dsh-web-search-pro 0.1.0

package/lib/engines.d.ts

@@ -0,0 +1,71 @@
+/**
+ * Search engine backends for web-search-pro. Each engine is a plain object
+ * with { id, label, available(), search(query, count, signal) }. Routing,
+ * caching, and persistence live in router.ts.
+ * @module web-search-pro/engines
+ */
+import type { WebSearchSource, WebRuntime } from '@deepseek-ai/dsh-web';
+import type { BrowserService } from './browser-service.ts';
+import type { CustomPlatformSpec } from './config.ts';
+export interface SearchOutcome {
+    /** Provider-generated answer/summary text, when any. */
+    content?: string;
+    sources: WebSearchSource[];
+}
+export interface Engine {
+    id: string;
+    label: string;
+    /** Cheap local availability check; must not do network I/O. */
+    available(): boolean;
+    search(query: string, count: number, signal?: AbortSignal): Promise<SearchOutcome>;
+}
+export declare class EngineError extends Error {
+    readonly code: string;
+    readonly retryable: boolean;
+    constructor(message: string, code: string, retryable?: boolean);
+}
+export interface EngineDeps {
+    web?: WebRuntime;
+    exaApiKey?: string;
+    jinaApiKey?: string;
+    enableCli: boolean;
+    opencliEnabled: boolean;
+    agentReachEnabled: boolean;
+    /** Browser service (dsh-browser) for Playwright platform search + bundled opencli. */
+    browser?: BrowserService;
+    /** Per-platform selector overrides (settings.yaml `platformRules`). */
+    platformRules?: Record<string, {
+        item: string;
+        title: string;
+        link: string;
+        text?: string;
+    }>;
+    /** User-defined custom platforms (settings.yaml `customPlatforms`). */
+    customPlatforms?: Record<string, CustomPlatformSpec>;
+    /** True when this call originates from the ctx.web provider (avoid seam recursion). */
+    skipSeam: boolean;
+}
+export declare function seamEngine(deps: EngineDeps): Engine;
+export declare function exaEngine(deps: EngineDeps): Engine;
+export declare function ddgEngine(): Engine;
+export declare function bingEngine(): Engine;
+/** Parse RSS/Atom XML into sources (used by bing engine and rss platform). */
+export declare function parseRss(xml: string, count?: number): WebSearchSource[];
+export declare function jinaSearchEngine(deps: EngineDeps): Engine;
+export declare function githubEngine(deps: EngineDeps): Engine;
+export declare function githubCodeEngine(deps: EngineDeps): Engine;
+export declare function githubIssuesEngine(deps: EngineDeps): Engine;
+export declare function bilibiliEngine(deps: EngineDeps): Engine;
+export declare function v2exEngine(): Engine;
+export declare function youtubeEngine(deps: EngineDeps): Engine;
+export declare function opencliEngine(platform: string, deps: EngineDeps): Engine;
+export declare function agentReachEngine(platform: string, deps: EngineDeps): Engine;
+export declare function arxivEngine(): Engine;
+export declare function pubmedEngine(): Engine;
+export declare function customPlatformEngine(id: string, spec: CustomPlatformSpec, deps: EngineDeps): Engine;
+export declare function playwrightPlatformEngine(platform: string, deps: EngineDeps): Engine;
+export declare function rssEngine(url: string): Engine;
+/** Build the ordered engine list for a platform search. */
+export declare function platformEngines(platform: string, deps: EngineDeps): Engine[];
+export declare const SEARCH_ENGINE_IDS: readonly ["seam", "exa", "ddg", "bing", "jina", "github", "bilibili", "v2ex", "youtube", "arxiv", "pubmed"];
+export declare const PLATFORM_IDS: readonly ["github", "github-code", "github-issues", "bilibili", "youtube", "v2ex", "xiaohongshu", "twitter", "reddit", "instagram", "facebook", "rss", "zhihu", "weibo", "douban", "tieba", "douyin", "kuaishou", "arxiv", "pubmed"];

package/lib/engines.js

@@ -0,0 +1,562 @@
+/**
+ * Search engine backends for web-search-pro. Each engine is a plain object
+ * with { id, label, available(), search(query, count, signal) }. Routing,
+ * caching, and persistence live in router.ts.
