Add Toasteful source plugin (English)

plugins/english/toasteful.ts

@@ -0,0 +1,352 @@
+import { fetchApi } from '@libs/fetch';
+import { Plugin } from '@/types/plugin';
+import { Filters, FilterTypes } from '@libs/filterInputs';
+import { load as loadCheerio } from 'cheerio';
+import { defaultCover } from '@libs/defaultCover';
+import { NovelStatus } from '@libs/novelStatus';
+
+type FeedLink = { rel: string; href: string };
+type FeedEntry = {
+  title: { $t: string };
+  link: FeedLink[];
+  published?: { $t: string };
+  updated?: { $t: string };
+  category?: { term: string }[];
+  content?: { $t: string };
+};
+type FeedResponse = {
+  feed?: {
+    openSearch$totalResults?: { $t: string };
+    entry?: FeedEntry[];
+  };
+};
+
+const NOVEL_LABEL = 'Web Novel';
+const PAGE_SIZE = 25;
+const CHAPTER_PATH = /^\/\d{4}\/\d{2}\/[^/]+\.html$/;
+const PLACEHOLDER_COVER = 'No_image_available';
+
+class Toasteful implements Plugin.PluginBase {
+  id = 'toasteful';
+  name = 'Toasteful';
+  site = 'https://www.toasteful.com';
+  icon = 'src/en/toasteful/icon.png';
+  version = '1.0.1';
+
+  filters: Filters = {
+    genres: {
+      label: 'Genres',
+      value: [],
+      options: [
+        'Action',
+        'Adventure',
+        'Comedy',
+        'Drama',
+        'Ecchi',
+        'Fantasy',
+        'Harem',
+        'Isekai',
+        'Reincarnation',
+        'Romance',
+        'Slice of Life',
+        'Slow Life',
+        'Supernatural',
+        'Overpowered',
+        'Villain',
+      ].map(label => ({ label, value: label })),
+      type: FilterTypes.CheckboxGroup,
+    },
+  };
+
+  private async request(url: string) {
+    const response = await fetchApi(url);
+    if (!response.ok) {
+      throw Object.assign(new Error(`Toasteful: HTTP ${response.status}.`), {
+        status: response.status,
+      });
+    }
+    return response;
+  }
+
+  private feedUrl(labels: string[], params: Record<string, string>): string {
+    const path = labels.map(label => encodeURIComponent(label)).join('/');
+    const query = new URLSearchParams({
+      'alt': 'json',
+      ...params,
+    }).toString();
+    return `${this.site}/feeds/posts/default/-/${path}?${query}`;
+  }
+
+  private entryToNovel(entry: FeedEntry): Plugin.NovelItem | undefined {
+    try {
+      const href = entry.link?.find(link => link.rel === 'alternate')?.href;
+      if (!href) return undefined;
+      const name = entry.title?.$t?.replace(/^\(NEW!!!\)\s*/, '').trim();
+      if (!name) return undefined;
+      const resolved = this.resolveUrl(href);
+      let cover: string | undefined;
+      const html = entry.content?.$t;
+      if (html) {
+        const src = loadCheerio(html)('img').first().attr('src');
+        if (src && !src.includes(PLACEHOLDER_COVER)) cover = src;
+      }
+      return {
+        name,
+        path: new URL(resolved).pathname,
+        cover: cover || defaultCover,
+      };
+    } catch {
+      return undefined;
+    }
+  }
+
+  async popularNovels(
+    pageNo: number,
+    {
+      showLatestNovels,
+      filters,
+    }: Plugin.PopularNovelsOptions<typeof this.filters>,
+  ): Promise<Plugin.NovelItem[]> {
+    const labels = [NOVEL_LABEL, ...(filters?.genres?.value || [])];
+    const params: Record<string, string> = {
+      'max-results': String(PAGE_SIZE),
+      'start-index': String((pageNo - 1) * PAGE_SIZE + 1),
+    };
+    if (showLatestNovels) params['orderby'] = 'updated';
+    const data: FeedResponse = await (
+      await this.request(this.feedUrl(labels, params))
+    ).json();
+    const entries = data.feed?.entry || [];
+    const novels: Plugin.NovelItem[] = [];
+    for (const entry of entries) {
+      const novel = this.entryToNovel(entry);
+      if (novel) novels.push(novel);
+    }
+    return novels;
+  }
+
+  async parseNovel(novelPath: string): Promise<Plugin.SourceNovel> {
+    const url = this.resolveUrl(novelPath);
+    const $ = loadCheerio(await (await this.request(url)).text());
+    const name =
+      $('h3.post-title').text().trim() ||
+      $('meta[property="og:title"]').attr('content')?.trim() ||
+      'Untitled';
+
+    const coverSrc =
+      $('link[rel="image_src"]').attr('href') ||
+      $('.post-body img').first().attr('src');
+    const cover =
+      coverSrc && !coverSrc.includes(PLACEHOLDER_COVER)
+        ? coverSrc
