[ArchiveTools, SimpleSort] Add magic number matching, make archive scanning more accurate

This commit is contained in:
Matt Nadareski
2016-06-17 20:03:07 -07:00
parent 18ce3860d0
commit 1e95a3515e
3 changed files with 124 additions and 97 deletions

View File

@@ -63,5 +63,16 @@ namespace SabreTools.Helper
public static long GigaByte = (long)Math.Pow(KiloByte, 2); public static long GigaByte = (long)Math.Pow(KiloByte, 2);
public static long TeraByte = (long)Math.Pow(KiloByte, 2); public static long TeraByte = (long)Math.Pow(KiloByte, 2);
public static long PetaByte = (long)Math.Pow(KiloByte, 2); public static long PetaByte = (long)Math.Pow(KiloByte, 2);
// Magic numbers as strings
public static string SevenZipSig = "377ABCAF271C";
public static string GzSig = "1F8B";
public static string RarSig = "526172211A0700";
public static string RarFiveSig = "526172211A070100";
public static string TarSig = "7573746172202000";
public static string TarZeroSig = "7573746172003030";
public static string ZipSig = "504B0304";
public static string ZipSigEmpty = "504B0506";
public static string ZipSigSpanned = "504B0708";
} }
} }

View File

@@ -454,19 +454,48 @@ namespace SabreTools.Helper
{ {
ArchiveType? outtype = null; ArchiveType? outtype = null;
IReader reader = null; // Read the first bytes of the file and get the magic numbe
try try
{ {
reader = ReaderFactory.Open(File.OpenRead(input)); byte[] magic = new byte[8];
outtype = reader.ArchiveType; using (BinaryReader br = new BinaryReader(File.OpenRead(input)))
{
magic = br.ReadBytes(8);
}
// Convert it to an uppercase string
string mstr = string.Empty;
for (int i = 0; i < magic.Length; i++)
{
mstr += BitConverter.ToString(new byte[] { magic[i] });
}
mstr = mstr.ToUpperInvariant();
// Now try to match it to a known signature
if (mstr.StartsWith(Constants.SevenZipSig))
{
outtype = ArchiveType.SevenZip;
}
else if (mstr.StartsWith(Constants.GzSig))
{
outtype = ArchiveType.GZip;
}
else if (mstr.StartsWith(Constants.RarSig) || mstr.StartsWith(Constants.RarFiveSig))
{
outtype = ArchiveType.Rar;
}
else if (mstr.StartsWith(Constants.TarSig) || mstr.StartsWith(Constants.TarZeroSig))
{
outtype = ArchiveType.Tar;
}
else if (mstr.StartsWith(Constants.ZipSig) || mstr.StartsWith(Constants.ZipSigEmpty) || mstr.StartsWith(Constants.ZipSigSpanned))
{
outtype = ArchiveType.Zip;
}
} }
catch (Exception) catch (Exception)
{ {
// Don't log archive open errors // Don't log file open errors
}
finally
{
reader?.Dispose();
} }
return outtype; return outtype;

