brister/jsmr
1
0
Fork 0
Code Issues 1 Pull Requests Actions Packages Projects Releases 1 Wiki Activity

Updated scanner logic to handle thumb VueJS components. Removed uneeded DLSiteWork fields.
All checks were successful
ci / build-test (push) Successful in 2m21s
Details
ci / publish-image (push) Has been skipped
Details

Browse Source
This commit is contained in:
Brian Bicknell
2026-02-28 22:20:24 -05:00
parent ca7ffa1730
commit 704a6fc433
9 changed files with 316 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
JSMR.Application/Scanning/Contracts/DLSiteWork.cs
View File

@@ -4,7 +4,7 @@ namespace JSMR.Application.Scanning.Contracts;
public class DLSiteWork
{
public DLSiteWorkType Type { get; set; }
//public DLSiteWorkType Type { get; set; }
public DLSiteWorkCategory Category { get; set; }
public required string ProductName { get; set; }
public required string ProductId { get; set; }

10
JSMR.Application/Scanning/Contracts/DLSiteWorkType.cs
View File

@@ -1,7 +1,7 @@
namespace JSMR.Application.Scanning.Contracts;
public enum DLSiteWorkType
{
Released,
Announced
}
//public enum DLSiteWorkType
//{
// Released,
// Announced
//}

29
JSMR.Infrastructure/Scanning/Models/DLSiteHtmlDocument.cs
View File

@@ -2,37 +2,26 @@
namespace JSMR.Infrastructure.Scanning.Models;
public class DLSiteHtmlDocument
public class DLSiteHtmlDocument(HtmlDocument document)
{
private readonly HtmlNodeCollection _workColumns;
private readonly HtmlNodeCollection _workColumnRights;
private readonly HtmlNodeCollection _workThumbs;
private readonly HtmlNodeCollection _workColumns = document.DocumentNode.SelectNodes("//dl[@class='work_1col']");
private readonly HtmlNodeCollection _workColumnRights = document.DocumentNode.SelectNodes("//td[contains(@class, 'work_1col_right')]");
private readonly HtmlNodeCollection _workThumbs = document.DocumentNode.SelectNodes("//div[@class='work_thumb']");
public HtmlNode PageTotalNode { get; } = document.DocumentNode.SelectNodes("//div[@class='page_total']/strong")[0];
public HtmlNode PageTotalNode { get; }
public DLSiteHtmlDocument(HtmlDocument document)
public DLSiteHtmlNode[] GetDLSiteNodes()
{
_workColumns = document.DocumentNode.SelectNodes("//dl[@class='work_1col']");
//_workColumnRights = document.DocumentNode.SelectNodes("//td[@class='work_1col_right']");
_workColumnRights = document.DocumentNode.SelectNodes("//td[contains(@class, 'work_1col_right')]");
_workThumbs = document.DocumentNode.SelectNodes("//div[@class='work_thumb']");
PageTotalNode = document.DocumentNode.SelectNodes("//div[@class='page_total']/strong")[0];
}
public List<DLSiteHtmlNode> GetDLSiteNodes()
{
var nodes = new List<DLSiteHtmlNode>();
List<DLSiteHtmlNode> nodes = [];
if (_workColumns.Count != _workColumnRights.Count || _workColumns.Count != _workThumbs.Count)
throw new Exception("Work column node counts do not match!");
for (int i = 0; i < _workColumns.Count; i++)
{
var node = new DLSiteHtmlNode(_workColumns[i], _workColumnRights[i], _workThumbs[i]);
DLSiteHtmlNode node = new(_workColumns[i], _workColumnRights[i], _workThumbs[i]);
nodes.Add(node);
}
return nodes;
return [.. nodes];
}
}

