@jackwener/opencli 1.6.2 → 1.6.3

package/src/clis/douban/reviews.ts

@@ -1,127 +0,0 @@
-import { cli, Strategy } from '../../registry.js';
-import type { IPage } from '../../types.js';
-import { getSelfUid, DoubanReview } from './utils.js';
-
-cli({
-  site: 'douban',
-  name: 'reviews',
-  description: '导出个人影评',
-  domain: 'movie.douban.com',
-  strategy: Strategy.COOKIE,
-  args: [
-    { name: 'limit', type: 'int', default: 20, help: '导出数量' },
-    { name: 'uid', help: '用户ID，不填则使用当前登录账号' },
-    { name: 'full', type: 'bool', default: false, help: '获取完整影评内容' },
-  ],
-  columns: ['movieTitle', 'title', 'myRating', 'votes', 'content', 'url'],
-  func: async (page: IPage, kwargs: { limit?: number; uid?: string; full?: boolean }) => {
-    const { limit = 20, uid: providedUid, full = false } = kwargs;
-    
-    const uid = providedUid || await getSelfUid(page);
-    const reviews = await fetchReviews(page, uid, limit, full);
-    
-    return reviews;
-  },
-});
-
-async function fetchReviews(
-  page: IPage,
-  uid: string,
-  limit: number,
-  full: boolean,
-): Promise<DoubanReview[]> {
-  const reviews: DoubanReview[] = [];
-  let start = 0;
-  const pageSize = 20;
-  
-  while (true) {
-    const url = `https://movie.douban.com/people/${uid}/reviews?start=${start}&sort=time`;
-    
-    await page.goto(url);
-    
-    await page.wait({ time: 1 });
-    
-    const data = await page.evaluate(`
-      () => {
-        const reviews = [];
-        
-        document.querySelectorAll('.tlst').forEach(el => {
-          const movieLinkEl = el.querySelector('.ilst a');
-          const reviewTitleEl = el.querySelector('.nlst a[title]');
-          const ratingEl = el.querySelector('.clst span[class*="allstar"]');
-          const contentEl = el.querySelector('.review-short span');
-          const votesEl = el.querySelector('.review-short .pl span');
-          
-          const movieHref = movieLinkEl?.href || '';
-          const movieId = movieHref.match(/subject\\/(\\d+)/)?.[1] || '';
-          const movieTitle = movieLinkEl?.getAttribute('title') || movieLinkEl?.textContent?.trim() || '';
-          
-          const reviewHref = reviewTitleEl?.href || '';
-          const reviewId = reviewHref.match(/reviews\\/(\\d+)/)?.[1] || '';
-          const title = reviewTitleEl?.textContent?.trim() || '';
-          
-          let myRating = 0;
-          if (ratingEl) {
-            const cls = ratingEl.className || '';
-            const ratingMatch = cls.match(/allstar(\\d)0/);
-            if (ratingMatch) {
-              myRating = parseInt(ratingMatch[1], 10) * 2;
-            }
-          }
-          
-          const votesText = votesEl?.textContent || '';
-          const votesMatch = votesText.match(/(\\d+)/);
-          const votes = votesMatch ? parseInt(votesMatch[1], 10) : 0;
-          
-          reviews.push({
-            reviewId,
-            movieId,
-            movieTitle,
-            title,
-            content: contentEl?.textContent?.trim() || '',
-            myRating,
-            createdAt: '',
-            votes,
-            url: reviewHref,
-          });
-        });
-        
-        return reviews;
-      }
-    `) as DoubanReview[];
-    
-    reviews.push(...data);
-    
-    if (data.length < pageSize) break;
-    if (limit > 0 && reviews.length >= limit) break;
-    
-    start += pageSize;
-  }
-  
-  const result = reviews.slice(0, limit > 0 ? limit : undefined);
-  
-  if (full && result.length > 0) {
-    for (const review of result) {
-      if (review.url) {
-        const fullContent = await fetchFullReview(page, review.url);
-        review.content = fullContent;
-      }
-    }
-  }
-  
-  return result;
-}
-
-async function fetchFullReview(page: IPage, reviewUrl: string): Promise<string> {
-  await page.goto(reviewUrl);
-  await page.wait({ time: 1 });
-  
-  const content = await page.evaluate(`
-    () => {
-      const contentEl = document.querySelector('.review-content');
-      return contentEl?.textContent?.trim() || '';
-    }
-  `) as string;
-  
-  return content;
-}

