_btopt
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 23,
+                    chainLog: 23,
+                    hashLog: 22,
+                    searchLog: 5,
+                    minMatch: 4,
+                    targetLength: 64,
+                    strategy: ZSTD_strategy.ZSTD_btopt
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 23,
+                    chainLog: 23,
+                    hashLog: 22,
+                    searchLog: 6,
+                    minMatch: 3,
+                    targetLength: 64,
+                    strategy: ZSTD_strategy.ZSTD_btultra
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 23,
+                    chainLog: 24,
+                    hashLog: 22,
+                    searchLog: 7,
+                    minMatch: 3,
+                    targetLength: 256,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 25,
+                    chainLog: 25,
+                    hashLog: 23,
+                    searchLog: 7,
+                    minMatch: 3,
+                    targetLength: 256,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 26,
+                    chainLog: 26,
+                    hashLog: 24,
+                    searchLog: 7,
+                    minMatch: 3,
+                    targetLength: 512,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 27,
+                    chainLog: 27,
+                    hashLog: 25,
+                    searchLog: 9,
+                    minMatch: 3,
+                    targetLength: 999,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+            },
+            new ZSTD_compressionParameters[23]
+            {
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 12,
+                    hashLog: 13,
+                    searchLog: 1,
+                    minMatch: 5,
+                    targetLength: 1,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 13,
+                    hashLog: 14,
+                    searchLog: 1,
+                    minMatch: 6,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 14,
+                    hashLog: 14,
+                    searchLog: 1,
+                    minMatch: 5,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_dfast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 16,
+                    hashLog: 16,
+                    searchLog: 1,
+                    minMatch: 4,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_dfast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 16,
+                    hashLog: 17,
+                    searchLog: 3,
+                    minMatch: 5,
+                    targetLength: 2,
+                    strategy: ZSTD_strategy.ZSTD_greedy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 17,
+                    hashLog: 18,
+                    searchLog: 5,
+                    minMatch: 5,
+                    targetLength: 2,
+                    strategy: ZSTD_strategy.ZSTD_greedy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 3,
+                    minMatch: 5,
+                    targetLength: 4,
+                    strategy: ZSTD_strategy.ZSTD_lazy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 4,
+                    minMatch: 4,
+                    targetLength: 4,
+                    strategy: ZSTD_strategy.ZSTD_lazy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 4,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 5,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 6,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 5,
+                    minMatch: 4,
+                    targetLength: 12,
+                    strategy: ZSTD_strategy.ZSTD_btlazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 19,
+                    hashLog: 19,
+                    searchLog: 7,
+                    minMatch: 4,
+                    targetLength: 12,
+                    strategy: ZSTD_strategy.ZSTD_btlazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 4,
+                    minMatch: 4,
+                    targetLength: 16,
+                    strategy: ZSTD_strategy.ZSTD_btopt
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 4,
+                    minMatch: 3,
+                    targetLength: 32,
+                    strategy: ZSTD_strategy.ZSTD_btopt
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 18,
+                    hashLog: 19,
+                    searchLog: 6,
+                    minMatch: 3,
+                    targetLength: 128,
+                    strategy: ZSTD_strategy.ZSTD_btopt
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 19,
+                    hashLog: 19,
+                    searchLog: 6,
+                    minMatch: 3,
+                    targetLength: 128,
+                    strategy: ZSTD_strategy.ZSTD_btultra
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 19,
+                    hashLog: 19,
+                    searchLog: 8,
+                    minMatch: 3,
+                    targetLength: 256,
+                    strategy: ZSTD_strategy.ZSTD_btultra
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 19,
+                    hashLog: 19,
+                    searchLog: 6,
+                    minMatch: 3,
+                    targetLength: 128,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 19,
+                    hashLog: 19,
+                    searchLog: 8,
+                    minMatch: 3,
+                    targetLength: 256,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 19,
+                    hashLog: 19,
+                    searchLog: 10,
+                    minMatch: 3,
+                    targetLength: 512,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 19,
+                    hashLog: 19,
+                    searchLog: 12,
+                    minMatch: 3,
+                    targetLength: 512,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 18,
+                    chainLog: 19,
+                    hashLog: 19,
+                    searchLog: 13,
+                    minMatch: 3,
+                    targetLength: 999,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+            },
+            new ZSTD_compressionParameters[23]
+            {
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 12,
+                    hashLog: 12,
+                    searchLog: 1,
+                    minMatch: 5,
+                    targetLength: 1,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 12,
+                    hashLog: 13,
+                    searchLog: 1,
+                    minMatch: 6,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 13,
+                    hashLog: 15,
+                    searchLog: 1,
+                    minMatch: 5,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 15,
+                    hashLog: 16,
+                    searchLog: 2,
+                    minMatch: 5,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_dfast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 17,
+                    hashLog: 17,
+                    searchLog: 2,
+                    minMatch: 4,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_dfast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 16,
+                    hashLog: 17,
+                    searchLog: 3,
+                    minMatch: 4,
+                    targetLength: 2,
+                    strategy: ZSTD_strategy.ZSTD_greedy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 16,
+                    hashLog: 17,
+                    searchLog: 3,
+                    minMatch: 4,
+                    targetLength: 4,
+                    strategy: ZSTD_strategy.ZSTD_lazy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 16,
+                    hashLog: 17,
+                    searchLog: 3,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 16,
+                    hashLog: 17,
+                    searchLog: 4,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 16,
+                    hashLog: 17,
+                    searchLog: 5,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 16,
+                    hashLog: 17,
+                    searchLog: 6,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 17,
+                    hashLog: 17,
+                    searchLog: 5,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_btlazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 7,
+                    minMatch: 4,
+                    targetLength: 12,
+                    strategy: ZSTD_strategy.ZSTD_btlazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 3,
+                    minMatch: 4,
+                    targetLength: 12,
+                    strategy: ZSTD_strategy.ZSTD_btopt
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 4,
+                    minMatch: 3,
+                    targetLength: 32,
+                    strategy: ZSTD_strategy.ZSTD_btopt
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 6,
+                    minMatch: 3,
+                    targetLength: 256,
+                    strategy: ZSTD_strategy.ZSTD_btopt
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 6,
+                    minMatch: 3,
+                    targetLength: 128,
+                    strategy: ZSTD_strategy.ZSTD_btultra
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 8,
+                    minMatch: 3,
+                    targetLength: 256,
+                    strategy: ZSTD_strategy.ZSTD_btultra
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 10,
+                    minMatch: 3,
+                    targetLength: 512,
+                    strategy: ZSTD_strategy.ZSTD_btultra
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 5,
+                    minMatch: 3,
+                    targetLength: 256,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 7,
+                    minMatch: 3,
+                    targetLength: 512,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 9,
+                    minMatch: 3,
+                    targetLength: 512,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 17,
+                    chainLog: 18,
+                    hashLog: 17,
+                    searchLog: 11,
+                    minMatch: 3,
+                    targetLength: 999,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+            },
+            new ZSTD_compressionParameters[23]
+            {
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 12,
+                    hashLog: 13,
+                    searchLog: 1,
+                    minMatch: 5,
+                    targetLength: 1,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 14,
+                    hashLog: 15,
+                    searchLog: 1,
+                    minMatch: 5,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 14,
+                    hashLog: 15,
+                    searchLog: 1,
+                    minMatch: 4,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_fast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 14,
+                    hashLog: 15,
+                    searchLog: 2,
+                    minMatch: 4,
+                    targetLength: 0,
+                    strategy: ZSTD_strategy.ZSTD_dfast
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 14,
+                    hashLog: 14,
+                    searchLog: 4,
+                    minMatch: 4,
+                    targetLength: 2,
+                    strategy: ZSTD_strategy.ZSTD_greedy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 14,
+                    hashLog: 14,
+                    searchLog: 3,
+                    minMatch: 4,
+                    targetLength: 4,
+                    strategy: ZSTD_strategy.ZSTD_lazy
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 14,
+                    hashLog: 14,
+                    searchLog: 4,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 14,
+                    hashLog: 14,
+                    searchLog: 6,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 14,
+                    hashLog: 14,
+                    searchLog: 8,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_lazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 14,
+                    searchLog: 5,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_btlazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 14,
+                    searchLog: 9,
+                    minMatch: 4,
+                    targetLength: 8,
+                    strategy: ZSTD_strategy.ZSTD_btlazy2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 14,
+                    searchLog: 3,
+                    minMatch: 4,
+                    targetLength: 12,
+                    strategy: ZSTD_strategy.ZSTD_btopt
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 14,
+                    searchLog: 4,
+                    minMatch: 3,
+                    targetLength: 24,
+                    strategy: ZSTD_strategy.ZSTD_btopt
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 14,
+                    searchLog: 5,
+                    minMatch: 3,
+                    targetLength: 32,
+                    strategy: ZSTD_strategy.ZSTD_btultra
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 15,
+                    searchLog: 6,
+                    minMatch: 3,
+                    targetLength: 64,
+                    strategy: ZSTD_strategy.ZSTD_btultra
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 15,
+                    searchLog: 7,
+                    minMatch: 3,
+                    targetLength: 256,
+                    strategy: ZSTD_strategy.ZSTD_btultra
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 15,
+                    searchLog: 5,
+                    minMatch: 3,
+                    targetLength: 48,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 15,
+                    searchLog: 6,
+                    minMatch: 3,
+                    targetLength: 128,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 15,
+                    searchLog: 7,
+                    minMatch: 3,
+                    targetLength: 256,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 15,
+                    searchLog: 8,
+                    minMatch: 3,
+                    targetLength: 256,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 15,
+                    searchLog: 8,
+                    minMatch: 3,
+                    targetLength: 512,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 15,
+                    searchLog: 9,
+                    minMatch: 3,
+                    targetLength: 512,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+                new ZSTD_compressionParameters(
+                    windowLog: 14,
+                    chainLog: 15,
+                    hashLog: 15,
+                    searchLog: 10,
+                    minMatch: 3,
+                    targetLength: 999,
+                    strategy: ZSTD_strategy.ZSTD_btultra2
+                ),
+            },
+        };
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/Compiler.cs

@@ -0,0 +1,61 @@
+using System.Runtime.CompilerServices;
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    /* @return 1 if @u is a 2^n value, 0 otherwise
+     * useful to check a value is valid for alignment restrictions */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static int ZSTD_isPower2(nuint u)
+    {
+        return (u & u - 1) == 0 ? 1 : 0;
+    }
+
+    /**
+     * Helper function to perform a wrapped pointer difference without triggering
+     * UBSAN.
+     *
+     * @returns lhs - rhs with wrapping
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nint ZSTD_wrappedPtrDiff(byte* lhs, byte* rhs)
+    {
+        return (nint)(lhs - rhs);
+    }
+
+    /**
+     * Helper function to perform a wrapped pointer add without triggering UBSAN.
+     *
+     * @return ptr + add with wrapping
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static byte* ZSTD_wrappedPtrAdd(byte* ptr, nint add)
+    {
+        return ptr + add;
+    }
+
+    /**
+     * Helper function to perform a wrapped pointer subtraction without triggering
+     * UBSAN.
+     *
+     * @return ptr - sub with wrapping
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static byte* ZSTD_wrappedPtrSub(byte* ptr, nint sub)
+    {
+        return ptr - sub;
+    }
+
+    /**
+     * Helper function to add to a pointer that works around C's undefined behavior
+     * of adding 0 to NULL.
+     *
+     * @returns `ptr + add` except it defines `NULL + 0 == NULL`.
+     */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static byte* ZSTD_maybeNullPtrAdd(byte* ptr, nint add)
+    {
+        return add > 0 ? ptr + add : ptr;
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/Cover.cs

