@monoes/monomindcli 2.14.0 → 2.14.1

package/dist/src/knowledge/html-extract.js

@@ -0,0 +1,286 @@
+/**
+ * Dependency-free HTML → readable Markdown-ish text (RCL-01).
+ *
+ * WHY NO DEPENDENCY — the capture path has to work in an `npx monomind`
+ * install with nothing extra resolved, and the extension already runs
+ * Readability against the live DOM. What lands here is the fallback for an
+ * archive nobody cleaned: a small, predictable extractor beats a large one we
+ * cannot install.
+ *
+ * WHY BOILERPLATE STRIPPING IS THE POINT — a site's nav, header, footer and
+ * cookie banner are byte-identical on every page. Leave them in and every
+ * document from that site chunks into near-identical text, so their embeddings
+ * cluster on the chrome rather than the article and recall collapses. Dropping
+ * a little real content is a far cheaper error than keeping the chrome.
+ *
+ * @module v1/cli/knowledge/html-extract
+ */
+import { BLOCK_TAGS, CELL_TAGS, CHROME_ROLES, CHROME_TAGS, CHROME_WORDS, decodeEntities, NON_TEXT_TAGS, RAW_TEXT_TAGS, VOID_TAGS, } from './html-tags.js';
+export { decodeEntities };
+const ATTR_RE = /([a-zA-Z_:][-a-zA-Z0-9_:.]*)\s*(?:=\s*("[^"]*"|'[^']*'|[^\s"'`=<>]+))?/g;
+function parseAttrs(source) {
+    const attrs = {};
+    ATTR_RE.lastIndex = 0;
+    let m = ATTR_RE.exec(source);
+    while (m) {
+        const raw = m[2] ?? '';
+        attrs[m[1].toLowerCase()] = decodeEntities(raw.replace(/^["']|["']$/g, ''));
+        m = ATTR_RE.exec(source);
+    }
+    return attrs;
+}
+function* tokenize(html) {
+    let i = 0;
+    while (i < html.length) {
+        const lt = html.indexOf('<', i);
+        if (lt === -1) {
+            yield { t: 'text', value: html.slice(i) };
+            return;
+        }
+        if (lt > i)
+            yield { t: 'text', value: html.slice(i, lt) };
+        if (html.startsWith('<!--', lt)) {
+            const end = html.indexOf('-->', lt + 4);
+            i = end === -1 ? html.length : end + 3;
+            continue;
+        }
+        if (html.startsWith('<![CDATA[', lt)) {
+            const end = html.indexOf(']]>', lt + 9);
+            yield { t: 'text', value: html.slice(lt + 9, end === -1 ? undefined : end) };
+            i = end === -1 ? html.length : end + 3;
+            continue;
+        }
+        if (html.startsWith('<!', lt) || html.startsWith('<?', lt)) {
+            const end = html.indexOf('>', lt);
+            i = end === -1 ? html.length : end + 1;
+            continue;
+        }
+        if (html.startsWith('</', lt)) {
+            const end = html.indexOf('>', lt);
+            if (end === -1) {
+                i = html.length;
+                continue;
+            }
+            const name = /^[a-zA-Z][a-zA-Z0-9:-]*/.exec(html.slice(lt + 2, end).trim())?.[0];
+            if (name)
+                yield { t: 'close', name: name.toLowerCase() };
+            i = end + 1;
+            continue;
+        }
+        const nameMatch = /^<([a-zA-Z][a-zA-Z0-9:-]*)/.exec(html.slice(lt, lt + 80));
+        if (!nameMatch) {
+            // A bare `<` in prose, e.g. "a < b". Keep it as text.
+            yield { t: 'text', value: '<' };
+            i = lt + 1;
+            continue;
+        }
+        const name = nameMatch[1].toLowerCase();
+        // Walk to the tag's `>` while honouring quoted attribute values, so
+        // `alt="a > b"` does not end the tag early.
