fix(ar): empty chapter bodies and Riwyat novel metadata (Galaxy Novels, Riwyat, Kol Novel, Free Kol Novel)

plugins/arabic/galaxynovels.ts

@@ -21,19 +21,10 @@ type ChaptersIndex = {
   chapters: ChapterJSON[];
 };
 
-type ChapterContentResponse = {
-  schema: number;
-  data: {
-    content_html: string;
-    display_title: string;
-    navigation: { next_url: string; previous_url: string };
-  };
-};
-
 class GalaxyNovels implements Plugin.PluginBase {
   id = 'galaxynovels';
   name = 'Galaxy Novels';
-  version = '1.1.0';
+  version = '1.1.1';
   icon = 'src/ar/galaxynovels/icon.png';
   site = 'https://galaxynovels.com/';
 
@@ -62,6 +53,11 @@ class GalaxyNovels implements Plugin.PluginBase {
 
   private baseUrl = 'https://galaxynovels.com';
 
+  private toChapterPath(url: string | undefined): string {
+    if (!url) return '';
+    return url.startsWith('http') ? new URL(url).pathname : url;
+  }
+
   private async fetchHtml(url: string): Promise<string> {
     const res = await fetchApi(url);
     if (!res.ok) {
@@ -106,8 +102,7 @@ class GalaxyNovels implements Plugin.PluginBase {
       const coverLink = $el.find('a.wor-novel-card__cover');
       const href = coverLink.attr('href');
       const img = $el.find('img.wor-cover-img');
-      const cover =
-        img.attr('data-src') || img.attr('src') || undefined;
+      const cover = img.attr('data-src') || img.attr('src') || undefined;
       const title = $el.find('h3 a').text().trim();
 
       if (!href || !title) return;
@@ -162,7 +157,7 @@ class GalaxyNovels implements Plugin.PluginBase {
 
         chapters = index.chapters.map(ch => ({
           name: ch.label + (ch.title ? `: ${ch.title}` : ''),
-          path: `${novelPath}chapter-${ch.id}/`,
+          path: this.toChapterPath(ch.url) || `${novelPath}chapter-${ch.id}/`,
           chapterNumber: ch.position,
           releaseTime: ch.date_iso?.split('T')[0] || '',
         }));
@@ -174,17 +169,21 @@ class GalaxyNovels implements Plugin.PluginBase {
     if (chapters.length === 0) {
       $('article.wor-novel-chapter-item').each((_, el) => {
         const $el = $(el);
-        const chapterLink = $el.find('h3 a').attr('href') || $el.find('a.wor-novel-chapter-item__num').attr('href');
-        const chapterName = $el.find('h3 a').text().trim() || $el.find('a.wor-novel-chapter-item__num').text().trim();
-        const chapterId = $el.attr('data-chapter-id');
+        const chapterLink =
+          $el.find('h3 a').attr('href') ||
+          $el.find('a.wor-novel-chapter-item__num').attr('href');
+        const chapterName =
+          $el.find('h3 a').text().trim() ||
+          $el.find('a.wor-novel-chapter-item__num').text().trim();
         const timeEl = $el.find('time');
         const releaseTime = timeEl.attr('datetime')?.split('T')[0] || '';
 
         if (!chapterLink) return;
 
-        const path = chapterId
-          ? `${novelPath}chapter-${chapterId}/`
-          : new URL(chapterLink, this.site).pathname;
+        // Only the full permalink resolves on the site; the data-chapter-id
+        // attribute holds the WordPress post id, and …/chapter-<post-id>/
+        // answers 404.
+        const path = this.toChapterPath(chapterLink);
         const numMatch = path.match(/chapter-(\d+)/);
         const chapterNumber = numMatch ? parseInt(numMatch[1]) : 0;
 
@@ -210,27 +209,27 @@ class GalaxyNovels implements Plugin.PluginBase {
   }
 
   async parseChapter(chapterPath: string): Promise<string> {
-    const idMatch = chapterPath.match(/chapter-(\d+)/);
-    const chapterId = idMatch?.[1];
+    const url = `${this.baseUrl}${chapterPath}`;
+    const html = await this.fetchHtml(url);
+    const $ = loadCheerio(html);
 
-    if (chapterId) {
-      const apiUrl = `${this.baseUrl}/wp-json/wor-reader-app/v1/chapters/${chapterId}`;
-      try {
-        const response = await this.fetchJson<ChapterContentResponse>(apiUrl);
-        if (response.data?.content_html) {
-          return response.data.content_html;
-        }
-      } catch {
-        // fallback to HTML
+    // The body is server-rendered inside the article marked up as
+    // https://schema.org/Chapter, as the element with itemprop="text". Pick
+    // the first candidate that actually carries text: a themed wrapper can be
+    // present but empty, and cheerio selections are always truthy.
+    for (const selector of [
+      'article[itemtype*="Chapter"] [itemprop="text"]',
+      '[itemprop="text"]',
+      '[data-wor-reader-text]',
+      '.wor-reader-text-surface',
+    ]) {
+      const candidate = $(selector);
+      if (candidate.length && candidate.text().trim()) {
+        return candidate.html() || '';
       }
     }
 