15
JSMR.Infrastructure/Scanning/Models/DLSiteHtmlNode.cs
View File

@@ -19,7 +19,8 @@ public class DLSiteHtmlNode
public HtmlNode? SalesDateNode { get; private set; }
public HtmlNode DownloadsNode { get; private set; }
public HtmlNode? StarRatingNode { get; private set; }
public HtmlNode ImageNode { get; private set; }
public HtmlNode? ImageNode { get; private set; }
public HtmlNode? ThumbWithNgFilterBlockNode { get; private set; }
public HtmlNode[] GenreNodes { get; private set; }
public HtmlNode[] SearchTagNodes { get; private set; }
public HtmlNode[] CreatorNodes { get; private set; }
@@ -55,7 +56,8 @@ public class DLSiteHtmlNode
//InitializeSalesAndDownloadsNodes();
StarRatingNode = GetStarRatingNode();
ImageNode = GetImageNode();
ImageNode = TryGetImageNode();
ThumbWithNgFilterBlockNode = ThumbNode.SelectSingleNode(".//thumb-with-ng-filter-block");
}
private HtmlNode[] GetGenreNodes()
@@ -165,10 +167,13 @@ public class DLSiteHtmlNode
// }
//}
private HtmlNode GetImageNode()
private HtmlNode? TryGetImageNode()
{
HtmlNode linkNode = ThumbNode.SelectNodes(".//a")[0];
HtmlNode? linkNode = ThumbNode.SelectSingleNode(".//a");
return linkNode.SelectNodes(".//img")[0];
if (linkNode is null)
return null;
return linkNode.SelectSingleNode(".//img");
}
}

23
JSMR.Infrastructure/Scanning/ScannerUtilities.cs
View File

@@ -1,4 +1,5 @@
using HtmlAgilityPack;
using System.Text.Json;
using System.Web;
namespace JSMR.Infrastructure.Scanning;
@@ -45,4 +46,26 @@ public static class ScannerUtilities
return imageSource;
}
public static string[] ParseJavaScriptArray(string value)
{
try
{
string json = NormalizeJavaScriptArray(value);
return JsonSerializer.Deserialize<string[]>(json) ?? [];
}
catch
{
return [.. value
.Trim('[', ']')
.Split(',', StringSplitOptions.RemoveEmptyEntries)
.Select(x => x.Trim().Trim('\'', '"'))];
}
}
private static string NormalizeJavaScriptArray(string input)
{
return input.Trim().Replace('\'', '"');
}
}