@@ -0,0 +1,444 @@
+using static SharpCompress.Compressors.ZStandard.UnsafeHelper;
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    private static int g_displayLevel = 0;
+
+    /**
+     * Returns the sum of the sample sizes.
+     */
+    private static nuint COVER_sum(nuint* samplesSizes, uint nbSamples)
+    {
+        nuint sum = 0;
+        uint i;
+        for (i = 0; i < nbSamples; ++i)
+        {
+            sum += samplesSizes[i];
+        }
+
+        return sum;
+    }
+
+    /**
+     * Warns the user when their corpus is too small.
+     */
+    private static void COVER_warnOnSmallCorpus(nuint maxDictSize, nuint nbDmers, int displayLevel)
+    {
+        double ratio = nbDmers / (double)maxDictSize;
+        if (ratio >= 10)
+        {
+            return;
+        }
+    }
+
+    /**
+     * Computes the number of epochs and the size of each epoch.
+     * We will make sure that each epoch gets at least 10 * k bytes.
+     *
+     * The COVER algorithms divide the data up into epochs of equal size and
+     * select one segment from each epoch.
+     *
+     * @param maxDictSize The maximum allowed dictionary size.
+     * @param nbDmers     The number of dmers we are training on.
+     * @param k           The parameter k (segment size).
+     * @param passes      The target number of passes over the dmer corpus.
+     *                    More passes means a better dictionary.
+     */
+    private static COVER_epoch_info_t COVER_computeEpochs(
+        uint maxDictSize,
+        uint nbDmers,
+        uint k,
+        uint passes
+    )
+    {
+        uint minEpochSize = k * 10;
+        COVER_epoch_info_t epochs;
+        epochs.num = 1 > maxDictSize / k / passes ? 1 : maxDictSize / k / passes;
+        epochs.size = nbDmers / epochs.num;
+        if (epochs.size >= minEpochSize)
+        {
+            assert(epochs.size * epochs.num <= nbDmers);
+            return epochs;
+        }
+
+        epochs.size = minEpochSize < nbDmers ? minEpochSize : nbDmers;
+        epochs.num = nbDmers / epochs.size;
+        assert(epochs.size * epochs.num <= nbDmers);
+        return epochs;
+    }
+
+    /**
+     *  Checks total compressed size of a dictionary
+     */
+    private static nuint COVER_checkTotalCompressedSize(
+        ZDICT_cover_params_t parameters,
+        nuint* samplesSizes,
+        byte* samples,
+        nuint* offsets,
+        nuint nbTrainSamples,
+        nuint nbSamples,
+        byte* dict,
+        nuint dictBufferCapacity
+    )
+    {
+        nuint totalCompressedSize = unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+        /* Pointers */
+        ZSTD_CCtx_s* cctx;
+        ZSTD_CDict_s* cdict;
+        void* dst;
+        /* Local variables */
+        nuint dstCapacity;
+        nuint i;
+        {
+            nuint maxSampleSize = 0;
+            i = parameters.splitPoint < 1 ? nbTrainSamples : 0;
+            for (; i < nbSamples; ++i)
+            {
+                maxSampleSize = samplesSizes[i] > maxSampleSize ? samplesSizes[i] : maxSampleSize;
+            }
+
+            dstCapacity = ZSTD_compressBound(maxSampleSize);
+            dst = malloc(dstCapacity);
+        }
+
+        cctx = ZSTD_createCCtx();
+        cdict = ZSTD_createCDict(dict, dictBufferCapacity, parameters.zParams.compressionLevel);
+        if (dst == null || cctx == null || cdict == null)
+        {
+            goto _compressCleanup;
+        }
+
+        totalCompressedSize = dictBufferCapacity;
+        i = parameters.splitPoint < 1 ? nbTrainSamples : 0;
+        for (; i < nbSamples; ++i)
+        {
+            nuint size = ZSTD_compress_usingCDict(
+                cctx,
+                dst,
+                dstCapacity,
+                samples + offsets[i],
+                samplesSizes[i],
+                cdict
+            );
+            if (ERR_isError(size))
+            {
+                totalCompressedSize = size;
+                goto _compressCleanup;
+            }
+
+            totalCompressedSize += size;
+        }
+
+        _compressCleanup:
+        ZSTD_freeCCtx(cctx);
+        ZSTD_freeCDict(cdict);
+        if (dst != null)
+        {
+            free(dst);
+        }
+
+        return totalCompressedSize;
+    }
+
+    /**
+     * Initialize the `COVER_best_t`.
+     */
+    private static void COVER_best_init(COVER_best_s* best)
+    {
+        if (best == null)
+            return;
+        SynchronizationWrapper.Init(&best->mutex);
+        best->liveJobs = 0;
+        best->dict = null;
+        best->dictSize = 0;
+        best->compressedSize = unchecked((nuint)(-1));
+        best->parameters = new ZDICT_cover_params_t();
+    }
+
+    /**
+     * Wait until liveJobs == 0.
+     */
+    private static void COVER_best_wait(COVER_best_s* best)
+    {
+        if (best == null)
+        {
+            return;
+        }
+
+        SynchronizationWrapper.Enter(&best->mutex);
+        while (best->liveJobs != 0)
+        {
+            SynchronizationWrapper.Wait(&best->mutex);
+        }
+
+        SynchronizationWrapper.Exit(&best->mutex);
+    }
+
+    /**
+     * Call COVER_best_wait() and then destroy the COVER_best_t.
+     */
+    private static void COVER_best_destroy(COVER_best_s* best)
+    {
+        if (best == null)
+        {
+            return;
+        }
+
+        COVER_best_wait(best);
+        if (best->dict != null)
+        {
+            free(best->dict);
+        }
+
+        SynchronizationWrapper.Free(&best->mutex);
+    }
+
+    /**
+     * Called when a thread is about to be launched.
+     * Increments liveJobs.
+     */
+    private static void COVER_best_start(COVER_best_s* best)
+    {
+        if (best == null)
+        {
+            return;
+        }
+
+        SynchronizationWrapper.Enter(&best->mutex);
+        ++best->liveJobs;
+        SynchronizationWrapper.Exit(&best->mutex);
+    }
+
+    /**
+     * Called when a thread finishes executing, both on error or success.
+     * Decrements liveJobs and signals any waiting threads if liveJobs == 0.
+     * If this dictionary is the best so far save it and its parameters.
+     */
+    private static void COVER_best_finish(
+        COVER_best_s* best,
+        ZDICT_cover_params_t parameters,
+        COVER_dictSelection selection
+    )
+    {
+        void* dict = selection.dictContent;
+        nuint compressedSize = selection.totalCompressedSize;
+        nuint dictSize = selection.dictSize;
+        if (best == null)
+        {
+            return;
+        }
+
+        {
+            nuint liveJobs;
+            SynchronizationWrapper.Enter(&best->mutex);
+            --best->liveJobs;
+            liveJobs = best->liveJobs;
+            if (compressedSize < best->compressedSize)
+            {
+                if (best->dict == null || best->dictSize < dictSize)
+                {
+                    if (best->dict != null)
+                    {
+                        free(best->dict);
+                    }
+
+                    best->dict = malloc(dictSize);
+                    if (best->dict == null)
+                    {
+                        best->compressedSize = unchecked(
+                            (nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC)
+                        );
+                        best->dictSize = 0;
+                        SynchronizationWrapper.Pulse(&best->mutex);
+                        SynchronizationWrapper.Exit(&best->mutex);
+                        return;
+                    }
+                }
+
+                if (dict != null)
+                {
+                    memcpy(best->dict, dict, (uint)dictSize);
+                    best->dictSize = dictSize;
+                    best->parameters = parameters;
+                    best->compressedSize = compressedSize;
+                }
+            }
+
+            if (liveJobs == 0)
+            {
+                SynchronizationWrapper.PulseAll(&best->mutex);
+            }
+
+            SynchronizationWrapper.Exit(&best->mutex);
+        }
+    }
+
+    private static COVER_dictSelection setDictSelection(byte* buf, nuint s, nuint csz)
+    {
+        COVER_dictSelection ds;
+        ds.dictContent = buf;
+        ds.dictSize = s;
+        ds.totalCompressedSize = csz;
+        return ds;
+    }
+
+    /**
+     * Error function for COVER_selectDict function. Returns a struct where
+     * return.totalCompressedSize is a ZSTD error.
+     */
+    private static COVER_dictSelection COVER_dictSelectionError(nuint error)
+    {
+        return setDictSelection(null, 0, error);
+    }
+
+    /**
+     * Error function for COVER_selectDict function. Checks if the return
+     * value is an error.
+     */
+    private static uint COVER_dictSelectionIsError(COVER_dictSelection selection)
+    {
+        return ERR_isError(selection.totalCompressedSize) || selection.dictContent == null
+            ? 1U
+            : 0U;
+    }
+
+    /**
+     * Always call after selectDict is called to free up used memory from
+     * newly created dictionary.
+     */
+    private static void COVER_dictSelectionFree(COVER_dictSelection selection)
+    {
+        free(selection.dictContent);
+    }
+
+    /**
+     * Called to finalize the dictionary and select one based on whether or not
+     * the shrink-dict flag was enabled. If enabled the dictionary used is the
+     * smallest dictionary within a specified regression of the compressed size
+     * from the largest dictionary.
+     */
+    private static COVER_dictSelection COVER_selectDict(
+        byte* customDictContent,
+        nuint dictBufferCapacity,
+        nuint dictContentSize,
+        byte* samplesBuffer,
+        nuint* samplesSizes,
+        uint nbFinalizeSamples,
+        nuint nbCheckSamples,
+        nuint nbSamples,
+        ZDICT_cover_params_t @params,
+        nuint* offsets,
+        nuint totalCompressedSize
+    )
+    {
+        nuint largestDict = 0;
+        nuint largestCompressed = 0;
+        byte* customDictContentEnd = customDictContent + dictContentSize;
+        byte* largestDictbuffer = (byte*)malloc(dictBufferCapacity);
+        byte* candidateDictBuffer = (byte*)malloc(dictBufferCapacity);
+        double regressionTolerance = (double)@params.shrinkDictMaxRegression / 100 + 1;
+        if (largestDictbuffer == null || candidateDictBuffer == null)
+        {
+            free(largestDictbuffer);
+            free(candidateDictBuffer);
+            return COVER_dictSelectionError(dictContentSize);
+        }
+
+        memcpy(largestDictbuffer, customDictContent, (uint)dictContentSize);
+        dictContentSize = ZDICT_finalizeDictionary(
+            largestDictbuffer,
+            dictBufferCapacity,
+            customDictContent,
+            dictContentSize,
+            samplesBuffer,
+            samplesSizes,
+            nbFinalizeSamples,
+            @params.zParams
+        );
+        if (ZDICT_isError(dictContentSize))
+        {
+            free(largestDictbuffer);
+            free(candidateDictBuffer);
+            return COVER_dictSelectionError(dictContentSize);
+        }
+
+        totalCompressedSize = COVER_checkTotalCompressedSize(
+            @params,
+            samplesSizes,
+            samplesBuffer,
+            offsets,
+            nbCheckSamples,
+            nbSamples,
+            largestDictbuffer,
+            dictContentSize
+        );
+        if (ERR_isError(totalCompressedSize))
+        {
+            free(largestDictbuffer);
+            free(candidateDictBuffer);
+            return COVER_dictSelectionError(totalCompressedSize);
+        }
+
+        if (@params.shrinkDict == 0)
+        {
+            free(candidateDictBuffer);
+            return setDictSelection(largestDictbuffer, dictContentSize, totalCompressedSize);
+        }
+
+        largestDict = dictContentSize;
+        largestCompressed = totalCompressedSize;
+        dictContentSize = 256;
+        while (dictContentSize < largestDict)
+        {
+            memcpy(candidateDictBuffer, largestDictbuffer, (uint)largestDict);
+            dictContentSize = ZDICT_finalizeDictionary(
+                candidateDictBuffer,
+                dictBufferCapacity,
+                customDictContentEnd - dictContentSize,
+                dictContentSize,
+                samplesBuffer,
+                samplesSizes,
+                nbFinalizeSamples,
+                @params.zParams
+            );
+            if (ZDICT_isError(dictContentSize))
+            {
+                free(largestDictbuffer);
+                free(candidateDictBuffer);
+                return COVER_dictSelectionError(dictContentSize);
+            }
+
+            totalCompressedSize = COVER_checkTotalCompressedSize(
+                @params,
+                samplesSizes,
+                samplesBuffer,
+                offsets,
+                nbCheckSamples,
+                nbSamples,
+                candidateDictBuffer,
+                dictContentSize
+            );
+            if (ERR_isError(totalCompressedSize))
+            {
+                free(largestDictbuffer);
+                free(candidateDictBuffer);
+                return COVER_dictSelectionError(totalCompressedSize);
+            }
+
+            if (totalCompressedSize <= largestCompressed * regressionTolerance)
+            {
+                free(largestDictbuffer);
+                return setDictSelection(candidateDictBuffer, dictContentSize, totalCompressedSize);
+            }
+
+            dictContentSize *= 2;
+        }
+
+        dictContentSize = largestDict;
+        totalCompressedSize = largestCompressed;
+        free(candidateDictBuffer);
+        return setDictSelection(largestDictbuffer, dictContentSize, totalCompressedSize);
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/DTableDesc.cs

@@ -0,0 +1,12 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/*-***************************/
+/*  generic DTableDesc       */
+/*-***************************/
+public struct DTableDesc
+{
+    public byte maxTableLog;
+    public byte tableType;
+    public byte tableLog;
+    public byte reserved;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/EStats_ress_t.cs

@@ -0,0 +1,13 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public unsafe struct EStats_ress_t
+{
+    /* dictionary */
+    public ZSTD_CDict_s* dict;
+
+    /* working context */
+    public ZSTD_CCtx_s* zc;
+
+    /* must be ZSTD_BLOCKSIZE_MAX allocated */
+    public void* workPlace;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/EntropyCommon.cs

