@page2ai/core 0.1.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Igor Saevets
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,128 @@
+# @page2ai/core
+
+Extract clean Markdown from any web page HTML. Works in Node.js (via linkedom) and in browsers. Powers the [Page2AI extension](https://github.com/igorsaevets/page2ai-extension) and the [`@page2ai/mcp` server](https://github.com/igorsaevets/page2ai-mcp).
+
+- **Node.js**: `htmlToMarkdown(html, opts)` or `fetchAndConvert(url, opts)` — zero external API calls, uses linkedom for parsing
+- **Browser**: exports `BrowserAdapter` and the `PageAdapter` interface; the full extract pipeline lives in the extension repo (planned for v0.2 migration)
+- **Zero data collection**: no telemetry, no analytics, no third-party calls
+- **MIT** licensed
+
+## Install
+
+```bash
+npm install @page2ai/core
+```
+
+## Quick start
+
+```javascript
+import { htmlToMarkdown, fetchAndConvert } from '@page2ai/core';
+
+// From an HTML string
+const html = `<article><h1>Hello</h1><p>World</p></article>`;
+const { markdown, title, charCount } = htmlToMarkdown(html, {
+  baseUrl: 'https://example.com/hello',
+  includeFrontmatter: true,
+});
+console.log(markdown);
+
+// From a URL
+const result = await fetchAndConvert('https://docs.anthropic.com/en/api/messages', {
+  timeoutMs: 15000,
+});
+console.log(result.markdown);
+```
+
+## API
+
+### `htmlToMarkdown(html, opts?)`
+
+Converts an HTML string to Markdown. Synchronous.
+
+- `html` (string) — the HTML to parse
+- `opts.baseUrl` (string) — URL used to resolve relative links (default: `'about:blank'`)
+- `opts.profile` (string) — extraction profile: `'auto' | 'docs' | 'marketing' | 'research' | 'dashboard' | 'wordpress-marketing'` (default: `'auto'`, v0.2)
+- `opts.includeFrontmatter` (boolean) — emit YAML frontmatter block (default: `true`)
+- `opts.includeImages` (boolean) — emit image links (default: `true`)
+
+Returns `{ markdown, title, baseUrl, charCount, extractedAt }`.
+
+### `fetchAndConvert(url, opts?)`
+
+Fetches a URL and converts it to Markdown. Async.
+
+- `url` (string) — the URL to fetch
+- `opts.timeoutMs` (number) — abort after N ms (default: `15000`)
+- `opts.userAgent` (string) — custom user agent (default: `page2ai-core/0.1`)
+- ...all `htmlToMarkdown` options
+
+Uses Node 18+ built-in `fetch()`. Follows redirects; passes the final URL to `htmlToMarkdown` as `baseUrl` unless overridden.
+
+## What it does
+
+- Parses HTML with [linkedom](https://github.com/WebReflection/linkedom) (fast, browser-DOM-fidelity, no layout engine)
+- Picks a content root (`main article` > `article` > `main` > `[role="main"]` > `body`)
+- Emits: headings (`# `), paragraphs, lists (ordered/unordered, nested), links `[text](url)`, images `![alt](src)`, code blocks (with language tag when detectable), tables (`| a | b |`), blockquotes, inline formatting (bold, italic, code)
+- Extracts metadata into YAML frontmatter: `title`, `source`, `captured_at`, `language`, `description`, `canonical`, `og_title`, `og_description`, `author`, `published`, `extractor`, `extractor_version`
+
+## What it does NOT do (v0.1)
+
+- No tab/dropdown expansion (browser-only interactive machinery)
+- No lazy-load activation (Node has no runtime images)
+- No `<details>` auto-expansion
+- No computed-style-based visibility filtering (linkedom has no layout engine)
+- No badge detection via font weight
+- No shadow DOM traversal (linkedom has no shadow DOM support)
+- No pseudo-element `::before` / `::after` content
+
+**All the above ship with the Chrome/Firefox extension**, which runs the full `lib/core/*` pipeline against a live browser DOM. See [github.com/igorsaevets/page2ai-extension](https://github.com/igorsaevets/page2ai-extension).
+
+## Roadmap
+
+- **v0.1** (current) — Node basic renderer, browser adapter interface published
+- **v0.2** — full port of `lib/core/html-to-md.ts` via PageAdapter, browser entry migrated from extension repo
+- **v0.3** — profile auto-detection in Node path, structured-data extraction
+- **v0.4** — official-markdown fallback (llms.txt, .md sibling URLs), MDX handling
+
+## Architecture
+
+Two implementations behind a single `PageAdapter` interface:
+
+```
+src/
+├── shared/           # No DOM assumptions — works with any adapter
+│   ├── page-adapter.ts    # PageAdapter interface + defaults + nodeCssEscape
+│   ├── utils.ts           # pure helpers (absUrl, escapeMd, etc.)
+│   ├── md-postprocess.ts  # pure string transforms
+│   ├── types.ts           # all TypeScript interfaces
+│   ├── constants.ts       # DEFAULTS, BLOCK_TAGS, PII patterns
+│   └── profiles.ts        # profile presets
+├── node/             # LinkedomAdapter + htmlToMarkdown + basic-renderer
+│   ├── index.ts
+│   ├── linkedom-adapter.ts
+│   └── basic-renderer.ts
+└── browser/          # BrowserAdapter (v0.1) + extract() (v0.2)
+    ├── index.ts
+    └── browser-adapter.ts
+```
+
+## Development
+
+```bash
+git clone https://github.com/igorsaevets/page2ai-core
+cd page2ai-core
+npm install
+npm run build     # tsc → dist/
+npm test          # vitest
+npm run typecheck # tsc --noEmit
+```
+
+## Related
+
+- [Page2AI Chrome extension](https://github.com/igorsaevets/page2ai-extension) — the human-facing browser extension
+- [`@page2ai/mcp`](https://github.com/igorsaevets/page2ai-mcp) — MCP server for Claude Desktop, Cursor, Windsurf, Zed
+- [Software Heritage archive](https://archive.softwareheritage.org/) — SWHID `swh:1:snp:05123c51ef9e7c0aeb06f42b1263c07a8d26999a`
+
+## License
+
+MIT — Copyright (c) 2026 Igor Saevets. See [LICENSE](./LICENSE).

