@sovovs/bycli 2.1.37 → 2.1.39

package/clis/52pojie/search.js

@@ -0,0 +1,32 @@
+import { cli, Strategy } from '@sovovs/bycli/registry';
+import { ArgumentError, CommandExecutionError } from '@sovovs/bycli/errors';
+import { emptySearchResults, requireBoundedInteger, requireSearchQuery, runBrowserStep } from '../_shared/search-adapter.js';
+
+const SORTS = ['relevance', 'latest', 'replies', 'views'];
+
+export function buildSearchUrl({ keyword, limit = 10, page = 1, section, sort }) {
+  const url = new URL('https://www.52pojie.cn/search.php'); url.searchParams.set('srchtxt', keyword); url.searchParams.set('page', String(page)); url.searchParams.set('perpage', String(limit));
+  if (section) url.searchParams.set('section', section); if (sort) url.searchParams.set('sort', sort); return url.toString();
+}
+
+function extractItems() {
+  const body = document.body?.textContent || ''; const blocked = /验证码|访问频繁|captcha|access denied/i.test(body); const items = []; const seen = new Set(); const numberFrom = (card, selector) => { const value = (card.querySelector(selector)?.textContent || '').replace(/[^\d.]/g, ''); return value ? Number(value) : null; };
+  for (const card of document.querySelectorAll('.forum-item, .search-list li, .sltm')) {
+    const anchor = card.querySelector('.thread-title[href], h3 a[href], h2 a[href]'); if (!anchor || !/^https?:/i.test(anchor.href) || seen.has(anchor.href)) continue; seen.add(anchor.href); const time = card.querySelector('time');
+    items.push({ title: (anchor.textContent || '').trim(), url: anchor.href, snippet: (card.querySelector('.summary, .desc, p')?.textContent || '').trim(), displayUrl: new URL(anchor.href).hostname + new URL(anchor.href).pathname, source: '52pojie', resultType: 'thread', author: (card.querySelector('.author, .username')?.textContent || '').trim() || null, publishedAt: time?.getAttribute('datetime') || null, score: null, extra: { replies: numberFrom(card, '.replies, .reply'), views: numberFrom(card, '.views, .view'), section: (card.querySelector('.section, .forum')?.textContent || '').trim() || null } });
+  }
+  return { blocked, items };
+}
+
+export const command = cli({
+  site: '52pojie', name: 'search', access: 'read', description: 'Search 52pojie', domain: 'www.52pojie.cn', strategy: Strategy.PUBLIC, browser: true,
+  args: [{ name: 'keyword', positional: true, required: true, help: 'Search query' }, { name: 'limit', type: 'int', default: 10, help: 'Number of threads (1-50)' }, { name: 'page', type: 'int', default: 1, help: 'Result page number' }, { name: 'section', help: 'Forum section identifier' }, { name: 'sort', help: 'Sort order: relevance, latest, replies, views' }],
+  columns: ['rank', 'title', 'url', 'snippet', 'displayUrl', 'source', 'resultType', 'author', 'publishedAt', 'score', 'extra'],
+  func: async (page, kwargs) => {
+    const keyword = requireSearchQuery(kwargs.keyword); const limit = requireBoundedInteger(kwargs.limit, 10, 1, 50, '--limit'); const pageNumber = requireBoundedInteger(kwargs.page, 1, 1, 100, '--page'); if (kwargs.sort && !SORTS.includes(String(kwargs.sort))) throw new ArgumentError(`--sort must be one of: ${SORTS.join(', ')}`);
+    await runBrowserStep('52pojie search navigation', () => page.goto(buildSearchUrl({ keyword, limit, page: pageNumber, section: kwargs.section, sort: kwargs.sort }))); await page.wait({ selector: '.forum-item, .search-list, .sltm', timeout: 8 }).catch(() => page.wait(2).catch(() => {})); const data = await runBrowserStep('52pojie search extraction', () => page.evaluate(`(${extractItems.toString()})()`));