@@ -0,0 +1,447 @@
+using System.Runtime.CompilerServices;
+using static SharpCompress.Compressors.ZStandard.UnsafeHelper;
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    /*===   Version   ===*/
+    private static uint FSE_versionNumber()
+    {
+        return 0 * 100 * 100 + 9 * 100 + 0;
+    }
+
+    /*===   Error Management   ===*/
+    private static bool FSE_isError(nuint code)
+    {
+        return ERR_isError(code);
+    }
+
+    private static string FSE_getErrorName(nuint code)
+    {
+        return ERR_getErrorName(code);
+    }
+
+    /* Error Management */
+    private static bool HUF_isError(nuint code)
+    {
+        return ERR_isError(code);
+    }
+
+    private static string HUF_getErrorName(nuint code)
+    {
+        return ERR_getErrorName(code);
+    }
+
+    /*-**************************************************************
+     *  FSE NCount encoding-decoding
+     ****************************************************************/
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint FSE_readNCount_body(
+        short* normalizedCounter,
+        uint* maxSVPtr,
+        uint* tableLogPtr,
+        void* headerBuffer,
+        nuint hbSize
+    )
+    {
+        byte* istart = (byte*)headerBuffer;
+        byte* iend = istart + hbSize;
+        byte* ip = istart;
+        int nbBits;
+        int remaining;
+        int threshold;
+        uint bitStream;
+        int bitCount;
+        uint charnum = 0;
+        uint maxSV1 = *maxSVPtr + 1;
+        int previous0 = 0;
+        if (hbSize < 8)
+        {
+            sbyte* buffer = stackalloc sbyte[8];
+            /* This function only works when hbSize >= 8 */
+            memset(buffer, 0, sizeof(sbyte) * 8);
+            memcpy(buffer, headerBuffer, (uint)hbSize);
+            {
+                nuint countSize = FSE_readNCount(
+                    normalizedCounter,
+                    maxSVPtr,
+                    tableLogPtr,
+                    buffer,
+                    sizeof(sbyte) * 8
+                );
+                if (FSE_isError(countSize))
+                    return countSize;
+                if (countSize > hbSize)
+                    return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+                return countSize;
+            }
+        }
+
+        assert(hbSize >= 8);
+        memset(normalizedCounter, 0, (*maxSVPtr + 1) * sizeof(short));
+        bitStream = MEM_readLE32(ip);
+        nbBits = (int)((bitStream & 0xF) + 5);
+        if (nbBits > 15)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_tableLog_tooLarge));
+        bitStream >>= 4;
+        bitCount = 4;
+        *tableLogPtr = (uint)nbBits;
+        remaining = (1 << nbBits) + 1;
+        threshold = 1 << nbBits;
+        nbBits++;
+        for (; ; )
+        {
+            if (previous0 != 0)
+            {
+                /* Count the number of repeats. Each time the
+                 * 2-bit repeat code is 0b11 there is another
+                 * repeat.
+                 * Avoid UB by setting the high bit to 1.
+                 */
+                int repeats = (int)(ZSTD_countTrailingZeros32(~bitStream | 0x80000000) >> 1);
+                while (repeats >= 12)
+                {
+                    charnum += 3 * 12;
+                    if (ip <= iend - 7)
+                    {
+                        ip += 3;
+                    }
+                    else
+                    {
+                        bitCount -= (int)(8 * (iend - 7 - ip));
+                        bitCount &= 31;
+                        ip = iend - 4;
+                    }
+
+                    bitStream = MEM_readLE32(ip) >> bitCount;
+                    repeats = (int)(ZSTD_countTrailingZeros32(~bitStream | 0x80000000) >> 1);
+                }
+
+                charnum += (uint)(3 * repeats);
+                bitStream >>= 2 * repeats;
+                bitCount += 2 * repeats;
+                assert((bitStream & 3) < 3);
+                charnum += bitStream & 3;
+                bitCount += 2;
+                if (charnum >= maxSV1)
+                    break;
+                if (ip <= iend - 7 || ip + (bitCount >> 3) <= iend - 4)
+                {
+                    assert(bitCount >> 3 <= 3);
+                    ip += bitCount >> 3;
+                    bitCount &= 7;
+                }
+                else
+                {
+                    bitCount -= (int)(8 * (iend - 4 - ip));
+                    bitCount &= 31;
+                    ip = iend - 4;
+                }
+
+                bitStream = MEM_readLE32(ip) >> bitCount;
+            }
+
+            {
+                int max = 2 * threshold - 1 - remaining;
+                int count;
+                if ((bitStream & (uint)(threshold - 1)) < (uint)max)
+                {
+                    count = (int)(bitStream & (uint)(threshold - 1));
+                    bitCount += nbBits - 1;
+                }
+                else
+                {
+                    count = (int)(bitStream & (uint)(2 * threshold - 1));
+                    if (count >= threshold)
+                        count -= max;
+                    bitCount += nbBits;
+                }
+
+                count--;
+                if (count >= 0)
+                {
+                    remaining -= count;
+                }
+                else
+                {
+                    assert(count == -1);
+                    remaining += count;
+                }
+
+                normalizedCounter[charnum++] = (short)count;
+                previous0 = count == 0 ? 1 : 0;
+                assert(threshold > 1);
+                if (remaining < threshold)
+                {
+                    if (remaining <= 1)
+                        break;
+                    nbBits = (int)(ZSTD_highbit32((uint)remaining) + 1);
+                    threshold = 1 << nbBits - 1;
+                }
+
+                if (charnum >= maxSV1)
+                    break;
+                if (ip <= iend - 7 || ip + (bitCount >> 3) <= iend - 4)
+                {
+                    ip += bitCount >> 3;
+                    bitCount &= 7;
+                }
+                else
+                {
+                    bitCount -= (int)(8 * (iend - 4 - ip));
+                    bitCount &= 31;
+                    ip = iend - 4;
+                }
+
+                bitStream = MEM_readLE32(ip) >> bitCount;
+            }
+        }
+
+        if (remaining != 1)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+        if (charnum > maxSV1)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_maxSymbolValue_tooSmall));
+        if (bitCount > 32)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+        *maxSVPtr = charnum - 1;
+        ip += bitCount + 7 >> 3;
+        return (nuint)(ip - istart);
+    }
+
+    /* Avoids the FORCE_INLINE of the _body() function. */
+    private static nuint FSE_readNCount_body_default(
+        short* normalizedCounter,
+        uint* maxSVPtr,
+        uint* tableLogPtr,
+        void* headerBuffer,
+        nuint hbSize
+    )
+    {
+        return FSE_readNCount_body(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize);
+    }
+
+    /*! FSE_readNCount_bmi2():
+     * Same as FSE_readNCount() but pass bmi2=1 when your CPU supports BMI2 and 0 otherwise.
+     */
+    private static nuint FSE_readNCount_bmi2(
+        short* normalizedCounter,
+        uint* maxSVPtr,
+        uint* tableLogPtr,
+        void* headerBuffer,
+        nuint hbSize,
+        int bmi2
+    )
+    {
+        return FSE_readNCount_body_default(
+            normalizedCounter,
+            maxSVPtr,
+            tableLogPtr,
+            headerBuffer,
+            hbSize
+        );
+    }
+
+    /*! FSE_readNCount():
+    Read compactly saved 'normalizedCounter' from 'rBuffer'.
+    @return : size read from 'rBuffer',
+    or an errorCode, which can be tested using FSE_isError().
+    maxSymbolValuePtr[0] and tableLogPtr[0] will also be updated with their respective values */
+    private static nuint FSE_readNCount(
+        short* normalizedCounter,
+        uint* maxSVPtr,
+        uint* tableLogPtr,
+        void* headerBuffer,
+        nuint hbSize
+    )
+    {
+        return FSE_readNCount_bmi2(
+            normalizedCounter,
+            maxSVPtr,
+            tableLogPtr,
+            headerBuffer,
+            hbSize,
+            0
+        );
+    }
+
+    /*! HUF_readStats() :
+    Read compact Huffman tree, saved by HUF_writeCTable().
+    `huffWeight` is destination buffer.
+    `rankStats` is assumed to be a table of at least HUF_TABLELOG_MAX U32.
+    @return : size read from `src` , or an error Code .
+    Note : Needed by HUF_readCTable() and HUF_readDTableX?() .
+     */
+    private static nuint HUF_readStats(
+        byte* huffWeight,
+        nuint hwSize,
+        uint* rankStats,
+        uint* nbSymbolsPtr,
+        uint* tableLogPtr,
+        void* src,
+        nuint srcSize
+    )
+    {
+        uint* wksp = stackalloc uint[219];
+        return HUF_readStats_wksp(
+            huffWeight,
+            hwSize,
+            rankStats,
+            nbSymbolsPtr,
+            tableLogPtr,
+            src,
+            srcSize,
+            wksp,
+            sizeof(uint) * 219,
+            0
+        );
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint HUF_readStats_body(
+        byte* huffWeight,
+        nuint hwSize,
+        uint* rankStats,
+        uint* nbSymbolsPtr,
+        uint* tableLogPtr,
+        void* src,
+        nuint srcSize,
+        void* workSpace,
+        nuint wkspSize,
+        int bmi2
+    )
+    {
+        uint weightTotal;
+        byte* ip = (byte*)src;
+        nuint iSize;
+        nuint oSize;
+        if (srcSize == 0)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_srcSize_wrong));
+        iSize = ip[0];
+        if (iSize >= 128)
+        {
+            oSize = iSize - 127;
+            iSize = (oSize + 1) / 2;
+            if (iSize + 1 > srcSize)
+                return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_srcSize_wrong));
+            if (oSize >= hwSize)
+                return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+            ip += 1;
+            {
+                uint n;
+                for (n = 0; n < oSize; n += 2)
+                {
+                    huffWeight[n] = (byte)(ip[n / 2] >> 4);
+                    huffWeight[n + 1] = (byte)(ip[n / 2] & 15);
+                }
+            }
+        }
+        else
+        {
+            if (iSize + 1 > srcSize)
+                return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_srcSize_wrong));
+            oSize = FSE_decompress_wksp_bmi2(
+                huffWeight,
+                hwSize - 1,
+                ip + 1,
+                iSize,
+                6,
+                workSpace,
+                wkspSize,
+                bmi2
+            );
+            if (FSE_isError(oSize))
+                return oSize;
+        }
+
+        memset(rankStats, 0, (12 + 1) * sizeof(uint));
+        weightTotal = 0;
+        {
+            uint n;
+            for (n = 0; n < oSize; n++)
+            {
+                if (huffWeight[n] > 12)
+                    return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+                rankStats[huffWeight[n]]++;
+                weightTotal += (uint)(1 << huffWeight[n] >> 1);
+            }
+        }
+
+        if (weightTotal == 0)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+        {
+            uint tableLog = ZSTD_highbit32(weightTotal) + 1;
+            if (tableLog > 12)
+                return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+            *tableLogPtr = tableLog;
+            {
+                uint total = (uint)(1 << (int)tableLog);
+                uint rest = total - weightTotal;
+                uint verif = (uint)(1 << (int)ZSTD_highbit32(rest));
+                uint lastWeight = ZSTD_highbit32(rest) + 1;
+                if (verif != rest)
+                    return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+                huffWeight[oSize] = (byte)lastWeight;
+                rankStats[lastWeight]++;
+            }
+        }
+
+        if (rankStats[1] < 2 || (rankStats[1] & 1) != 0)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+        *nbSymbolsPtr = (uint)(oSize + 1);
+        return iSize + 1;
+    }
+
+    /* Avoids the FORCE_INLINE of the _body() function. */
+    private static nuint HUF_readStats_body_default(
+        byte* huffWeight,
+        nuint hwSize,
+        uint* rankStats,
+        uint* nbSymbolsPtr,
+        uint* tableLogPtr,
+        void* src,
+        nuint srcSize,
+        void* workSpace,
+        nuint wkspSize
+    )
+    {
+        return HUF_readStats_body(
+            huffWeight,
+            hwSize,
+            rankStats,
+            nbSymbolsPtr,
+            tableLogPtr,
+            src,
+            srcSize,
+            workSpace,
+            wkspSize,
+            0
+        );
+    }
+
+    private static nuint HUF_readStats_wksp(
+        byte* huffWeight,
+        nuint hwSize,
+        uint* rankStats,
+        uint* nbSymbolsPtr,
+        uint* tableLogPtr,
+        void* src,
+        nuint srcSize,
+        void* workSpace,
+        nuint wkspSize,
+        int flags
+    )
+    {
+        return HUF_readStats_body_default(
+            huffWeight,
+            hwSize,
+            rankStats,
+            nbSymbolsPtr,
+            tableLogPtr,
+            src,
+            srcSize,
+            workSpace,
+            wkspSize
+        );
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/ErrorPrivate.cs

@@ -0,0 +1,110 @@
+using System.Runtime.CompilerServices;
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static bool ERR_isError(nuint code)
+    {
+        return code > unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_maxCode));
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static ZSTD_ErrorCode ERR_getErrorCode(nuint code)
+    {
+        if (!ERR_isError(code))
+            return 0;
+        return (ZSTD_ErrorCode)(0 - code);
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static string ERR_getErrorName(nuint code)
+    {
+        return ERR_getErrorString(ERR_getErrorCode(code));
+    }
+
+    /*-****************************************
+     *  Error Strings
+     ******************************************/
+    private static string ERR_getErrorString(ZSTD_ErrorCode code)
+    {
+        const string notErrorCode = "Unspecified error code";
+        switch (code)
+        {
+            case ZSTD_ErrorCode.ZSTD_error_no_error:
+                return "No error detected";
+            case ZSTD_ErrorCode.ZSTD_error_GENERIC:
+                return "Error (generic)";
+            case ZSTD_ErrorCode.ZSTD_error_prefix_unknown:
+                return "Unknown frame descriptor";
+            case ZSTD_ErrorCode.ZSTD_error_version_unsupported:
+                return "Version not supported";
+            case ZSTD_ErrorCode.ZSTD_error_frameParameter_unsupported:
+                return "Unsupported frame parameter";
+            case ZSTD_ErrorCode.ZSTD_error_frameParameter_windowTooLarge:
+                return "Frame requires too much memory for decoding";
+            case ZSTD_ErrorCode.ZSTD_error_corruption_detected:
+                return "Data corruption detected";
+            case ZSTD_ErrorCode.ZSTD_error_checksum_wrong:
+                return "Restored data doesn't match checksum";
+            case ZSTD_ErrorCode.ZSTD_error_literals_headerWrong:
+                return "Header of Literals' block doesn't respect format specification";
+            case ZSTD_ErrorCode.ZSTD_error_parameter_unsupported:
+                return "Unsupported parameter";
+            case ZSTD_ErrorCode.ZSTD_error_parameter_combination_unsupported:
+                return "Unsupported combination of parameters";
+            case ZSTD_ErrorCode.ZSTD_error_parameter_outOfBound:
+                return "Parameter is out of bound";
+            case ZSTD_ErrorCode.ZSTD_error_init_missing:
+                return "Context should be init first";
+            case ZSTD_ErrorCode.ZSTD_error_memory_allocation:
+                return "Allocation error : not enough memory";
+            case ZSTD_ErrorCode.ZSTD_error_workSpace_tooSmall:
+                return "workSpace buffer is not large enough";
+            case ZSTD_ErrorCode.ZSTD_error_stage_wrong:
+                return "Operation not authorized at current processing stage";
+            case ZSTD_ErrorCode.ZSTD_error_tableLog_tooLarge:
+                return "tableLog requires too much memory : unsupported";
+            case ZSTD_ErrorCode.ZSTD_error_maxSymbolValue_tooLarge:
+                return "Unsupported max Symbol Value : too large";
+            case ZSTD_ErrorCode.ZSTD_error_maxSymbolValue_tooSmall:
+                return "Specified maxSymbolValue is too small";
+            case ZSTD_ErrorCode.ZSTD_error_cannotProduce_uncompressedBlock:
+                return "This mode cannot generate an uncompressed block";
+            case ZSTD_ErrorCode.ZSTD_error_stabilityCondition_notRespected:
+                return "pledged buffer stability condition is not respected";
+            case ZSTD_ErrorCode.ZSTD_error_dictionary_corrupted:
+                return "Dictionary is corrupted";
+            case ZSTD_ErrorCode.ZSTD_error_dictionary_wrong:
+                return "Dictionary mismatch";
+            case ZSTD_ErrorCode.ZSTD_error_dictionaryCreation_failed:
+                return "Cannot create Dictionary from provided samples";
+            case ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall:
+                return "Destination buffer is too small";
+            case ZSTD_ErrorCode.ZSTD_error_srcSize_wrong:
+                return "Src size is incorrect";
+            case ZSTD_ErrorCode.ZSTD_error_dstBuffer_null:
+                return "Operation on NULL destination buffer";
+            case ZSTD_ErrorCode.ZSTD_error_noForwardProgress_destFull:
+                return "Operation made no progress over multiple calls, due to output buffer being full";
+            case ZSTD_ErrorCode.ZSTD_error_noForwardProgress_inputEmpty:
+                return "Operation made no progress over multiple calls, due to input being empty";
+            case ZSTD_ErrorCode.ZSTD_error_frameIndex_tooLarge:
+                return "Frame index is too large";
+            case ZSTD_ErrorCode.ZSTD_error_seekableIO:
+                return "An I/O error occurred when reading/seeking";
+            case ZSTD_ErrorCode.ZSTD_error_dstBuffer_wrong:
+                return "Destination buffer is wrong";
+            case ZSTD_ErrorCode.ZSTD_error_srcBuffer_wrong:
+                return "Source buffer is wrong";
+            case ZSTD_ErrorCode.ZSTD_error_sequenceProducer_failed:
+                return "Block-level external sequence producer returned an error code";
+            case ZSTD_ErrorCode.ZSTD_error_externalSequences_invalid:
+                return "External sequences are not valid";
+            case ZSTD_ErrorCode.ZSTD_error_maxCode:
+            default:
+                return notErrorCode;
+        }
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/EstimatedBlockSize.cs

@@ -0,0 +1,7 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public struct EstimatedBlockSize
+{
+    public nuint estLitSize;
+    public nuint estBlockSize;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FASTCOVER_accel_t.cs

@@ -0,0 +1,19 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/*-*************************************
+ * Acceleration
+ ***************************************/
+public struct FASTCOVER_accel_t
+{
+    /* Percentage of training samples used for ZDICT_finalizeDictionary */
+    public uint finalize;
+
+    /* Number of dmer skipped between each dmer counted in computeFrequency */
+    public uint skip;
+
+    public FASTCOVER_accel_t(uint finalize, uint skip)
+    {
+        this.finalize = finalize;
+        this.skip = skip;
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FASTCOVER_ctx_t.cs

@@ -0,0 +1,19 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/*-*************************************
+ * Context
+ ***************************************/
+public unsafe struct FASTCOVER_ctx_t
+{
+    public byte* samples;
+    public nuint* offsets;
+    public nuint* samplesSizes;
+    public nuint nbSamples;
+    public nuint nbTrainSamples;
+    public nuint nbTestSamples;
+    public nuint nbDmers;
+    public uint* freqs;
+    public uint d;
+    public uint f;
+    public FASTCOVER_accel_t accelParams;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FASTCOVER_tryParameters_data_s.cs

@@ -0,0 +1,12 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/**
+ * Parameters for FASTCOVER_tryParameters().
+ */
+public unsafe struct FASTCOVER_tryParameters_data_s
+{
+    public FASTCOVER_ctx_t* ctx;
+    public COVER_best_s* best;
+    public nuint dictBufferCapacity;
+    public ZDICT_cover_params_t parameters;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FPStats.cs

@@ -0,0 +1,7 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public struct FPStats
+{
+    public Fingerprint pastEvents;
+    public Fingerprint newEvents;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FSE_CState_t.cs

@@ -0,0 +1,16 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/* *****************************************
+ *  FSE symbol compression API
+ *******************************************/
+/*!
+This API consists of small unitary functions, which highly benefit from being inlined.
+Hence their body are included in next section.
+ */
+public unsafe struct FSE_CState_t
+{
+    public nint value;
+    public void* stateTable;
+    public void* symbolTT;
+    public uint stateLog;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FSE_DState_t.cs

@@ -0,0 +1,12 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/* *****************************************
+ *  FSE symbol decompression API
+ *******************************************/
+public unsafe struct FSE_DState_t
+{
+    public nuint state;
+
+    /* precise table may vary, depending on U16 */
+    public void* table;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FSE_DTableHeader.cs

@@ -0,0 +1,8 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/* ======    Decompression    ====== */
+public struct FSE_DTableHeader
+{
+    public ushort tableLog;
+    public ushort fastMode;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FSE_DecompressWksp.cs

@@ -0,0 +1,6 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public unsafe struct FSE_DecompressWksp
+{
+    public fixed short ncount[256];
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FSE_decode_t.cs

@@ -0,0 +1,8 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public struct FSE_decode_t
+{
+    public ushort newState;
+    public byte symbol;
+    public byte nbBits;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FSE_repeat.cs

@@ -0,0 +1,13 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public enum FSE_repeat
+{
+    /**< Cannot use the previous table */
+    FSE_repeat_none,
+
+    /**< Can use the previous table but it must be checked */
+    FSE_repeat_check,
+
+    /**< Can use the previous table and it is assumed to be valid */
+    FSE_repeat_valid,
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FSE_symbolCompressionTransform.cs