+ * @module web-search-pro/engines
+ */
+import { httpGet, runCli, jsYaml, stripTags, capText, decodeRedirectUrl } from "./util.js";
+import { PLATFORM_SEARCH_SPECS, parseCookieString } from "./platform-search.js";
+export class EngineError extends Error {
+    code;
+    retryable;
+    constructor(message, code, retryable = true) {
+        super(message);
+        this.code = code;
+        this.retryable = retryable;
+        this.name = 'EngineError';
+    }
+}
+function withTimeout(promise, ms, label) {
+    return new Promise((resolve, reject) => {
+        const timer = setTimeout(() => reject(new EngineError(label + ' timed out', 'ENGINE_TIMEOUT')), ms);
+        promise.then((v) => { clearTimeout(timer); resolve(v); }, (e) => { clearTimeout(timer); reject(e); });
+    });
+}
+// ── ctx.web seam (DeepSeek native search) ───────────────────────────────────
+export function seamEngine(deps) {
+    return {
+        id: 'seam',
+        label: 'DeepSeek 原生搜索 (ctx.web)',
+        available: () => !!deps.web && !deps.skipSeam,
+        async search(query, count, signal) {
+            if (!deps.web)
+                throw new EngineError('ctx.web seam unavailable', 'ENGINE_UNAVAILABLE');
+            const result = await withTimeout(deps.web.search({ query, maxResults: count }, signal), 45_000, 'seam search');
+            return { sources: [...result.sources], ...result.content !== undefined ? { content: result.content } : {} };
+        },
+    };
+}
+// ── Exa (API key) ────────────────────────────────────────────────────────────
+export function exaEngine(deps) {
+    const key = () => deps.exaApiKey || process.env.EXA_API_KEY;
+    return {
+        id: 'exa',
+        label: 'Exa',
+        available: () => (key()?.length ?? 0) > 0,
+        async search(query, count, signal) {
+            const res = await httpGet('https://api.exa.ai/search', {
+                method: 'POST',
+                headers: { 'x-api-key': key(), 'content-type': 'application/json' },
+                body: JSON.stringify({ query, numResults: Math.min(count, 10), type: 'auto', contents: { text: false, highlights: true } }),
+                signal,
+                timeoutMs: 30_000,
+            });
+            if (!res.ok)
+                throw new EngineError('Exa API error HTTP ' + res.status, 'ENGINE_ERROR');
+            const data = JSON.parse(res.text);
+            const sources = (data.results ?? []).map(r => ({
+                url: r.url ?? '',
+                ...r.title ? { title: r.title } : {},
+                ...(r.highlights?.length ? { snippet: capText(r.highlights.join(' '), 400) } : {}),
+                ...r.publishedDate ? { publishedAt: r.publishedDate } : {},
+            })).filter(s => s.url.length > 0);
+            return { sources };
+        },
+    };
+}
+// ── DuckDuckGo HTML (no key) ────────────────────────────────────────────────
+export function ddgEngine() {
+    return {
+        id: 'ddg',
+        label: 'DuckDuckGo',
+        available: () => true,
+        async search(query, count, signal) {
+            const res = await httpGet('https://html.duckduckgo.com/html/?q=' + encodeURIComponent(query), { signal, timeoutMs: 30_000 });
+            if (!res.ok)
+                throw new EngineError('DuckDuckGo HTTP ' + res.status, 'ENGINE_ERROR');
+            const sources = [];
+            const blockRe = /<a[^>]*class="[^"]*result__a[^"]*"[^>]*href="([^"]+)"[^>]*>([\s\S]*?)<\/a>[\s\S]*?(?:<a[^>]*class="[^"]*result__snippet[^"]*"[^>]*>([\s\S]*?)<\/a>)?/g;
+            let m;
+            while ((m = blockRe.exec(res.text)) !== null) {
+                const rawHref = m[1] ?? '';
+                const url = decodeRedirectUrl(rawHref);
+                const title = stripTags(m[2] ?? '').trim();
+                const snippet = m[3] ? stripTags(m[3]).trim() : undefined;
+                if (!/^https?:\/\//i.test(url) || title.length < 2)
+                    continue;
+                sources.push({ url, ...title ? { title } : {}, ...snippet ? { snippet: capText(snippet, 400) } : {} });
+                if (sources.length >= count)
+                    break;
+            }
+            if (!sources.length)