+    if (data?.blocked) throw new CommandExecutionError('52pojie search was blocked by a verification page', 'Complete the verification in the browser and retry.'); if (!data?.items?.length) throw emptySearchResults('52pojie', keyword); return data.items.slice(0, limit).map((item, index) => ({ rank: (pageNumber - 1) * limit + index + 1, ...item }));
+  },
+});
+
+export const __test__ = { command, buildSearchUrl };

package/clis/baidu/search.js

@@ -0,0 +1,68 @@
+import { cli, Strategy } from '@sovovs/bycli/registry';
+import { ArgumentError, CommandExecutionError } from '@sovovs/bycli/errors';
+import { emptySearchResults, requireBoundedInteger, requireNonNegativeInteger, requireSearchQuery, runBrowserStep, toHttpsUrl } from '../_shared/search-adapter.js';
+
+const TYPES = ['web', 'news', 'image', 'video'];
+const FILETYPES = ['pdf', 'doc', 'xls', 'ppt', 'rtf', 'all'];
+
+export function buildSearchUrl({ keyword, limit = 10, page = 1, site, filetype, platform, time }) {
+  const url = new URL('https://www.baidu.com/s');
+  url.searchParams.set('wd', keyword);
+  url.searchParams.set('pn', String((page - 1) * limit));
+  if (site) url.searchParams.set('si', site);
+  if (filetype) url.searchParams.set('ft', filetype);
+  if (platform) url.searchParams.set('ie', platform === 'mobile' ? 'utf-8' : 'utf-8');
+  if (time) url.searchParams.set('gpc', `stf=${time}`);
+  return url.toString();
+}
+
+function validateChoice(value, choices, flag) {
+  if (value !== undefined && !choices.includes(String(value))) throw new ArgumentError(`--${flag} must be one of: ${choices.join(', ')}`);
+  return value === undefined ? undefined : String(value);
+}
+
+function extractItems() {
+  const blocked = /验证码|安全验证|百度安全验证|访问过于频繁|请完成验证/.test(document.body?.textContent || '');
+  const items = [];
+  const seen = new Set();
+  for (const card of document.querySelectorAll('.result, .c-container')) {
+    const anchor = card.querySelector('h3 a, h3 a[href]');
+    if (!anchor) continue;
+    const url = anchor.href || '';
+    if (!/^https?:/i.test(url) || seen.has(url)) continue;
+    seen.add(url);
+    items.push({ title: (anchor.textContent || '').trim(), url, snippet: (card.querySelector('.c-abstract, .content-right_8Zs40, .c-span-last')?.textContent || '').trim(), displayUrl: (card.querySelector('.c-showurl, .c-color-gray2')?.textContent || '').trim(), resultType: 'web', extra: {} });
+  }
+  return { blocked, items };
+}
+
+export const command = cli({
+  site: 'baidu', name: 'search', access: 'read', description: 'Search Baidu', domain: 'www.baidu.com', strategy: Strategy.PUBLIC, browser: true,
+  args: [
+    { name: 'keyword', positional: true, required: true, help: 'Search query' },
+    { name: 'limit', type: 'int', default: 10, help: 'Number of results (1-50)' },
+    { name: 'page', type: 'int', default: 1, help: 'Result page number' },
+    { name: 'site', help: 'Restrict results to a domain' },
+    { name: 'filetype', help: 'File type: pdf, doc, xls, ppt, rtf, all' },
+    { name: 'platform', help: 'Client platform: pc or mobile' },
+    { name: 'time', help: 'Recent time filter understood by Baidu' },
+  ],
+  columns: ['rank', 'title', 'url', 'snippet', 'displayUrl', 'source', 'resultType', 'author', 'publishedAt', 'score', 'extra'],
+  func: async (page, kwargs) => {
+    const keyword = requireSearchQuery(kwargs.keyword);
+    const limit = requireBoundedInteger(kwargs.limit, 10, 1, 50, '--limit');
