Add and convert Archive.org

This commit is contained in:
Matt Nadareski
2023-09-08 17:26:37 -04:00
parent cf9de66a85
commit 3e5bd7113c
4 changed files with 265 additions and 0 deletions

View File

@@ -0,0 +1,120 @@
using System;
using System.Linq;
using SabreTools.Models.ArchiveDotOrg;
namespace SabreTools.Serialization.CrossModel
{
public partial class ArchiveDotOrg : IModelSerializer<Models.ArchiveDotOrg.Files, Models.Metadata.MetadataFile>
{
/// <inheritdoc/>
#if NET48
public Models.ArchiveDotOrg.Files Deserialize(Models.Metadata.MetadataFile obj)
#else
public Models.ArchiveDotOrg.Files? Deserialize(Models.Metadata.MetadataFile? obj)
#endif
{
if (obj == null)
return null;
var files = new Models.ArchiveDotOrg.Files();
var machines = obj.Read<Models.Metadata.Machine[]>(Models.Metadata.MetadataFile.MachineKey);
if (machines != null && machines.Any())
{
files.File = machines
.Where(m => m != null)
.SelectMany(ConvertFromInternalModel)
.ToArray();
}
return files;
}
/// <summary>
/// Convert from <cref="Models.Metadata.Machine"/> to an array of <cref="Models.ArchiveDotOrg.File"/>
/// </summary>
private static Models.ArchiveDotOrg.File[] ConvertFromInternalModel(Models.Metadata.Machine item)
{
var roms = item.Read<Models.Metadata.Rom[]>(Models.Metadata.Machine.RomKey);
if (roms == null)
return Array.Empty<File>();
return roms
.Where(r => r != null)
.Select(ConvertFromInternalModel).ToArray();
}
/// <summary>
/// Convert from <cref="Models.Metadata.Rom"/> to <cref="Models.ArchiveDotOrg.File"/>
/// </summary>
private static Models.ArchiveDotOrg.File ConvertFromInternalModel(Models.Metadata.Rom item)
{
var file = new Models.ArchiveDotOrg.File
{
Name = item.ReadString(Models.Metadata.Rom.NameKey),
Source = item.ReadString(Models.Metadata.Rom.SourceKey),
BitTorrentMagnetHash = item.ReadString(Models.Metadata.Rom.BitTorrentMagnetHashKey),
LastModifiedTime = item.ReadString(Models.Metadata.Rom.LastModifiedTimeKey),
Size = item.ReadString(Models.Metadata.Rom.SizeKey),
MD5 = item.ReadString(Models.Metadata.Rom.MD5Key),
CRC32 = item.ReadString(Models.Metadata.Rom.CRCKey),
SHA1 = item.ReadString(Models.Metadata.Rom.SHA1Key),
FileCount = item.ReadString(Models.Metadata.Rom.FileCountKey),
Format = item.ReadString(Models.Metadata.Rom.FormatKey),
Original = item.ReadString(Models.Metadata.Rom.OriginalKey),
Summation = item.ReadString(Models.Metadata.Rom.SummationKey),
MatrixNumber = item.ReadString(Models.Metadata.Rom.MatrixNumberKey),
CollectionCatalogNumber = item.ReadString(Models.Metadata.Rom.CollectionCatalogNumberKey),
Comment = item.ReadString(Models.Metadata.Rom.CommentKey),
ASRDetectedLang = item.ReadString(Models.Metadata.Rom.ASRDetectedLangKey),
ASRDetectedLangConf = item.ReadString(Models.Metadata.Rom.ASRDetectedLangConfKey),
ASRTranscribedLang = item.ReadString(Models.Metadata.Rom.ASRTranscribedLangKey),
WhisperASRModuleVersion = item.ReadString(Models.Metadata.Rom.WhisperASRModuleVersionKey),
WhisperModelHash = item.ReadString(Models.Metadata.Rom.WhisperModelHashKey),
WhisperModelName = item.ReadString(Models.Metadata.Rom.WhisperModelNameKey),
WhisperVersion = item.ReadString(Models.Metadata.Rom.WhisperVersionKey),