@@ -0,0 +1,10 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+/* *****************************************
+ *  Implementation of inlined functions
+ *******************************************/
+public struct FSE_symbolCompressionTransform
+{
+    public int deltaFindState;
+    public uint deltaNbBits;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/Fastcover.cs

@@ -0,0 +1,761 @@
+using static SharpCompress.Compressors.ZStandard.UnsafeHelper;
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    /*-*************************************
+     * Hash Functions
+     ***************************************/
+    /**
+     * Hash the d-byte value pointed to by p and mod 2^f into the frequency vector
+     */
+    private static nuint FASTCOVER_hashPtrToIndex(void* p, uint f, uint d)
+    {
+        if (d == 6)
+        {
+            return ZSTD_hash6Ptr(p, f);
+        }
+
+        return ZSTD_hash8Ptr(p, f);
+    }
+
+    private static readonly FASTCOVER_accel_t* FASTCOVER_defaultAccelParameters = GetArrayPointer(
+        new FASTCOVER_accel_t[11]
+        {
+            new FASTCOVER_accel_t(finalize: 100, skip: 0),
+            new FASTCOVER_accel_t(finalize: 100, skip: 0),
+            new FASTCOVER_accel_t(finalize: 50, skip: 1),
+            new FASTCOVER_accel_t(finalize: 34, skip: 2),
+            new FASTCOVER_accel_t(finalize: 25, skip: 3),
+            new FASTCOVER_accel_t(finalize: 20, skip: 4),
+            new FASTCOVER_accel_t(finalize: 17, skip: 5),
+            new FASTCOVER_accel_t(finalize: 14, skip: 6),
+            new FASTCOVER_accel_t(finalize: 13, skip: 7),
+            new FASTCOVER_accel_t(finalize: 11, skip: 8),
+            new FASTCOVER_accel_t(finalize: 10, skip: 9),
+        }
+    );
+
+    /*-*************************************
+     *  Helper functions
+     ***************************************/
+    /**
+     * Selects the best segment in an epoch.
+     * Segments of are scored according to the function:
+     *
+     * Let F(d) be the frequency of all dmers with hash value d.
+     * Let S_i be hash value of the dmer at position i of segment S which has length k.
+     *
+     *     Score(S) = F(S_1) + F(S_2) + ... + F(S_{k-d+1})
+     *
+     * Once the dmer with hash value d is in the dictionary we set F(d) = 0.
+     */
+    private static COVER_segment_t FASTCOVER_selectSegment(
+        FASTCOVER_ctx_t* ctx,
+        uint* freqs,
+        uint begin,
+        uint end,
+        ZDICT_cover_params_t parameters,
+        ushort* segmentFreqs
+    )
+    {
+        /* Constants */
+        uint k = parameters.k;
+        uint d = parameters.d;
+        uint f = ctx->f;
+        uint dmersInK = k - d + 1;
+        /* Try each segment (activeSegment) and save the best (bestSegment) */
+        COVER_segment_t bestSegment = new COVER_segment_t
+        {
+            begin = 0,
+            end = 0,
+            score = 0,
+        };
+        COVER_segment_t activeSegment;
+        activeSegment.begin = begin;
+        activeSegment.end = begin;
+        activeSegment.score = 0;
+        while (activeSegment.end < end)
+        {
+            /* Get hash value of current dmer */
+            nuint idx = FASTCOVER_hashPtrToIndex(ctx->samples + activeSegment.end, f, d);
+            if (segmentFreqs[idx] == 0)
+            {
+                activeSegment.score += freqs[idx];
+            }
+
+            activeSegment.end += 1;
+            segmentFreqs[idx] += 1;
+            if (activeSegment.end - activeSegment.begin == dmersInK + 1)
+            {
+                /* Get hash value of the dmer to be eliminated from active segment */
+                nuint delIndex = FASTCOVER_hashPtrToIndex(ctx->samples + activeSegment.begin, f, d);
+                segmentFreqs[delIndex] -= 1;
+                if (segmentFreqs[delIndex] == 0)
+                {
+                    activeSegment.score -= freqs[delIndex];
+                }
+
+                activeSegment.begin += 1;
+            }
+
+            if (activeSegment.score > bestSegment.score)
+            {
+                bestSegment = activeSegment;
+            }
+        }
+
+        while (activeSegment.begin < end)
+        {
+            nuint delIndex = FASTCOVER_hashPtrToIndex(ctx->samples + activeSegment.begin, f, d);
+            segmentFreqs[delIndex] -= 1;
+            activeSegment.begin += 1;
+        }
+
+        {
+            /*  Zero the frequency of hash value of each dmer covered by the chosen segment. */
+            uint pos;
+            for (pos = bestSegment.begin; pos != bestSegment.end; ++pos)
+            {
+                nuint i = FASTCOVER_hashPtrToIndex(ctx->samples + pos, f, d);
+                freqs[i] = 0;
+            }
+        }
+
+        return bestSegment;
+    }
+
+    private static int FASTCOVER_checkParameters(
+        ZDICT_cover_params_t parameters,
+        nuint maxDictSize,
+        uint f,
+        uint accel
+    )
+    {
+        if (parameters.d == 0 || parameters.k == 0)
+        {
+            return 0;
+        }
+
+        if (parameters.d != 6 && parameters.d != 8)
+        {
+            return 0;
+        }
+
+        if (parameters.k > maxDictSize)
+        {
+            return 0;
+        }
+
+        if (parameters.d > parameters.k)
+        {
+            return 0;
+        }
+
+        if (f > 31 || f == 0)
+        {
+            return 0;
+        }
+
+        if (parameters.splitPoint <= 0 || parameters.splitPoint > 1)
+        {
+            return 0;
+        }
+
+        if (accel > 10 || accel == 0)
+        {
+            return 0;
+        }
+
+        return 1;
+    }
+
+    /**
+     * Clean up a context initialized with `FASTCOVER_ctx_init()`.
+     */
+    private static void FASTCOVER_ctx_destroy(FASTCOVER_ctx_t* ctx)
+    {
+        if (ctx == null)
+            return;
+        free(ctx->freqs);
+        ctx->freqs = null;
+        free(ctx->offsets);
+        ctx->offsets = null;
+    }
+
+    /**
+     * Calculate for frequency of hash value of each dmer in ctx->samples
+     */
+    private static void FASTCOVER_computeFrequency(uint* freqs, FASTCOVER_ctx_t* ctx)
+    {
+        uint f = ctx->f;
+        uint d = ctx->d;
+        uint skip = ctx->accelParams.skip;
+        uint readLength = d > 8 ? d : 8;
+        nuint i;
+        assert(ctx->nbTrainSamples >= 5);
+        assert(ctx->nbTrainSamples <= ctx->nbSamples);
+        for (i = 0; i < ctx->nbTrainSamples; i++)
+        {
+            /* start of current dmer */
+            nuint start = ctx->offsets[i];
+            nuint currSampleEnd = ctx->offsets[i + 1];
+            while (start + readLength <= currSampleEnd)
+            {
+                nuint dmerIndex = FASTCOVER_hashPtrToIndex(ctx->samples + start, f, d);
+                freqs[dmerIndex]++;
+                start = start + skip + 1;
+            }
+        }
+    }
+
+    /**
+     * Prepare a context for dictionary building.
+     * The context is only dependent on the parameter `d` and can be used multiple
+     * times.
+     * Returns 0 on success or error code on error.
+     * The context must be destroyed with `FASTCOVER_ctx_destroy()`.
+     */
+    private static nuint FASTCOVER_ctx_init(
+        FASTCOVER_ctx_t* ctx,
+        void* samplesBuffer,
+        nuint* samplesSizes,
+        uint nbSamples,
+        uint d,
+        double splitPoint,
+        uint f,
+        FASTCOVER_accel_t accelParams
+    )
+    {
+        byte* samples = (byte*)samplesBuffer;
+        nuint totalSamplesSize = COVER_sum(samplesSizes, nbSamples);
+        /* Split samples into testing and training sets */
+        uint nbTrainSamples = splitPoint < 1 ? (uint)(nbSamples * splitPoint) : nbSamples;
+        uint nbTestSamples = splitPoint < 1 ? nbSamples - nbTrainSamples : nbSamples;
+        nuint trainingSamplesSize =
+            splitPoint < 1 ? COVER_sum(samplesSizes, nbTrainSamples) : totalSamplesSize;
+        nuint testSamplesSize =
+            splitPoint < 1
+                ? COVER_sum(samplesSizes + nbTrainSamples, nbTestSamples)
+                : totalSamplesSize;
+        if (
+            totalSamplesSize < (d > sizeof(ulong) ? d : sizeof(ulong))
+            || totalSamplesSize >= (sizeof(nuint) == 8 ? unchecked((uint)-1) : 1 * (1U << 30))
+        )
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_srcSize_wrong));
+        }
+
+        if (nbTrainSamples < 5)
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_srcSize_wrong));
+        }
+
+        if (nbTestSamples < 1)
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_srcSize_wrong));
+        }
+
+        *ctx = new FASTCOVER_ctx_t
+        {
+            samples = samples,
+            samplesSizes = samplesSizes,
+            nbSamples = nbSamples,
+            nbTrainSamples = nbTrainSamples,
+            nbTestSamples = nbTestSamples,
+            nbDmers = trainingSamplesSize - (d > sizeof(ulong) ? d : sizeof(ulong)) + 1,
+            d = d,
+            f = f,
+            accelParams = accelParams,
+            offsets = (nuint*)calloc(nbSamples + 1, (ulong)sizeof(nuint)),
+        };
+        if (ctx->offsets == null)
+        {
+            FASTCOVER_ctx_destroy(ctx);
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_memory_allocation));
+        }
+
+        {
+            uint i;
+            ctx->offsets[0] = 0;
+            assert(nbSamples >= 5);
+            for (i = 1; i <= nbSamples; ++i)
+            {
+                ctx->offsets[i] = ctx->offsets[i - 1] + samplesSizes[i - 1];
+            }
+        }
+
+        ctx->freqs = (uint*)calloc((ulong)1 << (int)f, sizeof(uint));
+        if (ctx->freqs == null)
+        {
+            FASTCOVER_ctx_destroy(ctx);
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_memory_allocation));
+        }
+
+        FASTCOVER_computeFrequency(ctx->freqs, ctx);
+        return 0;
+    }
+
+    /**
+     * Given the prepared context build the dictionary.
+     */
+    private static nuint FASTCOVER_buildDictionary(
+        FASTCOVER_ctx_t* ctx,
+        uint* freqs,
+        void* dictBuffer,
+        nuint dictBufferCapacity,
+        ZDICT_cover_params_t parameters,
+        ushort* segmentFreqs
+    )
+    {
+        byte* dict = (byte*)dictBuffer;
+        nuint tail = dictBufferCapacity;
+        /* Divide the data into epochs. We will select one segment from each epoch. */
+        COVER_epoch_info_t epochs = COVER_computeEpochs(
+            (uint)dictBufferCapacity,
+            (uint)ctx->nbDmers,
+            parameters.k,
+            1
+        );
+        const nuint maxZeroScoreRun = 10;
+        nuint zeroScoreRun = 0;
+        nuint epoch;
+        for (epoch = 0; tail > 0; epoch = (epoch + 1) % epochs.num)
+        {
+            uint epochBegin = (uint)(epoch * epochs.size);
+            uint epochEnd = epochBegin + epochs.size;
+            nuint segmentSize;
+            /* Select a segment */
+            COVER_segment_t segment = FASTCOVER_selectSegment(
+                ctx,
+                freqs,
+                epochBegin,
+                epochEnd,
+                parameters,
+                segmentFreqs
+            );
+            if (segment.score == 0)
+            {
+                if (++zeroScoreRun >= maxZeroScoreRun)
+                {
+                    break;
+                }
+
+                continue;
+            }
+
+            zeroScoreRun = 0;
+            segmentSize =
+                segment.end - segment.begin + parameters.d - 1 < tail
+                    ? segment.end - segment.begin + parameters.d - 1
+                    : tail;
+            if (segmentSize < parameters.d)
+            {
+                break;
+            }
+
+            tail -= segmentSize;
+            memcpy(dict + tail, ctx->samples + segment.begin, (uint)segmentSize);
+        }
+
+        return tail;
+    }
+
+    /**
+     * Tries a set of parameters and updates the COVER_best_t with the results.
+     * This function is thread safe if zstd is compiled with multithreaded support.
+     * It takes its parameters as an *OWNING* opaque pointer to support threading.
+     */
+    private static void FASTCOVER_tryParameters(void* opaque)
+    {
+        /* Save parameters as local variables */
+        FASTCOVER_tryParameters_data_s* data = (FASTCOVER_tryParameters_data_s*)opaque;
+        FASTCOVER_ctx_t* ctx = data->ctx;
+        ZDICT_cover_params_t parameters = data->parameters;
+        nuint dictBufferCapacity = data->dictBufferCapacity;
+        nuint totalCompressedSize = unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+        /* Initialize array to keep track of frequency of dmer within activeSegment */
+        ushort* segmentFreqs = (ushort*)calloc((ulong)1 << (int)ctx->f, sizeof(ushort));
+        /* Allocate space for hash table, dict, and freqs */
+        byte* dict = (byte*)malloc(dictBufferCapacity);
+        COVER_dictSelection selection = COVER_dictSelectionError(
+            unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC))
+        );
+        uint* freqs = (uint*)malloc(((ulong)1 << (int)ctx->f) * sizeof(uint));
+        if (segmentFreqs == null || dict == null || freqs == null)
+        {
+            goto _cleanup;
+        }
+
+        memcpy(freqs, ctx->freqs, (uint)(((ulong)1 << (int)ctx->f) * sizeof(uint)));
+        {
+            nuint tail = FASTCOVER_buildDictionary(
+                ctx,
+                freqs,
+                dict,
+                dictBufferCapacity,
+                parameters,
+                segmentFreqs
+            );
+            uint nbFinalizeSamples = (uint)(ctx->nbTrainSamples * ctx->accelParams.finalize / 100);
+            selection = COVER_selectDict(
+                dict + tail,
+                dictBufferCapacity,
+                dictBufferCapacity - tail,
+                ctx->samples,
+                ctx->samplesSizes,
+                nbFinalizeSamples,
+                ctx->nbTrainSamples,
+                ctx->nbSamples,
+                parameters,
+                ctx->offsets,
+                totalCompressedSize
+            );
+            if (COVER_dictSelectionIsError(selection) != 0)
+            {
+                goto _cleanup;
+            }
+        }
+
+        _cleanup:
+        free(dict);
+        COVER_best_finish(data->best, parameters, selection);
+        free(data);
+        free(segmentFreqs);
+        COVER_dictSelectionFree(selection);
+        free(freqs);
+    }
+
+    private static void FASTCOVER_convertToCoverParams(
+        ZDICT_fastCover_params_t fastCoverParams,
+        ZDICT_cover_params_t* coverParams
+    )
+    {
+        coverParams->k = fastCoverParams.k;
+        coverParams->d = fastCoverParams.d;
+        coverParams->steps = fastCoverParams.steps;
+        coverParams->nbThreads = fastCoverParams.nbThreads;
+        coverParams->splitPoint = fastCoverParams.splitPoint;
+        coverParams->zParams = fastCoverParams.zParams;
+        coverParams->shrinkDict = fastCoverParams.shrinkDict;
+    }
+
+    private static void FASTCOVER_convertToFastCoverParams(
+        ZDICT_cover_params_t coverParams,
+        ZDICT_fastCover_params_t* fastCoverParams,
+        uint f,
+        uint accel
+    )
+    {
+        fastCoverParams->k = coverParams.k;
+        fastCoverParams->d = coverParams.d;
+        fastCoverParams->steps = coverParams.steps;
+        fastCoverParams->nbThreads = coverParams.nbThreads;
+        fastCoverParams->splitPoint = coverParams.splitPoint;
+        fastCoverParams->f = f;
+        fastCoverParams->accel = accel;
+        fastCoverParams->zParams = coverParams.zParams;
+        fastCoverParams->shrinkDict = coverParams.shrinkDict;
+    }
+
+    /*! ZDICT_trainFromBuffer_fastCover():
+     *  Train a dictionary from an array of samples using a modified version of COVER algorithm.
+     *  Samples must be stored concatenated in a single flat buffer `samplesBuffer`,
+     *  supplied with an array of sizes `samplesSizes`, providing the size of each sample, in order.
+     *  d and k are required.
+     *  All other parameters are optional, will use default values if not provided
+     *  The resulting dictionary will be saved into `dictBuffer`.
+     * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
+     *          or an error code, which can be tested with ZDICT_isError().
+     *          See ZDICT_trainFromBuffer() for details on failure modes.
+     *  Note: ZDICT_trainFromBuffer_fastCover() requires 6 * 2^f bytes of memory.
+     *  Tips: In general, a reasonable dictionary has a size of ~ 100 KB.
+     *        It's possible to select smaller or larger size, just by specifying `dictBufferCapacity`.
+     *        In general, it's recommended to provide a few thousands samples, though this can vary a lot.
+     *        It's recommended that total size of all samples be about ~x100 times the target size of dictionary.
+     */
+    public static nuint ZDICT_trainFromBuffer_fastCover(
+        void* dictBuffer,
+        nuint dictBufferCapacity,
+        void* samplesBuffer,
+        nuint* samplesSizes,
+        uint nbSamples,
+        ZDICT_fastCover_params_t parameters
+    )
+    {
+        byte* dict = (byte*)dictBuffer;
+        FASTCOVER_ctx_t ctx;
+        ZDICT_cover_params_t coverParams;
+        FASTCOVER_accel_t accelParams;
+        g_displayLevel = (int)parameters.zParams.notificationLevel;
+        parameters.splitPoint = 1;
+        parameters.f = parameters.f == 0 ? 20 : parameters.f;
+        parameters.accel = parameters.accel == 0 ? 1 : parameters.accel;
+        coverParams = new ZDICT_cover_params_t();
+        FASTCOVER_convertToCoverParams(parameters, &coverParams);
+        if (
+            FASTCOVER_checkParameters(
+                coverParams,
+                dictBufferCapacity,
+                parameters.f,
+                parameters.accel
+            ) == 0
+        )
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_parameter_outOfBound));
+        }
+
+        if (nbSamples == 0)
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_srcSize_wrong));
+        }
+
+        if (dictBufferCapacity < 256)
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall));
+        }
+
+        accelParams = FASTCOVER_defaultAccelParameters[parameters.accel];
+        {
+            nuint initVal = FASTCOVER_ctx_init(
+                &ctx,
+                samplesBuffer,
+                samplesSizes,
+                nbSamples,
+                coverParams.d,
+                parameters.splitPoint,
+                parameters.f,
+                accelParams
+            );
+            if (ERR_isError(initVal))
+            {
+                return initVal;
+            }
+        }
+
+        COVER_warnOnSmallCorpus(dictBufferCapacity, ctx.nbDmers, g_displayLevel);
+        {
+            /* Initialize array to keep track of frequency of dmer within activeSegment */
+            ushort* segmentFreqs = (ushort*)calloc((ulong)1 << (int)parameters.f, sizeof(ushort));
+            nuint tail = FASTCOVER_buildDictionary(
+                &ctx,
+                ctx.freqs,
+                dictBuffer,
+                dictBufferCapacity,
+                coverParams,
+                segmentFreqs
+            );
+            uint nbFinalizeSamples = (uint)(ctx.nbTrainSamples * ctx.accelParams.finalize / 100);
+            nuint dictionarySize = ZDICT_finalizeDictionary(
+                dict,
+                dictBufferCapacity,
+                dict + tail,
+                dictBufferCapacity - tail,
+                samplesBuffer,
+                samplesSizes,
+                nbFinalizeSamples,
+                coverParams.zParams
+            );
+            if (!ERR_isError(dictionarySize)) { }
+
+            FASTCOVER_ctx_destroy(&ctx);
+            free(segmentFreqs);
+            return dictionarySize;
+        }
+    }
+
+    /*! ZDICT_optimizeTrainFromBuffer_fastCover():
+     * The same requirements as above hold for all the parameters except `parameters`.
+     * This function tries many parameter combinations (specifically, k and d combinations)
+     * and picks the best parameters. `*parameters` is filled with the best parameters found,
+     * dictionary constructed with those parameters is stored in `dictBuffer`.
+     * All of the parameters d, k, steps, f, and accel are optional.
+     * If d is non-zero then we don't check multiple values of d, otherwise we check d = {6, 8}.
+     * if steps is zero it defaults to its default value.
+     * If k is non-zero then we don't check multiple values of k, otherwise we check steps values in [50, 2000].
+     * If f is zero, default value of 20 is used.
+     * If accel is zero, default value of 1 is used.
+     *
+     * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
+     *          or an error code, which can be tested with ZDICT_isError().
+     *          On success `*parameters` contains the parameters selected.
+     *          See ZDICT_trainFromBuffer() for details on failure modes.
+     * Note: ZDICT_optimizeTrainFromBuffer_fastCover() requires about 6 * 2^f bytes of memory for each thread.
+     */
+    public static nuint ZDICT_optimizeTrainFromBuffer_fastCover(
+        void* dictBuffer,
+        nuint dictBufferCapacity,
+        void* samplesBuffer,
+        nuint* samplesSizes,
+        uint nbSamples,
+        ZDICT_fastCover_params_t* parameters
+    )
+    {
+        ZDICT_cover_params_t coverParams;
+        FASTCOVER_accel_t accelParams;
+        /* constants */
+        uint nbThreads = parameters->nbThreads;
+        double splitPoint = parameters->splitPoint <= 0 ? 0.75 : parameters->splitPoint;
+        uint kMinD = parameters->d == 0 ? 6 : parameters->d;
+        uint kMaxD = parameters->d == 0 ? 8 : parameters->d;
+        uint kMinK = parameters->k == 0 ? 50 : parameters->k;
+        uint kMaxK = parameters->k == 0 ? 2000 : parameters->k;
+        uint kSteps = parameters->steps == 0 ? 40 : parameters->steps;
+        uint kStepSize = (kMaxK - kMinK) / kSteps > 1 ? (kMaxK - kMinK) / kSteps : 1;
+        uint kIterations = (1 + (kMaxD - kMinD) / 2) * (1 + (kMaxK - kMinK) / kStepSize);
+        uint f = parameters->f == 0 ? 20 : parameters->f;
+        uint accel = parameters->accel == 0 ? 1 : parameters->accel;
+        const uint shrinkDict = 0;
+        /* Local variables */
+        int displayLevel = (int)parameters->zParams.notificationLevel;
+        uint iteration = 1;
+        uint d;
+        uint k;
+        COVER_best_s best;
+        void* pool = null;
+        int warned = 0;
+        if (splitPoint <= 0 || splitPoint > 1)
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_parameter_outOfBound));
+        }
+
+        if (accel == 0 || accel > 10)
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_parameter_outOfBound));
+        }
+
+        if (kMinK < kMaxD || kMaxK < kMinK)
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_parameter_outOfBound));
+        }
+
+        if (nbSamples == 0)
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_srcSize_wrong));
+        }
+
+        if (dictBufferCapacity < 256)
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall));
+        }
+
+        if (nbThreads > 1)
+        {
+            pool = POOL_create(nbThreads, 1);
+            if (pool == null)
+            {
+                return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_memory_allocation));
+            }
+        }
+
+        COVER_best_init(&best);
+        coverParams = new ZDICT_cover_params_t();
+        FASTCOVER_convertToCoverParams(*parameters, &coverParams);
+        accelParams = FASTCOVER_defaultAccelParameters[accel];
+        g_displayLevel = displayLevel == 0 ? 0 : displayLevel - 1;
+        for (d = kMinD; d <= kMaxD; d += 2)
+        {
+            /* Initialize the context for this value of d */
+            FASTCOVER_ctx_t ctx;
+            {
+                nuint initVal = FASTCOVER_ctx_init(
+                    &ctx,
+                    samplesBuffer,
+                    samplesSizes,
+                    nbSamples,
+                    d,
+                    splitPoint,
+                    f,
+                    accelParams
+                );
+                if (ERR_isError(initVal))
+                {
+                    COVER_best_destroy(&best);
+                    POOL_free(pool);
+                    return initVal;
+                }
+            }
+
+            if (warned == 0)
+            {
+                COVER_warnOnSmallCorpus(dictBufferCapacity, ctx.nbDmers, displayLevel);
+                warned = 1;
+            }
+
+            for (k = kMinK; k <= kMaxK; k += kStepSize)
+            {
+                /* Prepare the arguments */
+                FASTCOVER_tryParameters_data_s* data = (FASTCOVER_tryParameters_data_s*)malloc(
+                    (ulong)sizeof(FASTCOVER_tryParameters_data_s)
+                );
+                if (data == null)
+                {
+                    COVER_best_destroy(&best);
+                    FASTCOVER_ctx_destroy(&ctx);
+                    POOL_free(pool);
+                    return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_memory_allocation));
+                }
+
+                data->ctx = &ctx;
+                data->best = &best;
+                data->dictBufferCapacity = dictBufferCapacity;
+                data->parameters = coverParams;
+                data->parameters.k = k;
+                data->parameters.d = d;
+                data->parameters.splitPoint = splitPoint;
+                data->parameters.steps = kSteps;
+                data->parameters.shrinkDict = shrinkDict;
+                data->parameters.zParams.notificationLevel = (uint)g_displayLevel;
+                if (
+                    FASTCOVER_checkParameters(
+                        data->parameters,
+                        dictBufferCapacity,
+                        data->ctx->f,
+                        accel
+                    ) == 0
+                )
+                {
+                    free(data);
+                    continue;
+                }
+
+                COVER_best_start(&best);
+                if (pool != null)
+                {
+                    POOL_add(
+                        pool,
+                        (delegate* managed<void*, void>)(&FASTCOVER_tryParameters),
+                        data
+                    );
+                }
+                else
+                {
+                    FASTCOVER_tryParameters(data);
+                }
+
+                ++iteration;
+            }
+
+            COVER_best_wait(&best);
+            FASTCOVER_ctx_destroy(&ctx);
+        }
+
+        {
+            nuint dictSize = best.dictSize;
+            if (ERR_isError(best.compressedSize))
+            {
+                nuint compressedSize = best.compressedSize;
+                COVER_best_destroy(&best);
+                POOL_free(pool);
+                return compressedSize;
+            }
+
+            FASTCOVER_convertToFastCoverParams(best.parameters, parameters, f, accel);
+            memcpy(dictBuffer, best.dict, (uint)dictSize);
+            COVER_best_destroy(&best);
+            POOL_free(pool);
+            return dictSize;
+        }
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/Fingerprint.cs