+    const pageNumber = requireBoundedInteger(kwargs.page, 1, 1, 100, '--page');
+    const filetype = validateChoice(kwargs.filetype, FILETYPES, 'filetype');
+    const platform = validateChoice(kwargs.platform, ['pc', 'mobile'], 'platform');
+    const raw = await runBrowserStep('Baidu search navigation', () => page.goto(buildSearchUrl({ keyword, limit, page: pageNumber, site: kwargs.site, filetype, platform, time: kwargs.time })));
+    void raw;
+    await page.wait({ selector: '.result, .c-container', timeout: 8 }).catch(() => page.wait(2).catch(() => {}));
+    const data = await runBrowserStep('Baidu search extraction', () => page.evaluate(`(${extractItems.toString()})()`));
+    if (data?.blocked) throw new CommandExecutionError('Baidu search was blocked by a verification page', 'Complete the verification in the browser and retry.');
+    const items = data?.items || [];
+    if (!items.length) throw emptySearchResults('Baidu', keyword);
+    return items.slice(0, limit).map((item, index) => ({ rank: (pageNumber - 1) * limit + index + 1, ...item, source: 'baidu', author: null, publishedAt: null, score: null }));
+  },
+});
+
+export const __test__ = { command, buildSearchUrl };

package/clis/bing/search.js

@@ -0,0 +1,58 @@
+import { cli, Strategy } from '@sovovs/bycli/registry';
+import { ArgumentError, CommandExecutionError } from '@sovovs/bycli/errors';
+import { emptySearchResults, requireBoundedInteger, requireSearchQuery, runBrowserStep } from '../_shared/search-adapter.js';
+
+const FRESHNESS = ['day', 'week', 'month', 'year'];
+const ANSWERS = ['web', 'news', 'video', 'images'];
+const SAFE = ['off', 'moderate', 'strict'];
+
+export function buildSearchUrl({ keyword, limit = 10, page = 1, freshness, market, answer, safe }) {
+  const url = new URL('https://www.bing.com/search');
+  url.searchParams.set('q', keyword);
+  url.searchParams.set('count', String(limit));
+  url.searchParams.set('first', String((page - 1) * limit + 1));
+  if (freshness) url.searchParams.set('freshness', freshness);
+  if (market) url.searchParams.set('cc', String(market).split('-').pop().toUpperCase());
+  if (answer && answer !== 'web') url.searchParams.set('scope', answer);
+  if (safe) url.searchParams.set('safesearch', safe);
+  return url.toString();
+}
+
+function validate(value, choices, flag) {
+  if (value !== undefined && !choices.includes(String(value))) throw new ArgumentError(`--${flag} must be one of: ${choices.join(', ')}`);
+  return value === undefined ? undefined : String(value);
+}
+
+function extractItems() {
+  const blocked = /unusual traffic|verify you are human|captcha|access denied/i.test(document.body?.textContent || '');
+  const items = [];
+  const seen = new Set();
+  for (const card of document.querySelectorAll('li.b_algo')) {
+    const anchor = card.querySelector('h2 a[href]');
+    if (!anchor || !/^https?:/i.test(anchor.href) || seen.has(anchor.href)) continue;
+    seen.add(anchor.href);
+    items.push({ title: (anchor.textContent || '').trim(), url: anchor.href, snippet: (card.querySelector('.b_caption p')?.textContent || '').trim(), displayUrl: (card.querySelector('cite')?.textContent || '').trim(), resultType: 'web', extra: {} });
+  }
+  return { blocked, items };
+}
+
+export const command = cli({
+  site: 'bing', name: 'search', access: 'read', description: 'Search Bing', domain: 'www.bing.com', strategy: Strategy.PUBLIC, browser: true,
+  args: [
