mirror of
https://github.com/claunia/SabreTools.git
synced 2025-12-16 19:14:27 +00:00
Update archive.org deserialization test, fix issues
This commit is contained in:
Matt Nadareski
@@ -45,9 +45,42 @@ namespace SabreTools.Models.ArchiveDotOrg
|
||||
[XmlElement("summation")]
|
||||
public string? Summation { get; set; }
|
||||
|
||||
/// <remarks>Is this a set of defined values?</remarks>
|
||||
[XmlElement("rotation")]
|
||||
public long? Rotation { get; set; }
|
||||
[XmlElement("matrix_number")]
|
||||
public string? MatrixNumber { get; set; }
|
||||
|
||||
[XmlElement("collection-catalog-number")]
|
||||
public string? CollectionCatalogNumber { get; set; }
|
||||
|
||||
[XmlElement("publisher")]
|
||||
public string? Publisher { get; set; }
|
||||
|
||||
[XmlElement("comment")]
|
||||
public string? Comment { get; set; }
|
||||
|
||||
#region ASR-Related
|
||||
|
||||
[XmlElement("asr_detected_lang")]
|
||||
public string? ASRDetectedLang { get; set; }
|
||||
|
||||
[XmlElement("asr_detected_lang_conf")]
|
||||
public string? ASRDetectedLangConf { get; set; }
|
||||
|
||||
[XmlElement("asr_transcribed_lang")]
|
||||
public string? ASRTranscribedLang { get; set; }
|
||||
|
||||
[XmlElement("whisper_asr_module_version")]
|
||||
public string? WhisperASRModuleVersion { get; set; }
|
||||
|
||||
[XmlElement("whisper_model_hash")]
|
||||
public string? WhisperModelHash { get; set; }
|
||||
|
||||
[XmlElement("whisper_model_name")]
|
||||
public string? WhisperModelName { get; set; }
|
||||
|
||||
[XmlElement("whisper_version")]
|
||||
public string? WhisperVersion { get; set; }
|
||||
|
||||
#endregion
|
||||
|
||||
#region OCR-Related
|
||||
|
||||
@@ -131,6 +164,44 @@ namespace SabreTools.Models.ArchiveDotOrg
|
||||
|
||||
#endregion
|
||||
|
||||
#region Media-Related
|
||||
|
||||
[XmlElement("album")]
|
||||
public string? Album { get; set; }
|
||||
|
||||
[XmlElement("artist")]
|
||||
public string? Artist { get; set; }
|
||||
|
||||
[XmlElement("bitrate")]
|
||||
public long? Bitrate { get; set; }
|
||||
|
||||
[XmlElement("creator")]
|
||||
public string? Creator { get; set; }
|
||||
|
||||
[XmlElement("height")]
|
||||
public long? Height { get; set; }
|
||||
|
||||
[XmlElement("length")]
|
||||
public double? Length { get; set; }
|
||||
|
||||
[XmlElement("preview-image")]
|
||||
public string? PreviewImage { get; set; }
|
||||
|
||||
/// <remarks>Is this a set of defined values?</remarks>
|
||||
[XmlElement("rotation")]
|
||||
public long? Rotation { get; set; }
|
||||
|
||||
[XmlElement("title")]
|
||||
public string? Title { get; set; }
|
||||
|
||||
[XmlElement("track")]
|
||||
public long? Track { get; set; }
|
||||
|
||||
[XmlElement("width")]
|
||||
public long? Width { get; set; }
|
||||
|
||||
#endregion
|
||||
|
||||
#region DO NOT USE IN PRODUCTION
|
||||
|
||||
/// <remarks>Should be empty</remarks>
|
||||
|
||||
@@ -11,7 +11,11 @@ namespace SabreTools.Test.Parser
|
||||
public class DeserializationTests
|
||||
{
|
||||
[Theory]
|
||||
[InlineData("test-archivedotorg-files.xml", 22)]
|
||||
[InlineData("test-archivedotorg-files1.xml", 22)]
|
||||
[InlineData("test-archivedotorg-files2.xml", 13)]
|
||||
[InlineData("test-archivedotorg-files3.xml", 21)]
|
||||
[InlineData("test-archivedotorg-files4.xml", 19)]
|
||||
[InlineData("test-archivedotorg-files5.xml", 1390)]
|
||||
public void ArchiveDotOrgDeserializeTest(string path, long count)
|
||||
{
|
||||
// Open the file for reading
|
||||
|
||||
119
SabreTools.Test/TestData/test-archivedotorg-files2.xml
Normal file
119
SabreTools.Test/TestData/test-archivedotorg-files2.xml