ClothCoverDetectionModuleVersion = item.ReadString(Models.Metadata.Rom.ClothCoverDetectionModuleVersionKey),
hOCRCharToWordhOCRVersion = item.ReadString(Models.Metadata.Rom.hOCRCharToWordhOCRVersionKey),
hOCRCharToWordModuleVersion = item.ReadString(Models.Metadata.Rom.hOCRCharToWordModuleVersionKey),
hOCRFtsTexthOCRVersion = item.ReadString(Models.Metadata.Rom.hOCRFtsTexthOCRVersionKey),
hOCRFtsTextModuleVersion = item.ReadString(Models.Metadata.Rom.hOCRFtsTextModuleVersionKey),
hOCRPageIndexhOCRVersion = item.ReadString(Models.Metadata.Rom.hOCRPageIndexhOCRVersionKey),
hOCRPageIndexModuleVersion = item.ReadString(Models.Metadata.Rom.hOCRPageIndexModuleVersionKey),
TesseractOCR = item.ReadString(Models.Metadata.Rom.TesseractOCRKey),
TesseractOCRConverted = item.ReadString(Models.Metadata.Rom.TesseractOCRConvertedKey),
TesseractOCRDetectedLang = item.ReadString(Models.Metadata.Rom.TesseractOCRDetectedLangKey),
TesseractOCRDetectedLangConf = item.ReadString(Models.Metadata.Rom.TesseractOCRDetectedLangConfKey),
TesseractOCRDetectedScript = item.ReadString(Models.Metadata.Rom.TesseractOCRDetectedScriptKey),
TesseractOCRDetectedScriptConf = item.ReadString(Models.Metadata.Rom.TesseractOCRDetectedScriptConfKey),
TesseractOCRModuleVersion = item.ReadString(Models.Metadata.Rom.TesseractOCRModuleVersionKey),
TesseractOCRParameters = item.ReadString(Models.Metadata.Rom.TesseractOCRParametersKey),
PDFModuleVersion = item.ReadString(Models.Metadata.Rom.PDFModuleVersionKey),
WordConfidenceInterval0To10 = item.ReadString(Models.Metadata.Rom.WordConfidenceInterval0To10Key),
WordConfidenceInterval11To20 = item.ReadString(Models.Metadata.Rom.WordConfidenceInterval11To20Key),
WordConfidenceInterval21To30 = item.ReadString(Models.Metadata.Rom.WordConfidenceInterval21To30Key),
WordConfidenceInterval31To40 = item.ReadString(Models.Metadata.Rom.WordConfidenceInterval31To40Key),
WordConfidenceInterval41To50 = item.ReadString(Models.Metadata.Rom.WordConfidenceInterval41To50Key),
WordConfidenceInterval51To60 = item.ReadString(Models.Metadata.Rom.WordConfidenceInterval51To60Key),
WordConfidenceInterval61To70 = item.ReadString(Models.Metadata.Rom.WordConfidenceInterval61To70Key),
WordConfidenceInterval71To80 = item.ReadString(Models.Metadata.Rom.WordConfidenceInterval71To80Key),
WordConfidenceInterval81To90 = item.ReadString(Models.Metadata.Rom.WordConfidenceInterval81To90Key),
WordConfidenceInterval91To100 = item.ReadString(Models.Metadata.Rom.WordConfidenceInterval91To100Key),
Album = item.ReadString(Models.Metadata.Rom.AlbumKey),
Artist = item.ReadString(Models.Metadata.Rom.ArtistKey),
Bitrate = item.ReadString(Models.Metadata.Rom.BitrateKey),
Creator = item.ReadString(Models.Metadata.Rom.CreatorKey),
Height = item.ReadString(Models.Metadata.Rom.HeightKey),
Length = item.ReadString(Models.Metadata.Rom.LengthKey),
PreviewImage = item.ReadString(Models.Metadata.Rom.PreviewImageKey),
Rotation = item.ReadString(Models.Metadata.Rom.RotationKey),
Title = item.ReadString(Models.Metadata.Rom.TitleKey),
Track = item.ReadString(Models.Metadata.Rom.TrackKey),
Width = item.ReadString(Models.Metadata.Rom.WidthKey),
};
return file;
}
}
}