+        let j = lt + nameMatch[0].length;
+        let quote = '';
+        while (j < html.length) {
+            const c = html[j];
+            if (quote) {
+                if (c === quote)
+                    quote = '';
+            }
+            else if (c === '"' || c === "'") {
+                quote = c;
+            }
+            else if (c === '>') {
+                break;
+            }
+            j++;
+        }
+        const inner = html.slice(lt + nameMatch[0].length, j);
+        const selfClosing = inner.trimEnd().endsWith('/') || VOID_TAGS.has(name);
+        yield { t: 'open', tag: { name, attrs: parseAttrs(inner), selfClosing } };
+        i = j + 1;
+        if (RAW_TEXT_TAGS.has(name) && !inner.trimEnd().endsWith('/')) {
+            const rest = html.slice(i);
+            const close = new RegExp(`</${name}\\s*>`, 'i').exec(rest);
+            yield { t: 'text', value: close ? rest.slice(0, close.index) : rest };
+            yield { t: 'close', name };
+            i = close ? i + close.index + close[0].length : html.length;
+        }
+    }
+}
+// ── Drop decisions ─────────────────────────────────────────────────
+function classTokens(value) {
+    return value
+        .replace(/([a-z0-9])([A-Z])/g, '$1 $2')
+        .toLowerCase()
+        .split(/[^a-z0-9]+/)
+        .filter(Boolean);
+}
+function isChromeAttrs(attrs) {
+    const role = attrs.role?.trim().toLowerCase();
+    if (role?.split(/\s+/).some((r) => CHROME_ROLES.has(r)))
+        return true;
+    for (const key of ['class', 'id']) {
+        const value = attrs[key];
+        if (!value)
+            continue;
+        if (classTokens(value).some((w) => CHROME_WORDS.has(w)))
+            return true;
+    }
+    return false;
+}
+function isHidden(attrs) {
+    if (attrs['aria-hidden']?.trim().toLowerCase() === 'true')
+        return true;
+    if ('hidden' in attrs)
+        return true;
+    const style = attrs.style;
+    return !!style && /(?:^|;)\s*display\s*:\s*none/i.test(style);
+}
+/** The innermost open block decides how the buffered text renders — so a
+ *  `<p>` nested inside an `<li>` still comes out as a bullet, not a
+ *  paragraph. */
+function resolveKind(openBlocks) {
+    for (let i = openBlocks.length - 1; i >= 0; i--) {
+        const name = openBlocks[i];
+        if (/^h[1-6]$/.test(name))
+            return { kind: 'heading', level: Number(name[1]) };
+        if (name === 'li') {
+            let indent = -1;
+            for (let k = 0; k <= i; k++)
+                if (openBlocks[k] === 'li')
+                    indent++;
+            return { kind: 'li', indent: Math.max(0, indent) };
+        }
+        if (name === 'pre')
+            return { kind: 'pre' };
+        if (name === 'blockquote')
+            return { kind: 'quote' };
+    }
+    return { kind: 'para' };
+}
+function render(blocks) {
+    let out = '';
+    for (let i = 0; i < blocks.length; i++) {
+        const b = blocks[i];
+        let line;
+        if (b.kind === 'heading')
+            line = `${'#'.repeat(Math.min(6, b.level ?? 1))} ${b.text}`;
+        else if (b.kind === 'li')
+            line = `${'  '.repeat(b.indent ?? 0)}- ${b.text}`;
+        else if (b.kind === 'quote')
+            line = `> ${b.text}`;
+        else
+            line = b.text;
+        if (i === 0)
+            out += line;
+        else if (blocks[i - 1].kind === 'li' && b.kind === 'li')
+            out += `\n${line}`;
+        else
+            out += `\n\n${line}`;
+    }
+    return out.replace(/\n{3,}/g, '\n\n').trim();
+}
+/**
+ * Parse `html` into a title and clean block text.