-    const url = `${this.baseUrl}${chapterPath}`;
-    const html = await this.fetchHtml(url);
-    const $ = loadCheerio(html);
-    const content =
-      $('article.wor-chapter-content, .wor-chapter-text, .entry-content').html();
-    return content || '<p>Content not available.</p>';
+    return '<p>Content not available.</p>';
   }
 
   async searchNovels(
@@ -254,9 +253,7 @@ class GalaxyNovels implements Plugin.PluginBase {
 
     const term = searchTerm.toLowerCase();
     const filtered = searchIndex.items.filter(
-      n =>
-        n.t.toLowerCase().includes(term) ||
-        n.s.toLowerCase().includes(term),
+      n => n.t.toLowerCase().includes(term) || n.s.toLowerCase().includes(term),
     );
 
     const limit = 20;
@@ -265,9 +262,7 @@ class GalaxyNovels implements Plugin.PluginBase {
     return filtered.slice(offset, offset + limit).map(novel => ({
       name: novel.t,
       path: novel.u,
-      cover: novel.c.startsWith('http')
-        ? novel.c
-        : `${this.baseUrl}${novel.c}`,
+      cover: novel.c.startsWith('http') ? novel.c : `${this.baseUrl}${novel.c}`,
     }));
   }
 }

plugins/multisrc/lightnovelwp/template.ts

@@ -42,7 +42,7 @@ export class LightNovelWPPlugin implements Plugin.PluginBase {
     this.icon = `multisrc/lightnovelwp/${metadata.id.toLowerCase()}/icon.png`;
     this.site = metadata.sourceSite;
     const versionIncrements = metadata.options?.versionIncrements || 0;
-    this.version = `1.1.${10 + versionIncrements}`;
+    this.version = `1.2.${10 + versionIncrements}`;
     this.options = metadata.options ?? ({} as LightNovelWPOptions);
     this.filters = metadata.filters satisfies Filters;
 
@@ -220,7 +220,13 @@ export class LightNovelWPPlugin implements Plugin.PluginBase {
           isReadingChapter = true;
         } else if (isReadingChapter) {
           if (name === 'a' && tempChapter.path === undefined) {
-            tempChapter.path = attribs['href'].replace(baseURL, '').trim();
+            // Some installs link chapters with the absolute URL of the main
+            // site (free.kolnovel.com links kolnovel.com), and parseChapter
+            // re-prefixes this.site, so keep the path only.
+            tempChapter.path = attribs['href']
+              .replace(baseURL, '')
+              .replace(/^https?:\/\/[^/]+/, '')
+              .trim();
           } else if (attribs['class'] === 'epl-num') {
             isReadingChapterInfo = 1;
           } else if (attribs['class'] === 'epl-title') {
@@ -438,11 +444,20 @@ export class LightNovelWPPlugin implements Plugin.PluginBase {
         throw error;
       }
     }
+
+    // The body is the theme's `.epcontent` block, which newer installs render
+    // as an <article> instead of a <div> and no longer close with
+    // `<div class="bottomnav">`. The old regex sliced the raw document between
+    // those two wrappers, so both of those changes emptied every chapter;
+    // select the body itself and read its paragraphs instead.
+    const $ = load(data);
     return (
-      data
-        .match(/<div.*?class="epcontent ([^]*?)<div.*?class="?bottomnav/g)?.[0]
-        .match(/<p[^>]*>([^]*?)<\/p>/g)
-        ?.join('\n') || ''
+      $('.epcontent')
+        .first()
+        .find('p')
+        .map((_, el) => $.html(el) || '')
+        .get()
+        .join('\n') || ''
     );
   }
 

plugins/multisrc/madara/sources.json

@@ -137,7 +137,7 @@
     "options": {
       "useNewChapterEndpoint": true,
       "lang": "Arabic",
-      "customJs": "chapterText.find('span[style*=\"opacity: 0; position: fixed;\"],[role=\"presentation\"]').remove();"
+      "customJs": "chapterText.find('style,script,[inert],[data-nosnippet],[data-nhv-reader-promo]').remove();chapterText.find('span[style*=\"opacity: 0; position: fixed;\"],[role=\"presentation\"]').remove();"
     }
   },
   {

plugins/multisrc/madara/template.ts

@@ -46,7 +46,7 @@ export class MadaraPlugin implements Plugin.PluginBase {
     this.icon = `multisrc/madara/${metadata.id.toLowerCase()}/icon.png`;
     this.site = metadata.sourceSite;
     const versionIncrements = metadata.options?.versionIncrements || 0;
-    this.version = `2.2.${versionIncrements}`;
+    this.version = `2.3.${versionIncrements}`;
     this.options = metadata.options;
     this.filters = metadata.filters;
 