Normal file
@@ -0,0 +1,119 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<files>
|
||||
<file name="I have grown so used to you - J. Frank Hopkins.gif" source="derivative">
|
||||
<format>Animated GIF</format>
|
||||
<original>I have grown so used to you - J. Frank Hopkins.mp4</original>
|
||||
<mtime>1446681646</mtime>
|
||||
<size>83293</size>
|
||||
<md5>d60eddbc1d9915dc3a49889890c172d1</md5>
|
||||
<crc32>30fe3c3e</crc32>
|
||||
<sha1>80f7f6b523965b9dc4f997a775d03aab7f6dfcfb</sha1>
|
||||
</file>
|
||||
<file name="I have grown so used to you - J. Frank Hopkins.mp4" source="original">
|
||||
<mtime>1446681528</mtime>
|
||||
<size>69339907</size>
|
||||
<md5>867ad802b751623b409eb833a15d0b80</md5>
|
||||
<crc32>dbc81bfe</crc32>
|
||||
<sha1>5e860bdb6cad0b7f974ebe9ea391e26f4cd8c714</sha1>
|
||||
<format>MPEG4</format>
|
||||
<length>140.42</length>
|
||||
<height>720</height>
|
||||
<width>1280</width>
|
||||
</file>
|
||||
<file name="I have grown so used to you - J. Frank Hopkins.ogv" source="derivative">
|
||||
<format>Ogg Video</format>
|
||||
<original>I have grown so used to you - J. Frank Hopkins.mp4</original>
|
||||
<mtime>1446682043</mtime>
|
||||
<size>10548875</size>
|
||||
<md5>9a9dc6524b8ca0be68d1b0c3e9b2d0b2</md5>
|
||||
<crc32>5a17284d</crc32>
|
||||
<sha1>36b7e06adf4cf4f8fc159dba380dae9e632e410a</sha1>
|
||||
<length>140.42</length>
|
||||
<height>300</height>
|
||||
<width>533</width>
|
||||
</file>
|
||||
<file name="IHaveGrownSoUsedToYouJ.FrankHopkins.thumbs/I have grown so used to you - J. Frank Hopkins_000001.jpg" source="derivative">
|
||||
<format>Thumbnail</format>
|
||||
<original>I have grown so used to you - J. Frank Hopkins.mp4</original>
|
||||
<mtime>1446681658</mtime>
|
||||
<size>8382</size>
|
||||
<md5>dbb3838f9e348938ffabf8570809bd1b</md5>
|
||||
<crc32>ca0b9e96</crc32>
|
||||
<sha1>1efa60020eb9b0441256d00d51c1ef208cb50047</sha1>
|
||||
</file>
|
||||
<file name="IHaveGrownSoUsedToYouJ.FrankHopkins.thumbs/I have grown so used to you - J. Frank Hopkins_000045.jpg" source="derivative">
|
||||
<format>Thumbnail</format>
|
||||
<original>I have grown so used to you - J. Frank Hopkins.mp4</original>
|
||||
<mtime>1446681676</mtime>
|
||||
<size>8244</size>
|
||||
<md5>295c4e5fe1b56e0438008c60d28213b7</md5>
|
||||
<crc32>441108d4</crc32>
|
||||
<sha1>388801815d2aa0fa98c4a75589f73faeb958071f</sha1>
|
||||
</file>
|
||||
<file name="IHaveGrownSoUsedToYouJ.FrankHopkins.thumbs/I have grown so used to you - J. Frank Hopkins_000075.jpg" source="derivative">
|
||||
<format>Thumbnail</format>
|
||||
<original>I have grown so used to you - J. Frank Hopkins.mp4</original>
|
||||
<mtime>1446681693</mtime>
|
||||
<size>8321</size>
|
||||
<md5>94a0c9ae786011773c29be2bb844bd70</md5>
|
||||
<crc32>f09fea6a</crc32>
|
||||
<sha1>439476ad109f2f1edd0b85c2c04b5c6dfdfe0bce</sha1>
|
||||
</file>
|
||||
<file name="IHaveGrownSoUsedToYouJ.FrankHopkins.thumbs/I have grown so used to you - J. Frank Hopkins_000105.jpg" source="derivative">
|
||||
<format>Thumbnail</format>
|
||||
<original>I have grown so used to you - J. Frank Hopkins.mp4</original>
|
||||
<mtime>1446681710</mtime>
|
||||
<size>8342</size>
|
||||
<md5>a284cbd6a9288e29b7b03118c6658816</md5>
|
||||
<crc32>62785a42</crc32>
|
||||
<sha1>44dd6f1df0c681f410076035bc7b8b3bae81bff4</sha1>
|
||||
</file>
|
||||
<file name="IHaveGrownSoUsedToYouJ.FrankHopkins.thumbs/I have grown so used to you - J. Frank Hopkins_000135.jpg" source="derivative">
|
||||
<format>Thumbnail</format>
|
||||
<original>I have grown so used to you - J. Frank Hopkins.mp4</original>
|
||||