+ *
+ * Unbalanced markup is expected — real archives are full of it — so closing a
+ * tag pops back to its last matching open and a never-closed element simply
+ * stays open until its parent closes. Nothing here throws.
+ */
+export function extractHtmlDocument(html, opts = {}) {
+    const stripBoilerplate = opts.stripBoilerplate !== false;
+    const blocks = [];
+    const openBlocks = [];
+    let buf = '';
+    let title = null;
+    let inTitle = false;
+    // Drop state: the element being skipped and how deep we are inside nested
+    // elements of the same name.
+    let dropName = null;
+    let dropNest = 0;
+    const flush = () => {
+        const resolved = resolveKind(openBlocks);
+        let text = resolved.kind === 'pre' ? buf.replace(/[ \t]+$/gm, '') : buf.replace(/\s+/g, ' ');
+        buf = '';
+        text = text.replace(/\s*\|\s*$/, '').trim();
+        if (!text)
+            return;
+        blocks.push({ kind: resolved.kind, text, level: resolved.level, indent: resolved.indent });
+    };
+    for (const token of tokenize(html)) {
+        if (dropName) {
+            if (token.t === 'open' && token.tag.name === dropName && !token.tag.selfClosing)
+                dropNest++;
+            else if (token.t === 'close' && token.name === dropName && --dropNest === 0)
+                dropName = null;
+            continue;
+        }
+        if (token.t === 'text') {
+            if (inTitle) {
+                title = decodeEntities(token.value).replace(/\s+/g, ' ').trim() || null;
+            }
+            else {
+                buf += decodeEntities(token.value);
+            }
+            continue;
+        }
+        if (token.t === 'open') {
+            const { name, attrs, selfClosing } = token.tag;
+            if (name === 'title') {
+                // Captured for the document heading, never emitted into the body.
+                inTitle = true;
+                continue;
+            }
+            const drop = NON_TEXT_TAGS.has(name) ||
+                isHidden(attrs) ||
+                (stripBoilerplate && (CHROME_TAGS.has(name) || isChromeAttrs(attrs)));
+            if (drop) {
+                flush();
+                if (!selfClosing) {
+                    dropName = name;
+                    dropNest = 1;
+                }
+                continue;
+            }
+            if (BLOCK_TAGS.has(name)) {
+                flush();
+                if (!selfClosing)
+                    openBlocks.push(name);
+            }
+            if (name === 'img' && attrs.alt?.trim())
+                buf += ` ${attrs.alt.trim()} `;
+            continue;
+        }
+        // close
+        const { name } = token;
+        if (name === 'title') {
+            inTitle = false;
+            continue;
+        }
+        if (CELL_TAGS.has(name)) {
+            if (buf.trim())
+                buf = `${buf.trimEnd()} | `;
+            continue;
+        }
+        if (BLOCK_TAGS.has(name)) {
+            flush();
+            const idx = openBlocks.lastIndexOf(name);
+            if (idx !== -1)
+                openBlocks.splice(idx);
+        }
+    }
+    flush();
+    return { title, text: render(blocks) };
+}
+/**
+ * Clean text for an HTML document, with the `<title>` promoted to an `# h1`
+ * when the page has no top-level heading of its own.