@@ -0,0 +1,7 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public unsafe struct Fingerprint
+{
+    public fixed uint events[1024];
+    public nuint nbEvents;
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/Fse.cs

@@ -0,0 +1,198 @@
+using System.Runtime.CompilerServices;
+using static SharpCompress.Compressors.ZStandard.UnsafeHelper;
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void FSE_initCState(FSE_CState_t* statePtr, uint* ct)
+    {
+        void* ptr = ct;
+        ushort* u16ptr = (ushort*)ptr;
+        uint tableLog = MEM_read16(ptr);
+        statePtr->value = (nint)1 << (int)tableLog;
+        statePtr->stateTable = u16ptr + 2;
+        statePtr->symbolTT = ct + 1 + (tableLog != 0 ? 1 << (int)(tableLog - 1) : 1);
+        statePtr->stateLog = tableLog;
+    }
+
+    /*! FSE_initCState2() :
+     *   Same as FSE_initCState(), but the first symbol to include (which will be the last to be read)
+     *   uses the smallest state value possible, saving the cost of this symbol */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void FSE_initCState2(ref FSE_CState_t statePtr, uint* ct, uint symbol)
+    {
+        FSE_initCState(ref statePtr, ct);
+        {
+            FSE_symbolCompressionTransform symbolTT = (
+                (FSE_symbolCompressionTransform*)statePtr.symbolTT
+            )[symbol];
+            ushort* stateTable = (ushort*)statePtr.stateTable;
+            uint nbBitsOut = symbolTT.deltaNbBits + (1 << 15) >> 16;
+            statePtr.value = (nint)((nbBitsOut << 16) - symbolTT.deltaNbBits);
+            statePtr.value = stateTable[
+                (statePtr.value >> (int)nbBitsOut) + symbolTT.deltaFindState
+            ];
+        }
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void FSE_encodeSymbol(
+        ref nuint bitC_bitContainer,
+        ref uint bitC_bitPos,
+        ref FSE_CState_t statePtr,
+        uint symbol
+    )
+    {
+        FSE_symbolCompressionTransform symbolTT = (
+            (FSE_symbolCompressionTransform*)statePtr.symbolTT
+        )[symbol];
+        ushort* stateTable = (ushort*)statePtr.stateTable;
+        uint nbBitsOut = (uint)statePtr.value + symbolTT.deltaNbBits >> 16;
+        BIT_addBits(ref bitC_bitContainer, ref bitC_bitPos, (nuint)statePtr.value, nbBitsOut);
+        statePtr.value = stateTable[(statePtr.value >> (int)nbBitsOut) + symbolTT.deltaFindState];
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void FSE_flushCState(
+        ref nuint bitC_bitContainer,
+        ref uint bitC_bitPos,
+        ref sbyte* bitC_ptr,
+        sbyte* bitC_endPtr,
+        ref FSE_CState_t statePtr
+    )
+    {
+        BIT_addBits(
+            ref bitC_bitContainer,
+            ref bitC_bitPos,
+            (nuint)statePtr.value,
+            statePtr.stateLog
+        );
+        BIT_flushBits(ref bitC_bitContainer, ref bitC_bitPos, ref bitC_ptr, bitC_endPtr);
+    }
+
+    /* FSE_getMaxNbBits() :
+     * Approximate maximum cost of a symbol, in bits.
+     * Fractional get rounded up (i.e. a symbol with a normalized frequency of 3 gives the same result as a frequency of 2)
+     * note 1 : assume symbolValue is valid (<= maxSymbolValue)
+     * note 2 : if freq[symbolValue]==0, @return a fake cost of tableLog+1 bits */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint FSE_getMaxNbBits(void* symbolTTPtr, uint symbolValue)
+    {
+        FSE_symbolCompressionTransform* symbolTT = (FSE_symbolCompressionTransform*)symbolTTPtr;
+        return symbolTT[symbolValue].deltaNbBits + ((1 << 16) - 1) >> 16;
+    }
+
+    /* FSE_bitCost() :
+     * Approximate symbol cost, as fractional value, using fixed-point format (accuracyLog fractional bits)
+     * note 1 : assume symbolValue is valid (<= maxSymbolValue)
+     * note 2 : if freq[symbolValue]==0, @return a fake cost of tableLog+1 bits */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint FSE_bitCost(
+        void* symbolTTPtr,
+        uint tableLog,
+        uint symbolValue,
+        uint accuracyLog
+    )
+    {
+        FSE_symbolCompressionTransform* symbolTT = (FSE_symbolCompressionTransform*)symbolTTPtr;
+        uint minNbBits = symbolTT[symbolValue].deltaNbBits >> 16;
+        uint threshold = minNbBits + 1 << 16;
+        assert(tableLog < 16);
+        assert(accuracyLog < 31 - tableLog);
+        {
+            uint tableSize = (uint)(1 << (int)tableLog);
+            uint deltaFromThreshold = threshold - (symbolTT[symbolValue].deltaNbBits + tableSize);
+            /* linear interpolation (very approximate) */
+            uint normalizedDeltaFromThreshold =
+                deltaFromThreshold << (int)accuracyLog >> (int)tableLog;
+            uint bitMultiplier = (uint)(1 << (int)accuracyLog);
+            assert(symbolTT[symbolValue].deltaNbBits + tableSize <= threshold);
+            assert(normalizedDeltaFromThreshold <= bitMultiplier);
+            return (minNbBits + 1) * bitMultiplier - normalizedDeltaFromThreshold;
+        }
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void FSE_initDState(ref FSE_DState_t DStatePtr, ref BIT_DStream_t bitD, uint* dt)
+    {
+        void* ptr = dt;
+        FSE_DTableHeader* DTableH = (FSE_DTableHeader*)ptr;
+        DStatePtr.state = BIT_readBits(bitD.bitContainer, ref bitD.bitsConsumed, DTableH->tableLog);
+        BIT_reloadDStream(
+            ref bitD.bitContainer,
+            ref bitD.bitsConsumed,
+            ref bitD.ptr,
+            bitD.start,
+            bitD.limitPtr
+        );
+        DStatePtr.table = dt + 1;
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static byte FSE_peekSymbol(FSE_DState_t* DStatePtr)
+    {
+        FSE_decode_t DInfo = ((FSE_decode_t*)DStatePtr->table)[DStatePtr->state];
+        return DInfo.symbol;
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void FSE_updateState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD)
+    {
+        FSE_decode_t DInfo = ((FSE_decode_t*)DStatePtr->table)[DStatePtr->state];
+        uint nbBits = DInfo.nbBits;
+        nuint lowBits = BIT_readBits(bitD, nbBits);
+        DStatePtr->state = DInfo.newState + lowBits;
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static byte FSE_decodeSymbol(
+        ref FSE_DState_t DStatePtr,
+        nuint bitD_bitContainer,
+        ref uint bitD_bitsConsumed
+    )
+    {
+        FSE_decode_t DInfo = ((FSE_decode_t*)DStatePtr.table)[DStatePtr.state];
+        uint nbBits = DInfo.nbBits;
+        byte symbol = DInfo.symbol;
+        nuint lowBits = BIT_readBits(bitD_bitContainer, ref bitD_bitsConsumed, nbBits);
+        DStatePtr.state = DInfo.newState + lowBits;
+        return symbol;
+    }
+
+    /*! FSE_decodeSymbolFast() :
+    unsafe, only works if no symbol has a probability > 50% */
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static byte FSE_decodeSymbolFast(
+        ref FSE_DState_t DStatePtr,
+        nuint bitD_bitContainer,
+        ref uint bitD_bitsConsumed
+    )
+    {
+        FSE_decode_t DInfo = ((FSE_decode_t*)DStatePtr.table)[DStatePtr.state];
+        uint nbBits = DInfo.nbBits;
+        byte symbol = DInfo.symbol;
+        nuint lowBits = BIT_readBitsFast(bitD_bitContainer, ref bitD_bitsConsumed, nbBits);
+        DStatePtr.state = DInfo.newState + lowBits;
+        return symbol;
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static uint FSE_endOfDState(FSE_DState_t* DStatePtr)
+    {
+        return DStatePtr->state == 0 ? 1U : 0U;
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static void FSE_initCState(ref FSE_CState_t statePtr, uint* ct)
+    {
+        void* ptr = ct;
+        ushort* u16ptr = (ushort*)ptr;
+        uint tableLog = MEM_read16(ptr);
+        statePtr.value = (nint)1 << (int)tableLog;
+        statePtr.stateTable = u16ptr + 2;
+        statePtr.symbolTT = ct + 1 + (tableLog != 0 ? 1 << (int)(tableLog - 1) : 1);
+        statePtr.stateLog = tableLog;
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FseCompress.cs