<mtime>1446681726</mtime>
|
||||
<size>8586</size>
|
||||
<md5>0f0d08441fc414794063b7e07d5d0947</md5>
|
||||
<crc32>dc6e44cf</crc32>
|
||||
<sha1>cfac85033a4d0a45cc1a5c8ffd2fb37263ecfa8e</sha1>
|
||||
</file>
|
||||
<file name="IHaveGrownSoUsedToYouJ.FrankHopkins_archive.torrent" source="metadata">
|
||||
<btih>65d70d3788efdc9fba994ac2abf4a82c3f3d3864</btih>
|
||||
<mtime>1688963806</mtime>
|
||||
<size>5514</size>
|
||||
<md5>5279f6df87259e51a00ef537cc91ce9b</md5>
|
||||
<crc32>eedeca57</crc32>
|
||||
<sha1>26830f2b391cd8ed60f5d793d87016a0f1496bdc</sha1>
|
||||
<format>Archive BitTorrent</format>
|
||||
</file>
|
||||
<file name="IHaveGrownSoUsedToYouJ.FrankHopkins_files.xml" source="original">
|
||||
<format>Metadata</format>
|
||||
<md5>954d2f3d58262e80c87a2f623a59200e</md5>
|
||||
<summation>md5</summation>
|
||||
</file>
|
||||
<file name="IHaveGrownSoUsedToYouJ.FrankHopkins_meta.sqlite" source="original">
|
||||
<mtime>1446681548</mtime>
|
||||
<size>12288</size>
|
||||
<md5>1da7d006f8f27b4fc9a1fd74941095cf</md5>
|
||||
<crc32>1308a8d9</crc32>
|
||||
<sha1>f31849351ed620108d7f3bb4af5aa5e40b4cdb7b</sha1>
|
||||
<format>Metadata</format>
|
||||
</file>
|
||||
<file name="IHaveGrownSoUsedToYouJ.FrankHopkins_meta.xml" source="original">
|
||||
<mtime>1682350692</mtime>
|
||||
<size>1161</size>
|
||||
<format>Metadata</format>
|
||||
<md5>740d1aa686af09659361ce8c10b6fcfe</md5>
|
||||
<crc32>3d03262a</crc32>
|
||||
<sha1>9272d98773615f94040f799d2341ba45c13d6218</sha1>
|
||||
</file>
|
||||
<file name="__ia_thumb.jpg" source="original">
|
||||
<mtime>1656366463</mtime>
|
||||
<size>9378</size>
|
||||
<md5>34dd181b0af401cfc3205b503ab63f48</md5>
|
||||
<crc32>72e288dc</crc32>
|
||||
<sha1>6bafe89be5f45e68eb641ae3aca960baa759bfe5</sha1>
|
||||
<format>Item Tile</format>
|
||||
<rotation>0</rotation>
|
||||
</file>
|
||||
</files>
|
||||
215
SabreTools.Test/TestData/test-archivedotorg-files3.xml
Normal file
215
SabreTools.Test/TestData/test-archivedotorg-files3.xml
Normal file
@@ -0,0 +1,215 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<files>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a.jpg" source="derivative">
|
||||
<format>JPEG</format>
|
||||
<original>78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a.tif</original>
|
||||
<mtime>1501911616</mtime>
|
||||
<size>1326621</size>
|
||||
<md5>b50a8cc01a720df337cebbcb69e1cddf</md5>
|
||||
<crc32>b50088ba</crc32>
|
||||
<sha1>672d2922dff5a0954558b97d13301394870a6cc5</sha1>
|
||||
</file>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a.tif" source="original">
|
||||
<mtime>1501911368</mtime>
|
||||
<size>37016416</size>
|
||||
<md5>77176ab4e12418558aefa6f4b709737a</md5>
|
||||
<crc32>3bbd199a</crc32>
|
||||
<sha1>64beda28ffd02a67b4c69e33e4b18a808b63453e</sha1>
|
||||
<format>TIFF</format>
|
||||
<matrix_number>69660BB</matrix_number>
|
||||
<collection-catalog-number>GBIA0011964A</collection-catalog-number>
|
||||
<publisher>Decca</publisher>
|
||||
</file>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a.zip" source="original">
|
||||
<mtime>1684102561</mtime>
|
||||
<size>575486142</size>
|
||||
<md5>bd7f9d1d873b870c8b83f8b82ee6c503</md5>
|
||||
<crc32>d1186398</crc32>
|
||||
<sha1>1405bf17b31c0f25652399e820faa001922930a2</sha1>
|
||||
<format>ZIP</format>
|
||||
<filecount>9</filecount>
|
||||
</file>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a_archive.torrent" source="metadata">
|
||||
<btih>2abc74cbab976a03ca6afdcffb9fecfbf19c6f70</btih>
|
||||
<mtime>1688008658</mtime>
|
||||
<size>31306</size>
|
||||
<md5>6cc1a4bac4d031f48ffe313a39ffc65f</md5>
|
||||