@@ -189,13 +189,17 @@ export class MadaraPlugin implements Plugin.PluginBase {
         loadedCheerio('.post-title h1').text().trim() ||
         loadedCheerio('#manga-title h1').text().trim() ||
         loadedCheerio('.manga-title').text().trim() ||
+        loadedCheerio('h1.nhv-novel-title').text().trim() ||
         '',
     };
 
     novel.cover =
       loadedCheerio('.summary_image > a > img').attr('data-lazy-src') ||
       loadedCheerio('.summary_image > a > img').attr('data-src') ||
       loadedCheerio('.summary_image > a > img').attr('src') ||
+      loadedCheerio('.nhv-novel-cover img').attr('data-lazy-src') ||
+      loadedCheerio('.nhv-novel-cover img').attr('data-src') ||
+      loadedCheerio('.nhv-novel-cover img').attr('src') ||
       defaultCover;
 
     loadedCheerio('.post-content_item, .post-content').each(function () {
@@ -251,6 +255,36 @@ export class MadaraPlugin implements Plugin.PluginBase {
       }
     });
 
+    // Fallback for the custom "NHV" theme (e.g. cenele.com), which renders
+    // novel pages without the classic Madara detail blocks. Every field is
+    // only filled when the selectors above left it empty, so sources on the
+    // classic theme keep their existing behaviour.
+    {
+      if (!novel.author)
+        novel.author =
+          loadedCheerio('.nhv-novel-meta a[href*="cont-author"]')
+            .first()
+            .text()
+            .trim() ||
+          loadedCheerio('a[href*="cont-author"]').first().text().trim();
+      if (!novel.genres)
+        novel.genres = loadedCheerio('.nhv-novel-genres a')
+          .map((i, el) => loadedCheerio(el).text())
+          .get()
+          .join(', ');
+      const nhvStatus = loadedCheerio('.nhv-novel-status').text().trim();
+      if (!novel.status && nhvStatus)
+        novel.status = nhvStatus.includes('مستمرة')
+          ? NovelStatus.Ongoing
+          : NovelStatus.Completed;
+      if (!novel.rating) {
+        const nhvRating = parseFloat(
+          loadedCheerio('.nhv-simple-rating__avg').text().trim(),
+        );
+
+        if (!isNaN(nhvRating)) novel.rating = nhvRating;
+      }
+    }
     // Checks for "Madara NovelHub" version
     {
       if (!novel.genres)
@@ -292,6 +326,11 @@ export class MadaraPlugin implements Plugin.PluginBase {
         .join('\n\n')
         .trim() ||
       loadedCheerio('.manga-excerpt p')
+        .map((i, el) => loadedCheerio(el).text())
+        .get()
+        .join('\n\n')
+        .trim() ||
+      loadedCheerio('.nhv-novel-synopsis p')
         .map((i, el) => loadedCheerio(el).text())
         .get()
         .join('\n\n')
@@ -385,11 +424,29 @@ export class MadaraPlugin implements Plugin.PluginBase {
 
   async parseChapter(chapterPath: string): Promise<string> {
     const loadedCheerio = await this.getCheerio(this.site + chapterPath, false);
-    const chapterText =
-      loadedCheerio('.text-left') ||
-      loadedCheerio('.text-right') ||
-      loadedCheerio('.entry-content') ||
-      loadedCheerio('.c-blog-post > div > div:nth-child(2)');
+
+    // A cheerio selection is always truthy, so the previous
+    // `$('.text-left') || $('.text-right') || …` chain never looked past the
+    // first selector: on a site that no longer renders `.text-left` it
+    // returned an empty body no matter what else the page offered. Walk the
+    // candidates and keep the first one that actually holds text.
+    let chapterText = loadedCheerio('.__no-chapter-content__');
+    for (const selector of [
+      '.text-left',
+      '.text-right',
+      '.text-content',
+      '.text-chapter-content',
+      'novel-chapter',
+      '.reading-content',
+      '.entry-content',
+      '.c-blog-post > div > div:nth-child(2)',
+    ]) {
+      const candidate = loadedCheerio(selector);
+      if (candidate.text().trim()) {
+        chapterText = candidate;
+        break;
+      }
+    }
 
     if (this.options?.customJs) {
       try {