2023-08-10 11:35:32 -04:00
|
|
|
using System;
|
2023-08-09 21:46:32 -04:00
|
|
|
using System.Linq;
|
|
|
|
|
using SabreTools.Models.ArchiveDotOrg;
|
|
|
|
|
|
|
|
|
|
namespace SabreTools.Serialization
|
|
|
|
|
{
|
|
|
|
|
/// <summary>
|
|
|
|
|
/// XML deserializer for archive.org metadata files
|
|
|
|
|
/// </summary>
|
|
|
|
|
public partial class ArchiveDotOrg : XmlSerializer<Files>
|
|
|
|
|
{
|
|
|
|
|
#region Internal
|
|
|
|
|
|
2023-08-10 00:59:36 -04:00
|
|
|
/// <summary>
|
2023-09-04 23:51:37 -04:00
|
|
|
/// Convert from <cref="Models.Metadata.MetadataFile"/> to an array of <cref="Models.ArchiveDotOrg.Files"/>
|
2023-08-10 00:59:36 -04:00
|
|
|
/// </summary>
|
2023-09-04 23:51:37 -04:00
|
|
|
public static Files? ConvertFromInternalModel(Models.Metadata.MetadataFile? item)
|
2023-08-10 00:59:36 -04:00
|
|
|
{
|
|
|
|
|
if (item == null)
|
|
|
|
|
return null;
|
|
|
|
|
|
|
|
|
|
var files = new Files();
|
|
|
|
|
|
2023-09-04 23:51:37 -04:00
|
|
|
var machines = item.Read<Models.Metadata.Machine[]>(Models.Metadata.MetadataFile.MachineKey);
|
2023-08-10 00:59:36 -04:00
|
|
|
if (machines != null && machines.Any())
|
2023-08-10 11:35:32 -04:00
|
|
|
{
|
|
|
|
|
files.File = machines
|
|
|
|
|
.Where(m => m != null)
|
|
|
|
|
.SelectMany(ConvertFromInternalModel)
|
|
|
|
|
.ToArray();
|
|
|
|
|
}
|
2023-08-10 00:59:36 -04:00
|
|
|
|
|
|
|
|
return files;
|
|
|
|
|
}
|
|
|
|
|
|
2023-08-09 21:46:32 -04:00
|
|
|
/// <summary>
|
2023-09-04 23:51:37 -04:00
|
|
|
/// Convert from <cref="Models.Metadata.Machine"/> to an array of <cref="Models.ArchiveDotOrg.File"/>
|
2023-08-09 21:46:32 -04:00
|
|
|
/// </summary>
|
2023-09-04 23:51:37 -04:00
|
|
|
private static File[] ConvertFromInternalModel(Models.Metadata.Machine item)
|
2023-08-09 21:46:32 -04:00
|
|
|
{
|
2023-09-04 23:51:37 -04:00
|
|
|
var roms = item.Read<Models.Metadata.Rom[]>(Models.Metadata.Machine.RomKey);
|
2023-08-10 11:35:32 -04:00
|
|
|
if (roms == null)
|
|
|
|
|
return Array.Empty<File>();
|
|
|
|
|
|
|
|
|
|
return roms
|
|
|
|
|
.Where(r => r != null)
|
|
|
|
|
.Select(ConvertFromInternalModel).ToArray();
|
2023-08-09 21:46:32 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// <summary>
|
2023-09-04 23:51:37 -04:00
|
|
|
/// Convert from <cref="Models.Metadata.Rom"/> to <cref="Models.ArchiveDotOrg.File"/>
|
2023-08-09 21:46:32 -04:00
|
|
|
/// </summary>
|
2023-09-04 23:51:37 -04:00
|
|
|
private static File ConvertFromInternalModel(Models.Metadata.Rom item)
|
2023-08-09 21:46:32 -04:00
|
|
|
{
|
2023-08-09 21:55:57 -04:00
|
|
|
var file = new File
|
2023-08-09 21:46:32 -04:00
|
|
|
{
|
2023-09-04 23:51:37 -04:00
|
|
|
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),
|
2023-08-09 21:46:32 -04:00
|
|
|
|
2023-09-04 23:51:37 -04:00
|
|
|
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),
|
2023-08-09 21:46:32 -04:00
|
|
|
|
2023-09-04 23:51:37 -04:00
|
|
|
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),
|
2023-08-09 21:46:32 -04:00
|
|
|
|
2023-09-04 23:51:37 -04:00
|
|
|
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),
|
2023-08-09 21:46:32 -04:00
|
|
|
};
|
2023-08-09 21:55:57 -04:00
|
|
|
return file;
|
2023-08-09 21:46:32 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#endregion
|
|
|
|
|
}
|
|
|
|
|
}
|