package/dist/browser/browser-adapter.d.ts

@@ -0,0 +1,15 @@
+import type { AdapterCapabilities, ComputedStyleLike, PageAdapter } from '../shared/page-adapter.js';
+export declare class BrowserAdapter implements PageAdapter {
+    readonly capabilities: Readonly<AdapterCapabilities>;
+    private cache;
+    getComputedStyle(el: Element): ComputedStyleLike;
+    getPseudoContent(el: Element, pseudo: '::before' | '::after'): string;
+    innerText(el: Element): string;
+    shadowRoot(el: Element): ShadowRoot | null;
+    currentSrc(img: Element): string;
+    baseUrl(): string;
+    cssEscape(s: string): string;
+    document(): Document;
+    resetCache(): void;
+}
+//# sourceMappingURL=browser-adapter.d.ts.map

package/dist/browser/browser-adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"browser-adapter.d.ts","sourceRoot":"","sources":["../../src/browser/browser-adapter.ts"],"names":[],"mappings":"AASA,OAAO,KAAK,EACV,mBAAmB,EACnB,iBAAiB,EACjB,WAAW,EACZ,MAAM,2BAA2B,CAAC;AAEnC,qBAAa,cAAe,YAAW,WAAW;IAChD,QAAQ,CAAC,YAAY,EAAE,QAAQ,CAAC,mBAAmB,CAAC,CAMjD;IAEH,OAAO,CAAC,KAAK,CAA+C;IAE5D,gBAAgB,CAAC,EAAE,EAAE,OAAO,GAAG,iBAAiB;IAehD,gBAAgB,CAAC,EAAE,EAAE,OAAO,EAAE,MAAM,EAAE,UAAU,GAAG,SAAS,GAAG,MAAM;IAgBrE,SAAS,CAAC,EAAE,EAAE,OAAO,GAAG,MAAM;IAI9B,UAAU,CAAC,EAAE,EAAE,OAAO,GAAG,UAAU,GAAG,IAAI;IAI1C,UAAU,CAAC,GAAG,EAAE,OAAO,GAAG,MAAM;IAIhC,OAAO,IAAI,MAAM;IAIjB,SAAS,CAAC,CAAC,EAAE,MAAM,GAAG,MAAM;IAK5B,QAAQ,IAAI,QAAQ;IAIpB,UAAU,IAAI,IAAI;CAGnB"}