package/src/clis/douban/search.ts

@@ -1,17 +0,0 @@
-import { cli, Strategy } from '../../registry.js';
-import { searchDouban } from './utils.js';
-
-cli({
-  site: 'douban',
-  name: 'search',
-  description: '搜索豆瓣电影、图书或音乐',
-  domain: 'search.douban.com',
-  strategy: Strategy.COOKIE,
-  args: [
-    { name: 'type', default: 'movie', choices: ['movie', 'book', 'music'], help: '搜索类型（movie=电影, book=图书, music=音乐）' },
-    { name: 'keyword', required: true, positional: true, help: '搜索关键词' },
-    { name: 'limit', type: 'int', default: 20, help: '返回结果数量' },
-  ],
-  columns: ['rank', 'title', 'rating', 'abstract', 'url'],
-  func: async (page, args) => searchDouban(page, args.type, args.keyword, Number(args.limit) || 20),
-});

package/src/clis/douban/subject.yaml

@@ -1,107 +0,0 @@
-site: douban
-name: subject
-description: 获取电影详情
-domain: movie.douban.com
-strategy: cookie
-browser: true
-
-args:
-  id:
-    positional: true
-    required: true
-    type: str
-    description: 电影 ID
-
-pipeline:
-  - navigate: https://movie.douban.com/subject/${{ args.id }}
-
-  - evaluate: |
-      (async () => {
-        const id = '${{ args.id }}';
-
-        // Wait for page to load
-        await new Promise(r => setTimeout(r, 2000));
-
-        // Extract title - v:itemreviewed contains "中文名 OriginalName"
-        const titleEl = document.querySelector('span[property="v:itemreviewed"]');
-        const fullTitle = titleEl?.textContent?.trim() || '';
-
-        // Split title and originalTitle
-        // Douban format: "中文名 OriginalName" - split by first space that separates CJK from non-CJK
-        let title = fullTitle;
-        let originalTitle = '';
-        const titleMatch = fullTitle.match(/^([\u4e00-\u9fff\u3000-\u303f\uff00-\uffef]+(?:\s*[\u4e00-\u9fff\u3000-\u303f\uff00-\uffef·：:！？]+)*)\s+(.+)$/);
-        if (titleMatch) {
-          title = titleMatch[1].trim();
-          originalTitle = titleMatch[2].trim();
-        }
-
-        // Extract year
-        const yearEl = document.querySelector('.year');
-        const year = yearEl?.textContent?.trim().replace(/[()（）]/g, '') || '';
-
-        // Extract rating
-        const ratingEl = document.querySelector('strong[property="v:average"]');
-        const rating = parseFloat(ratingEl?.textContent || '0');
-
-        // Extract rating count
-        const ratingCountEl = document.querySelector('span[property="v:votes"]');
-        const ratingCount = parseInt(ratingCountEl?.textContent || '0', 10);
-
-        // Extract genres
-        const genreEls = document.querySelectorAll('span[property="v:genre"]');
-        const genres = Array.from(genreEls).map(el => el.textContent?.trim()).filter(Boolean).join(',');
-
-        // Extract directors
-        const directorEls = document.querySelectorAll('a[rel="v:directedBy"]');
-        const directors = Array.from(directorEls).map(el => el.textContent?.trim()).filter(Boolean).join(',');
-
-        // Extract casts
-        const castEls = document.querySelectorAll('a[rel="v:starring"]');
-        const casts = Array.from(castEls).slice(0, 5).map(el => el.textContent?.trim()).filter(Boolean);
-
-        // Extract info section for country and duration
-        const infoEl = document.querySelector('#info');
-        const infoText = infoEl?.textContent || '';
-
-        // Extract country/region from #info as list
-        let country = [];
-        const countryMatch = infoText.match(/制片国家\/地区:\s*([^\n]+)/);
-        if (countryMatch) {
-          country = countryMatch[1].trim().split(/\s*\/\s*/).filter(Boolean);
-        }
-
-        // Extract duration from #info as pure number in min
-        const durationEl = document.querySelector('span[property="v:runtime"]');
-        let durationRaw = durationEl?.textContent?.trim() || '';
-        if (!durationRaw) {
-          const durationMatch = infoText.match(/片长:\s*([^\n]+)/);
-          if (durationMatch) {
-            durationRaw = durationMatch[1].trim();
-          }
-        }
-        const durationNumMatch = durationRaw.match(/(\d+)/);
-        const duration = durationNumMatch ? parseInt(durationNumMatch[1], 10) : null;
-
-        // Extract summary
-        const summaryEl = document.querySelector('span[property="v:summary"]');
-        const summary = summaryEl?.textContent?.trim() || '';
-
-        return [{
-          id,
-          title,
-          originalTitle,
-          year,
-          rating,
-          ratingCount,
-          genres,
-          directors,
-          casts,
-          country,
-          duration,
-          summary: summary.substring(0, 200),
-          url: `https://movie.douban.com/subject/${id}`
-        }];
-      })()
-
-columns: [id, title, originalTitle, year, rating, ratingCount, genres, directors, casts, country, duration, summary, url]

