refactor: Update Read From Net to Use Desktop UA, URL Method, and Inline Tags in parseChapter (#2117)

plugins/english/readfrom.ts

@@ -4,14 +4,22 @@ import { Filters } from '@libs/filterInputs';
 import { CheerioAPI, load as loadCheerio } from 'cheerio';
 import { defaultCover } from '@libs/defaultCover';
 
+const pluginHeaders = {
+  'User-Agent':
+    'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/146.0.0.0 Safari/537.36',
+};
+
 class ReadFromPlugin implements Plugin.PluginBase {
   id = 'readfrom';
   name = 'Read From Net';
   icon = 'src/en/readfrom/icon.png';
-  site = 'https://readfrom.net';
-  version = '1.0.2';
+  site = 'https://readfrom.net/';
+  version = '1.1.0';
   filters: Filters | undefined = undefined;
-  imageRequestInit?: Plugin.ImageRequestInit | undefined = undefined;
+  headers = new Headers(pluginHeaders);
+  imageRequestInit: Plugin.ImageRequestInit = {
+    headers: pluginHeaders,
+  };
 
   //flag indicates whether access to LocalStorage, SesesionStorage is required.
   webStorageUtilized?: boolean;
@@ -34,6 +42,9 @@ class ReadFromPlugin implements Plugin.PluginBase {
       (isSearch ? 'div.text' : '#dle-content') + ' > article.box',
     )
       .map((i, el) => {
+        const $el = loadedCheerio(el);
+        const novelPath = $el.find('h2.title a').attr('href');
+        if (!novelPath) return;
         const summary = loadedCheerio(el).find(
           isSearch ? 'div.text5' : 'div.text3',
         )[0];
@@ -41,11 +52,7 @@ class ReadFromPlugin implements Plugin.PluginBase {
         loadedCheerio(summary).find('a').remove();
         return {
           name: loadedCheerio(el).find('h2.title').text().trim(),
-          path: loadedCheerio(el)
-            .find('h2.title > a')
-            .attr('href')!
-            .replace('https://readfrom.net/', '')
-            .replace(/^\//, ''),
+          path: new URL(novelPath, this.site).pathname.substring(1),
           cover: loadedCheerio(el).find('img').attr('src') || defaultCover,
           summary:
             loadedCheerio(summary).text().trim() +
@@ -78,21 +85,20 @@ class ReadFromPlugin implements Plugin.PluginBase {
 
   async popularNovels(
     pageNo: number,
-    {
-      showLatestNovels,
-      filters,
-    }: Plugin.PopularNovelsOptions<typeof this.filters>,
+    { showLatestNovels }: Plugin.PopularNovelsOptions<typeof this.filters>,
   ) {
     const type = showLatestNovels ? 'last_added_books' : 'allbooks';
-    const res = await fetchApi(
-      'https://readfrom.net/' + type + '/page/' + pageNo + '/',
-    );
+    const res = await fetchApi(this.site + type + '/page/' + pageNo, {
+      headers: this.headers,
+    });
     const text = await res.text();
     return this.parseNovels(loadCheerio(text));
   }
 
   async parseNovel(novelPath: string): Promise<Plugin.SourceNovel> {
-    const data = await fetchApi('https://readfrom.net/' + novelPath);
+    const data = await fetchApi(this.site + novelPath, {
+      headers: this.headers,
+    });
     const text = await data.text();
     const loadedCheerio = loadCheerio(text);
 
@@ -109,25 +115,22 @@ class ReadFromPlugin implements Plugin.PluginBase {
       loadedCheerio('article.box > div > center > div > a > img').attr('src') ||
       defaultCover;
 