package/dist/browser/browser-adapter.js

@@ -0,0 +1,74 @@
+// BrowserAdapter: PageAdapter implementation for real browsers (Chrome/Firefox
+// extensions). Thin wrapper around window.getComputedStyle, HTMLElement.innerText,
+// Element.shadowRoot, HTMLImageElement.currentSrc, CSS.escape, location.href.
+//
+// Per-instance computedStyleCache (Spark P20 fix — the extension previously had
+// a module-level WeakMap which leaked in long-lived contexts and races on
+// concurrent extractions). Callers should create ONE BrowserAdapter per
+// extraction call and discard when done.
+export class BrowserAdapter {
+    capabilities = Object.freeze({
+        canComputeStyle: true,
+        canPseudoContent: true,
+        canShadowDom: true,
+        canInnerText: true,
+        canCurrentSrc: true,
+    });
+    cache = new WeakMap();
+    getComputedStyle(el) {
+        let s = this.cache.get(el);
+        if (!s) {
+            s = window.getComputedStyle(el);
+            this.cache.set(el, s);
+        }
+        return {
+            display: s.display,
+            visibility: s.visibility,
+            opacity: s.opacity,
+            fontSize: s.fontSize,
+            fontWeight: s.fontWeight,
+        };
+    }
+    getPseudoContent(el, pseudo) {
+        try {
+            const s = window.getComputedStyle(el, pseudo);
+            let c = s.getPropertyValue('content') || '';
+            if (!c || c === 'none' || c === 'normal')
+                return '';
+            c = c.trim();
+            if ((c.startsWith('"') && c.endsWith('"')) || (c.startsWith("'") && c.endsWith("'"))) {
+                c = c.slice(1, -1);
+            }
+            if (/counter\(|counters\(/i.test(c))
+                return '';
+            return c.replace(/ /g, ' ').replace(/[ \t\r\n]+/g, ' ').trim();
+        }
+        catch {
+            return '';
+        }
+    }
+    innerText(el) {
+        return el.innerText || el.textContent || '';
+    }
+    shadowRoot(el) {
+        return el.shadowRoot ?? null;
+    }
+    currentSrc(img) {
+        return img.currentSrc || '';
+    }
+    baseUrl() {
+        return location.href;
+    }
+    cssEscape(s) {
+        if (typeof CSS !== 'undefined' && CSS.escape)
+            return CSS.escape(s);
+        return s.replace(/([^\w-])/g, '\\$1');
+    }
+    document() {
+        return document;
+    }
+    resetCache() {
+        this.cache = new WeakMap();
+    }
+}
+//# sourceMappingURL=browser-adapter.js.map