@@ -0,0 +1,782 @@
+using System;
+using System.Runtime.InteropServices;
+using static SharpCompress.Compressors.ZStandard.UnsafeHelper;
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    /* FSE_buildCTable_wksp() :
+     * Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
+     * wkspSize should be sized to handle worst case situation, which is `1<<max_tableLog * sizeof(FSE_FUNCTION_TYPE)`
+     * workSpace must also be properly aligned with FSE_FUNCTION_TYPE requirements
+     */
+    private static nuint FSE_buildCTable_wksp(
+        uint* ct,
+        short* normalizedCounter,
+        uint maxSymbolValue,
+        uint tableLog,
+        void* workSpace,
+        nuint wkspSize
+    )
+    {
+        uint tableSize = (uint)(1 << (int)tableLog);
+        uint tableMask = tableSize - 1;
+        void* ptr = ct;
+        ushort* tableU16 = (ushort*)ptr + 2;
+        /* header */
+        void* FSCT = (uint*)ptr + 1 + (tableLog != 0 ? tableSize >> 1 : 1);
+        FSE_symbolCompressionTransform* symbolTT = (FSE_symbolCompressionTransform*)FSCT;
+        uint step = (tableSize >> 1) + (tableSize >> 3) + 3;
+        uint maxSV1 = maxSymbolValue + 1;
+        /* size = maxSV1 */
+        ushort* cumul = (ushort*)workSpace;
+        /* size = tableSize */
+        byte* tableSymbol = (byte*)(cumul + (maxSV1 + 1));
+        uint highThreshold = tableSize - 1;
+        assert(((nuint)workSpace & 1) == 0);
+        if (
+            sizeof(uint)
+                * ((maxSymbolValue + 2 + (1UL << (int)tableLog)) / 2 + sizeof(ulong) / sizeof(uint))
+            > wkspSize
+        )
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_tableLog_tooLarge));
+        tableU16[-2] = (ushort)tableLog;
+        tableU16[-1] = (ushort)maxSymbolValue;
+        assert(tableLog < 16);
+        {
+            uint u;
+            cumul[0] = 0;
+            for (u = 1; u <= maxSV1; u++)
+            {
+                if (normalizedCounter[u - 1] == -1)
+                {
+                    cumul[u] = (ushort)(cumul[u - 1] + 1);
+                    tableSymbol[highThreshold--] = (byte)(u - 1);
+                }
+                else
+                {
+                    assert(normalizedCounter[u - 1] >= 0);
+                    cumul[u] = (ushort)(cumul[u - 1] + (ushort)normalizedCounter[u - 1]);
+                    assert(cumul[u] >= cumul[u - 1]);
+                }
+            }
+
+            cumul[maxSV1] = (ushort)(tableSize + 1);
+        }
+
+        if (highThreshold == tableSize - 1)
+        {
+            /* size = tableSize + 8 (may write beyond tableSize) */
+            byte* spread = tableSymbol + tableSize;
+            {
+                const ulong add = 0x0101010101010101UL;
+                nuint pos = 0;
+                ulong sv = 0;
+                uint s;
+                for (s = 0; s < maxSV1; ++s, sv += add)
+                {
+                    int i;
+                    int n = normalizedCounter[s];
+                    MEM_write64(spread + pos, sv);
+                    for (i = 8; i < n; i += 8)
+                    {
+                        MEM_write64(spread + pos + i, sv);
+                    }
+
+                    assert(n >= 0);
+                    pos += (nuint)n;
+                }
+            }
+
+            {
+                nuint position = 0;
+                nuint s;
+                /* Experimentally determined optimal unroll */
+                const nuint unroll = 2;
+                assert(tableSize % unroll == 0);
+                for (s = 0; s < tableSize; s += unroll)
+                {
+                    nuint u;
+                    for (u = 0; u < unroll; ++u)
+                    {
+                        nuint uPosition = position + u * step & tableMask;
+                        tableSymbol[uPosition] = spread[s + u];
+                    }
+
+                    position = position + unroll * step & tableMask;
+                }
+
+                assert(position == 0);
+            }
+        }
+        else
+        {
+            uint position = 0;
+            uint symbol;
+            for (symbol = 0; symbol < maxSV1; symbol++)
+            {
+                int nbOccurrences;
+                int freq = normalizedCounter[symbol];
+                for (nbOccurrences = 0; nbOccurrences < freq; nbOccurrences++)
+                {
+                    tableSymbol[position] = (byte)symbol;
+                    position = position + step & tableMask;
+                    while (position > highThreshold)
+                        position = position + step & tableMask;
+                }
+            }
+
+            assert(position == 0);
+        }
+
+        {
+            uint u;
+            for (u = 0; u < tableSize; u++)
+            {
+                /* note : static analyzer may not understand tableSymbol is properly initialized */
+                byte s = tableSymbol[u];
+                tableU16[cumul[s]++] = (ushort)(tableSize + u);
+            }
+        }
+
+        {
+            uint total = 0;
+            uint s;
+            for (s = 0; s <= maxSymbolValue; s++)
+            {
+                switch (normalizedCounter[s])
+                {
+                    case 0:
+                        symbolTT[s].deltaNbBits = (tableLog + 1 << 16) - (uint)(1 << (int)tableLog);
+                        break;
+                    case -1:
+                    case 1:
+                        symbolTT[s].deltaNbBits = (tableLog << 16) - (uint)(1 << (int)tableLog);
+                        assert(total <= 2147483647);
+                        symbolTT[s].deltaFindState = (int)(total - 1);
+                        total++;
+                        break;
+                    default:
+                        assert(normalizedCounter[s] > 1);
+
+                        {
+                            uint maxBitsOut =
+                                tableLog - ZSTD_highbit32((uint)normalizedCounter[s] - 1);
+                            uint minStatePlus = (uint)normalizedCounter[s] << (int)maxBitsOut;
+                            symbolTT[s].deltaNbBits = (maxBitsOut << 16) - minStatePlus;
+                            symbolTT[s].deltaFindState = (int)(total - (uint)normalizedCounter[s]);
+                            total += (uint)normalizedCounter[s];
+                        }
+
+                        break;
+                }
+            }
+        }
+
+        return 0;
+    }
+
+    /*-**************************************************************
+     *  FSE NCount encoding
+     ****************************************************************/
+    private static nuint FSE_NCountWriteBound(uint maxSymbolValue, uint tableLog)
+    {
+        nuint maxHeaderSize = ((maxSymbolValue + 1) * tableLog + 4 + 2) / 8 + 1 + 2;
+        return maxSymbolValue != 0 ? maxHeaderSize : 512;
+    }
+
+    private static nuint FSE_writeNCount_generic(
+        void* header,
+        nuint headerBufferSize,
+        short* normalizedCounter,
+        uint maxSymbolValue,
+        uint tableLog,
+        uint writeIsSafe
+    )
+    {
+        byte* ostart = (byte*)header;
+        byte* @out = ostart;
+        byte* oend = ostart + headerBufferSize;
+        int nbBits;
+        int tableSize = 1 << (int)tableLog;
+        int remaining;
+        int threshold;
+        uint bitStream = 0;
+        int bitCount = 0;
+        uint symbol = 0;
+        uint alphabetSize = maxSymbolValue + 1;
+        int previousIs0 = 0;
+        bitStream += tableLog - 5 << bitCount;
+        bitCount += 4;
+        remaining = tableSize + 1;
+        threshold = tableSize;
+        nbBits = (int)tableLog + 1;
+        while (symbol < alphabetSize && remaining > 1)
+        {
+            if (previousIs0 != 0)
+            {
+                uint start = symbol;
+                while (symbol < alphabetSize && normalizedCounter[symbol] == 0)
+                    symbol++;
+                if (symbol == alphabetSize)
+                    break;
+                while (symbol >= start + 24)
+                {
+                    start += 24;
+                    bitStream += 0xFFFFU << bitCount;
+                    if (writeIsSafe == 0 && @out > oend - 2)
+                        return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall));
+                    @out[0] = (byte)bitStream;
+                    @out[1] = (byte)(bitStream >> 8);
+                    @out += 2;
+                    bitStream >>= 16;
+                }
+
+                while (symbol >= start + 3)
+                {
+                    start += 3;
+                    bitStream += 3U << bitCount;
+                    bitCount += 2;
+                }
+
+                bitStream += symbol - start << bitCount;
+                bitCount += 2;
+                if (bitCount > 16)
+                {
+                    if (writeIsSafe == 0 && @out > oend - 2)
+                        return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall));
+                    @out[0] = (byte)bitStream;
+                    @out[1] = (byte)(bitStream >> 8);
+                    @out += 2;
+                    bitStream >>= 16;
+                    bitCount -= 16;
+                }
+            }
+
+            {
+                int count = normalizedCounter[symbol++];
+                int max = 2 * threshold - 1 - remaining;
+                remaining -= count < 0 ? -count : count;
+                count++;
+                if (count >= threshold)
+                    count += max;
+                bitStream += (uint)count << bitCount;
+                bitCount += nbBits;
+                bitCount -= count < max ? 1 : 0;
+                previousIs0 = count == 1 ? 1 : 0;
+                if (remaining < 1)
+                    return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+                while (remaining < threshold)
+                {
+                    nbBits--;
+                    threshold >>= 1;
+                }
+            }
+
+            if (bitCount > 16)
+            {
+                if (writeIsSafe == 0 && @out > oend - 2)
+                    return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall));
+                @out[0] = (byte)bitStream;
+                @out[1] = (byte)(bitStream >> 8);
+                @out += 2;
+                bitStream >>= 16;
+                bitCount -= 16;
+            }
+        }
+
+        if (remaining != 1)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+        assert(symbol <= alphabetSize);
+        if (writeIsSafe == 0 && @out > oend - 2)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall));
+        @out[0] = (byte)bitStream;
+        @out[1] = (byte)(bitStream >> 8);
+        @out += (bitCount + 7) / 8;
+        assert(@out >= ostart);
+        return (nuint)(@out - ostart);
+    }
+
+    /*! FSE_writeNCount():
+    Compactly save 'normalizedCounter' into 'buffer'.
+    @return : size of the compressed table,
+    or an errorCode, which can be tested using FSE_isError(). */
+    private static nuint FSE_writeNCount(
+        void* buffer,
+        nuint bufferSize,
+        short* normalizedCounter,
+        uint maxSymbolValue,
+        uint tableLog
+    )
+    {
+        if (tableLog > 14 - 2)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_tableLog_tooLarge));
+        if (tableLog < 5)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+        if (bufferSize < FSE_NCountWriteBound(maxSymbolValue, tableLog))
+            return FSE_writeNCount_generic(
+                buffer,
+                bufferSize,
+                normalizedCounter,
+                maxSymbolValue,
+                tableLog,
+                0
+            );
+        return FSE_writeNCount_generic(
+            buffer,
+            bufferSize,
+            normalizedCounter,
+            maxSymbolValue,
+            tableLog,
+            1
+        );
+    }
+
+    /* provides the minimum logSize to safely represent a distribution */
+    private static uint FSE_minTableLog(nuint srcSize, uint maxSymbolValue)
+    {
+        uint minBitsSrc = ZSTD_highbit32((uint)srcSize) + 1;
+        uint minBitsSymbols = ZSTD_highbit32(maxSymbolValue) + 2;
+        uint minBits = minBitsSrc < minBitsSymbols ? minBitsSrc : minBitsSymbols;
+        assert(srcSize > 1);
+        return minBits;
+    }
+
+    /* *****************************************
+     *  FSE advanced API
+     ***************************************** */
+    private static uint FSE_optimalTableLog_internal(
+        uint maxTableLog,
+        nuint srcSize,
+        uint maxSymbolValue,
+        uint minus
+    )
+    {
+        uint maxBitsSrc = ZSTD_highbit32((uint)(srcSize - 1)) - minus;
+        uint tableLog = maxTableLog;
+        uint minBits = FSE_minTableLog(srcSize, maxSymbolValue);
+        assert(srcSize > 1);
+        if (tableLog == 0)
+            tableLog = 13 - 2;
+        if (maxBitsSrc < tableLog)
+            tableLog = maxBitsSrc;
+        if (minBits > tableLog)
+            tableLog = minBits;
+        if (tableLog < 5)
+            tableLog = 5;
+        if (tableLog > 14 - 2)
+            tableLog = 14 - 2;
+        return tableLog;
+    }
+
+    /*! FSE_optimalTableLog():
+    dynamically downsize 'tableLog' when conditions are met.
+    It saves CPU time, by using smaller tables, while preserving or even improving compression ratio.
+    @return : recommended tableLog (necessarily <= 'maxTableLog') */
+    private static uint FSE_optimalTableLog(uint maxTableLog, nuint srcSize, uint maxSymbolValue)
+    {
+        return FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 2);
+    }
+
+    /* Secondary normalization method.
+    To be used when primary method fails. */
+    private static nuint FSE_normalizeM2(
+        short* norm,
+        uint tableLog,
+        uint* count,
+        nuint total,
+        uint maxSymbolValue,
+        short lowProbCount
+    )
+    {
+        const short NOT_YET_ASSIGNED = -2;
+        uint s;
+        uint distributed = 0;
+        uint ToDistribute;
+        /* Init */
+        uint lowThreshold = (uint)(total >> (int)tableLog);
+        uint lowOne = (uint)(total * 3 >> (int)(tableLog + 1));
+        for (s = 0; s <= maxSymbolValue; s++)
+        {
+            if (count[s] == 0)
+            {
+                norm[s] = 0;
+                continue;
+            }
+
+            if (count[s] <= lowThreshold)
+            {
+                norm[s] = lowProbCount;
+                distributed++;
+                total -= count[s];
+                continue;
+            }
+
+            if (count[s] <= lowOne)
+            {
+                norm[s] = 1;
+                distributed++;
+                total -= count[s];
+                continue;
+            }
+
+            norm[s] = NOT_YET_ASSIGNED;
+        }
+
+        ToDistribute = (uint)(1 << (int)tableLog) - distributed;
+        if (ToDistribute == 0)
+            return 0;
+        if (total / ToDistribute > lowOne)
+        {
+            lowOne = (uint)(total * 3 / (ToDistribute * 2));
+            for (s = 0; s <= maxSymbolValue; s++)
+            {
+                if (norm[s] == NOT_YET_ASSIGNED && count[s] <= lowOne)
+                {
+                    norm[s] = 1;
+                    distributed++;
+                    total -= count[s];
+                    continue;
+                }
+            }
+
+            ToDistribute = (uint)(1 << (int)tableLog) - distributed;
+        }
+
+        if (distributed == maxSymbolValue + 1)
+        {
+            /* all values are pretty poor;
+            probably incompressible data (should have already been detected);
+            find max, then give all remaining points to max */
+            uint maxV = 0,
+                maxC = 0;
+            for (s = 0; s <= maxSymbolValue; s++)
+                if (count[s] > maxC)
+                {
+                    maxV = s;
+                    maxC = count[s];
+                }
+
+            norm[maxV] += (short)ToDistribute;
+            return 0;
+        }
+
+        if (total == 0)
+        {
+            for (s = 0; ToDistribute > 0; s = (s + 1) % (maxSymbolValue + 1))
+                if (norm[s] > 0)
+                {
+                    ToDistribute--;
+                    norm[s]++;
+                }
+
+            return 0;
+        }
+
+        {
+            ulong vStepLog = 62 - tableLog;
+            ulong mid = (1UL << (int)(vStepLog - 1)) - 1;
+            /* scale on remaining */
+            ulong rStep = (((ulong)1 << (int)vStepLog) * ToDistribute + mid) / (uint)total;
+            ulong tmpTotal = mid;
+            for (s = 0; s <= maxSymbolValue; s++)
+            {
+                if (norm[s] == NOT_YET_ASSIGNED)
+                {
+                    ulong end = tmpTotal + count[s] * rStep;
+                    uint sStart = (uint)(tmpTotal >> (int)vStepLog);
+                    uint sEnd = (uint)(end >> (int)vStepLog);
+                    uint weight = sEnd - sStart;
+                    if (weight < 1)
+                        return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+                    norm[s] = (short)weight;
+                    tmpTotal = end;
+                }
+            }
+        }
+
+        return 0;
+    }
+
+#if NET7_0_OR_GREATER
+    private static ReadOnlySpan<uint> Span_rtbTable =>
+        new uint[8] { 0, 473195, 504333, 520860, 550000, 700000, 750000, 830000 };
+    private static uint* rtbTable =>
+        (uint*)
+            System.Runtime.CompilerServices.Unsafe.AsPointer(
+                ref MemoryMarshal.GetReference(Span_rtbTable)
+            );
+#else
+
+    private static readonly uint* rtbTable = GetArrayPointer(
+        new uint[8] { 0, 473195, 504333, 520860, 550000, 700000, 750000, 830000 }
+    );
+#endif
+    /*! FSE_normalizeCount():
+    normalize counts so that sum(count[]) == Power_of_2 (2^tableLog)
+    'normalizedCounter' is a table of short, of minimum size (maxSymbolValue+1).
+    useLowProbCount is a boolean parameter which trades off compressed size for
+    faster header decoding. When it is set to 1, the compressed data will be slightly
+    smaller. And when it is set to 0, FSE_readNCount() and FSE_buildDTable() will be
+    faster. If you are compressing a small amount of data (< 2 KB) then useLowProbCount=0
+    is a good default, since header deserialization makes a big speed difference.
+    Otherwise, useLowProbCount=1 is a good default, since the speed difference is small.
+    @return : tableLog,
+    or an errorCode, which can be tested using FSE_isError() */
+    private static nuint FSE_normalizeCount(
+        short* normalizedCounter,
+        uint tableLog,
+        uint* count,
+        nuint total,
+        uint maxSymbolValue,
+        uint useLowProbCount
+    )
+    {
+        if (tableLog == 0)
+            tableLog = 13 - 2;
+        if (tableLog < 5)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+        if (tableLog > 14 - 2)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_tableLog_tooLarge));
+        if (tableLog < FSE_minTableLog(total, maxSymbolValue))
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+        {
+            short lowProbCount = (short)(useLowProbCount != 0 ? -1 : 1);
+            ulong scale = 62 - tableLog;
+            /* <== here, one division ! */
+            ulong step = ((ulong)1 << 62) / (uint)total;
+            ulong vStep = 1UL << (int)(scale - 20);
+            int stillToDistribute = 1 << (int)tableLog;
+            uint s;
+            uint largest = 0;
+            short largestP = 0;
+            uint lowThreshold = (uint)(total >> (int)tableLog);
+            for (s = 0; s <= maxSymbolValue; s++)
+            {
+                if (count[s] == total)
+                    return 0;
+                if (count[s] == 0)
+                {
+                    normalizedCounter[s] = 0;
+                    continue;
+                }
+
+                if (count[s] <= lowThreshold)
+                {
+                    normalizedCounter[s] = lowProbCount;
+                    stillToDistribute--;
+                }
+                else
+                {
+                    short proba = (short)(count[s] * step >> (int)scale);
+                    if (proba < 8)
+                    {
+                        ulong restToBeat = vStep * rtbTable[proba];
+                        proba += (short)(
+                            count[s] * step - ((ulong)proba << (int)scale) > restToBeat ? 1 : 0
+                        );
+                    }
+
+                    if (proba > largestP)
+                    {
+                        largestP = proba;
+                        largest = s;
+                    }
+
+                    normalizedCounter[s] = proba;
+                    stillToDistribute -= proba;
+                }
+            }
+
+            if (-stillToDistribute >= normalizedCounter[largest] >> 1)
+            {
+                /* corner case, need another normalization method */
+                nuint errorCode = FSE_normalizeM2(
+                    normalizedCounter,
+                    tableLog,
+                    count,
+                    total,
+                    maxSymbolValue,
+                    lowProbCount
+                );
+                if (ERR_isError(errorCode))
+                    return errorCode;
+            }
+            else
+                normalizedCounter[largest] += (short)stillToDistribute;
+        }
+
+        return tableLog;
+    }
+
+    /* fake FSE_CTable, for rle input (always same symbol) */
+    private static nuint FSE_buildCTable_rle(uint* ct, byte symbolValue)
+    {
+        void* ptr = ct;
+        ushort* tableU16 = (ushort*)ptr + 2;
+        void* FSCTptr = (uint*)ptr + 2;
+        FSE_symbolCompressionTransform* symbolTT = (FSE_symbolCompressionTransform*)FSCTptr;
+        tableU16[-2] = 0;
+        tableU16[-1] = symbolValue;
+        tableU16[0] = 0;
+        tableU16[1] = 0;
+        symbolTT[symbolValue].deltaNbBits = 0;
+        symbolTT[symbolValue].deltaFindState = 0;
+        return 0;
+    }
+
+    private static nuint FSE_compress_usingCTable_generic(
+        void* dst,
+        nuint dstSize,
+        void* src,
+        nuint srcSize,
+        uint* ct,
+        uint fast
+    )
+    {
+        byte* istart = (byte*)src;
+        byte* iend = istart + srcSize;
+        byte* ip = iend;
+        BIT_CStream_t bitC;
+        System.Runtime.CompilerServices.Unsafe.SkipInit(out bitC);
+        FSE_CState_t CState1,
+            CState2;
+        System.Runtime.CompilerServices.Unsafe.SkipInit(out CState1);
+        System.Runtime.CompilerServices.Unsafe.SkipInit(out CState2);
+        if (srcSize <= 2)
+            return 0;
+        {
+            nuint initError = BIT_initCStream(ref bitC, dst, dstSize);
+            if (ERR_isError(initError))
+                return 0;
+        }
+
+        nuint bitC_bitContainer = bitC.bitContainer;
+        uint bitC_bitPos = bitC.bitPos;
+        sbyte* bitC_ptr = bitC.ptr;
+        sbyte* bitC_endPtr = bitC.endPtr;
+        if ((srcSize & 1) != 0)
+        {
+            FSE_initCState2(ref CState1, ct, *--ip);
+            FSE_initCState2(ref CState2, ct, *--ip);
+            FSE_encodeSymbol(ref bitC_bitContainer, ref bitC_bitPos, ref CState1, *--ip);
+            if (fast != 0)
+                BIT_flushBitsFast(
+                    ref bitC_bitContainer,
+                    ref bitC_bitPos,
+                    ref bitC_ptr,
+                    bitC_endPtr
+                );
+            else
+                BIT_flushBits(ref bitC_bitContainer, ref bitC_bitPos, ref bitC_ptr, bitC_endPtr);
+        }
+        else
+        {
+            FSE_initCState2(ref CState2, ct, *--ip);
+            FSE_initCState2(ref CState1, ct, *--ip);
+        }
+
+        srcSize -= 2;
+        if (sizeof(nuint) * 8 > (14 - 2) * 4 + 7 && (srcSize & 2) != 0)
+        {
+            FSE_encodeSymbol(ref bitC_bitContainer, ref bitC_bitPos, ref CState2, *--ip);
+            FSE_encodeSymbol(ref bitC_bitContainer, ref bitC_bitPos, ref CState1, *--ip);
+            if (fast != 0)
+                BIT_flushBitsFast(
+                    ref bitC_bitContainer,
+                    ref bitC_bitPos,
+                    ref bitC_ptr,
+                    bitC_endPtr
+                );
+            else
+                BIT_flushBits(ref bitC_bitContainer, ref bitC_bitPos, ref bitC_ptr, bitC_endPtr);
+        }
+
+        while (ip > istart)
+        {
+            FSE_encodeSymbol(ref bitC_bitContainer, ref bitC_bitPos, ref CState2, *--ip);
+            if (sizeof(nuint) * 8 < (14 - 2) * 2 + 7)
+                if (fast != 0)
+                    BIT_flushBitsFast(
+                        ref bitC_bitContainer,
+                        ref bitC_bitPos,
+                        ref bitC_ptr,
+                        bitC_endPtr
+                    );
+                else
+                    BIT_flushBits(
+                        ref bitC_bitContainer,
+                        ref bitC_bitPos,
+                        ref bitC_ptr,
+                        bitC_endPtr
+                    );
+            FSE_encodeSymbol(ref bitC_bitContainer, ref bitC_bitPos, ref CState1, *--ip);
+            if (sizeof(nuint) * 8 > (14 - 2) * 4 + 7)
+            {
+                FSE_encodeSymbol(ref bitC_bitContainer, ref bitC_bitPos, ref CState2, *--ip);
+                FSE_encodeSymbol(ref bitC_bitContainer, ref bitC_bitPos, ref CState1, *--ip);
+            }
+
+            if (fast != 0)
+                BIT_flushBitsFast(
+                    ref bitC_bitContainer,
+                    ref bitC_bitPos,
+                    ref bitC_ptr,
+                    bitC_endPtr
+                );
+            else
+                BIT_flushBits(ref bitC_bitContainer, ref bitC_bitPos, ref bitC_ptr, bitC_endPtr);
+        }
+
+        FSE_flushCState(
+            ref bitC_bitContainer,
+            ref bitC_bitPos,
+            ref bitC_ptr,
+            bitC_endPtr,
+            ref CState2
+        );
+        FSE_flushCState(
+            ref bitC_bitContainer,
+            ref bitC_bitPos,
+            ref bitC_ptr,
+            bitC_endPtr,
+            ref CState1
+        );
+        return BIT_closeCStream(
+            ref bitC_bitContainer,
+            ref bitC_bitPos,
+            bitC_ptr,
+            bitC_endPtr,
+            bitC.startPtr
+        );
+    }
+
+    /*! FSE_compress_usingCTable():
+    Compress `src` using `ct` into `dst` which must be already allocated.
+    @return : size of compressed data (<= `dstCapacity`),
+    or 0 if compressed data could not fit into `dst`,
+    or an errorCode, which can be tested using FSE_isError() */
+    private static nuint FSE_compress_usingCTable(
+        void* dst,
+        nuint dstSize,
+        void* src,
+        nuint srcSize,
+        uint* ct
+    )
+    {
+        uint fast = dstSize >= srcSize + (srcSize >> 7) + 4 + (nuint)sizeof(nuint) ? 1U : 0U;
+        if (fast != 0)
+            return FSE_compress_usingCTable_generic(dst, dstSize, src, srcSize, ct, 1);
+        else
+            return FSE_compress_usingCTable_generic(dst, dstSize, src, srcSize, ct, 0);
+    }
+
+    /*-*****************************************
+     *  Tool functions
+     ******************************************/
+    private static nuint FSE_compressBound(nuint size)
+    {
+        return 512 + (size + (size >> 7) + 4 + (nuint)sizeof(nuint));
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/FseDecompress.cs