-    novel.chapters = loadedCheerio(loadedCheerio('div.pages').get()[0])
+    const rawChapters = loadedCheerio('div.pages')
+      .first()
       .find('> a')
-      .map((i, el) => {
-        return {
-          name: loadedCheerio(el).text().trim(),
-          path: loadedCheerio(el)
-            .attr('href')!
-            .replace('https://readfrom.net/', '')
-            .replace(/^\//, ''),
-          // releaseTime: '',
-          chapterNumber: i + 2,
-        };
-      })
-      .toArray();
-    novel.chapters.unshift({
-      name: '1',
-      path: novelPath,
-      chapterNumber: 1,
-    });
+      .toArray()
+      .flatMap(el => {
+        const href = loadedCheerio(el).attr('href');
+        if (!href) return [];
+
+        const path = new URL(href, this.site).pathname.substring(1);
+        return path ? [{ name: loadedCheerio(el).text().trim(), path }] : [];
+      });
+
+    novel.chapters = [
+      { name: '1', path: novelPath, chapterNumber: 1 },
+      ...rawChapters.map((ch, i) => ({ ...ch, chapterNumber: i + 2 })),
+    ];
 
     let moreNovelInfo = this.loadedNovelCache.find(
       novel => novel.path === novelPath,
@@ -156,43 +159,45 @@ class ReadFromPlugin implements Plugin.PluginBase {
   }
 
   async parseChapter(chapterPath: string): Promise<string> {
-    const data = await fetchApi('https://readfrom.net/' + chapterPath);
-    const text = await data.text();
-    const loadedCheerio = loadCheerio(text);
-    loadedCheerio('#textToRead > span:empty').remove();
-    loadedCheerio('#textToRead > center').remove();
-
-    const textToRead = loadedCheerio('#textToRead');
-
-    let paragraph: string[] = [];
-    const chapterHtml: string[] = [];
-
-    textToRead.contents().each((_, element) => {
-      switch (element.type) {
-        case 'text':
-          const content = element.data.trim();
-          if (content) {
-            paragraph.push(content);
-          }
-          break;
-
-        case 'tag':
-          if (paragraph.length > 0) {
-            chapterHtml.push(`<p>${paragraph.join(' ')}</p>`);
-            paragraph = [];
-          }
-          if (element.tagName !== 'br') {
-            chapterHtml.push(loadedCheerio.html(element));
-          }
-          break;
-      }
+    const response = await fetchApi(this.site + chapterPath, {
+      headers: this.headers,
     });
+    const $ = loadCheerio(await response.text());
+    $('#textToRead > span:empty, #textToRead > center').remove();
 
-    // Close any remaining paragraph
-    if (paragraph.length > 0) {
-      chapterHtml.push(`<p>${paragraph.join(' ')}</p>`);
+    const chapterHtml: string[] = [];
+    let p: string[] = [];
+
+    const allowed = new Set(['b', 'i', 'u', 'strong', 'em', 'a']);
+    const flush = () =>
+      p.length && (chapterHtml.push(`<p>${p.join(' ').trim()}</p>`), (p = []));
+
+    for (const el of $('#textToRead').contents().toArray()) {
+      switch (el.type) {
+        case 'comment':
+          continue;
+        case 'text':
+          if (el.data.trim()) {
+            // Convert _text_ to <i>text</i>
+            const jbText = el.data.trim().replace(/_([^_]+)_/g, '<i>$1</i>');
+            p.push(jbText);
+          }
+          continue;
+        case 'tag':
+          if (allowed.has(el.name)) {
+            p.push($.html(el));
+            continue;
+          }
+          if (el.name === 'br') {
+            flush();
+            continue;
+          }
+      }
+      flush();
+      chapterHtml.push($.html(el));
     }
 
+    flush();
     return chapterHtml.join('');
   }
 
@@ -201,12 +206,13 @@ class ReadFromPlugin implements Plugin.PluginBase {
     const res = await fetchApi(
       'https://readfrom.net/build_in_search/?q=' +
         encodeURIComponent(searchTerm),
+      { headers: this.headers },
     );
     const text = await res.text();
     return this.parseNovels(loadCheerio(text), true);
   }
 
-  resolveUrl = (path: string, isNovel?: boolean) => this.site + '/' + path;
+  // resolveUrl = (path: string, isNovel?: boolean) => this.site + '/' + path;
 }
 
 export default new ReadFromPlugin();