package/dist/browser/browser-adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"browser-adapter.js","sourceRoot":"","sources":["../../src/browser/browser-adapter.ts"],"names":[],"mappings":"AAAA,+EAA+E;AAC/E,mFAAmF;AACnF,8EAA8E;AAC9E,EAAE;AACF,gFAAgF;AAChF,0EAA0E;AAC1E,wEAAwE;AACxE,yCAAyC;AAQzC,MAAM,OAAO,cAAc;IAChB,YAAY,GAAkC,MAAM,CAAC,MAAM,CAAC;QACnE,eAAe,EAAE,IAAI;QACrB,gBAAgB,EAAE,IAAI;QACtB,YAAY,EAAE,IAAI;QAClB,YAAY,EAAE,IAAI;QAClB,aAAa,EAAE,IAAI;KACpB,CAAC,CAAC;IAEK,KAAK,GAAG,IAAI,OAAO,EAAgC,CAAC;IAE5D,gBAAgB,CAAC,EAAW;QAC1B,IAAI,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAC3B,IAAI,CAAC,CAAC,EAAE,CAAC;YACP,CAAC,GAAG,MAAM,CAAC,gBAAgB,CAAC,EAAE,CAAC,CAAC;YAChC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC;QACxB,CAAC;QACD,OAAO;YACL,OAAO,EAAE,CAAC,CAAC,OAAO;YAClB,UAAU,EAAE,CAAC,CAAC,UAAU;YACxB,OAAO,EAAE,CAAC,CAAC,OAAO;YAClB,QAAQ,EAAE,CAAC,CAAC,QAAQ;YACpB,UAAU,EAAE,CAAC,CAAC,UAAU;SACzB,CAAC;IACJ,CAAC;IAED,gBAAgB,CAAC,EAAW,EAAE,MAA8B;QAC1D,IAAI,CAAC;YACH,MAAM,CAAC,GAAG,MAAM,CAAC,gBAAgB,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC;YAC9C,IAAI,CAAC,GAAG,CAAC,CAAC,gBAAgB,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;YAC5C,IAAI,CAAC,CAAC,IAAI,CAAC,KAAK,MAAM,IAAI,CAAC,KAAK,QAAQ;gBAAE,OAAO,EAAE,CAAC;YACpD,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;YACb,IAAI,CAAC,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC;gBACrF,CAAC,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;YACrB,CAAC;YACD,IAAI,uBAAuB,CAAC,IAAI,CAAC,CAAC,CAAC;gBAAE,OAAO,EAAE,CAAC;YAC/C,OAAO,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC,OAAO,CAAC,aAAa,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;QACjE,CAAC;QAAC,MAAM,CAAC;YACP,OAAO,EAAE,CAAC;QACZ,CAAC;IACH,CAAC;IAED,SAAS,CAAC,EAAW;QACnB,OAAQ,EAAkB,CAAC,SAAS,IAAI,EAAE,CAAC,WAAW,IAAI,EAAE,CAAC;IAC/D,CAAC;IAED,UAAU,CAAC,EAAW;QACpB,OAAO,EAAE,CAAC,UAAU,IAAI,IAAI,CAAC;IAC/B,CAAC;IAED,UAAU,CAAC,GAAY;QACrB,OAAQ,GAAwB,CAAC,UAAU,IAAI,EAAE,CAAC;IACpD,CAAC;IAED,OAAO;QACL,OAAO,QAAQ,CAAC,IAAI,CAAC;IACvB,CAAC;IAED,SAAS,CAAC,CAAS;QACjB,IAAI,OAAO,GAAG,KAAK,WAAW,IAAI,GAAG,CAAC,MAAM;YAAE,OAAO,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;QACnE,OAAO,CAAC,CAAC,OAAO,CAAC,WAAW,EAAE,MAAM,CAAC,CAAC;IACxC,CAAC;IAED,QAAQ;QACN,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,UAAU;QACR,IAAI,CAAC,KAAK,GAAG,IAAI,OAAO,EAAE,CAAC;IAC7B,CAAC;CACF"}

package/dist/browser/index.d.ts

@@ -0,0 +1,3 @@
+export { BrowserAdapter } from './browser-adapter.js';
+export type { AdapterCapabilities, ComputedStyleLike, PageAdapter, } from '../shared/page-adapter.js';
+//# sourceMappingURL=index.d.ts.map

package/dist/browser/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/browser/index.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,YAAY,EACV,mBAAmB,EACnB,iBAAiB,EACjB,WAAW,GACZ,MAAM,2BAA2B,CAAC"}

package/dist/browser/index.js

@@ -0,0 +1,7 @@
+// @page2ai/core browser entry point.
+//
+// v0.1 SCOPE: exports the BrowserAdapter only. The full extract() pipeline
+// still lives in the Page2AI extension repo (lib/core/*). v0.2 will migrate
+// the extension to import from this package.
+export { BrowserAdapter } from './browser-adapter.js';
+//# sourceMappingURL=index.js.map

package/dist/browser/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/browser/index.ts"],"names":[],"mappings":"AAAA,qCAAqC;AACrC,EAAE;AACF,2EAA2E;AAC3E,4EAA4E;AAC5E,6CAA6C;AAE7C,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC"}