+    { name: 'keyword', positional: true, required: true, help: 'Search query' }, { name: 'limit', type: 'int', default: 10, help: 'Number of results (1-50)' }, { name: 'page', type: 'int', default: 1, help: 'Result page number' },
+    { name: 'freshness', help: 'Time range: day, week, month, year' }, { name: 'market', help: 'Market code such as en-US or zh-CN' }, { name: 'answer', help: 'Result scope: web, news, video, images' }, { name: 'safe', help: 'Safe search: off, moderate, strict' },
+  ],
+  columns: ['rank', 'title', 'url', 'snippet', 'displayUrl', 'source', 'resultType', 'author', 'publishedAt', 'score', 'extra'],
+  func: async (page, kwargs) => {
+    const keyword = requireSearchQuery(kwargs.keyword); const limit = requireBoundedInteger(kwargs.limit, 10, 1, 50, '--limit'); const pageNumber = requireBoundedInteger(kwargs.page, 1, 1, 100, '--page');
+    const freshness = validate(kwargs.freshness, FRESHNESS, 'freshness'); const answer = validate(kwargs.answer, ANSWERS, 'answer'); const safe = validate(kwargs.safe, SAFE, 'safe');
+    await runBrowserStep('Bing search navigation', () => page.goto(buildSearchUrl({ keyword, limit, page: pageNumber, freshness, market: kwargs.market, answer, safe })));
+    await page.wait({ selector: 'li.b_algo', timeout: 8 }).catch(() => page.wait(2).catch(() => {}));
+    const data = await runBrowserStep('Bing search extraction', () => page.evaluate(`(${extractItems.toString()})()`));
+    if (data?.blocked) throw new CommandExecutionError('Bing search was blocked by a verification page', 'Complete the verification in the browser and retry.');
+    if (!data?.items?.length) throw emptySearchResults('Bing', keyword);
+    return data.items.slice(0, limit).map((item, index) => ({ rank: (pageNumber - 1) * limit + index + 1, ...item, source: 'bing', author: null, publishedAt: null, score: null }));
+  },
+});
+
+export const __test__ = { command, buildSearchUrl };

package/clis/csdn/search.js

@@ -0,0 +1,34 @@
+import { cli, Strategy } from '@sovovs/bycli/registry';
+import { ArgumentError, CommandExecutionError } from '@sovovs/bycli/errors';
+import { emptySearchResults, requireBoundedInteger, requireSearchQuery, runBrowserStep } from '../_shared/search-adapter.js';
+
+const TYPES = ['all', 'blog', 'download', 'course']; const SORTS = ['relevance', 'latest', 'hot']; const TIMES = ['day', 'week', 'month', 'year'];
+
+export function buildSearchUrl({ keyword, limit = 10, page = 1, contentType = 'all', sort, time }) {
+  const url = new URL('https://so.csdn.net/so/search'); url.searchParams.set('q', keyword); url.searchParams.set('p', String(page)); url.searchParams.set('t', contentType); url.searchParams.set('size', String(limit));
+  if (sort) url.searchParams.set('sort', sort); if (time) url.searchParams.set('time', time); return url.toString();
+}
+
+function extractItems() {
+  const blocked = /登录|验证码|访问异常|captcha|sign in/i.test(document.body?.textContent || '') && !document.querySelector('.search-result-info, .search-result'); const items = []; const seen = new Set();
+  for (const card of document.querySelectorAll('.search-result-info, .search-result')) {
+    const anchor = card.querySelector('h3 a[href], h4 a[href]'); if (!anchor || !/^https?:/i.test(anchor.href) || seen.has(anchor.href)) continue;
+    seen.add(anchor.href); const time = card.querySelector('time'); items.push({ rank: items.length + 1, title: (anchor.textContent || '').trim(), url: anchor.href, snippet: (card.querySelector('.search-result-desc, p')?.textContent || '').trim(), displayUrl: new URL(anchor.href).hostname + new URL(anchor.href).pathname, source: 'csdn', resultType: 'article', author: (card.querySelector('.author, .user-name')?.textContent || '').trim() || null, publishedAt: time?.getAttribute('datetime') || null, score: null, extra: {} });