+                throw new EngineError('DuckDuckGo returned no results (may be rate-limited)', 'ENGINE_EMPTY', true);
+            return { sources };
+        },
+    };
+}
+// ── Bing RSS (no key) ───────────────────────────────────────────────────────
+export function bingEngine() {
+    return {
+        id: 'bing',
+        label: 'Bing',
+        available: () => true,
+        async search(query, count, signal) {
+            const res = await httpGet('https://www.bing.com/search?q=' + encodeURIComponent(query) + '&format=rss&count=' + Math.min(count, 20), { signal, timeoutMs: 30_000 });
+            if (!res.ok)
+                throw new EngineError('Bing HTTP ' + res.status, 'ENGINE_ERROR');
+            const sources = parseRss(res.text, count);
+            if (!sources.length)
+                throw new EngineError('Bing returned no results', 'ENGINE_EMPTY', true);
+            return { sources };
+        },
+    };
+}
+/** Parse RSS/Atom XML into sources (used by bing engine and rss platform). */
+export function parseRss(xml, count = 20) {
+    const sources = [];
+    const itemRe = /<(item|entry)[^>]*>([\s\S]*?)<\/(?:item|entry)>/gi;
+    let m;
+    while ((m = itemRe.exec(xml)) !== null) {
+        const block = m[2] ?? '';
+        const grab = (tag) => {
+            const t = new RegExp('<' + tag + '(?:[^>]*)>([\\s\\S]*?)<\\/' + tag + '>', 'i').exec(block);
+            return t ? decodeCdata(stripTags(t[1])) : undefined;
+        };
+        const linkMatch = /<link[^>]*href="([^"]+)"/i.exec(block) ?? /<link[^>]*>([\s\S]*?)<\/link>/i.exec(block);
+        const title = grab('title');
+        // Atom feeds (arXiv) use <id> as the canonical URL.
+        const idMatch = /<id[^>]*>([\s\S]*?)<\/id>/i.exec(block);
+        const link = linkMatch ? (linkMatch[1] ?? stripTags(linkMatch[2] ?? '')) : (idMatch ? stripTags(idMatch[1] ?? '') : undefined);
+        const description = grab('description') ?? grab('summary') ?? grab('content');
+        const pubDate = grab('pubDate') ?? grab('published') ?? grab('updated');
+        if (!link || !/^https?:\/\//i.test(link))
+            continue;
+        sources.push({
+            url: link,
+            ...title && title.length > 1 ? { title } : {},
+            ...description && description.length > 1 ? { snippet: capText(description, 400) } : {},
+            ...pubDate ? { publishedAt: pubDate } : {},
+        });
+        if (sources.length >= count)
+            break;
+    }
+    return sources;
+}
+function decodeCdata(s) {
+    const m = /<!\[CDATA\[([\s\S]*?)\]\]>/.exec(s);
+    return m ? m[1] : s;
+}
+// ── Jina AI search / reader (optional key) ──────────────────────────────────
+export function jinaSearchEngine(deps) {
+    const key = () => deps.jinaApiKey || process.env.JINA_API_KEY;
+    return {
+        id: 'jina',
+        label: 'Jina AI',
+        available: () => true,
+        async search(query, count, signal) {
+            const headers = {};
+            const k = key();
+            if (k)
+                headers['authorization'] = 'Bearer ' + k;
+            const res = await httpGet('https://s.jina.ai/?q=' + encodeURIComponent(query), { headers, signal, timeoutMs: 30_000 });
+            if (res.status === 401 && !k)
+                throw new EngineError('Jina AI requires an API key (set jinaApiKey or $JINA_API_KEY)', 'ENGINE_UNAVAILABLE', false);
+            if (!res.ok)
+                throw new EngineError('Jina search HTTP ' + res.status, 'ENGINE_ERROR');
+            const sources = [];
+            const lineRe = /^\s*(\d+)\.\s*\[([^\]]+)\]\(([^)]+)\)(?:[：:\-—]?\s*([\s\S]*?))?$/gm;
+            let m;
+            while ((m = lineRe.exec(res.text)) !== null) {
+                const url = m[3] ?? '';
+                if (!/^https?:\/\//i.test(url))
+                    continue;
+                sources.push({
+                    url,
+                    ...(m[2] ?? '').trim() ? { title: (m[2] ?? '').trim() } : {},
+                    ...(m[4] ?? '').trim() ? { snippet: capText((m[4] ?? '').trim(), 400) } : {},
+                });
+                if (sources.length >= count)
+                    break;
+            }
+            if (!sources.length) {
+                // Jina may return a plain markdown list without numbering.