View File

@@ -1,5 +1,4 @@
using SabreTools.Helper; using SabreTools.Helper;
using SharpCompress.Archive;
using SharpCompress.Common; using SharpCompress.Common;
using System; using System;
using System.Collections.Generic; using System.Collections.Generic;
@@ -14,7 +13,7 @@ namespace SabreTools
private List<string> _inputs; private List<string> _inputs;
private string _outdir; private string _outdir;
private string _tempdir; private string _tempdir;
private bool _externalScan; private bool _quickScan;
private ArchiveScanLevel _7z; private ArchiveScanLevel _7z;
private ArchiveScanLevel _gz; private ArchiveScanLevel _gz;
private ArchiveScanLevel _rar; private ArchiveScanLevel _rar;
@@ -28,20 +27,20 @@ namespace SabreTools
/// <param name="inputs">List of input files/folders to check</param> /// <param name="inputs">List of input files/folders to check</param>
/// <param name="outdir">Output directory to use to build to</param> /// <param name="outdir">Output directory to use to build to</param>
/// <param name="tempdir">Temporary directory for archive extraction</param> /// <param name="tempdir">Temporary directory for archive extraction</param>
/// <param name="externalScan">True to enable external scanning of archives, false otherwise</param> /// <param name="quickScan">True to enable external scanning of archives, false otherwise</param>
/// <param name="sevenzip">Integer representing the archive handling level for 7z</param> /// <param name="sevenzip">Integer representing the archive handling level for 7z</param>
/// <param name="gz">Integer representing the archive handling level for GZip</param> /// <param name="gz">Integer representing the archive handling level for GZip</param>
/// <param name="rar">Integer representing the archive handling level for RAR</param> /// <param name="rar">Integer representing the archive handling level for RAR</param>
/// <param name="zip">Integer representing the archive handling level for Zip</param> /// <param name="zip">Integer representing the archive handling level for Zip</param>
/// <param name="logger">Logger object for file and console output</param> /// <param name="logger">Logger object for file and console output</param>
public SimpleSort(Dat datdata, List<string> inputs, string outdir, string tempdir, public SimpleSort(Dat datdata, List<string> inputs, string outdir, string tempdir,
bool externalScan, int sevenzip, int gz, int rar, int zip, Logger logger) bool quickScan, int sevenzip, int gz, int rar, int zip, Logger logger)
{ {
_datdata = datdata; _datdata = datdata;
_inputs = inputs; _inputs = inputs;
_outdir = (outdir == "" ? "Rebuild" : outdir); _outdir = (outdir == "" ? "Rebuild" : outdir);
_tempdir = (tempdir == "" ? "__TEMP__" : tempdir); _tempdir = (tempdir == "" ? "__TEMP__" : tempdir);
_externalScan = externalScan; _quickScan = quickScan;
_7z = (ArchiveScanLevel)(sevenzip < 0 || sevenzip > 2 ? 0 : sevenzip); _7z = (ArchiveScanLevel)(sevenzip < 0 || sevenzip > 2 ? 0 : sevenzip);
_gz = (ArchiveScanLevel)(gz < 0 || gz > 2 ? 0 : gz); _gz = (ArchiveScanLevel)(gz < 0 || gz > 2 ? 0 : gz);
_rar = (ArchiveScanLevel)(rar < 0 || rar > 2 ? 0 : rar); _rar = (ArchiveScanLevel)(rar < 0 || rar > 2 ? 0 : rar);
@@ -85,7 +84,7 @@ namespace SabreTools
// Set all default values // Set all default values
bool help = false, bool help = false,
externalScan = false, quickScan = false,
simpleSort = true; simpleSort = true;
int sevenzip = 0, int sevenzip = 0,
gz = 2, gz = 2,
@@ -108,7 +107,7 @@ namespace SabreTools
break; break;
case "-qs": case "-qs":
case "--quick": case "--quick":
externalScan = true; quickScan = true;
break; break;
default: default:
string temparg = arg.Replace("\"", "").Replace("file://", ""); string temparg = arg.Replace("\"", "").Replace("file://", "");
@@ -200,7 +199,7 @@ namespace SabreTools
{ {
if (datfiles.Count > 0) if (datfiles.Count > 0)
{ {
InitSimpleSort(datfiles, inputs, outdir, tempdir, externalScan, sevenzip, gz, rar, zip, logger); InitSimpleSort(datfiles, inputs, outdir, tempdir, quickScan, sevenzip, gz, rar, zip, logger);
} }
else else
{ {
@@ -228,14 +227,14 @@ namespace SabreTools
/// <param name="inputs">List of input files/folders to check</param> /// <param name="inputs">List of input files/folders to check</param>
/// <param name="outdir">Output directory to use to build to</param> /// <param name="outdir">Output directory to use to build to</param>
/// <param name="tempdir">Temporary directory for archive extraction</param> /// <param name="tempdir">Temporary directory for archive extraction</param>
/// <param name="externalScan">True to enable external scanning of archives, false otherwise</param> /// <param name="quickScan">True to enable external scanning of archives, false otherwise</param>
/// <param name="sevenzip">Integer representing the archive handling level for 7z</param> /// <param name="sevenzip">Integer representing the archive handling level for 7z</param>
/// <param name="gz">Integer representing the archive handling level for GZip</param> /// <param name="gz">Integer representing the archive handling level for GZip</param>