+  }
+  return { blocked, items };
+}
+
+export const command = cli({
+  site: 'csdn', name: 'search', access: 'read', description: 'Search CSDN', domain: 'so.csdn.net', strategy: Strategy.PUBLIC, browser: true,
+  args: [{ name: 'keyword', positional: true, required: true, help: 'Search query' }, { name: 'limit', type: 'int', default: 10, help: 'Number of results (1-50)' }, { name: 'page', type: 'int', default: 1, help: 'Result page number' }, { name: 'content-type', help: 'Content type: all, blog, download, course' }, { name: 'sort', help: 'Sort order: relevance, latest, hot' }, { name: 'time', help: 'Time range: day, week, month, year' }],
+  columns: ['rank', 'title', 'url', 'snippet', 'displayUrl', 'source', 'resultType', 'author', 'publishedAt', 'score', 'extra'],
+  func: async (page, kwargs) => {
+    const keyword = requireSearchQuery(kwargs.keyword); const limit = requireBoundedInteger(kwargs.limit, 10, 1, 50, '--limit'); const pageNumber = requireBoundedInteger(kwargs.page, 1, 1, 100, '--page'); const contentType = kwargs['content-type'] ?? kwargs.contentType ?? 'all';
+    if (!TYPES.includes(String(contentType))) throw new ArgumentError(`--content-type must be one of: ${TYPES.join(', ')}`); if (kwargs.sort && !SORTS.includes(String(kwargs.sort))) throw new ArgumentError(`--sort must be one of: ${SORTS.join(', ')}`); if (kwargs.time && !TIMES.includes(String(kwargs.time))) throw new ArgumentError(`--time must be one of: ${TIMES.join(', ')}`);
+    await runBrowserStep('CSDN search navigation', () => page.goto(buildSearchUrl({ keyword, limit, page: pageNumber, contentType, sort: kwargs.sort, time: kwargs.time }))); await page.wait({ selector: '.search-result-info, .search-result', timeout: 8 }).catch(() => page.wait(2).catch(() => {}));
+    const data = await runBrowserStep('CSDN search extraction', () => page.evaluate(`(${extractItems.toString()})()`)); if (data?.blocked) throw new CommandExecutionError('CSDN search requires login or was blocked', 'Open CSDN in the browser, complete any prompt, and retry.'); if (!data?.items?.length) throw emptySearchResults('CSDN', keyword);
+    return data.items.slice(0, limit).map((item, index) => ({ ...item, rank: (pageNumber - 1) * limit + index + 1 }));
+  },
+});
+
+export const __test__ = { command, buildSearchUrl };

package/clis/gitlab/search.js

@@ -0,0 +1,36 @@
+import { cli, Strategy } from '@sovovs/bycli/registry';
+import { ArgumentError, CommandExecutionError } from '@sovovs/bycli/errors';
+import { emptySearchResults, requireBoundedInteger, requireSearchQuery, runBrowserStep } from '../_shared/search-adapter.js';
+
+const SCOPES = ['projects', 'issues', 'merge_requests', 'commits', 'blobs', 'users'];
+const ORDER = ['created_at', 'updated_at', 'latest_activity_at']; const SORT = ['asc', 'desc'];
+
+export function buildSearchUrl({ keyword, limit = 10, page = 1, scope = 'projects', orderBy, sort }) {
+  const url = new URL('https://gitlab.com/search'); url.searchParams.set('search', keyword); url.searchParams.set('page', String(page)); url.searchParams.set('per_page', String(limit));
+  if (scope) url.searchParams.set('scope', scope); if (orderBy) url.searchParams.set('order_by', orderBy); if (sort) url.searchParams.set('sort', sort); return url.toString();
+}
+
+function extractItems(scope) {