40
JSMR.Infrastructure/Scanning/VoiceWorksScanner.cs
View File

@@ -7,6 +7,7 @@ using JSMR.Domain.ValueObjects;
using JSMR.Infrastructure.Http;
using JSMR.Infrastructure.Scanning.Models;
using System.Globalization;
using System.Text.Json;
using System.Text.RegularExpressions;
namespace JSMR.Infrastructure.Scanning;
@@ -22,7 +23,7 @@ public abstract class VoiceWorksScanner(IHtmlLoader htmlLoader) : IVoiceWorksSca
public async Task<IReadOnlyList<DLSiteWork>> ScanPageAsync(VoiceWorkScanOptions options, CancellationToken cancellationToken = default)
{
DLSiteHtmlDocument document = await GetDLSiteHtmlCollectionAsync(options, cancellationToken);
List<DLSiteHtmlNode> nodes = document.GetDLSiteNodes();
DLSiteHtmlNode[] nodes = document.GetDLSiteNodes();
return GetDLSiteWorks(nodes, options);
}
@@ -52,7 +53,7 @@ public abstract class VoiceWorksScanner(IHtmlLoader htmlLoader) : IVoiceWorksSca
return filterBuilder.BuildSearchQuery(options.PageNumber, options.PageSize);
}
private List<DLSiteWork> GetDLSiteWorks(List<DLSiteHtmlNode> nodes, VoiceWorkScanOptions options)
private List<DLSiteWork> GetDLSiteWorks(DLSiteHtmlNode[] nodes, VoiceWorkScanOptions options)
{
var works = new List<DLSiteWork>();
@@ -73,8 +74,7 @@ public abstract class VoiceWorksScanner(IHtmlLoader htmlLoader) : IVoiceWorksSca
{
string productUrl = node.ProductLinkNode.Attributes["href"].Value;
string makerUrl = node.MakerLinkNode.Attributes["href"].Value;
string imageSource = ScannerUtilities.GetImageSource(node.ImageNode);
string imageUrl = imageSource.Replace("_sam.jpg", "_main.jpg").Replace("_sam.gif", "_main.gif");
(string imageSource, string imageUrl) = TryGetImageSourceAndUrl(node);
ScannedRating? rating = GetScannedRating(node.StarRatingNode);
DLSiteWork work = new()
@@ -89,7 +89,7 @@ public abstract class VoiceWorksScanner(IHtmlLoader htmlLoader) : IVoiceWorksSca
Creators = ScannerUtilities.GetStringListFromNodes(node.CreatorNodes),
SmallImageUrl = imageSource,
ImageUrl = imageUrl,
Type = imageUrl.Contains("ana/doujin") ? DLSiteWorkType.Announced : DLSiteWorkType.Released,
//Type = imageUrl.Contains("ana/doujin") ? DLSiteWorkType.Announced : DLSiteWorkType.Released,
StarRating = rating?.Score,
Votes = rating?.Votes,
AgeRating = GetAgeRating(node.GenreNodes)
@@ -113,6 +113,36 @@ public abstract class VoiceWorksScanner(IHtmlLoader htmlLoader) : IVoiceWorksSca
return work;
}
private static (string, string) TryGetImageSourceAndUrl(DLSiteHtmlNode node)
{
if (node.ThumbWithNgFilterBlockNode is not null)
{
string candidates = node.ThumbWithNgFilterBlockNode.GetAttributeValue(":thumb-candidates", string.Empty);
string[] imageUrls = ScannerUtilities.ParseJavaScriptArray(candidates);
if (imageUrls.Length == 0)
{
throw new Exception("No thumb candidartes found");
}
string imageSource = imageUrls[0];
string imageUrl = imageSource.Replace("_sam.jpg", "_main.jpg").Replace("_sam.gif", "_main.gif");
return (imageSource, imageUrl);
}
else if (node.ImageNode is not null)
{
string imageSource = ScannerUtilities.GetImageSource(node.ImageNode);
string imageUrl = imageSource.Replace("_sam.jpg", "_main.jpg").Replace("_sam.gif", "_main.gif");
return (imageSource, imageUrl);
}
else
{
throw new Exception("Unable to find image source and/or url");
}
}
private static AgeRating GetAgeRating(HtmlNode[] genreNodes)
{
List<string> genres = ScannerUtilities.GetStringListFromNodes(genreNodes);

1
JSMR.Tests/JSMR.Tests.csproj
View File

@@ -15,6 +15,7 @@
<ItemGroup>
<EmbeddedResource Include="Integrations\DLSite\Product-Info.json" />
<EmbeddedResource Include="Scanning\English-Page-Updated.html" />
<EmbeddedResource Include="Scanning\Japanese-Page-Updated.html" />
<EmbeddedResource Include="Scanning\Japanese-Page.html" />
<EmbeddedResource Include="Scanning\English-Page.html" />
</ItemGroup>