/// <param name="rar">Integer representing the archive handling level for RAR</param> /// <param name="rar">Integer representing the archive handling level for RAR</param>
/// <param name="zip">Integer representing the archive handling level for Zip</param> /// <param name="zip">Integer representing the archive handling level for Zip</param>
/// <param name="logger">Logger object for file and console output</param> /// <param name="logger">Logger object for file and console output</param>
private static void InitSimpleSort(List<string> datfiles, List<string> inputs, string outdir, string tempdir, private static void InitSimpleSort(List<string> datfiles, List<string> inputs, string outdir, string tempdir,
bool externalScan, int sevenzip, int gz, int rar, int zip, Logger logger) bool quickScan, int sevenzip, int gz, int rar, int zip, Logger logger)
{ {
// Add all of the input DATs into one huge internal DAT // Add all of the input DATs into one huge internal DAT
Dat datdata = new Dat(); Dat datdata = new Dat();
@@ -244,7 +243,7 @@ namespace SabreTools
datdata = DatTools.Parse(datfile, 0, 0, datdata, logger); datdata = DatTools.Parse(datfile, 0, 0, datdata, logger);
} }
SimpleSort ss = new SimpleSort(datdata, inputs, outdir, tempdir, externalScan, sevenzip, gz, rar, zip, logger); SimpleSort ss = new SimpleSort(datdata, inputs, outdir, tempdir, quickScan, sevenzip, gz, rar, zip, logger);
ss.RebuildToFolder(); ss.RebuildToFolder();
} }
@@ -329,34 +328,37 @@ namespace SabreTools
input = Path.GetFullPath(input); input = Path.GetFullPath(input);
_logger.User("Beginning processing of '" + input + "'"); _logger.User("Beginning processing of '" + input + "'");
// If we have an archive, scan it if necessary // Get if the file should be scanned internally and externally
bool shouldscan = true; bool shouldExternalScan = true;
try bool shouldInternalScan = true;
{
IArchive temp = ArchiveFactory.Open(input); ArchiveType? archiveType = ArchiveTools.GetCurrentArchiveType(input, _logger);
switch (temp.Type) switch (archiveType)
{ {
case null:
shouldExternalScan = true;
shouldInternalScan = false;
break;
case ArchiveType.GZip: case ArchiveType.GZip:
shouldscan = (_gz != ArchiveScanLevel.Internal); shouldExternalScan = (_gz != ArchiveScanLevel.Internal);
shouldInternalScan = (_gz != ArchiveScanLevel.External);
break; break;
case ArchiveType.Rar: case ArchiveType.Rar:
shouldscan = (_rar != ArchiveScanLevel.Internal); shouldExternalScan = (_rar != ArchiveScanLevel.Internal);
shouldInternalScan = (_rar != ArchiveScanLevel.External);
break; break;
case ArchiveType.SevenZip: case ArchiveType.SevenZip:
shouldscan = (_7z != ArchiveScanLevel.Internal); shouldExternalScan = (_7z != ArchiveScanLevel.Internal);
shouldInternalScan = (_7z != ArchiveScanLevel.External);
break; break;
case ArchiveType.Zip: case ArchiveType.Zip:
shouldscan = (_zip != ArchiveScanLevel.Internal); shouldExternalScan = (_zip != ArchiveScanLevel.Internal);
shouldInternalScan = (_zip != ArchiveScanLevel.External);
break; break;
} }
}
catch
{
shouldscan = true;
}
// Hash and match the external files // Hash and match the external files
if (shouldscan) if (shouldExternalScan)
{ {
Rom rom = RomTools.GetSingleFileInfo(input); Rom rom = RomTools.GetSingleFileInfo(input);
@@ -421,8 +423,11 @@ namespace SabreTools
} }
} }
// If we should scan the file as an archive
if (shouldInternalScan)
{
// If external scanning is enabled, use that method instead // If external scanning is enabled, use that method instead
if (_externalScan) if (_quickScan)
{ {
_logger.Log("Beginning quick scan of contents from '" + input + "'"); _logger.Log("Beginning quick scan of contents from '" + input + "'");
List<Rom> internalRomData = ArchiveTools.GetArchiveFileInfo(input, _logger); List<Rom> internalRomData = ArchiveTools.GetArchiveFileInfo(input, _logger);
@@ -458,25 +463,6 @@ namespace SabreTools
} }
else else
{ {
// If the file isn't an archive, skip out sooner
if (ArchiveTools.GetCurrentArchiveType(input, _logger) == null)
{
// Remove the current file if we are in recursion so it's not picked up in the next step
if (recurse)
{
try
{
File.Delete(input);
}
catch (Exception)
{
// Don't log file deletion errors
}
}
return success;
}
// Now, if the file is a supported archive type, also run on all files within // Now, if the file is a supported archive type, also run on all files within
bool encounteredErrors = !ArchiveTools.ExtractArchive(input, _tempdir, _7z, _gz, _rar, _zip, _logger); bool encounteredErrors = !ArchiveTools.ExtractArchive(input, _tempdir, _7z, _gz, _rar, _zip, _logger);
@@ -503,6 +489,7 @@ namespace SabreTools
} }
} }
} }
}
return success; return success;
} }
@@ -557,7 +544,7 @@ namespace SabreTools
List<Rom> roms = new List<Rom>(); List<Rom> roms = new List<Rom>();
// If we are in quickscan, get the list of roms that way // If we are in quickscan, get the list of roms that way
if (_externalScan) if (_quickScan)
{ {
roms = ArchiveTools.GetArchiveFileInfo(Path.GetFullPath(archive), _logger); roms = ArchiveTools.GetArchiveFileInfo(Path.GetFullPath(archive), _logger);
} }