+  const blocked = /sign in|login|access denied|captcha/i.test(document.body?.textContent || '') && !document.querySelector('.search-result-row, .search-results'); const items = []; const seen = new Set();
+  for (const card of document.querySelectorAll('.search-result-row, .search-result-item, .search-results li')) {
+    const anchor = card.querySelector('a.gl-link[href], a[href*="gitlab.com/"]'); if (!anchor || !/^https?:/i.test(anchor.href) || seen.has(anchor.href)) continue;
+    seen.add(anchor.href); const time = card.querySelector('time'); const type = scope === 'issues' ? 'issue' : scope === 'merge_requests' ? 'merge_request' : scope === 'projects' ? 'project' : scope || 'result';
+    items.push({ title: (anchor.textContent || '').trim(), url: anchor.href, snippet: (card.querySelector('.description, .description p, p')?.textContent || '').trim(), displayUrl: new URL(anchor.href).hostname + new URL(anchor.href).pathname, source: 'gitlab', resultType: type, author: (card.querySelector('.author, [data-testid="author"]')?.textContent || '').trim() || null, publishedAt: time?.getAttribute('datetime') || null, score: null, extra: {} });
+  }
+  return { blocked, items };
+}
+
+export const command = cli({
+  site: 'gitlab', name: 'search', access: 'read', description: 'Search GitLab', domain: 'gitlab.com', strategy: Strategy.PUBLIC, browser: true,
+  args: [{ name: 'keyword', positional: true, required: true, help: 'Search query' }, { name: 'limit', type: 'int', default: 10, help: 'Number of results (1-50)' }, { name: 'page', type: 'int', default: 1, help: 'Result page number' }, { name: 'scope', help: 'Search scope: projects, issues, merge_requests, commits, blobs, users' }, { name: 'order-by', help: 'Order by: created_at, updated_at, latest_activity_at' }, { name: 'sort', help: 'Sort direction: asc or desc' }],
+  columns: ['rank', 'title', 'url', 'snippet', 'displayUrl', 'source', 'resultType', 'author', 'publishedAt', 'score', 'extra'],
+  func: async (page, kwargs) => {
+    const keyword = requireSearchQuery(kwargs.keyword); const limit = requireBoundedInteger(kwargs.limit, 10, 1, 50, '--limit'); const pageNumber = requireBoundedInteger(kwargs.page, 1, 1, 100, '--page'); const scope = kwargs.scope || 'projects'; const orderBy = kwargs['order-by'] ?? kwargs.orderBy; const sort = kwargs.sort;
+    if (!SCOPES.includes(String(scope))) throw new ArgumentError(`--scope must be one of: ${SCOPES.join(', ')}`); if (orderBy && !ORDER.includes(String(orderBy))) throw new ArgumentError(`--order-by must be one of: ${ORDER.join(', ')}`); if (sort && !SORT.includes(String(sort))) throw new ArgumentError('--sort must be one of: asc, desc');
+    await runBrowserStep('GitLab search navigation', () => page.goto(buildSearchUrl({ keyword, limit, page: pageNumber, scope, orderBy, sort }))); await page.wait({ selector: '.search-result-row, .search-results', timeout: 8 }).catch(() => page.wait(2).catch(() => {}));
+    const data = await runBrowserStep('GitLab search extraction', () => page.evaluate(`(${extractItems.toString()})(${JSON.stringify(scope)})`)); if (data?.blocked) throw new CommandExecutionError('GitLab search requires sign-in or was blocked', 'Open GitLab in the browser, sign in if needed, and retry.'); if (!data?.items?.length) throw emptySearchResults('GitLab', keyword);
+    return data.items.slice(0, limit).map((item, index) => ({ rank: (pageNumber - 1) * limit + index + 1, ...item }));
+  },
+});
+
+export const __test__ = { command, buildSearchUrl };