+ */
+export function htmlToMarkdown(html, opts = {}) {
+    const doc = extractHtmlDocument(html, opts);
+    if (!doc.title)
+        return doc.text;
+    if (/^# (?!#)/m.test(doc.text))
+        return doc.text;
+    return doc.text ? `# ${doc.title}\n\n${doc.text}` : `# ${doc.title}`;
+}
+//# sourceMappingURL=html-extract.js.map

package/dist/src/knowledge/html-extract.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"html-extract.js","sourceRoot":"","sources":["../../../src/knowledge/html-extract.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAEH,OAAO,EACL,UAAU,EACV,SAAS,EACT,YAAY,EACZ,WAAW,EACX,YAAY,EACZ,cAAc,EACd,aAAa,EACb,aAAa,EACb,SAAS,GACV,MAAM,gBAAgB,CAAC;AAExB,OAAO,EAAE,cAAc,EAAE,CAAC;AA2B1B,MAAM,OAAO,GAAG,yEAAyE,CAAC;AAE1F,SAAS,UAAU,CAAC,MAAc;IAChC,MAAM,KAAK,GAA2B,EAAE,CAAC;IACzC,OAAO,CAAC,SAAS,GAAG,CAAC,CAAC;IACtB,IAAI,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC7B,OAAO,CAAC,EAAE,CAAC;QACT,MAAM,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QACvB,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC,GAAG,cAAc,CAAC,GAAG,CAAC,OAAO,CAAC,cAAc,EAAE,EAAE,CAAC,CAAC,CAAC;QAC5E,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC3B,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,QAAQ,CAAC,CAAC,QAAQ,CAAC,IAAY;IAC7B,IAAI,CAAC,GAAG,CAAC,CAAC;IACV,OAAO,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC;QACvB,MAAM,EAAE,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC;QAChC,IAAI,EAAE,KAAK,CAAC,CAAC,EAAE,CAAC;YACd,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC;YAC1C,OAAO;QACT,CAAC;QACD,IAAI,EAAE,GAAG,CAAC;YAAE,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC;QAE1D,IAAI,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC;YAChC,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,GAAG,CAAC,CAAC,CAAC;YACxC,CAAC,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC;YACvC,SAAS;QACX,CAAC;QACD,IAAI,IAAI,CAAC,UAAU,CAAC,WAAW,EAAE,EAAE,CAAC,EAAE,CAAC;YACrC,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,GAAG,CAAC,CAAC,CAAC;YACxC,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC;YAC7E,CAAC,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC;YACvC,SAAS;QACX,CAAC;QACD,IAAI,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,EAAE,CAAC,EAAE,CAAC;YAC3D,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;YAClC,CAAC,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC;YACvC,SAAS;QACX,CAAC;QACD,IAAI,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,EAAE,CAAC,EAAE,CAAC;YAC9B,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;YAClC,IAAI,GAAG,KAAK,CAAC,CAAC,EAAE,CAAC;gBACf,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC;gBAChB,SAAS;YACX,CAAC;YACD,MAAM,IAAI,GAAG,yBAAyB,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;YACjF,IAAI,IAAI;gBAAE,MAAM,EAAE,CAAC,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,WAAW,EAAE,EAAE,CAAC;YACzD,CAAC,GAAG,GAAG,GAAG,CAAC,CAAC;YACZ,SAAS;QACX,CAAC;QAED,MAAM,SAAS,GAAG,4BAA4B,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC;QAC7E,IAAI,CAAC,SAAS,EAAE,CAAC;YACf,sDAAsD;YACtD,