package/src/clis/douban/top250.yaml

@@ -1,70 +0,0 @@
-site: douban
-name: top250
-description: 豆瓣电影 Top250
-domain: movie.douban.com
-strategy: cookie
-browser: true
-
-args:
-  limit:
-    type: int
-    default: 250
-    description: 返回结果数量
-
-pipeline:
-  - navigate: https://movie.douban.com/top250
-
-  - evaluate: |
-      async () => {
-        const results = [];
-        const limit = ${{ args.limit }};
-
-        const parsePage = (doc) => {
-          const items = doc.querySelectorAll('.item');
-          for (const item of items) {
-            if (results.length >= limit) break;
-
-            const rankEl = item.querySelector('.pic em');
-            const linkEl = item.querySelector('a');
-            const titleEl = item.querySelector('.title');
-            const ratingEl = item.querySelector('.rating_num');
-
-            const href = linkEl?.href || '';
-            const matchResult = href.match(/subject\/(\d+)/);
-            const id = matchResult ? matchResult[1] : '';
-
-            const title = titleEl?.textContent?.trim() || '';
-            const rank = parseInt(rankEl?.textContent || '0', 10);
-            const rating = ratingEl?.textContent?.trim() || '';
-
-            if (id && title) {
-              results.push({
-                rank: rank || results.length + 1,
-                id,
-                title,
-                rating: rating ? parseFloat(rating) : 0,
-                url: href
-              });
-            }
-          }
-        };
-
-        parsePage(document);
-
-        for (let start = 25; start < 250 && results.length < limit; start += 25) {
-          const resp = await fetch(`https://movie.douban.com/top250?start=${start}`);
-          if (!resp.ok) break;
-          const html = await resp.text();
-          if (!html) break;
-
-          const doc = new DOMParser().parseFromString(html, 'text/html');
-          parsePage(doc);
-          await new Promise(r => setTimeout(r, 150));
-        }
-
-        return results;
-      }
-
-  - limit: ${{ args.limit }}
-
-columns: [rank, id, title, rating, url]