package/dist/node/basic-renderer.d.ts

@@ -0,0 +1,9 @@
+import type { PageAdapter } from '../shared/page-adapter.js';
+interface RenderOpts {
+    baseUrl: string;
+    includeFrontmatter: boolean;
+    includeImages: boolean;
+}
+export declare const renderBasicMarkdown: (doc: Document, adapter: PageAdapter, opts: RenderOpts) => string;
+export {};
+//# sourceMappingURL=basic-renderer.d.ts.map

package/dist/node/basic-renderer.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"basic-renderer.d.ts","sourceRoot":"","sources":["../../src/node/basic-renderer.ts"],"names":[],"mappings":"AAUA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAC;AAS7D,UAAU,UAAU;IAClB,OAAO,EAAE,MAAM,CAAC;IAChB,kBAAkB,EAAE,OAAO,CAAC;IAC5B,aAAa,EAAE,OAAO,CAAC;CACxB;AAUD,eAAO,MAAM,mBAAmB,GAC9B,KAAK,QAAQ,EACb,SAAS,WAAW,EACpB,MAAM,UAAU,KACf,MAyBF,CAAC"}

package/dist/node/basic-renderer.js

@@ -0,0 +1,347 @@
+// Basic Markdown renderer for the v0.1 Node build. Covers the essentials:
+// headings, paragraphs, lists, links, images, code blocks, tables, blockquotes,
+// inline formatting, YAML frontmatter, plus STATIC tab discovery emitting
+// '## Tab: {label}' sections (Spark P20 fix — prevents Python+TypeScript panel
+// concatenation on docs.anthropic.com, platform.openai.com, Mintlify, etc.).
+// NOT a port of the full extension renderer.
+//
+// v0.2 will replace this with the full lib/core/html-to-md.ts pipeline via
+// PageAdapter, once the browser renderer is refactored to accept an adapter.
+import { absUrl, chooseCodeFence, cleanInline, escapeMd, escapeMdTableCell } from '../shared/utils.js';
+const SKIP_TAGS = new Set(['SCRIPT', 'STYLE', 'NOSCRIPT', 'TEMPLATE', 'SVG', 'CANVAS']);
+const INLINE_TAGS = new Set(['A', 'B', 'STRONG', 'I', 'EM', 'CODE', 'SPAN', 'SMALL', 'MARK', 'U', 'S', 'DEL', 'INS', 'SUB', 'SUP', 'ABBR', 'CITE', 'Q', 'DFN', 'KBD', 'SAMP', 'VAR', 'TIME']);
+const KNOWN_LANGS = new Set(['bash', 'sh', 'shell', 'python', 'py', 'javascript', 'js', 'typescript', 'ts', 'tsx', 'jsx', 'json', 'yaml', 'yml', 'toml', 'html', 'css', 'sql', 'go', 'rust', 'java', 'kotlin', 'swift', 'c', 'cpp', 'csharp', 'php', 'ruby', 'markdown', 'md', 'mermaid', 'dockerfile', 'text']);
+const yq = (s) => '"' + String(s).replace(/"/g, '\\"') + '"';
+export const renderBasicMarkdown = (doc, adapter, opts) => {
+    const parts = [];
+    if (opts.includeFrontmatter) {
+        parts.push(...buildFrontmatter(doc, opts.baseUrl));
+    }
+    const title = cleanInline(doc.querySelector('title')?.textContent || '');
+    if (title)
+        parts.push(`# ${title}`, '');
+    const root = pickContentRoot(doc);
+    if (root && root.children) {
+        const tabGroups = discoverTabGroups(root);
+        const capturedPanelEls = new WeakSet();
+        tabGroups.forEach((g) => g.panels.forEach((p) => capturedPanelEls.add(p.el)));
+        const lines = [];
+        void adapter;
+        walkWithTabs(root, lines, 0, opts, tabGroups, capturedPanelEls);
+        parts.push(...lines);
+    }
+    else {
+        parts.push('', '<!-- AI: page has no extractable body content -->', '');
+    }
+    return normalizeBlank(parts.join('\n')).trim() + '\n';
+};
+const buildFrontmatter = (doc, baseUrl) => {
+    const title = cleanInline(doc.querySelector('title')?.textContent || '');
+    const description = cleanInline(doc.querySelector('meta[name="description"]')?.getAttribute('content') || '');
+    const lang = doc.documentElement?.getAttribute('lang') || 'unknown';
+    const canonical = absUrl(doc.querySelector('link[rel="canonical"]')?.getAttribute('href') || '', baseUrl);
+    const ogTitle = cleanInline(doc.querySelector('meta[property="og:title"]')?.getAttribute('content') || '');
+    const ogDescription = cleanInline(doc.querySelector('meta[property="og:description"]')?.getAttribute('content') || '');