@@ -0,0 +1,462 @@
+using System.Runtime.CompilerServices;
+using static SharpCompress.Compressors.ZStandard.UnsafeHelper;
+
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public static unsafe partial class Methods
+{
+    private static nuint FSE_buildDTable_internal(
+        uint* dt,
+        short* normalizedCounter,
+        uint maxSymbolValue,
+        uint tableLog,
+        void* workSpace,
+        nuint wkspSize
+    )
+    {
+        /* because *dt is unsigned, 32-bits aligned on 32-bits */
+        void* tdPtr = dt + 1;
+        FSE_decode_t* tableDecode = (FSE_decode_t*)tdPtr;
+        ushort* symbolNext = (ushort*)workSpace;
+        byte* spread = (byte*)(symbolNext + maxSymbolValue + 1);
+        uint maxSV1 = maxSymbolValue + 1;
+        uint tableSize = (uint)(1 << (int)tableLog);
+        uint highThreshold = tableSize - 1;
+        if (sizeof(short) * (maxSymbolValue + 1) + (1UL << (int)tableLog) + 8 > wkspSize)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_maxSymbolValue_tooLarge));
+        if (maxSymbolValue > 255)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_maxSymbolValue_tooLarge));
+        if (tableLog > 14 - 2)
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_tableLog_tooLarge));
+        {
+            FSE_DTableHeader DTableH;
+            DTableH.tableLog = (ushort)tableLog;
+            DTableH.fastMode = 1;
+            {
+                short largeLimit = (short)(1 << (int)(tableLog - 1));
+                uint s;
+                for (s = 0; s < maxSV1; s++)
+                {
+                    if (normalizedCounter[s] == -1)
+                    {
+                        tableDecode[highThreshold--].symbol = (byte)s;
+                        symbolNext[s] = 1;
+                    }
+                    else
+                    {
+                        if (normalizedCounter[s] >= largeLimit)
+                            DTableH.fastMode = 0;
+                        symbolNext[s] = (ushort)normalizedCounter[s];
+                    }
+                }
+            }
+
+            memcpy(dt, &DTableH, (uint)sizeof(FSE_DTableHeader));
+        }
+
+        if (highThreshold == tableSize - 1)
+        {
+            nuint tableMask = tableSize - 1;
+            nuint step = (tableSize >> 1) + (tableSize >> 3) + 3;
+            {
+                const ulong add = 0x0101010101010101UL;
+                nuint pos = 0;
+                ulong sv = 0;
+                uint s;
+                for (s = 0; s < maxSV1; ++s, sv += add)
+                {
+                    int i;
+                    int n = normalizedCounter[s];
+                    MEM_write64(spread + pos, sv);
+                    for (i = 8; i < n; i += 8)
+                    {
+                        MEM_write64(spread + pos + i, sv);
+                    }
+
+                    pos += (nuint)n;
+                }
+            }
+
+            {
+                nuint position = 0;
+                nuint s;
+                const nuint unroll = 2;
+                assert(tableSize % unroll == 0);
+                for (s = 0; s < tableSize; s += unroll)
+                {
+                    nuint u;
+                    for (u = 0; u < unroll; ++u)
+                    {
+                        nuint uPosition = position + u * step & tableMask;
+                        tableDecode[uPosition].symbol = spread[s + u];
+                    }
+
+                    position = position + unroll * step & tableMask;
+                }
+
+                assert(position == 0);
+            }
+        }
+        else
+        {
+            uint tableMask = tableSize - 1;
+            uint step = (tableSize >> 1) + (tableSize >> 3) + 3;
+            uint s,
+                position = 0;
+            for (s = 0; s < maxSV1; s++)
+            {
+                int i;
+                for (i = 0; i < normalizedCounter[s]; i++)
+                {
+                    tableDecode[position].symbol = (byte)s;
+                    position = position + step & tableMask;
+                    while (position > highThreshold)
+                        position = position + step & tableMask;
+                }
+            }
+
+            if (position != 0)
+                return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+        }
+
+        {
+            uint u;
+            for (u = 0; u < tableSize; u++)
+            {
+                byte symbol = tableDecode[u].symbol;
+                uint nextState = symbolNext[symbol]++;
+                tableDecode[u].nbBits = (byte)(tableLog - ZSTD_highbit32(nextState));
+                tableDecode[u].newState = (ushort)(
+                    (nextState << tableDecode[u].nbBits) - tableSize
+                );
+            }
+        }
+
+        return 0;
+    }
+
+    private static nuint FSE_buildDTable_wksp(
+        uint* dt,
+        short* normalizedCounter,
+        uint maxSymbolValue,
+        uint tableLog,
+        void* workSpace,
+        nuint wkspSize
+    )
+    {
+        return FSE_buildDTable_internal(
+            dt,
+            normalizedCounter,
+            maxSymbolValue,
+            tableLog,
+            workSpace,
+            wkspSize
+        );
+    }
+
+    /*-*******************************************************
+     *  Decompression (Byte symbols)
+     *********************************************************/
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint FSE_decompress_usingDTable_generic(
+        void* dst,
+        nuint maxDstSize,
+        void* cSrc,
+        nuint cSrcSize,
+        uint* dt,
+        uint fast
+    )
+    {
+        byte* ostart = (byte*)dst;
+        byte* op = ostart;
+        byte* omax = op + maxDstSize;
+        byte* olimit = omax - 3;
+        BIT_DStream_t bitD;
+        System.Runtime.CompilerServices.Unsafe.SkipInit(out bitD);
+        FSE_DState_t state1;
+        System.Runtime.CompilerServices.Unsafe.SkipInit(out state1);
+        FSE_DState_t state2;
+        System.Runtime.CompilerServices.Unsafe.SkipInit(out state2);
+        {
+            /* Init */
+            nuint _var_err__ = BIT_initDStream(ref bitD, cSrc, cSrcSize);
+            if (ERR_isError(_var_err__))
+                return _var_err__;
+        }
+
+        FSE_initDState(ref state1, ref bitD, dt);
+        FSE_initDState(ref state2, ref bitD, dt);
+        nuint bitD_bitContainer = bitD.bitContainer;
+        uint bitD_bitsConsumed = bitD.bitsConsumed;
+        sbyte* bitD_ptr = bitD.ptr;
+        sbyte* bitD_start = bitD.start;
+        sbyte* bitD_limitPtr = bitD.limitPtr;
+        if (
+            BIT_reloadDStream(
+                ref bitD_bitContainer,
+                ref bitD_bitsConsumed,
+                ref bitD_ptr,
+                bitD_start,
+                bitD_limitPtr
+            ) == BIT_DStream_status.BIT_DStream_overflow
+        )
+        {
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_corruption_detected));
+        }
+
+        for (
+            ;
+            BIT_reloadDStream(
+                ref bitD_bitContainer,
+                ref bitD_bitsConsumed,
+                ref bitD_ptr,
+                bitD_start,
+                bitD_limitPtr
+            ) == BIT_DStream_status.BIT_DStream_unfinished
+                && op < olimit;
+            op += 4
+        )
+        {
+            op[0] =
+                fast != 0
+                    ? FSE_decodeSymbolFast(ref state1, bitD_bitContainer, ref bitD_bitsConsumed)
+                    : FSE_decodeSymbol(ref state1, bitD_bitContainer, ref bitD_bitsConsumed);
+            if ((14 - 2) * 2 + 7 > sizeof(nuint) * 8)
+                BIT_reloadDStream(
+                    ref bitD_bitContainer,
+                    ref bitD_bitsConsumed,
+                    ref bitD_ptr,
+                    bitD_start,
+                    bitD_limitPtr
+                );
+            op[1] =
+                fast != 0
+                    ? FSE_decodeSymbolFast(ref state2, bitD_bitContainer, ref bitD_bitsConsumed)
+                    : FSE_decodeSymbol(ref state2, bitD_bitContainer, ref bitD_bitsConsumed);
+            if ((14 - 2) * 4 + 7 > sizeof(nuint) * 8)
+            {
+                if (
+                    BIT_reloadDStream(
+                        ref bitD_bitContainer,
+                        ref bitD_bitsConsumed,
+                        ref bitD_ptr,
+                        bitD_start,
+                        bitD_limitPtr
+                    ) > BIT_DStream_status.BIT_DStream_unfinished
+                )
+                {
+                    op += 2;
+                    break;
+                }
+            }
+
+            op[2] =
+                fast != 0
+                    ? FSE_decodeSymbolFast(ref state1, bitD_bitContainer, ref bitD_bitsConsumed)
+                    : FSE_decodeSymbol(ref state1, bitD_bitContainer, ref bitD_bitsConsumed);
+            if ((14 - 2) * 2 + 7 > sizeof(nuint) * 8)
+                BIT_reloadDStream(
+                    ref bitD_bitContainer,
+                    ref bitD_bitsConsumed,
+                    ref bitD_ptr,
+                    bitD_start,
+                    bitD_limitPtr
+                );
+            op[3] =
+                fast != 0
+                    ? FSE_decodeSymbolFast(ref state2, bitD_bitContainer, ref bitD_bitsConsumed)
+                    : FSE_decodeSymbol(ref state2, bitD_bitContainer, ref bitD_bitsConsumed);
+        }
+
+        while (true)
+        {
+            if (op > omax - 2)
+                return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall));
+            *op++ =
+                fast != 0
+                    ? FSE_decodeSymbolFast(ref state1, bitD_bitContainer, ref bitD_bitsConsumed)
+                    : FSE_decodeSymbol(ref state1, bitD_bitContainer, ref bitD_bitsConsumed);
+            if (
+                BIT_reloadDStream(
+                    ref bitD_bitContainer,
+                    ref bitD_bitsConsumed,
+                    ref bitD_ptr,
+                    bitD_start,
+                    bitD_limitPtr
+                ) == BIT_DStream_status.BIT_DStream_overflow
+            )
+            {
+                *op++ =
+                    fast != 0
+                        ? FSE_decodeSymbolFast(ref state2, bitD_bitContainer, ref bitD_bitsConsumed)
+                        : FSE_decodeSymbol(ref state2, bitD_bitContainer, ref bitD_bitsConsumed);
+                break;
+            }
+
+            if (op > omax - 2)
+                return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_dstSize_tooSmall));
+            *op++ =
+                fast != 0
+                    ? FSE_decodeSymbolFast(ref state2, bitD_bitContainer, ref bitD_bitsConsumed)
+                    : FSE_decodeSymbol(ref state2, bitD_bitContainer, ref bitD_bitsConsumed);
+            if (
+                BIT_reloadDStream(
+                    ref bitD_bitContainer,
+                    ref bitD_bitsConsumed,
+                    ref bitD_ptr,
+                    bitD_start,
+                    bitD_limitPtr
+                ) == BIT_DStream_status.BIT_DStream_overflow
+            )
+            {
+                *op++ =
+                    fast != 0
+                        ? FSE_decodeSymbolFast(ref state1, bitD_bitContainer, ref bitD_bitsConsumed)
+                        : FSE_decodeSymbol(ref state1, bitD_bitContainer, ref bitD_bitsConsumed);
+                break;
+            }
+        }
+
+        assert(op >= ostart);
+        return (nuint)(op - ostart);
+    }
+
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    private static nuint FSE_decompress_wksp_body(
+        void* dst,
+        nuint dstCapacity,
+        void* cSrc,
+        nuint cSrcSize,
+        uint maxLog,
+        void* workSpace,
+        nuint wkspSize,
+        int bmi2
+    )
+    {
+        byte* istart = (byte*)cSrc;
+        byte* ip = istart;
+        uint tableLog;
+        uint maxSymbolValue = 255;
+        FSE_DecompressWksp* wksp = (FSE_DecompressWksp*)workSpace;
+        nuint dtablePos = (nuint)(sizeof(FSE_DecompressWksp) / sizeof(uint));
+        uint* dtable = (uint*)workSpace + dtablePos;
+        if (wkspSize < (nuint)sizeof(FSE_DecompressWksp))
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_GENERIC));
+        {
+            nuint NCountLength = FSE_readNCount_bmi2(
+                wksp->ncount,
+                &maxSymbolValue,
+                &tableLog,
+                istart,
+                cSrcSize,
+                bmi2
+            );
+            if (ERR_isError(NCountLength))
+                return NCountLength;
+            if (tableLog > maxLog)
+                return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_tableLog_tooLarge));
+            assert(NCountLength <= cSrcSize);
+            ip += NCountLength;
+            cSrcSize -= NCountLength;
+        }
+
+        if (
+            (
+                (ulong)(1 + (1 << (int)tableLog) + 1)
+                + (
+                    sizeof(short) * (maxSymbolValue + 1)
+                    + (1UL << (int)tableLog)
+                    + 8
+                    + sizeof(uint)
+                    - 1
+                ) / sizeof(uint)
+                + (255 + 1) / 2
+                + 1
+            ) * sizeof(uint)
+            > wkspSize
+        )
+            return unchecked((nuint)(-(int)ZSTD_ErrorCode.ZSTD_error_tableLog_tooLarge));
+        assert(
+            (nuint)(sizeof(FSE_DecompressWksp) + (1 + (1 << (int)tableLog)) * sizeof(uint))
+                <= wkspSize
+        );
+        workSpace =
+            (byte*)workSpace
+            + sizeof(FSE_DecompressWksp)
+            + (1 + (1 << (int)tableLog)) * sizeof(uint);
+        wkspSize -= (nuint)(sizeof(FSE_DecompressWksp) + (1 + (1 << (int)tableLog)) * sizeof(uint));
+        {
+            nuint _var_err__ = FSE_buildDTable_internal(
+                dtable,
+                wksp->ncount,
+                maxSymbolValue,
+                tableLog,
+                workSpace,
+                wkspSize
+            );
+            if (ERR_isError(_var_err__))
+                return _var_err__;
+        }
+
+        {
+            void* ptr = dtable;
+            FSE_DTableHeader* DTableH = (FSE_DTableHeader*)ptr;
+            uint fastMode = DTableH->fastMode;
+            if (fastMode != 0)
+                return FSE_decompress_usingDTable_generic(
+                    dst,
+                    dstCapacity,
+                    ip,
+                    cSrcSize,
+                    dtable,
+                    1
+                );
+            return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, dtable, 0);
+        }
+    }
+
+    /* Avoids the FORCE_INLINE of the _body() function. */
+    private static nuint FSE_decompress_wksp_body_default(
+        void* dst,
+        nuint dstCapacity,
+        void* cSrc,
+        nuint cSrcSize,
+        uint maxLog,
+        void* workSpace,
+        nuint wkspSize
+    )
+    {
+        return FSE_decompress_wksp_body(
+            dst,
+            dstCapacity,
+            cSrc,
+            cSrcSize,
+            maxLog,
+            workSpace,
+            wkspSize,
+            0
+        );
+    }
+
+    private static nuint FSE_decompress_wksp_bmi2(
+        void* dst,
+        nuint dstCapacity,
+        void* cSrc,
+        nuint cSrcSize,
+        uint maxLog,
+        void* workSpace,
+        nuint wkspSize,
+        int bmi2
+    )
+    {
+        return FSE_decompress_wksp_body_default(
+            dst,
+            dstCapacity,
+            cSrc,
+            cSrcSize,
+            maxLog,
+            workSpace,
+            wkspSize
+        );
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/HIST_checkInput_e.cs