183
JSMR.Tests/Scanning/Japanese-Page-Updated.html Normal file
View File

@@ -0,0 +1,183 @@
<html lang="ja-JP">
<body>
<div id="container">
<div id="wrapper">
<div id="main">
<div id="main_inner">
<div>
<div data-toggle="found" class="sort_box border_b pb10">
<div class="page_total">
<strong>6670</strong>
<span>件中</span>
<strong>1～30</strong>
<span>件目</span>
</div>
</div>
</div>
<div id="search_result_list" class="loading_display_open" data-toggle="found">
<table class="work_1col_table n_worklist" cellspacing="0">
<tbody>
<!-- RJ01536422 (With Image) -->
<tr data-list_item_product_id="RJ01536422" class=" ">
<td class="work_1col_thumb">
<div class="work_thumb">
<div class="work_thumb_inner" data-vue-component="thumb-img-popup">
<thumb-with-ng-filter-block ref="popup_img"
link="https://www.dlsite.com/maniax/announce/=/product_id/RJ01536422.html"
:thumb-candidates="['//img.dlsite.jp/resize/images2/ana/doujin/RJ01537000/RJ01536422_ana_img_main_240x240.webp','//img.dlsite.jp/resize/images2/ana/doujin/RJ01537000/RJ01536422_ana_img_main_240x240.jpg']"
alt="珈琲屋 綴 / いつもいつでも〜Alone with you〜 [喫茶綴]"
@mouseenter="showPopupImg">
</thumb-with-ng-filter-block>
<div v-cloak class="work_img_popover">
<img src="data:image/gif;base64,R0lGODlhAQABAGAAACH5BAEKAP8ALAAAAAABAAEAAAgEAP8FBAA7"
:src="is_show ? '//img.dlsite.jp/modpub/images2/ana/doujin/RJ01537000/RJ01536422_ana_img_main.jpg' : 'data:image/gif;base64,R0lGODlhAQABAGAAACH5BAEKAP8ALAAAAAABAAEAAAgEAP8FBAA7'"
alt="珈琲屋 綴 / いつもいつでも〜Alone with you〜 [喫茶綴]">
</div>
</div>
<div class="work_category type_SOU">
<a href="https://www.dlsite.com/maniax/fsr/=/work_type/SOU">ボイス・ASMR</a>
</div>
</td>
<td>
<dl class="work_1col">
<dt class="work_name">
<p class="expected_date">2026年12月下旬 発売予定</p>
<div class="icon_wrap"></div>
<a href="https://www.dlsite.com/maniax/announce/=/product_id/RJ01536422.html">
珈琲屋 綴 / いつもいつでも〜Alone with you〜
</a>
</dt>
<dd class="maker_name">
<a href="https://www.dlsite.com/maniax/circle/profile/=/maker_id/RG36156.html">喫茶綴</a>
<span class="separator">/</span>
<span class="author">
<a href="https://www.dlsite.com/maniax/fsr/=/keyword_creater/%22%E9%87%8E%E4%B8%8A%E8%8F%9C%E6%9C%88%22/ana_flg/all"
class="">野上菜月</a>
</span>
</dd>
<dd class="work_text">
珈琲に特化した喫茶店、喫茶綴、外伝。『珈琲屋 綴』の従業員、綴明日菜が、大好きな珈琲と、貴方との時間を大切に育みます。珈琲に特化した喫茶店、喫茶綴の外伝です。CV:野上菜月様
</dd>
<dd class="work_genre">
<span class="icon_GEN" title="全年齢">全年齢</span><span data-vue-component="product-coupon" data-product_id="RJ01536422" v-cloak></span>
<input type="hidden" class="__product_attributes"
name="__product_attributes" id="_RJ01536422"
value="RG36156,male,SOU,JPN,SND,497,056,496,008,442,058"
disabled="disabled">
</dd>
<dd class="search_tag">
<a href="https://www.dlsite.com/maniax/fsr/=/genre/497/from/work.genre">ASMR</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/056/from/work.genre">癒し</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/058/from/work.genre">オールハッピー</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/496/from/work.genre">バイノーラル/ダミヘ</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/008/from/work.genre">日常/生活</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/442/from/work.genre">耳かき</a>
</dd>
</dl>
</td>
<td class="work_1col_right">
<ul>
<li class="sales_date">
予告開始日:&nbsp;2026年01月01日
</li>
<li class="work_dl clear">
<div data-vue-component="product-wishlist-count"
data-product_id="RJ01536422" v-cloak></div>
</li>
</ul>
<div data-vue-component="product-item" data-product_id="RJ01536422"