<crc32>fddb76db</crc32>
|
||||
<sha1>3c880e5814be545914d7a1a61239bfe821505b7c</sha1>
|
||||
<format>Archive BitTorrent</format>
|
||||
</file>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a_files.xml" source="original">
|
||||
<format>Metadata</format>
|
||||
<md5>7bc8555ee3daef91f32efcfe143cf29b</md5>
|
||||
<summation>md5</summation>
|
||||
</file>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a_itemimage.jpg" source="original">
|
||||
<mtime>1501911384</mtime>
|
||||
<size>335881</size>
|
||||
<md5>929d45f445e54b72d511a663e7324b86</md5>
|
||||
<crc32>5266d154</crc32>
|
||||
<sha1>f01dfce276cb1ca73148b964b290ee92063ea4b2</sha1>
|
||||
<format>Item Image</format>
|
||||
<rotation>0</rotation>
|
||||
<matrix_number>69660BB</matrix_number>
|
||||
<collection-catalog-number>GBIA0011964A</collection-catalog-number>
|
||||
<publisher>Decca</publisher>
|
||||
</file>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a_meta.sqlite" source="original">
|
||||
<mtime>1684102571</mtime>
|
||||
<size>72704</size>
|
||||
<format>Metadata</format>
|
||||
<md5>38976e41f16009ed5e00723fabc9a4e6</md5>
|
||||
<crc32>0e30e5ef</crc32>
|
||||
<sha1>2bcad0eb21e2f77b54883d736faefa33f3cf787e</sha1>
|
||||
</file>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a_meta.xml" source="original">
|
||||
<mtime>1688008490</mtime>
|
||||
<size>2917</size>
|
||||
<format>Metadata</format>
|
||||
<md5>c5326c9a2198274fc206032da9301193</md5>
|
||||
<crc32>ca42acd0</crc32>
|
||||
<sha1>6eac8fecd728dafabef951da785ed2200d351dea</sha1>
|
||||
</file>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a_reviews.xml" source="original">
|
||||
<mtime>1659982866</mtime>
|
||||
<size>948</size>
|
||||
<md5>b67e874d814cf6988e2beb6b243ddc17</md5>
|
||||
<crc32>73b94665</crc32>
|
||||
<sha1>74a01e64dd82da073163f5f457525ab3b84e7497</sha1>
|
||||
<format>Metadata</format>
|
||||
</file>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a_roles.json" source="original">
|
||||
<mtime>1501911433</mtime>
|
||||
<size>80</size>
|
||||
<md5>bd244561b4d7c40c7bdd2d71a401829d</md5>
|
||||
<crc32>e2515f39</crc32>
|
||||
<sha1>93eccc18266c503b84dd47840c41baf2b0a361a5</sha1>
|
||||
<format>JSON</format>
|
||||
<matrix_number>69660BB</matrix_number>
|
||||
<collection-catalog-number>GBIA0011964A</collection-catalog-number>
|
||||
<publisher>Decca</publisher>
|
||||
</file>
|
||||
<file name="78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a_thumb.jpg" source="derivative">
|
||||
<format>JPEG Thumb</format>
|
||||
<original>78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a.tif</original>
|
||||
<mtime>1501911615</mtime>
|
||||
<size>9138</size>
|
||||
<md5>952296a31ff7efd183e2b74cf253640a</md5>
|
||||
<crc32>b616ba9c</crc32>
|
||||
<sha1>153f229b49cd4434605f3739576586bf5d7c9582</sha1>
|
||||
</file>
|
||||
<file name="I Don_t Want to Set the World on Fire - Ink Spots-restored.afpk" source="derivative">
|
||||
<format>Columbia Peaks</format>
|
||||
<original>I Don_t Want to Set the World on Fire - Ink Spots-restored.flac</original>
|
||||
<mtime>1599685667</mtime>
|
||||
<size>29912</size>
|
||||
<md5>5383c38733d9e66b747ec8102ddd0a75</md5>
|
||||
<crc32>3bb2ece6</crc32>
|
||||
<sha1>64bce8d6decac71824583cf12853d3e2c97d0206</sha1>
|
||||
</file>
|
||||
<file name="I Don_t Want to Set the World on Fire - Ink Spots-restored.asr.srt" source="derivative">
|
||||
<format>SubRip</format>
|
||||
<original>I Don_t Want to Set the World on Fire - Ink Spots-restored.asr.vtt</original>
|
||||
<mtime>1688008632</mtime>
|
||||
<size>1973</size>
|
||||
<md5>63a556629e33cc233dba249218af648b</md5>