View File

@@ -0,0 +1,131 @@
using System.Linq;
namespace SabreTools.Serialization.CrossModel
{
public partial class ArchiveDotOrg : IModelSerializer<Models.ArchiveDotOrg.Files, Models.Metadata.MetadataFile>
{
/// <inheritdoc/>
#if NET48
public Models.Metadata.MetadataFile Serialize(Models.ArchiveDotOrg.Files item)
#else
public static Models.Metadata.MetadataFile? Serialize(Models.ArchiveDotOrg.Files? item)
#endif
{
if (item == null)
return null;
var metadataFile = new Models.Metadata.MetadataFile
{
[Models.Metadata.MetadataFile.HeaderKey] = ConvertHeaderToInternalModel(item),
};
if (item?.File != null && item.File.Any())
{
metadataFile[Models.Metadata.MetadataFile.MachineKey] = item.File
.Where(f => f != null)
.Select(ConvertMachineToInternalModel)
.ToArray();
}
return metadataFile;
}
/// <summary>
/// Convert from <cref="Models.ArchiveDotOrg.Files"/> to <cref="Models.Metadata.Header"/>
/// </summary>
private static Models.Metadata.Header ConvertHeaderToInternalModel(Models.ArchiveDotOrg.Files item)
{
var header = new Models.Metadata.Header
{
[Models.Metadata.Header.NameKey] = "archive.org",
};
return header;
}
/// <summary>
/// Convert from <cref="Models.ArchiveDotOrg.File"/> to <cref="Models.Metadata.Machine"/>
/// </summary>
private static Models.Metadata.Machine ConvertMachineToInternalModel(Models.ArchiveDotOrg.File item)
{
var machine = new Models.Metadata.Machine
{
[Models.Metadata.Machine.RomKey] = ConvertToInternalModel(item),
};
return machine;
}
/// <summary>
/// Convert from <cref="Models.ArchiveDotOrg.File"/> to <cref="Models.Metadata.Rom"/>
/// </summary>
private static Models.Metadata.Rom ConvertToInternalModel(Models.ArchiveDotOrg.File item)
{
var rom = new Models.Metadata.Rom
{
[Models.Metadata.Rom.NameKey] = item.Name,
[Models.Metadata.Rom.SourceKey] = item.Source,
[Models.Metadata.Rom.BitTorrentMagnetHashKey] = item.BitTorrentMagnetHash,
[Models.Metadata.Rom.LastModifiedTimeKey] = item.LastModifiedTime,
[Models.Metadata.Rom.SizeKey] = item.Size,
[Models.Metadata.Rom.MD5Key] = item.MD5,
[Models.Metadata.Rom.CRCKey] = item.CRC32,
[Models.Metadata.Rom.SHA1Key] = item.SHA1,
[Models.Metadata.Rom.FileCountKey] = item.FileCount,
[Models.Metadata.Rom.FormatKey] = item.Format,
[Models.Metadata.Rom.OriginalKey] = item.Original,
[Models.Metadata.Rom.SummationKey] = item.Summation,
[Models.Metadata.Rom.MatrixNumberKey] = item.MatrixNumber,
[Models.Metadata.Rom.CollectionCatalogNumberKey] = item.CollectionCatalogNumber,
[Models.Metadata.Rom.PublisherKey] = item.Publisher,
[Models.Metadata.Rom.CommentKey] = item.Comment,
[Models.Metadata.Rom.ASRDetectedLangKey] = item.ASRDetectedLang,
[Models.Metadata.Rom.ASRDetectedLangConfKey] = item.ASRDetectedLangConf,
[Models.Metadata.Rom.ASRTranscribedLangKey] = item.ASRTranscribedLang,
[Models.Metadata.Rom.WhisperASRModuleVersionKey] = item.WhisperASRModuleVersion,
[Models.Metadata.Rom.WhisperModelHashKey] = item.WhisperModelHash,
[Models.Metadata.Rom.WhisperModelNameKey] = item.WhisperModelName,
[Models.Metadata.Rom.WhisperVersionKey] = item.WhisperVersion,
[Models.Metadata.Rom.ClothCoverDetectionModuleVersionKey] = item.ClothCoverDetectionModuleVersion,