package/src/clis/douban/utils.test.ts

@@ -1,97 +0,0 @@
-import { describe, expect, it, vi } from 'vitest';
-import type { IPage } from '../../types.js';
-import {
-  getDoubanPhotoExtension,
-  loadDoubanSubjectPhotos,
-  normalizeDoubanSubjectId,
-  promoteDoubanPhotoUrl,
-  resolveDoubanPhotoAssetUrl,
-} from './utils.js';
-
-describe('douban utils', () => {
-  it('normalizes valid subject ids', () => {
-    expect(normalizeDoubanSubjectId(' 30382501 ')).toBe('30382501');
-  });
-
-  it('rejects invalid subject ids', () => {
-    expect(() => normalizeDoubanSubjectId('tt30382501')).toThrow('Invalid Douban subject ID');
-  });
-
-  it('promotes thumbnail urls to large photo urls', () => {
-    expect(
-      promoteDoubanPhotoUrl('https://img1.doubanio.com/view/photo/m/public/p2913450214.webp'),
-    ).toBe('https://img1.doubanio.com/view/photo/l/public/p2913450214.webp');
-
-    expect(
-      promoteDoubanPhotoUrl('https://img9.doubanio.com/view/photo/s_ratio_poster/public/p2578474613.jpg'),
-    ).toBe('https://img9.doubanio.com/view/photo/l/public/p2578474613.jpg');
-  });
-
-  it('rejects non-http photo urls during promotion', () => {
-    expect(promoteDoubanPhotoUrl('data:image/gif;base64,abc')).toBe('');
-  });
-
-  it('prefers lazy-loaded photo urls over data placeholders', () => {
-    expect(
-      resolveDoubanPhotoAssetUrl([
-        '',
-        'https://img1.doubanio.com/view/photo/m/public/p2913450214.webp',
-        'data:image/gif;base64,abc',
-      ], 'https://movie.douban.com/subject/30382501/photos?type=Rb'),
-    ).toBe('https://img1.doubanio.com/view/photo/m/public/p2913450214.webp');
-  });
-
-  it('drops unsupported non-http photo urls when no real image url exists', () => {
-    expect(
-      resolveDoubanPhotoAssetUrl(
-        ['data:image/gif;base64,abc', 'blob:https://movie.douban.com/example'],
-        'https://movie.douban.com/subject/30382501/photos?type=Rb',
-      ),
-    ).toBe('');
-  });
-
-  it('removes the default photo cap when scanning for an exact photo id', async () => {
-    const evaluate = vi.fn()
-      .mockResolvedValueOnce({ blocked: false, title: 'Some Movie', href: 'https://movie.douban.com/subject/30382501/photos?type=Rb' })
-      .mockResolvedValueOnce({
-        subjectId: '30382501',
-        subjectTitle: 'The Wandering Earth 2',
-        type: 'Rb',
-        photos: [
-          {
-            index: 731,
-            photoId: '2913450215',
-            title: 'Character poster',
-            imageUrl: 'https://img1.doubanio.com/view/photo/l/public/p2913450215.jpg',
-            thumbUrl: 'https://img1.doubanio.com/view/photo/m/public/p2913450215.jpg',
-            detailUrl: 'https://movie.douban.com/photos/photo/2913450215/',
-            page: 25,
-          },
-        ],
-      });
-    const page = {
-      goto: vi.fn().mockResolvedValue(undefined),
-      wait: vi.fn().mockResolvedValue(undefined),
-      evaluate,
-    } as unknown as IPage;
-
-    await loadDoubanSubjectPhotos(page, '30382501', {
-      type: 'Rb',
-      targetPhotoId: '2913450215',
-    });
-
-    const scanScript = evaluate.mock.calls[1]?.[0];
-    expect(scanScript).toContain('const targetPhotoId = "2913450215";');
-    expect(scanScript).toContain(`const limit = ${Number.MAX_SAFE_INTEGER};`);
-    expect(scanScript).toContain('for (let pageIndex = 0; photos.length < limit; pageIndex += 1)');
-  });
-
-  it('keeps image extensions when download urls contain query params', () => {
-    expect(
-      getDoubanPhotoExtension('https://img1.doubanio.com/view/photo/l/public/p2913450214.webp?foo=1'),
-    ).toBe('.webp');
-    expect(
-      getDoubanPhotoExtension('https://img1.doubanio.com/view/photo/l/public/p2913450214.jpeg'),
-    ).toBe('.jpeg');
-  });
-});