+                throw new EngineError('Jina returned no parseable results', 'ENGINE_EMPTY', true);
+            }
+            return { sources };
+        },
+    };
+}
+// ── GitHub (gh CLI) ─────────────────────────────────────────────────────────
+export function githubEngine(deps) {
+    return {
+        id: 'github',
+        label: 'GitHub',
+        available: () => deps.enableCli,
+        async search(query, count, signal) {
+            const res = await runCli('gh', ['search', 'repos', query, '--limit', String(Math.min(count, 15)), '--json', 'fullName,url,description,stargazersCount,language,updatedAt'], { timeoutMs: 30_000, signal });
+            if (res.code !== 0)
+                throw new EngineError('gh search failed: ' + res.stderr.trim().slice(0, 200), 'ENGINE_ERROR');
+            const rows = JSON.parse(res.stdout);
+            const sources = rows.map(r => {
+                const stars = r.stargazersCount != null ? ' ⭐' + r.stargazersCount : '';
+                const lang = r.language ? ' [' + r.language + ']' : '';
+                return {
+                    url: r.url ?? 'https://github.com/' + (r.fullName ?? ''),
+                    ...r.fullName ? { title: r.fullName } : {},
+                    ...(r.description ?? r.fullName) ? { snippet: capText((r.description ?? '') + stars + lang, 400) } : {},
+                };
+            });
+            return { sources };
+        },
+    };
+}
+// ── GitHub code / issues search (gh CLI) ────────────────────────────────────
+export function githubCodeEngine(deps) {
+    return {
+        id: 'github-code', label: 'GitHub 代码',
+        available: () => deps.enableCli,
+        async search(query, count, signal) {
+            const res = await runCli('gh', ['search', 'code', query, '--limit', String(Math.min(count, 15)), '--json', 'repository,path,url'], { timeoutMs: 30_000, signal });
+            if (res.code !== 0)
+                throw new EngineError('gh code search failed: ' + res.stderr.trim().slice(0, 200), 'ENGINE_ERROR');
+            const rows = JSON.parse(res.stdout);
+            const sources = rows.map(r => ({
+                url: r.url ?? '',
+                ...(r.path && r.repository?.nameWithOwner) ? { title: r.repository.nameWithOwner + ' / ' + r.path } : { title: r.path ?? 'code match' },
+                ...r.repository?.nameWithOwner ? { snippet: '仓库: ' + r.repository.nameWithOwner } : {},
+            })).filter(s => /^https?:\/\//i.test(s.url));
+            return { sources };
+        },
+    };
+}
+export function githubIssuesEngine(deps) {
+    return {
+        id: 'github-issues', label: 'GitHub Issues',
+        available: () => deps.enableCli,
+        async search(query, count, signal) {
+            const res = await runCli('gh', ['search', 'issues', query, '--limit', String(Math.min(count, 15)), '--json', 'repository,title,url,state,commentsCount'], { timeoutMs: 30_000, signal });
+            if (res.code !== 0)
+                throw new EngineError('gh issue search failed: ' + res.stderr.trim().slice(0, 200), 'ENGINE_ERROR');
+            const rows = JSON.parse(res.stdout);
+            const sources = rows.map(r => ({
+                url: r.url ?? '',
+                ...r.title ? { title: r.title } : {},
+                ...(r.state || r.repository?.nameWithOwner) ? { snippet: '[' + (r.state ?? '') + ']' + (r.repository?.nameWithOwner ? ' · ' + r.repository.nameWithOwner : '') } : {},
+            })).filter(s => /^https?:\/\//i.test(s.url));
+            return { sources };
+        },
+    };
+}
+// ── Bilibili (bili CLI) ─────────────────────────────────────────────────────
+export function bilibiliEngine(deps) {
+    return {
+        id: 'bilibili',
+        label: 'B站 (bili-cli)',
+        available: () => deps.enableCli,
+        async search(query, count, signal) {
+            const res = await runCli('bili', ['search', query, '--type', 'video', '-n', String(Math.min(count, 10))], { timeoutMs: 30_000, signal });