+    const author = cleanInline(doc.querySelector('meta[name="author"]')?.getAttribute('content') ||
+        doc.querySelector('meta[property="article:author"]')?.getAttribute('content') ||
+        '');
+    const publishedTime = cleanInline(doc.querySelector('meta[property="article:published_time"]')?.getAttribute('content') || '');
+    const fm = ['---'];
+    fm.push(`title: ${yq(title)}`);
+    fm.push(`source: ${yq(baseUrl)}`);
+    fm.push(`captured_at: ${yq(new Date().toISOString())}`);
+    fm.push(`language: ${yq(lang)}`);
+    if (description)
+        fm.push(`description: ${yq(description)}`);
+    if (canonical)
+        fm.push(`canonical: ${yq(canonical)}`);
+    if (ogTitle && ogTitle !== title)
+        fm.push(`og_title: ${yq(ogTitle)}`);
+    if (ogDescription && ogDescription !== description)
+        fm.push(`og_description: ${yq(ogDescription)}`);
+    if (author)
+        fm.push(`author: ${yq(author)}`);
+    if (publishedTime)
+        fm.push(`published: ${yq(publishedTime)}`);
+    fm.push(`extractor: ${yq('page2ai-core')}`);
+    fm.push(`extractor_version: ${yq('0.1.0')}`);
+    fm.push('---', '');
+    return fm;
+};
+const pickContentRoot = (doc) => {
+    const cand = doc.querySelector('main article') ||
+        doc.querySelector('article') ||
+        doc.querySelector('main') ||
+        doc.querySelector('[role="main"]') ||
+        doc.body ||
+        doc.documentElement ||
+        null;
+    return cand || null;
+};
+// Static tab discovery: aria-controls chains, role="tablist" siblings, and
+// data-state attribute pairs. Emits '## Tab: {label}' sections. Does NOT
+// click — for Node we treat each panel as ALREADY present in the DOM.
+const discoverTabGroups = (root) => {
+    const groups = [];
+    const seenContainers = new WeakSet();
+    // Path 1: [role="tablist"] with aria-controls chain
+    root.querySelectorAll('[role="tablist"]').forEach((tl) => {
+        if (seenContainers.has(tl))
+            return;
+        const tabs = [...tl.querySelectorAll('[role="tab"], button')];
+        const panels = [];
+        for (const tab of tabs) {
+            const label = cleanInline(tab.textContent || '');
+            if (!label)
+                continue;
+            const controls = tab.getAttribute('aria-controls');
+            if (!controls)
+                continue;
+            // linkedom supports getElementById via document.getElementById
+            const doc = tl.ownerDocument;
+            if (!doc)
+                continue;
+            const panel = doc.getElementById(controls);
+            if (!panel)
+                continue;
+            panels.push({ label, el: panel });
+        }
+        if (panels.length >= 2) {
+            groups.push({ container: tl, panels });
+            seenContainers.add(tl);
+        }
+    });
+    // Path 2: data-state="active"/"inactive" panels sharing a parent
+    root.querySelectorAll('[data-state="active"], [data-state="inactive"]').forEach((el) => {
+        if (el.getAttribute('role') !== 'tabpanel')
+            return;
+        const parent = el.parentElement;
+        if (!parent || seenContainers.has(parent))
+            return;
+        const siblingPanels = [...parent.children].filter((c) => c.getAttribute('role') === 'tabpanel');
+        if (siblingPanels.length < 2)
+            return;
+        const panels = siblingPanels.map((p) => ({
+            label: p.getAttribute('aria-labelledby') ||
+                p.getAttribute('aria-label') ||
+                p.getAttribute('data-value') ||
+                p.id ||
+                'Panel',
+            el: p,
+        }));
+        groups.push({ container: parent, panels });
+        seenContainers.add(parent);
+    });
+    return groups;
+};
+const walkWithTabs = (el, out, depth, opts, tabGroups, capturedPanelEls) => {