data-layout="fsr_announce_only" data-is_ana="1" class="btn_2col"
data-usesample="true" data-samples="[]">
<p class="work_free_sample">
<a href="https://www.dlsite.com/maniax/announce/=/product_id/RJ01536422.html"
class="btn_free_sample disabled" data-product-id="RJ01536422">無料サンプル</a>
</p>
<p class="work_favorite_xs">
<a href="https://www.dlsite.com/maniax/mypage/wishlist/=/product_id/RJ01536422.html"
class="btn_favorite _btn_favorite" title="お気に入りに追加">お気に入りに追加</a>
</p>
</div>
</td>
</tr>
<!-- RJ01393816 (No Image) -->
<tr data-list_item_product_id="RJ01393816" class=" ">
<td class="work_1col_thumb">
<div class="work_thumb">
<div class="work_thumb_inner" data-vue-component="thumb-img-popup">
<thumb-with-ng-filter-block ref="popup_img" link="https://www.dlsite.com/maniax/announce/=/product_id/RJ01393816.html" :thumb-candidates="['//www.dlsite.com/images/web/home/no_img_main.gif','//www.dlsite.com/images/web/home/no_img_main.gif']" alt="アダルトグッズショップの店長にオナ禁でオモチャにされる話 [平たい胸族]" @mouseenter="showPopupImg"></thumb-with-ng-filter-block>
</div>
<div class="work_category type_SOU"><a href="https://www.dlsite.com/maniax/fsr/=/work_type/SOU">ボイス・ASMR</a></div>
</td>
<td>
<dl class="work_1col">
<dt class="work_name">
<p class="expected_date">
2026年12月下旬 発売予定
</p>
<div class="icon_wrap">
</div>
<a href="https://www.dlsite.com/maniax/announce/=/product_id/RJ01393816.html">アダルトグッズショップの店長にオナ禁でオモチャにされる話</a>
</dt>
<dd class="maker_name">
<a href="https://www.dlsite.com/maniax/circle/profile/=/maker_id/RG01044380.html">平たい胸族</a>
</dd>
<dd class="work_text">アダルトグッズショップでダウナーなセンパイと仕事中にオナ禁サポートをしてサボっていたことが店長にバレてしまった。今度はセンパイの詩乃と店長のミチル、2人にオナ禁でオモチャにされることになってしまった。</dd>
<dd class="work_genre">
<span data-vue-component="product-coupon" data-product_id="RJ01393816" v-cloak></span>
<input type="hidden" class="__product_attributes" name="__product_attributes" id="_RJ01393816" value="RG01044380,adl,male,SOU,JPN,SND,497,118,183,187,158,496,440,448" disabled="disabled">
</dd>
<dd class="search_tag">
<a href="https://www.dlsite.com/maniax/fsr/=/genre/497/from/work.genre">ASMR</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/496/from/work.genre">バイノーラル/ダミヘ</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/448/from/work.genre">色仕掛け</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/440/from/work.genre">浮気</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/158/from/work.genre">百合</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/118/from/work.genre">レズ/女同士</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/187/from/work.genre">ツルペタ</a>
<a href="https://www.dlsite.com/maniax/fsr/=/genre/183/from/work.genre">貧乳/微乳</a>
</dd>
</dl>
</td>
<td class="work_1col_right">
<ul>
<li class="sales_date">
予告開始日:&nbsp;2025年05月15日
</li>
<li class="work_dl clear">
<div data-vue-component="product-wishlist-count" data-product_id="RJ01393816" v-cloak></div>
</li>
</ul>
<div data-vue-component="product-item" data-product_id="RJ01393816" data-layout="fsr_announce_only" data-is_ana="1" class="btn_2col" data-usesample="true" data-samples="[]">
<p class="work_free_sample"><a href="https://www.dlsite.com/maniax/announce/=/product_id/RJ01393816.html" class="btn_free_sample disabled" data-product-id="RJ01393816">無料サンプル</a></p>
<p class="work_favorite_xs"><a href="https://www.dlsite.com/maniax/mypage/wishlist/=/product_id/RJ01393816.html" class="btn_favorite _btn_favorite" title="お気に入りに追加">お気に入りに追加</a></p>
</div>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
</div>
</div>
</body>
</html>