+            if (res.code !== 0)
+                throw new EngineError('bili search failed: ' + res.stderr.trim().slice(0, 200), 'ENGINE_ERROR');
+            let data;
+            try {
+                data = jsYaml.load(res.stdout);
+            }
+            catch {
+                throw new EngineError('bili output not parseable', 'ENGINE_ERROR');
+            }
+            const items = data?.data ?? [];
+            const sources = items.filter(i => i.bvid).map(i => ({
+                url: 'https://www.bilibili.com/video/' + i.bvid,
+                ...i.title ? { title: i.title } : {},
+                ...(i.author || i.play != null || i.duration) ? { snippet: capText(['UP: ' + (i.author ?? ''), '播放: ' + i.play, i.duration ?? ''].filter(Boolean).join(' | '), 300) } : {},
+            }));
+            return { sources };
+        },
+    };
+}
+// ── V2EX (sov2ex community search API) ──────────────────────────────────────
+export function v2exEngine() {
+    return {
+        id: 'v2ex',
+        label: 'V2EX (sov2ex)',
+        available: () => true,
+        async search(query, count, signal) {
+            const res = await httpGet('https://www.sov2ex.com/api/search?q=' + encodeURIComponent(query) + '&size=' + Math.min(count, 15), { signal, timeoutMs: 25_000 });
+            if (!res.ok)
+                throw new EngineError('sov2ex HTTP ' + res.status, 'ENGINE_ERROR');
+            const parsed = JSON.parse(res.text);
+            // sov2ex returns the hits array at top level; keep a defensive fallback.
+            const rawHits = Array.isArray(parsed.hits)
+                ? parsed.hits
+                : (parsed.hits?.hits ?? []);
+            const sources = rawHits.map(h => {
+                const s = h._source;
+                const url = s?.id != null ? 'https://www.v2ex.com/t/' + s.id : undefined;
+                const created = typeof s?.created === 'number' ? new Date(s.created * 1000).toISOString().slice(0, 10) : s?.created;
+                return {
+                    url: url ?? '',
+                    ...s?.title ? { title: s.title } : {},
+                    ...(s?.content || s?.node?.title) ? { snippet: capText((s.content ?? '') + (s.node?.title ? ' [节点: ' + s.node.title + ']' : ''), 400) } : {},
+                    ...created ? { publishedAt: String(created) } : {},
+                };
+            }).filter(s => s.url.length > 0);
+            return { sources };
+        },
+    };
+}
+// ── YouTube (yt-dlp search) ─────────────────────────────────────────────────
+export function youtubeEngine(deps) {
+    return {
+        id: 'youtube',
+        label: 'YouTube (yt-dlp)',
+        available: () => deps.enableCli,
+        async search(query, count, signal) {
+            const n = Math.min(count, 10);
+            const res = await runCli('yt-dlp', ['ytsearch' + n + ':' + query, '--flat-playlist', '--skip-download', '--no-warnings', '--print', '%(id)s\t%(title)s\t%(channel)s\t%(view_count)s\t%(duration_string)s'], { timeoutMs: 60_000, signal });
+            if (res.code !== 0)
+                throw new EngineError('yt-dlp failed: ' + res.stderr.trim().slice(0, 200), 'ENGINE_ERROR');
+            const sources = [];
+            for (const line of res.stdout.split(/\r?\n/)) {
+                const [id, title, channel, views, duration] = line.split('\t');
+                if (!id || !title)
+                    continue;
+                const meta = [];
+                if (channel)
+                    meta.push(channel);
+                if (views && views !== 'None')
+                    meta.push(views + ' views');
+                if (duration)
+                    meta.push(duration);
+                sources.push({
+                    url: 'https://www.youtube.com/watch?v=' + id,
+                    title,
+                    ...meta.length ? { snippet: meta.join(' | ') } : {},