+    for (const child of el.children) {
+        renderNodeWithTabs(child, out, depth, opts, tabGroups, capturedPanelEls);
+    }
+};
+const renderNodeWithTabs = (el, out, depth, opts, tabGroups, capturedPanelEls) => {
+    // If THIS element is a captured tab panel, skip — it will be rendered as
+    // part of its tab group section, not inline.
+    if (capturedPanelEls.has(el))
+        return;
+    // If THIS element is a tab-group container, render its panels as sections.
+    const group = tabGroups.find((g) => g.container === el);
+    if (group) {
+        for (const panel of group.panels) {
+            out.push('', `### Tab: ${escapeMd(panel.label)}`, '');
+            walkWithTabs(panel.el, out, depth, opts, tabGroups, capturedPanelEls);
+        }
+        return;
+    }
+    renderNode(el, out, depth, opts, tabGroups, capturedPanelEls);
+};
+const renderNode = (el, out, depth, opts, tabGroups, capturedPanelEls) => {
+    const tag = el.tagName;
+    if (SKIP_TAGS.has(tag))
+        return;
+    if (el.getAttribute('aria-hidden') === 'true')
+        return;
+    const pfx = '   '.repeat(depth);
+    if (/^H[1-6]$/.test(tag)) {
+        const level = Number(tag.slice(1));
+        const text = renderInline(el, opts).trim();
+        if (text)
+            out.push('', `${'#'.repeat(level)} ${text}`, '');
+        return;
+    }
+    if (tag === 'PRE') {
+        renderCodeBlock(el, out);
+        return;
+    }
+    if (tag === 'P') {
+        const text = renderInline(el, opts).trim();
+        if (text)
+            out.push('', text, '');
+        return;
+    }
+    if (tag === 'A') {
+        const href = absUrl(el.getAttribute('href') || '', opts.baseUrl);
+        const text = renderInline(el, opts).trim() || cleanInline(el.textContent || '');
+        if (text && href)
+            out.push(`${pfx}[${escapeMd(text)}](${href})`);
+        else if (text)
+            out.push(`${pfx}${escapeMd(text)}`);
+        return;
+    }
+    if (tag === 'BLOCKQUOTE') {
+        const inner = [];
+        walkWithTabs(el, inner, depth, opts, tabGroups, capturedPanelEls);
+        const text = inner.join('\n').trim();
+        if (text) {
+            out.push('');
+            text.split('\n').forEach((l) => out.push(`> ${l}`));
+            out.push('');
+        }
+        return;
+    }
+    if (tag === 'UL' || tag === 'OL') {
+        const items = [...el.children].filter((c) => c.tagName === 'LI');
+        let i = 1;
+        for (const li of items) {
+            const marker = tag === 'OL' ? `${i}.` : '-';
+            const text = renderInline(li, opts).trim();
+            if (text)
+                out.push(`${pfx}${marker} ${text}`);
+            for (const sub of li.children) {
+                if (sub.tagName === 'UL' || sub.tagName === 'OL') {
+                    renderNodeWithTabs(sub, out, depth + 1, opts, tabGroups, capturedPanelEls);
+                }
+            }
+            i++;
+        }
+        return;
+    }
+    if (tag === 'TABLE') {
+        renderTable(el, out);
+        return;
+    }
+    if (tag === 'IMG') {
+        if (!opts.includeImages)
+            return;
+        const alt = cleanInline(el.getAttribute('alt') || '');
+        const src = absUrl(el.getAttribute('src') || el.getAttribute('data-src') || '', opts.baseUrl);
+        if (src)
+            out.push(`${pfx}![${escapeMd(alt || 'image')}](${src})`);
+        return;
+    }
+    if (tag === 'HR') {
+        out.push('', '---', '');
+        return;
+    }
+    if (tag === 'BR') {
+        out.push('');
+        return;
+    }
+    walkWithTabs(el, out, depth, opts, tabGroups, capturedPanelEls);
+};
+const renderInline = (el, opts) => {
+    const parts = [];
+    for (const node of el.childNodes) {
+        if (node.nodeType === 3 /* TEXT_NODE */) {
+            const t = cleanInline(node.textContent || '');