|
||||
<crc32>477180f5</crc32>
|
||||
<sha1>d82d685af20ce28d21ab1eaa2a5efbf1bf354261</sha1>
|
||||
</file>
|
||||
<file name="I Don_t Want to Set the World on Fire - Ink Spots-restored.asr.vtt" source="derivative">
|
||||
<format>Web Video Text Tracks</format>
|
||||
<original>I Don_t Want to Set the World on Fire - Ink Spots-restored_intermediate_asr.json</original>
|
||||
<mtime>1688008577</mtime>
|
||||
<size>1925</size>
|
||||
<md5>c1b6808ac66b1fb89b6bf1517f6e9ed1</md5>
|
||||
<crc32>349354b8</crc32>
|
||||
<sha1>5fe6aa67644046ab07d61ce3644d388f575649e9</sha1>
|
||||
</file>
|
||||
<file name="I Don_t Want to Set the World on Fire - Ink Spots-restored.flac" source="original">
|
||||
<mtime>1599685214</mtime>
|
||||
<size>48528837</size>
|
||||
<md5>2a2ab6af18cdad6f572bb634c4155a44</md5>
|
||||
<crc32>c1996233</crc32>
|
||||
<sha1>5cb6761e6e24c5545006dd127ce5e44483707982</sha1>
|
||||
<format>24bit Flac</format>
|
||||
<track>02</track>
|
||||
<length>184.55</length>
|
||||
<height>0</height>
|
||||
<width>0</width>
|
||||
<title>I Don't Want to Set the World on Fire (restored)</title>
|
||||
<creator>Durham</creator>
|
||||
<album>I Don't Want to Set the World on Fire</album>
|
||||
<artist>Ink Spots;Seiler;Marcus;Benjemen;Durham</artist>
|
||||
<comment>https://archive.org/details/78_i-dont-want-to-set-the-world-on-fire_ink-spots-seiler-marcus-benjemen-durham_gbia0011964a/I+Don%27t+Want+to+Set+the+World+on+Fire+-+Ink+Spots.flac\</comment>
|
||||
</file>
|
||||
<file name="I Don_t Want to Set the World on Fire - Ink Spots-restored.mp3" source="derivative">
|
||||
<creator>Ink Spots;Seiler;Marcus;Benjemen;Durham</creator>
|
||||
<title>I Don't Want to Set the World on Fire (restored)</title>
|
||||
<track>02</track>
|
||||
<album>I Don't Want to Set the World on Fire</album>
|
||||
<bitrate>125</bitrate>
|
||||
<format>VBR MP3</format>
|
||||
<original>I Don_t Want to Set the World on Fire - Ink Spots-restored.flac</original>
|
||||
<mtime>1599685494</mtime>
|
||||
<size>2996749</size>
|
||||
<md5>dc56351ccd39b09ced0a130fdcd8b2fc</md5>
|
||||
<crc32>38e842b0</crc32>
|
||||
<sha1>3e37080b5abde1be08f20072ca6fcd3f387dca15</sha1>
|
||||
<length>184.58</length>
|
||||
<height>640</height>
|
||||
<width>640</width>
|
||||
</file>
|
||||
<file name="I Don_t Want to Set the World on Fire - Ink Spots-restored.png" source="derivative">
|
||||
<format>PNG</format>
|
||||
<original>I Don_t Want to Set the World on Fire - Ink Spots-restored.flac</original>
|
||||
<mtime>1599685521</mtime>
|
||||
<size>35307</size>
|
||||
<md5>4bc62df3610dde296218e9782bf9cab1</md5>
|
||||
<crc32>ab0f45cf</crc32>
|
||||
<sha1>c3e0ce23e6990e1deabec942ff71e3178cf18310</sha1>
|
||||
</file>
|
||||
<file name="I Don_t Want to Set the World on Fire - Ink Spots-restored_intermediate_asr.json" source="derivative">
|
||||
<format>Intermediate ASR JSON</format>
|
||||
<original>I Don_t Want to Set the World on Fire - Ink Spots-restored_whisper_asr.json</original>
|
||||
<mtime>1688008526</mtime>
|
||||
<size>2672</size>
|
||||
<md5>5bdd50eff7abadfe6a7d764c20dc35d9</md5>
|
||||
<crc32>78a58328</crc32>
|
||||
<sha1>044c53e7cef9fd1536c484c71aa3e75b1306be3c</sha1>
|
||||
</file>
|
||||
<file name="I Don_t Want to Set the World on Fire - Ink Spots-restored_spectrogram.png" source="derivative">
|
||||
<format>Spectrogram</format>
|
||||
<original>I Don_t Want to Set the World on Fire - Ink Spots-restored.flac</original>
|
||||
<mtime>1599685546</mtime>
|
||||
<size>116769</size>
|
||||
<md5>2ea631450681b8d955a9fcf669891d4a</md5>
|
||||
<crc32>e32c035a</crc32>
|
||||