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,GAAG,EAAE,CAAC;YAChC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;YACX,SAAS;QACX,CAAC;QACD,MAAM,IAAI,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC;QAExC,oEAAoE;QACpE,4CAA4C;QAC5C,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC;QACjC,IAAI,KAAK,GAAG,EAAE,CAAC;QACf,OAAO,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC;YACvB,MAAM,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;YAClB,IAAI,KAAK,EAAE,CAAC;gBACV,IAAI,CAAC,KAAK,KAAK;oBAAE,KAAK,GAAG,EAAE,CAAC;YAC9B,CAAC;iBAAM,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC;gBAClC,KAAK,GAAG,CAAC,CAAC;YACZ,CAAC;iBAAM,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC;gBACrB,MAAM;YACR,CAAC;YACD,CAAC,EAAE,CAAC;QACN,CAAC;QACD,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,EAAE,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;QACtD,MAAM,WAAW,GAAG,KAAK,CAAC,OAAO,EAAE,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACzE,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,GAAG,EAAE,EAAE,IAAI,EAAE,KAAK,EAAE,UAAU,CAAC,KAAK,CAAC,EAAE,WAAW,EAAE,EAAE,CAAC;QAC1E,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QAEV,IAAI,aAAa,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC;YAC9D,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;YAC3B,MAAM,KAAK,GAAG,IAAI,MAAM,CAAC,KAAK,IAAI,OAAO,EAAE,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAC3D,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YACtE,MAAM,EAAE,CAAC,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC;YAC3B,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC;QAC9D,CAAC;IACH,CAAC;AACH,CAAC;AAED,sEAAsE;AAEtE,SAAS,WAAW,CAAC,KAAa;IAChC,OAAO,KAAK;SACT,OAAO,CAAC,oBAAoB,EAAE,OAAO,CAAC;SACtC,WAAW,EAAE;SACb,KAAK,CAAC,YAAY,CAAC;SACnB,MAAM,CAAC,OAAO,CAAC,CAAC;AACrB,CAAC;AAED,SAAS,aAAa,CAAC,KAA6B;IAClD,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,EAAE,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC;IAC9C,IAAI,IAAI,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,YAAY,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;QAAE,OAAO,IAAI,CAAC;IACrE,KAAK,MAAM,GAAG,IAAI,CAAC,OAAO,EAAE,IAAI,CAAC,EAAE,CAAC;QAClC,MAAM,KAAK,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC;QACzB,IAAI,CAAC,KAAK;YAAE,SAAS;QACrB,IAAI,WAAW,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,YAAY,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YAAE,OAAO,IAAI,CAAC;IACvE,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,SAAS,QAAQ,CAAC,KAA6B;IAC7C,IAAI,KAAK,CAAC,aAAa,CAAC,EAAE,IAAI,EAAE,CAAC,WAAW,EAAE,KAAK,MAAM;QAAE,OAAO,IAAI,CAAC;IACvE,IAAI,QAAQ,IAAI,KAAK;QAAE,OAAO,IAAI,CAAC;IACnC,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC;IAC1B,OAAO,CAAC,CAAC,KAAK,IAAI,+BAA+B,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;AAChE,CAAC;AAaD;;iBAEiB;AACjB,SAAS,WAAW,CAAC,UAAoB;IACvC,KAAK,IAAI,CAAC,GAAG,UAAU,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;QAChD,MAAM,IAAI,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC;QAC3B,IAAI,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC;YAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,KAAK,EAAE,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QAC9E,IAAI,IAAI,KAAK,IAAI,EAAE,CAAC;YAClB,IAAI,MAAM,GAAG,CAAC,CAAC,CAAC;YAChB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE;gBAAE,IAAI,UAAU,CAAC,CAAC,CAAC,KAAK,IAAI;oBAAE,MAAM,EAAE,CAAC;YAClE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,CAAC,EAAE,CAAC;QACrD,CAAC;QACD,IAAI,IAAI,KAAK,KAAK;YAAE,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC;QAC3C,IAAI,IAAI,KAAK,YAAY;YAAE,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;IACtD,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC;AAC1B,CAAC;AAED,SAAS,MAAM,CAAC,MAAe;IAC7B,IAAI,GAAG,GAAG,EAAE,CAAC;IACb,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACvC,MAAM,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;QACpB,IAAI,IAAY,CAAC;QACjB,IAAI,CAAC,CAAC,IAAI,KAAK,SAAS;YAAE,IAAI,GAAG,GAAG,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;aACjF,IAAI,CAAC,CAAC,IAAI,KAAK,IAAI;YAAE,IAAI,GAAG,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,CAAC;aACvE,IAAI,CAAC,CAAC,IAAI,KAAK,OAAO;YAAE,IAAI,GAAG,KAAK,CAAC,CAAC,IAAI,EAAE,CAAC;;YAC7C,IAAI,GAAG,CAAC,CAAC,IAAI,CAAC;QAEnB,IAAI,CAAC,KAAK,CAAC;YAAE,GAAG,IAAI,IAAI,CAAC;aACpB,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,KAAK,IAAI,IAAI,CAAC,CAAC,IAAI,KAAK,IAAI;YAAE,GAAG,IAAI,KAAK,IAAI,EAAE,CAAC;;YACvE,GAAG,IAAI,OAAO,IAAI,EAAE,CAAC;IAC5B,CAAC;IACD,OAAO,GAAG,CAAC,OAAO,CAAC,SAAS,EAAE,MAAM,CAAC,CAAC,IAAI,EAAE,CAAC;AAC/C,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,mBAAmB,CAAC,IAAY,EAAE,IAAI,GAAuB,EAAE;IAC7E,MAAM,gBAAgB,GAAG,IAAI,CAAC,gBAAgB,KAAK,KAAK,CAAC;IACzD,MAAM,MAAM,GAAY,EAAE,CAAC;IAC3B,MAAM,UAAU,GAAa,EAAE,CAAC;IAChC,IAAI,GAAG,GAAG,EAAE,CAAC;IACb,IAAI,KAAK,GAAkB,IAAI,CAAC;IAChC,IAAI,OAAO,GAAG,KAAK,CAAC;IAEpB,0EAA0E;IAC1E,6BAA6B;IAC7B,IAAI,QAAQ,GAAkB,IAAI,CAAC;IACnC,IAAI,QAAQ,GAAG,CAAC,CAAC;IAEjB,MAAM,KAAK,GAAG,GAAS,EAAE;QACvB,MAAM,QAAQ,GAAG,WAAW,CAAC,UAAU,CAAC,CAAC;QACzC,IAAI,IAAI,GAAG,QAAQ,CAAC,IAAI,KAAK,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;QAC7F,GAAG,GAAG,EAAE,CAAC;QACT,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;QAC5C,IAAI,CAAC,IAAI;YAAE,OAAO;QAClB,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,QAAQ,CAAC,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,QAAQ,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,CAAC,MAAM,EAAE,CAAC,CAAC;IAC7F,CAAC,CAAC;IAEF,KAAK,MAAM,KAAK,IAAI,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;QACnC,IAAI,QAAQ,EAAE,CAAC;YACb,IAAI,KAAK,CAAC,CAAC,KAAK,MAAM,IAAI,KAAK,CAAC,GAAG,CAAC,IAAI,KAAK,QAAQ,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,WAAW;gBAAE,QAAQ,EAAE,CAAC;iBACvF,IAAI,KAAK,CAAC,CAAC,KAAK,OAAO,IAAI,KAAK,CAAC,IAAI,KAAK,QAAQ,IAAI,EAAE,QAAQ,KAAK,CAAC;gBAAE,QAAQ,GAAG,IAAI,CAAC;YAC7F,SAAS;QACX,CAAC;QAED,IAAI,KAAK,CAAC,CAAC,KAAK,MAAM,EAAE,CAAC;YACvB,IAAI,OAAO,EAAE,CAAC;gBACZ,KAAK,GAAG,cAAc,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,IAAI,IAAI,CAAC;YAC1E,CAAC;iBAAM,CAAC;gBACN,GAAG,IAAI,cAAc,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;YACrC,CAAC;YACD,SAAS;QACX,CAAC;QAED,IAAI,KAAK,CAAC,CAAC,KAAK,MAAM,EAAE,CAAC;YACvB,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,WAAW,EAAE,GAAG,KAAK,CAAC,GAAG,CAAC;YAE/C,IAAI,IAAI,KAAK,OAAO,EAAE,CAAC;gBACrB,kEAAkE;gBAClE,OAAO,GAAG,IAAI,CAAC;gBACf,SAAS;YACX,CAAC;YAED,MAAM,IAAI,GACR,aAAa,CAAC,GAAG,CAAC,IAAI,CAAC;gBACvB,QAAQ,CAAC,KAAK,CAAC;gBACf,CAAC,gBAAgB,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,aAAa,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;YAExE,IAAI,IAAI,EAAE,CAAC;gBACT,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,WAAW,EAAE,CAAC;oBACjB,QAAQ,GAAG,IAAI,CAAC;oBAChB,QAAQ,GAAG,CAAC,CAAC;gBACf,CAAC;gBACD,SAAS;YACX,CAAC;YAED,IAAI,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;gBACzB,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,WAAW;oBAAE,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAC1C,CAAC;YACD,IAAI,IAAI,KAAK,KAAK,IAAI,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE;gBAAE,GAAG,IAAI,IAAI,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,GAAG,CAAC;YACxE,SAAS;QACX,CAAC;QAED,QAAQ;QACR,MAAM,EAAE,IAAI,EAAE,GAAG,KAAK,CAAC;QACvB,IAAI,IAAI,KAAK,OAAO,EAAE,CAAC;YACrB,OAAO,GAAG,KAAK,CAAC;YAChB,SAAS;QACX,CAAC;QACD,IAAI,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;YACxB,IAAI,GAAG,CAAC,IAAI,EAAE;gBAAE,GAAG,GAAG,GAAG,GAAG,CAAC,OAAO,EAAE,KAAK,CAAC;YAC5C,SAAS;QACX,CAAC;QACD,IAAI,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;YACzB,KAAK,EAAE,CAAC;YACR,MAAM,GAAG,GAAG,UAAU,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC;YACzC,IAAI,GAAG,KAAK,CAAC,CAAC;gBAAE,UAAU,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QACzC,CAAC;IACH,CAAC;IACD,KAAK,EAAE,CAAC;IAER,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,CAAC,MAAM,CAAC,EAAE,CAAC;AACzC,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,cAAc,CAAC,IAAY,EAAE,IAAI,GAAuB,EAAE;IACxE,MAAM,GAAG,GAAG,mBAAmB,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;IAC5C,IAAI,CAAC,GAAG,CAAC,KAAK;QAAE,OAAO,GAAG,CAAC,IAAI,CAAC;IAChC,IAAI,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC;QAAE,OAAO,GAAG,CAAC,IAAI,CAAC;IAChD,OAAO,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,KAAK,OAAO,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,KAAK,EAAE,CAAC;AACvE,CAAC"}

package/dist/src/knowledge/html-tags.d.ts

@@ -0,0 +1,30 @@
+/**
+ * Tag and entity tables for the HTML extractor (RCL-01).
+ *
+ * Split out of `html-extract.ts` purely for size: these are data, the file
+ * next door is the algorithm.
+ *
+ * @module v1/cli/knowledge/html-tags
+ */
+export declare const VOID_TAGS: Set<string>;
+/** Elements whose content is never prose. Dropped regardless of options. */
+export declare const NON_TEXT_TAGS: Set<string>;
+/** Structural chrome. Dropped only when `stripBoilerplate` is on. */
+export declare const CHROME_TAGS: Set<string>;
+/** ARIA landmark roles that mark chrome rather than content. */
+export declare const CHROME_ROLES: Set<string>;
+/**
+ * Class/id word list. Matched against `-`/`_`/space/camelCase-split tokens, so
+ * `site-header`, `cookieBanner` and `ad_slot` all hit and `content-wrapper`
+ * does not.