+            if (t)
+                parts.push(t);
+            continue;
+        }
+        if (node.nodeType !== 1 /* ELEMENT_NODE */)
+            continue;
+        const c = node;
+        const tag = c.tagName;
+        if (SKIP_TAGS.has(tag))
+            continue;
+        if (tag === 'BR') {
+            parts.push('\n');
+            continue;
+        }
+        if (tag === 'A') {
+            const href = absUrl(c.getAttribute('href') || '', opts.baseUrl);
+            const text = renderInline(c, opts) || cleanInline(c.textContent || '');
+            if (text && href)
+                parts.push(`[${escapeMd(text)}](${href})`);
+            else if (text)
+                parts.push(escapeMd(text));
+            continue;
+        }
+        if (tag === 'STRONG' || tag === 'B') {
+            const t = renderInline(c, opts) || cleanInline(c.textContent || '');
+            if (t)
+                parts.push(`**${t}**`);
+            continue;
+        }
+        if (tag === 'EM' || tag === 'I') {
+            const t = renderInline(c, opts) || cleanInline(c.textContent || '');
+            if (t)
+                parts.push(`*${t}*`);
+            continue;
+        }
+        if (tag === 'CODE') {
+            const t = cleanInline(c.textContent || '');
+            if (t)
+                parts.push(`\`${t}\``);
+            continue;
+        }
+        if (tag === 'IMG') {
+            const alt = cleanInline(c.getAttribute('alt') || '');
+            const src = absUrl(c.getAttribute('src') || '', opts.baseUrl);
+            if (src)
+                parts.push(`![${escapeMd(alt || 'image')}](${src})`);
+            continue;
+        }
+        if (INLINE_TAGS.has(tag)) {
+            const t = renderInline(c, opts) || cleanInline(c.textContent || '');
+            if (t)
+                parts.push(t);
+            continue;
+        }
+        const t = cleanInline(c.textContent || '');
+        if (t)
+            parts.push(t);
+    }
+    return parts.join(' ').replace(/\s+\n\s+/g, '\n');
+};
+const renderCodeBlock = (pre, out) => {
+    const codeEl = pre.querySelector('code') || pre;
+    const text = (codeEl.textContent || '').replace(/\r/g, '').replace(/^\n+|\n+$/g, '');
+    if (!text.trim())
+        return;
+    let lang = '';
+    const codeClass = codeEl.getAttribute('class') || pre.getAttribute('class') || '';
+    const langMatch = codeClass.match(/language-([a-z0-9_+#.-]+)/i);
+    if (langMatch) {
+        const l = langMatch[1].toLowerCase();
+        if (KNOWN_LANGS.has(l) || /^[a-z][a-z0-9]{0,15}$/.test(l))
+            lang = l;
+    }
+    const dataLang = pre.getAttribute('data-language') || codeEl.getAttribute('data-language') || '';
+    if (!lang && dataLang) {
+        const l = dataLang.toLowerCase();
+        if (KNOWN_LANGS.has(l) || /^[a-z][a-z0-9]{0,15}$/.test(l))
+            lang = l;
+    }
+    const fence = chooseCodeFence(text);
+    out.push('', `${fence}${lang}`);
+    text.split('\n').forEach((l) => out.push(l));
+    out.push(fence, '');
+};
+const renderTable = (tbl, out) => {
+    const rows = [...tbl.querySelectorAll('tr')];
+    if (!rows.length)
+        return;
+    const data = rows.map((tr) => [...tr.querySelectorAll('th, td')].map((c) => escapeMdTableCell(c.textContent || '')));
+    const width = Math.max(...data.map((r) => r.length));
+    const padded = data.map((r) => { while (r.length < width)
+        r.push(''); return r; });
+    out.push('');
+    const header = padded[0];
+    out.push(`| ${header.join(' | ')} |`);
+    out.push(`| ${header.map(() => '---').join(' | ')} |`);
+    padded.slice(1).forEach((r) => out.push(`| ${r.join(' | ')} |`));
+    out.push('');
+};
+const normalizeBlank = (s) => s.replace(/\n{3,}/g, '\n\n');
+//# sourceMappingURL=basic-renderer.js.map