53
JSMR.Tests/Scanning/VoiceWorkScannerTests.cs
View File

@@ -49,10 +49,58 @@ public class VoiceWorkScannerTests
result[0].Creators.ShouldBe(["柚木つばめ"]);
result[0].Genres.ShouldBe(["体験版"]);
result[0].Tags.ShouldBe(["バイノーラル/ダミヘ", "手コキ", "足コキ", "パイズリ", "言葉責め", "焦らし", "乳首責め", "本番なし"]);
result[0].Type.ShouldBe(DLSiteWorkType.Released);
result[0].Downloads.ShouldBe(1220);
}
[Fact]
public async Task Scan_With_Updated_Japanese_Locale()
{
string html = await ReadResourceAsync("Japanese-Page-Updated.html");
IHttpService httpService = Substitute.For<IHttpService>();
httpService.GetStringAsync(Arg.Any<string>(), CancellationToken.None)
.Returns(Task.FromResult(html));
HtmlLoader loader = new(httpService);
JapaneseVoiceWorksScanner scanner = new(loader);
VoiceWorkScanOptions options = new(
PageNumber: 1,
PageSize: 100,
ExcludeAIGeneratedWorks: true,
ExcludePartiallyAIGeneratedWorks: true,
ExcludedMakerIds: []
);
var result = await scanner.ScanPageAsync(options, CancellationToken.None);
result.Count.ShouldBe(2);
result[0].SalesDate.ShouldBeNull();
result[0].ExpectedDate.ShouldBe(new DateOnly(2026, 12, 21));
result[0].ProductId.ShouldBe("RJ01536422");
result[0].ProductName.ShouldBe("珈琲屋 綴 / いつもいつでも〜Alone with you〜");
result[0].Description.ShouldBe("珈琲に特化した喫茶店、喫茶綴、外伝。『珈琲屋 綴』の従業員、綴明日菜が、大好きな珈琲と、貴方との時間を大切に育みます。珈琲に特化した喫茶店、喫茶綴の外伝です。CV:野上菜月様");
result[0].Maker.ShouldBe("喫茶綴");
result[0].MakerId.ShouldBe("RG36156");
result[0].Creators.ShouldBe(["野上菜月"]);
result[0].Genres.ShouldBe(["全年齢"]);
result[0].Tags.ShouldBe(["ASMR", "癒し", "オールハッピー", "バイノーラル/ダミヘ", "日常/生活", "耳かき"]);
// TODO: Wishlist count?
result[1].SalesDate.ShouldBeNull();
result[1].ExpectedDate.ShouldBe(new DateOnly(2026, 12, 21));
result[1].ProductId.ShouldBe("RJ01393816");
result[1].ProductName.ShouldBe("アダルトグッズショップの店長にオナ禁でオモチャにされる話");
result[1].Description.ShouldBe("アダルトグッズショップでダウナーなセンパイと仕事中にオナ禁サポートをしてサボっていたことが店長にバレてしまった。今度はセンパイの詩乃と店長のミチル、2人にオナ禁でオモチャにされることになってしまった。");
result[1].Maker.ShouldBe("平たい胸族");
result[1].MakerId.ShouldBe("RG01044380");
result[1].Creators.ShouldBe([]);
result[1].Genres.ShouldBe([]);
result[1].Tags.ShouldBe(["ASMR", "バイノーラル/ダミヘ", "色仕掛け", "浮気", "百合", "レズ/女同士", "ツルペタ", "貧乳/微乳"]);
}
[Fact]
public async Task Scan_With_English_Locale()
{
@@ -88,13 +136,11 @@ public class VoiceWorkScannerTests
result[0].Creators.ShouldBe(["Some Creator"]);
result[0].Genres.ShouldBe(["Voice", "Trial version"]);
result[0].Tags.ShouldBe(["Male Protagonist", "Gal", "Uniform", "Harem", "Big Breasts", "Tanned Skin / Suntan"]);
result[0].Type.ShouldBe(DLSiteWorkType.Released);
result[0].Downloads.ShouldBe(1000);
result[1].ExpectedDate.ShouldBe(new DateOnly(2025, 10, 11));
result[1].SalesDate.ShouldBeNull();
result[1].ProductId.ShouldBe("RJ00000002");
result[1].Type.ShouldBe(DLSiteWorkType.Announced);
}
[Fact]
@@ -132,7 +178,6 @@ public class VoiceWorkScannerTests
result[0].Creators.ShouldBe(["沼倉愛美"]);
result[0].Genres.ShouldBe(["All Ages", "Trial version"]);
result[0].Tags.ShouldBe(["Moe", "Healing", "Binaural", "ASMR", "Ear Cleaning", "Slice of Life / Daily Living", "Heartwarming", "Whispering"]);
result[0].Type.ShouldBe(DLSiteWorkType.Released);
result[0].Downloads.ShouldBe(1);
}
}