@@ -0,0 +1,7 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public enum HIST_checkInput_e
+{
+    trustInput,
+    checkMaxSymbolValue,
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/HUF_CStream_t.cs

@@ -0,0 +1,22 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public unsafe struct HUF_CStream_t
+{
+    public _bitContainer_e__FixedBuffer bitContainer;
+    public _bitPos_e__FixedBuffer bitPos;
+    public byte* startPtr;
+    public byte* ptr;
+    public byte* endPtr;
+
+    public unsafe struct _bitContainer_e__FixedBuffer
+    {
+        public nuint e0;
+        public nuint e1;
+    }
+
+    public unsafe struct _bitPos_e__FixedBuffer
+    {
+        public nuint e0;
+        public nuint e1;
+    }
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/HUF_CTableHeader.cs

@@ -0,0 +1,8 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public unsafe struct HUF_CTableHeader
+{
+    public byte tableLog;
+    public byte maxSymbolValue;
+    public fixed byte unused[6];
+}

src/SharpCompress/Compressors/ZStandard/Unsafe/HUF_CompressWeightsWksp.cs

@@ -0,0 +1,9 @@
+namespace SharpCompress.Compressors.ZStandard.Unsafe;
+
+public unsafe struct HUF_CompressWeightsWksp
+{
+    public fixed uint CTable[59];
+    public fixed uint scratchBuffer[41];
+    public fixed uint count[13];
+    public fixed short norm[13];
+}