<sha1>fcb550c1596a93528802646f072cd1b19c1b139b</sha1>
|
||||
</file>
|
||||
<file name="I Don_t Want to Set the World on Fire - Ink Spots-restored_whisper_asr.json" source="derivative">
|
||||
<whisper_asr_module_version>20230611.01</whisper_asr_module_version>
|
||||
<whisper_version>20230314</whisper_version>
|
||||
<whisper_model_name>small.en</whisper_model_name>
|
||||
<whisper_model_hash>sha256:f953ad0fd29cacd07d5a9eda5624af0f6bcf2258be67c92b79389873d91e0872</whisper_model_hash>
|
||||
<asr_detected_lang>en</asr_detected_lang>
|
||||
<asr_detected_lang_conf>0.85381597280502</asr_detected_lang_conf>
|
||||
<asr_transcribed_lang>en</asr_transcribed_lang>
|
||||
<format>Whisper ASR JSON</format>
|
||||
<original>I Don_t Want to Set the World on Fire - Ink Spots-restored.mp3</original>
|
||||
<mtime>1688008489</mtime>
|
||||
<size>28875</size>
|
||||
<md5>3b7a44eb35530253c8f22e55542a9eed</md5>
|
||||
<crc32>c800afa0</crc32>
|
||||
<sha1>770e390e635b714058d148b06a92fffd3181e2ba</sha1>
|
||||
</file>
|
||||
<file name="__ia_thumb.jpg" source="original">
|
||||
<mtime>1634652950</mtime>
|
||||
<size>12863</size>
|
||||
<md5>bde829c30a7951d22eb88ad4c33f866c</md5>
|
||||
<crc32>e2abeb1a</crc32>
|
||||
<sha1>ac449c1db692c6c8e5af636012da5d275c000a4d</sha1>
|
||||
<format>Item Tile</format>
|
||||
<rotation>0</rotation>
|
||||
</file>
|
||||
</files>
|
||||
195
SabreTools.Test/TestData/test-archivedotorg-files4.xml
Normal file
195
SabreTools.Test/TestData/test-archivedotorg-files4.xml
Normal file
@@ -0,0 +1,195 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<files>
|
||||
<file name="__ia_thumb.jpg" source="original">
|
||||
<mtime>1688105563</mtime>
|
||||
<size>19642</size>
|
||||
<md5>2dc43777ab13cabd2324e25d8c69a790</md5>
|
||||
<crc32>e7f10d45</crc32>
|
||||
<sha1>1ed3cb724f1470cbb5056169fc731c642e36fc1b</sha1>
|
||||
<format>Item Tile</format>
|
||||
<rotation>0</rotation>
|
||||
</file>
|
||||
<file name="header/header_0000.jp2" source="original">
|
||||
<mtime>1686986524</mtime>
|
||||
<size>12013083</size>
|
||||
<md5>5928785725ea03932b3fd496d3f4c75e</md5>
|
||||
<crc32>53f8e9a7</crc32>
|
||||
<sha1>3af69d2c09d993d0f7bc81b7a2edfdd1f9b8446d</sha1>
|
||||
<format>JPEG 2000</format>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42.pdf" source="derivative">
|
||||
<pdf_module_version>0.0.23</pdf_module_version>
|
||||
<format>Text PDF</format>
|
||||
<original>per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_page_numbers.json</original>
|
||||
<mtime>1688105470</mtime>
|
||||
<size>98013235</size>
|
||||
<md5>8c206bd0aaf82698b3d502b233c6c2f8</md5>
|
||||
<crc32>ac64d257</crc32>
|
||||
<sha1>e6b7dfbde34f6a494f7795cd0edcb5676a8c37da</sha1>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_archive.torrent" source="metadata">
|
||||
<btih>bf04ea0f7c2dca19cf748e529611f497e6cba150</btih>
|
||||
<mtime>1688105603</mtime>
|
||||
<size>21259</size>
|
||||
<md5>0a3685a7ab0a0d0f9fdb4b913d31f8fd</md5>
|
||||
<crc32>70eb691b</crc32>
|
||||
<sha1>413c499bc001947b82030585c855958693cd2f78</sha1>
|
||||
<format>Archive BitTorrent</format>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_chocr.html.gz" source="derivative">
|
||||
<ocr>tesseract 5.3.0-3-g9920</ocr>
|
||||
<ocr_parameters>lang-eng;two-pass-disabled;pass-dpi-disabled;autonomous-mode-disabled;binarisation-method-otsu</ocr_parameters>
|
||||
<ocr_module_version>0.0.21</ocr_module_version>
|
||||
<ocr_detected_script>Latin</ocr_detected_script>
|
||||
<ocr_detected_script_conf>0.7458</ocr_detected_script_conf>
|
||||
<ocr_detected_lang>en</ocr_detected_lang>
|
||||