+                });
+                if (sources.length >= n)
+                    break;
+            }
+            if (!sources.length)
+                throw new EngineError('yt-dlp returned no results', 'ENGINE_EMPTY', true);
+            return { sources };
+        },
+    };
+}
+// ── OpenCLI platform search (reuses the user's logged-in browser session) ───
+const OPENCLI_PLATFORMS = {
+    xiaohongshu: 'xiaohongshu',
+    twitter: 'twitter',
+    reddit: 'reddit',
+    instagram: 'instagram',
+    facebook: 'facebook',
+};
+export function opencliEngine(platform, deps) {
+    const adapter = OPENCLI_PLATFORMS[platform];
+    return {
+        id: 'opencli-' + platform,
+        label: 'OpenCLI ' + platform,
+        available: () => deps.enableCli && deps.opencliEnabled && !!adapter && !!deps.browser,
+        async search(query, count, signal) {
+            if (!adapter || !deps.browser)
+                throw new EngineError('opencli bundled backend unavailable for ' + platform, 'ENGINE_UNAVAILABLE', false);
+            const res = await deps.browser.opencli([adapter, 'search', query, '-f', 'yaml'], { timeoutMs: 45_000, signal });
+            if (res.code !== 0) {
+                const msg = res.stderr.trim() || res.stdout.trim() || 'exit ' + res.code;
+                throw new EngineError('opencli ' + platform + ' search failed (browser session connected?): ' + msg.slice(0, 200), 'ENGINE_UNAVAILABLE', false);
+            }
+            let rows = [];
+            try {
+                const parsed = jsYaml.load(res.stdout);
+                rows = Array.isArray(parsed) ? parsed : (parsed && typeof parsed === 'object' ? Object.values(parsed).find(Array.isArray) ?? [] : []);
+            }
+            catch {
+                try {
+                    rows = JSON.parse(res.stdout);
+                }
+                catch { /* fallthrough */ }
+            }
+            const sources = rows.slice(0, count).map((r) => ({
+                url: String(r.url ?? r.link ?? r.href ?? ''),
+                ...(r.title ?? r.name ?? r.text) ? { title: String(r.title ?? r.name ?? r.text ?? '') } : {},
+                ...(r.description ?? r.snippet ?? r.desc ?? r.author ?? r.user) ? { snippet: capText(String(r.description ?? r.snippet ?? r.desc ?? r.author ?? r.user ?? ''), 400) } : {},
+            })).filter(s => /^https?:\/\//i.test(s.url));
+            if (!sources.length)
+                throw new EngineError('opencli ' + platform + ' returned no parseable results', 'ENGINE_EMPTY', false);
+            return { sources };
+        },
+    };
+}
+// ── agent-reach CLI backends (twitter etc.) ─────────────────────────────────
+export function agentReachEngine(platform, deps) {
+    if (platform === 'twitter') {
+        return {
+            id: 'agentreach-twitter',
+            label: 'agent-reach twitter-cli',
+            available: () => deps.enableCli && deps.agentReachEnabled && !!process.env.TWITTER_AUTH_TOKEN && !!process.env.TWITTER_CT0,
+            async search(query, count, signal) {
+                const res = await runCli('twitter', ['search', query, '-n', String(Math.min(count, 10))], { timeoutMs: 45_000, signal });
+                if (res.code !== 0)
+                    throw new EngineError('twitter search failed: ' + res.stderr.trim().slice(0, 200), 'ENGINE_ERROR');
+                const sources = [];
+                for (const line of res.stdout.split(/\r?\n/)) {
+                    const m = /(https?:\/\/[^\s]+)/.exec(line);
+                    if (!m)
+                        continue;
+                    const title = stripTags(line).replace(m[1], '').trim();
+                    if (title)
+                        sources.push({ url: m[1], title: capText(title, 200) });
+                    if (sources.length >= count)
+                        break;
+                }
+                return { sources };
+            },