+        : defaultCover;
+
+    const labels = $('span.post-labels a[rel="tag"]')
+      .toArray()
+      .map(el => $(el).text().trim())
+      .filter(Boolean);
+    const statusLabel = labels
+      .find(label => /completed|on-?going|dropped|hiatus/i.test(label))
+      ?.toLowerCase();
+    let status: string | undefined;
+    if (statusLabel) {
+      if (/completed/.test(statusLabel)) status = NovelStatus.Completed;
+      else if (/on-?going/.test(statusLabel)) status = NovelStatus.Ongoing;
+      else if (/dropped/.test(statusLabel)) status = NovelStatus.Cancelled;
+      else if (/hiatus/.test(statusLabel)) status = NovelStatus.OnHiatus;
+    }
+    const normalize = (text: string) =>
+      text.toLowerCase().replace(/[^a-z0-9]/g, '');
+    const novelLabel = normalize(name);
+    const genres = labels
+      .filter(
+        label =>
+          !/completed|on-?going|dropped|hiatus/i.test(label) &&
+          label.toLowerCase() !== NOVEL_LABEL.toLowerCase() &&
+          label.toLowerCase() !== 'light novel' &&
+          normalize(label) !== novelLabel,
+      )
+      .join(', ');
+
+    const authorText = $('.post-body')
+      .find('b')
+      .toArray()
+      .map(el => $(el).parent().text())
+      .find(text => /^author:/i.test(text.trim()));
+    const author = authorText
+      ? authorText
+          .trim()
+          .replace(/^author:\s*/i, '')
+          .split(/\n|genre:|raw\s*source:|raw:|status:|synopsis:/i)[0]
+          ?.trim() || undefined
+      : undefined;
+
+    const body = $('.post-body').first();
+    let summary = '';
+    const article = body.find('article').first();
+    if (article.length) {
+      const synopsis = article.clone();
+      synopsis.find('b').each((_, el) => {
+        if (/^synopsis:/i.test($(el).text().trim())) $(el).remove();
+      });
+      summary = synopsis.text().trim();
+    } else {
+      const paragraphs: string[] = [];
+      let inChapters = false;
+      body.find('p, a[href]').each((_, el) => {
+        const node = $(el);
+        if (node.is('a')) {
+          if (this.isChapterLink(node.attr('href'), url, node.text()))
+            inChapters = true;
+          return;
+        }
+        if (inChapters) return;
+        const text = node.text().trim();
+        if (!text || /^genre:/i.test(text) || /raw source:/i.test(text)) return;
+        paragraphs.push(text);
+      });
+      summary = paragraphs.join('\n\n');
+    }
+
+    const chapters: Plugin.ChapterItem[] = [];
+    const seen = new Set<string>();
+    let volume: string | undefined;
+    body.find('b, a[href]').each((_, el) => {
+      const node = $(el);
+      if (node.is('a')) {
+        const href = node.attr('href');
+        if (!this.isChapterLink(href, url, node.text())) return;
+        const chapterPath = new URL(href!, this.site).pathname;
+        if (seen.has(chapterPath)) return;
+        seen.add(chapterPath);
+        const chapterName = node.text().trim().replace(/\s+/g, ' ');
+        if (!chapterName) return;
+        const number = /chapter\s+(\d+(?:\.\d+)?)/i.exec(chapterName)?.[1];
+        chapters.push({
+          name: volume ? `${volume} - ${chapterName}` : chapterName,
+          path: chapterPath,
+          chapterNumber: number ? Number(number) : undefined,
+        });
+        return;
+      }
+      if (node.find('a[href]').length) return;
+      const text = node.text().trim().replace(/\s+/g, ' ');
+      if (/^(volume|afterstory)\b/i.test(text)) volume = text;
+    });
+
+    if (!chapters.length)
+      throw new Error('Toasteful: no chapters were found on this page.');
+
+    return {
+      name,
+      path: new URL(url).pathname,
+      cover,
+      summary: summary || undefined,
+      author,
+      genres: genres || undefined,
+      status,
+      chapters,
+    };
+  }
+
+  private isChapterLink(
+    href: string | undefined,
+    pageUrl: string,
+    linkText: string,
+  ): boolean {
+    if (!href) return false;
+    let path: string;
+    try {
+      const parsed = new URL(href, this.site);
+      if (parsed.origin !== this.site) return false;
+      path = parsed.pathname;
+    } catch {
+      return false;
+    }
+    if (!CHAPTER_PATH.test(path)) return false;
+    if (path === new URL(pageUrl).pathname) return false;
+    // Side-story hubs are linked as bare "here" pointers, not chapters.