<ocr_detected_lang_conf>1.0000</ocr_detected_lang_conf>
|
||||
<format>chOCR</format>
|
||||
<original>per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_jp2.zip</original>
|
||||
<mtime>1688102450</mtime>
|
||||
<size>11484860</size>
|
||||
<md5>8fcb1864fd839ef3dcc6039b669b8f36</md5>
|
||||
<crc32>87f7f712</crc32>
|
||||
<sha1>15abc19f5cf46ccc4e8f1a5796ce6302869bffe5</sha1>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_djvu.txt" source="derivative">
|
||||
<format>DjVuTXT</format>
|
||||
<original>per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_djvu.xml</original>
|
||||
<mtime>1688103469</mtime>
|
||||
<size>759494</size>
|
||||
<md5>febe8a4594d3ef07af32f3fbed4d2764</md5>
|
||||
<crc32>98136b35</crc32>
|
||||
<sha1>5fe48b18f19568906680f8dd4936ec81ec143f9e</sha1>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_djvu.xml" source="derivative">
|
||||
<format>Djvu XML</format>
|
||||
<original>per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_hocr.html</original>
|
||||
<mtime>1688103134</mtime>
|
||||
<size>10726635</size>
|
||||
<md5>e06813d99e1adf8d6fff350899b5f846</md5>
|
||||
<crc32>35fd1344</crc32>
|
||||
<sha1>418da8f3cd5a9f6aac998ee5f0fa68a828aa78a8</sha1>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_files.xml" source="original">
|
||||
<format>Metadata</format>
|
||||
<md5>640a9c2fe776c10777911979bc04994a</md5>
|
||||
<summation>md5</summation>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_frame-md.json" source="original">
|
||||
<mtime>1686986516</mtime>
|
||||
<size>224</size>
|
||||
<md5>0f68d8e584bd1aea80a889525a00b547</md5>
|
||||
<crc32>2e83a589</crc32>
|
||||
<sha1>7fe89fdd2aa69246f30c34c73e57814f62154994</sha1>
|
||||
<format>JSON</format>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_hocr.html" source="derivative">
|
||||
<hocr_char_to_word_module_version>1.1.0</hocr_char_to_word_module_version>
|
||||
<hocr_char_to_word_hocr_version>1.1.15</hocr_char_to_word_hocr_version>
|
||||
<ocr_parameters>lang-eng;two-pass-disabled;pass-dpi-disabled;autonomous-mode-disabled;binarisation-method-otsu</ocr_parameters>
|
||||
<ocr_module_version>0.0.21</ocr_module_version>
|
||||
<ocr_detected_script>Latin</ocr_detected_script>
|
||||
<ocr_detected_script_conf>0.7458</ocr_detected_script_conf>
|
||||
<ocr_detected_lang>en</ocr_detected_lang>
|
||||
<ocr_detected_lang_conf>1.0000</ocr_detected_lang_conf>
|
||||
<format>hOCR</format>
|
||||
<original>per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_chocr.html.gz</original>
|
||||
<mtime>1688102642</mtime>
|
||||
<size>20717059</size>
|
||||
<md5>2313d8cbe745bf4994b86be8f09b241b</md5>
|
||||
<crc32>e3b89eb2</crc32>
|
||||
<sha1>b5d6e6d025dca7315307aee1032e3470eb797ac4</sha1>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_hocr_pageindex.json.gz" source="derivative">
|
||||
<hocr_pageindex_module_version>1.0.0</hocr_pageindex_module_version>
|
||||
<hocr_pageindex_hocr_version>1.1.15</hocr_pageindex_hocr_version>
|
||||
<format>OCR Page Index</format>
|
||||
<original>per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_hocr.html</original>
|
||||
<mtime>1688103151</mtime>
|
||||
<size>203</size>
|
||||
<md5>aaf29479ec4d752ab5011ef1629a1de0</md5>
|
||||
<crc32>de6755b7</crc32>
|
||||
<sha1>d71e9705a5f1abb4eadcd920b06a961bf0eb293c</sha1>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_hocr_searchtext.txt.gz" source="derivative">
|
||||
<hocr_fts_text_module_version>1.1.0</hocr_fts_text_module_version>
|
||||
<hocr_fts_text_hocr_version>1.1.15</hocr_fts_text_hocr_version>
|
||||
<word_conf_0_10>5195</word_conf_0_10>
|
||||