+        };
+    }
+    return {
+        id: 'agentreach-' + platform,
+        label: 'agent-reach ' + platform,
+        available: () => false,
+        async search() {
+            throw new EngineError('agent-reach has no backend for ' + platform, 'ENGINE_UNAVAILABLE', false);
+        },
+    };
+}
+// ── Academic verticals (public APIs, no login) ─────────────────────────────
+export function arxivEngine() {
+    return {
+        id: 'arxiv', label: 'arXiv',
+        available: () => true,
+        async search(query, count, signal) {
+            const res = await httpGet('http://export.arxiv.org/api/query?search_query=all:' + encodeURIComponent(query) + '&start=0&max_results=' + Math.min(count, 20), { signal, timeoutMs: 30_000 });
+            if (!res.ok)
+                throw new EngineError('arXiv HTTP ' + res.status, 'ENGINE_ERROR');
+            const sources = parseRss(res.text, count);
+            if (!sources.length)
+                throw new EngineError('arXiv returned no results', 'ENGINE_EMPTY', true);
+            return { sources };
+        },
+    };
+}
+export function pubmedEngine() {
+    return {
+        id: 'pubmed', label: 'PubMed',
+        available: () => true,
+        async search(query, count, signal) {
+            const n = Math.min(Math.max(count, 1), 20);
+            const esearch = await httpGet('https://eutils.ncbi.nlm.nih.gov/entrez/eutils/esearch.fcgi?db=pubmed&term=' + encodeURIComponent(query) + '&retmax=' + n + '&retmode=json', { signal, timeoutMs: 30_000 });
+            if (!esearch.ok)
+                throw new EngineError('PubMed esearch HTTP ' + esearch.status, 'ENGINE_ERROR');
+            const ids = JSON.parse(esearch.text)?.esearchresult?.idlist ?? [];
+            if (!ids.length)
+                throw new EngineError('PubMed returned no results', 'ENGINE_EMPTY', true);
+            const sources = ids.map(id => ({ url: 'https://pubmed.ncbi.nlm.nih.gov/' + id + '/', title: 'PubMed ' + id }));
+            const esummary = await httpGet('https://eutils.ncbi.nlm.nih.gov/entrez/eutils/esummary.fcgi?db=pubmed&retmode=json&id=' + ids.join(','), { signal, timeoutMs: 30_000 });
+            if (esummary.ok) {
+                const result = JSON.parse(esummary.text)?.result ?? {};
+                return {
+                    sources: ids.map(id => {
+                        const doc = result[id];
+                        return {
+                            url: 'https://pubmed.ncbi.nlm.nih.gov/' + id + '/',
+                            ...doc?.title ? { title: String(doc.title) } : { title: 'PubMed ' + id },
+                            ...doc?.pubdate ? { publishedAt: String(doc.pubdate) } : {},
+                        };
+                    }),
+                };
+            }
+            return { sources };
+        },
+    };
+}
+// ── User-defined custom platform (url template + selectors + cookie) ───────
+export function customPlatformEngine(id, spec, deps) {
+    const searchSpec = {
+        id: 'custom-' + id,
+        label: spec.name,
+        url: () => spec.url,
+        item: spec.item,
+        title: spec.title,
+        link: spec.link,
+        ...spec.text ? { text: spec.text } : {},
+    };
+    return {
+        id: 'custom-' + id,
+        label: spec.name + ' (自定义)',
+        available: () => !!deps.browser,
+        async search(query, count, signal) {
+            if (!deps.browser)
+                throw new EngineError('custom platform search unavailable (no browser service)', 'ENGINE_UNAVAILABLE', false);
+            const url = spec.url.replace(/{query}/g, encodeURIComponent(query));
+            const cookies = spec.cookie ? parseCookieString(spec.cookie, url) : undefined;
+            const sources = await deps.browser.searchResults(url, searchSpec, { signal, count, cookies });
+            if (!sources.length)