+    if (linkText.trim().toLowerCase() === 'here') return false;
+    if (
+      !/\b(chapter|episode|part|prologue|epilogue|interlude|afterstory|volume)\b|\bside\s*story\b/i.test(
+        linkText,
+      )
+    )
+      return false;
+    return true;
+  }
+
+  async parseChapter(chapterPath: string): Promise<string> {
+    const url = this.resolveUrl(chapterPath);
+    const $ = loadCheerio(await (await this.request(url)).text());
+    const content = $('.content-post').first().length
+      ? $('.content-post').first()
+      : $('.post-body').first();
+    content.find('script,style,iframe,form,noscript,.ChapterNav').remove();
+    content
+      .find(
+        '.reader-toolbar,.js-notice,.adsbygoogle,.ad,.ads,[class^="ad-"],[class*=" ad-"]',
+      )
+      .remove();
+    content.find('a[href*="docs.google.com/forms"]').remove();
+    const junk =
+      /hello,?\s+we have a (new|other)|we also have a similar genre|please bookmark this series|report error/i;
+    content.find('div,p').each((_, el) => {
+      const node = $(el);
+      if (node.is('div') && node.find('p,div').length) return;
+      if (junk.test(node.text().trim())) node.remove();
+    });
+    const chapter = content.html() || '';
+    if (loadCheerio(chapter).text().trim().length < 200)
+      throw new Error('Toasteful: no readable chapter text was found.');
+    return chapter;
+  }
+
+  async searchNovels(
+    searchTerm: string,
+    pageNo: number,
+  ): Promise<Plugin.NovelItem[]> {
+    const tokens = searchTerm
+      .toLowerCase()
+      .split(/[^a-z0-9]+/)
+      .filter(Boolean);
+    if (!tokens.length) return [];
+    // Blogger ignores the /-/label path once q is present, so a q search
+    // returns chapter posts that crowd novels out of the window. Search the
+    // labeled novel list instead: one feed holds the whole catalogue, so
+    // matching novels can never be crowded out and paging is exact.
+    const novels: Plugin.NovelItem[] = [];
+    let start = 1;
+    for (let round = 0; round < 4; round++) {
+      const data: FeedResponse = await (
+        await this.request(
+          this.feedUrl([NOVEL_LABEL], {
+            'max-results': '150',
+            'start-index': String(start),
+          }),
+        )
+      ).json();
+      const entries = data.feed?.entry || [];
+      if (!entries.length) break;
+      for (const entry of entries) {
+        const novel = this.entryToNovel(entry);
+        if (!novel || novels.some(item => item.path === novel.path)) continue;
+        const name = novel.name.toLowerCase();
+        if (tokens.every(token => name.includes(token))) novels.push(novel);
+      }
+      const total = Number(data.feed?.openSearch$totalResults?.$t || 0);
+      start += entries.length;
+      if (!total || start > total) break;
+    }
+    return novels.slice((pageNo - 1) * PAGE_SIZE, pageNo * PAGE_SIZE);
+  }
+
+  resolveUrl(path: string): string {
+    const url = new URL(path, this.site);
+    if (url.origin !== this.site)
+      throw new Error('Toasteful: invalid page address.');
+    return url.href;
+  }
+}
+
+export default new Toasteful();