<word_conf_11_20>2390</word_conf_11_20>
|
||||
<word_conf_21_30>3139</word_conf_21_30>
|
||||
<word_conf_31_40>3258</word_conf_31_40>
|
||||
<word_conf_41_50>3191</word_conf_41_50>
|
||||
<word_conf_51_60>3213</word_conf_51_60>
|
||||
<word_conf_61_70>3776</word_conf_61_70>
|
||||
<word_conf_71_80>5252</word_conf_71_80>
|
||||
<word_conf_81_90>13153</word_conf_81_90>
|
||||
<word_conf_91_100>85216</word_conf_91_100>
|
||||
<format>OCR Search Text</format>
|
||||
<original>per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_hocr.html</original>
|
||||
<mtime>1688103216</mtime>
|
||||
<size>309695</size>
|
||||
<md5>3d209f772e41f56191cb37ac04f8965a</md5>
|
||||
<crc32>ea3470ec</crc32>
|
||||
<sha1>c425dbfdf9900a6f1cb594e4e4403333c2ef4f36</sha1>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_itemimage.png" source="original">
|
||||
<mtime>1686986511</mtime>
|
||||
<size>33159468</size>
|
||||
<md5>5e8e1ffe981317426081234ade24073e</md5>
|
||||
<crc32>b8456619</crc32>
|
||||
<sha1>2ffdc308b639959bbd13ddadbab8f8d4b847fbba</sha1>
|
||||
<format>Item Image</format>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_jp2.zip" source="original">
|
||||
<mtime>1686986495</mtime>
|
||||
<size>231100229</size>
|
||||
<md5>eac76d5be19d38caee6cdcee2b9ff936</md5>
|
||||
<crc32>c2ce4d1c</crc32>
|
||||
<sha1>858120c8724f345b206961abe508291565816b7d</sha1>
|
||||
<preview-image>n0</preview-image>
|
||||
<format>Single Page Processed JP2 ZIP</format>
|
||||
<filecount>13</filecount>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_meta.sqlite" source="original">
|
||||
<mtime>1686986699</mtime>
|
||||
<size>40960</size>
|
||||
<md5>385dd15968e9ce2a38ac48a2838e8c48</md5>
|
||||
<crc32>192d70cb</crc32>
|
||||
<sha1>09ab56c0b9b3eff6b0725364a4fd070705b80d2e</sha1>
|
||||
<format>Metadata</format>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_meta.xml" source="original">
|
||||
<mtime>1688105483</mtime>
|
||||
<size>3028</size>
|
||||
<md5>5ff8508ae78bdb3bbe56ab9de53b43bb</md5>
|
||||
<crc32>1b81e6e4</crc32>
|
||||
<sha1>30cc20a8cd9a9d726408f795b0fcc3db73459185</sha1>
|
||||
<format>Metadata</format>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_page_numbers.json" source="derivative">
|
||||
<format>Page Numbers JSON</format>
|
||||
<original>per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_djvu.xml</original>
|
||||
<mtime>1688103557</mtime>
|
||||
<size>2277</size>
|
||||
<md5>a1bc72e8504b217ccd46c678745982f4</md5>
|
||||
<crc32>634b90ea</crc32>
|
||||
<sha1>680343d1c42c0732cdc9c8b8c355e5985aab5d6a</sha1>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_rotations.json" source="original">
|
||||
<mtime>1686986519</mtime>
|
||||
<size>1334</size>
|
||||
<md5>37f40a3e763cda634832781f6f1d6775</md5>
|
||||
<crc32>328b653d</crc32>
|
||||
<sha1>f43ce2a767ee70f0c2cb2e004891f699213e27a1</sha1>
|
||||
<format>JSON</format>
|
||||
</file>
|
||||
<file name="per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_scandata.xml" source="derivative">
|
||||
<format>Scandata</format>
|
||||
<original>per_chicago-daily-tribune_the-chicago-daily-tribun_1883-04-13_42_djvu.xml</original>
|
||||
<mtime>1688103552</mtime>
|
||||
<size>4478</size>
|
||||
<md5>cfc96ac86bda784dd477131a9ce78d85</md5>
|
||||
<crc32>35d000b3</crc32>
|
||||
<sha1>45d344b009d902373579fd8939a1730b70a10ae4</sha1>
|
||||
</file>
|
||||
</files>
|
||||
11812
SabreTools.Test/TestData/test-archivedotorg-files5.xml
Normal file
11812
SabreTools.Test/TestData/test-archivedotorg-files5.xml
Normal file
File diff suppressed because it is too large
Load Diff
Reference in New Issue
Block a user