+                throw new EngineError('自定义平台 ' + spec.name + ' 未取到结果：检查 url 的 {query} 占位、item/title/link 选择器，或补充 cookie。', 'ENGINE_EMPTY', false);
+            return { sources };
+        },
+    };
+}
+// ── Chinese community search via Playwright (logged-in browser) ────────────
+export function playwrightPlatformEngine(platform, deps) {
+    const builtin = PLATFORM_SEARCH_SPECS[platform];
+    return {
+        id: 'playwright-' + platform,
+        label: (builtin?.label ?? platform) + ' (Playwright)',
+        available: () => !!builtin && !!deps.browser,
+        async search(query, count, signal) {
+            if (!builtin || !deps.browser)
+                throw new EngineError('playwright platform search unavailable for ' + platform, 'ENGINE_UNAVAILABLE', false);
+            const override = deps.platformRules?.[platform];
+            const spec = { ...builtin, ...override ?? {} };
+            const sources = await deps.browser.searchResults(spec.url(query), spec, { signal, count });
+            if (!sources.length) {
+                throw new EngineError(builtin.label + ' 未取到结果：该平台需要浏览器登录态（复用你已登录的浏览器）。运行 node scripts/save-login.mjs 登录一次并设置 dsh-browser 的 storageStatePath；或到 $DSH_HOME/settings.yaml 的 platformRules.' + platform + ' 微调结果选择器。', 'ENGINE_EMPTY', false);
+            }
+            return { sources };
+        },
+    };
+}
+// ── RSS feed (platform tool) ────────────────────────────────────────────────
+export function rssEngine(url) {
+    return {
+        id: 'rss',
+        label: 'RSS ' + url,
+        available: () => /^https?:\/\//i.test(url),
+        async search(_query, count, signal) {
+            const res = await httpGet(url, { signal, timeoutMs: 25_000 });
+            if (!res.ok)
+                throw new EngineError('RSS HTTP ' + res.status, 'ENGINE_ERROR');
+            const sources = parseRss(res.text, count);
+            if (!sources.length)
+                throw new EngineError('RSS feed has no items', 'ENGINE_EMPTY', false);
+            return { sources };
+        },
+    };
+}
+/** Build the ordered engine list for a platform search. */
+export function platformEngines(platform, deps) {
+    switch (platform) {
+        case 'github': return [githubEngine(deps)];
+        case 'github-code': return [githubCodeEngine(deps)];
+        case 'github-issues': return [githubIssuesEngine(deps)];
+        case 'bilibili': return [bilibiliEngine(deps)];
+        case 'youtube': return [youtubeEngine(deps)];
+        case 'v2ex': return [v2exEngine()];
+        case 'xiaohongshu': return [opencliEngine('xiaohongshu', deps)];
+        case 'twitter': return [opencliEngine('twitter', deps), agentReachEngine('twitter', deps)];
+        case 'reddit': return [opencliEngine('reddit', deps)];
+        case 'instagram': return [opencliEngine('instagram', deps)];
+        case 'facebook': return [opencliEngine('facebook', deps)];
+        // Chinese communities (MediaCrawler-style): Playwright drives the logged-in search page.
+        case 'arxiv': return [arxivEngine()];
+        case 'pubmed': return [pubmedEngine()];
+        case 'zhihu': return [playwrightPlatformEngine('zhihu', deps)];
+        case 'weibo': return [playwrightPlatformEngine('weibo', deps)];
+        case 'douban': return [playwrightPlatformEngine('douban', deps)];
+        case 'tieba': return [playwrightPlatformEngine('tieba', deps)];
+        case 'douyin': return [playwrightPlatformEngine('douyin', deps)];
+        case 'kuaishou': return [playwrightPlatformEngine('kuaishou', deps)];
+        default: return [];
+    }
+}
+export const SEARCH_ENGINE_IDS = ['seam', 'exa', 'ddg', 'bing', 'jina', 'github', 'bilibili', 'v2ex', 'youtube', 'arxiv', 'pubmed'];
+export const PLATFORM_IDS = ['github', 'github-code', 'github-issues', 'bilibili', 'youtube', 'v2ex', 'xiaohongshu', 'twitter', 'reddit', 'instagram', 'facebook', 'rss', 'zhihu', 'weibo', 'douban', 'tieba', 'douyin', 'kuaishou', 'arxiv', 'pubmed'];
+//# sourceMappingURL=engines.js.map