[Models.Metadata.Rom.hOCRCharToWordhOCRVersionKey] = item.hOCRCharToWordhOCRVersion,
[Models.Metadata.Rom.hOCRCharToWordModuleVersionKey] = item.hOCRCharToWordModuleVersion,
[Models.Metadata.Rom.hOCRFtsTexthOCRVersionKey] = item.hOCRFtsTexthOCRVersion,
[Models.Metadata.Rom.hOCRFtsTextModuleVersionKey] = item.hOCRFtsTextModuleVersion,
[Models.Metadata.Rom.hOCRPageIndexhOCRVersionKey] = item.hOCRPageIndexhOCRVersion,
[Models.Metadata.Rom.hOCRPageIndexModuleVersionKey] = item.hOCRPageIndexModuleVersion,
[Models.Metadata.Rom.TesseractOCRKey] = item.TesseractOCR,
[Models.Metadata.Rom.TesseractOCRConvertedKey] = item.TesseractOCRConverted,
[Models.Metadata.Rom.TesseractOCRDetectedLangKey] = item.TesseractOCRDetectedLang,
[Models.Metadata.Rom.TesseractOCRDetectedLangConfKey] = item.TesseractOCRDetectedLangConf,
[Models.Metadata.Rom.TesseractOCRDetectedScriptKey] = item.TesseractOCRDetectedScript,
[Models.Metadata.Rom.TesseractOCRDetectedScriptConfKey] = item.TesseractOCRDetectedScriptConf,
[Models.Metadata.Rom.TesseractOCRModuleVersionKey] = item.TesseractOCRModuleVersion,
[Models.Metadata.Rom.TesseractOCRParametersKey] = item.TesseractOCRParameters,
[Models.Metadata.Rom.PDFModuleVersionKey] = item.PDFModuleVersion,
[Models.Metadata.Rom.WordConfidenceInterval0To10Key] = item.WordConfidenceInterval0To10,
[Models.Metadata.Rom.WordConfidenceInterval11To20Key] = item.WordConfidenceInterval11To20,
[Models.Metadata.Rom.WordConfidenceInterval21To30Key] = item.WordConfidenceInterval21To30,
[Models.Metadata.Rom.WordConfidenceInterval31To40Key] = item.WordConfidenceInterval31To40,
[Models.Metadata.Rom.WordConfidenceInterval41To50Key] = item.WordConfidenceInterval41To50,
[Models.Metadata.Rom.WordConfidenceInterval51To60Key] = item.WordConfidenceInterval51To60,
[Models.Metadata.Rom.WordConfidenceInterval61To70Key] = item.WordConfidenceInterval61To70,
[Models.Metadata.Rom.WordConfidenceInterval71To80Key] = item.WordConfidenceInterval71To80,
[Models.Metadata.Rom.WordConfidenceInterval81To90Key] = item.WordConfidenceInterval81To90,
[Models.Metadata.Rom.WordConfidenceInterval91To100Key] = item.WordConfidenceInterval91To100,
[Models.Metadata.Rom.AlbumKey] = item.Album,
[Models.Metadata.Rom.ArtistKey] = item.Artist,
[Models.Metadata.Rom.BitrateKey] = item.Bitrate,
[Models.Metadata.Rom.CreatorKey] = item.Creator,
[Models.Metadata.Rom.HeightKey] = item.Height,
[Models.Metadata.Rom.LengthKey] = item.Length,
[Models.Metadata.Rom.PreviewImageKey] = item.PreviewImage,
[Models.Metadata.Rom.RotationKey] = item.Rotation,
[Models.Metadata.Rom.TitleKey] = item.Title,
[Models.Metadata.Rom.TrackKey] = item.Track,
[Models.Metadata.Rom.WidthKey] = item.Width,
};
return rom;
}
}
}

View File

@@ -0,0 +1,7 @@
namespace SabreTools.Serialization.Files
{
public partial class ArchiveDotOrg : XmlFile<Models.ArchiveDotOrg.Files>
{
// All serialization logic is in the base class
}
}

View File

@@ -0,0 +1,7 @@
namespace SabreTools.Serialization.Streams
{
public partial class ArchiveDotOrg : XmlFile<Models.ArchiveDotOrg.Files>
{
// All serialization logic is in the base class
}
}