+ */
+export declare const CHROME_WORDS: Set<string>;
+/** Elements that start a new output block. */
+export declare const BLOCK_TAGS: Set<string>;
+/** Table cells are joined into their row with ` | ` rather than broken out
+ *  into blocks of their own, so a row survives as one readable line. */
+export declare const CELL_TAGS: Set<string>;
+/** Elements whose content is raw text — `<` inside them is not markup. */
+export declare const RAW_TEXT_TAGS: Set<string>;
+export declare function decodeEntities(s: string): string;
+//# sourceMappingURL=html-tags.d.ts.map

package/dist/src/knowledge/html-tags.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"html-tags.d.ts","sourceRoot":"","sources":["../../../src/knowledge/html-tags.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,eAAO,MAAM,SAAS,aAepB,CAAC;AAEH,4EAA4E;AAC5E,eAAO,MAAM,aAAa,aAaxB,CAAC;AAEH,qEAAqE;AACrE,eAAO,MAAM,WAAW,aAUtB,CAAC;AAEH,gEAAgE;AAChE,eAAO,MAAM,YAAY,aAWvB,CAAC;AAEH;;;;GAIG;AACH,eAAO,MAAM,YAAY,aAmDvB,CAAC;AAEH,8CAA8C;AAC9C,eAAO,MAAM,UAAU,aAqCrB,CAAC;AAEH;wEACwE;AACxE,eAAO,MAAM,SAAS,aAAwB,CAAC;AAE/C,0EAA0E;AAC1E,eAAO,MAAM,aAAa,aAAoD,CAAC;AA+C/E,wBAAgB,cAAc,CAAC,CAAC,EAAE,MAAM,GAAG,MAAM,CAgBhD"}

package/dist/src/knowledge/html-tags.js

@@ -0,0 +1,229 @@
+/**
+ * Tag and entity tables for the HTML extractor (RCL-01).
+ *
+ * Split out of `html-extract.ts` purely for size: these are data, the file
+ * next door is the algorithm.
+ *
+ * @module v1/cli/knowledge/html-tags
+ */
+export const VOID_TAGS = new Set([
+    'area',
+    'base',
+    'br',
+    'col',
+    'embed',
+    'hr',
+    'img',
+    'input',
+    'link',
+    'meta',
+    'param',
+    'source',
+    'track',
+    'wbr',
+]);
+/** Elements whose content is never prose. Dropped regardless of options. */
+export const NON_TEXT_TAGS = new Set([
+    'script',
+    'style',
+    'noscript',
+    'svg',
+    'math',
+    'template',
+    'canvas',
+    'iframe',
+    'object',
+    'applet',
+    'audio',
+    'video',
+]);
+/** Structural chrome. Dropped only when `stripBoilerplate` is on. */
+export const CHROME_TAGS = new Set([
+    'nav',
+    'header',
+    'footer',
+    'aside',
+    'menu',
+    'dialog',
+    'button',
+    'select',
+    'textarea',
+]);
+/** ARIA landmark roles that mark chrome rather than content. */
+export const CHROME_ROLES = new Set([
+    'navigation',
+    'banner',
+    'contentinfo',
+    'complementary',
+    'search',
+    'menubar',
+    'menu',
+    'toolbar',
+    'dialog',
+    'alertdialog',
+]);
+/**
+ * Class/id word list. Matched against `-`/`_`/space/camelCase-split tokens, so
+ * `site-header`, `cookieBanner` and `ad_slot` all hit and `content-wrapper`
+ * does not.
+ */
+export const CHROME_WORDS = new Set([
+    'cookie',
+    'cookies',
+    'consent',
+    'gdpr',
+    'ccpa',
+    'cmp',
+    'banner',
+    'promo',
+    'promotion',
+    'ad',
+    'ads',
+    'advert',
+    'advertisement',
+    'adsbygoogle',
+    'sponsored',
+    'newsletter',
+    'subscribe',
+    'signup',
+    'paywall',
+    'popup',
+    'modal',
+    'overlay',
+    'lightbox',
+    'nav',
+    'navbar',
+    'navigation',
+    'menu',
+    'breadcrumb',
+    'breadcrumbs',
+    'masthead',
+    'topbar',
+    'toolbar',
+    'sidebar',
+    'skiplink',
+    'pagination',
+    'paginate',
+    'pager',
+    'share',
+    'sharing',
+    'social',
+    'socials',
+    'footer',
+    'header',
+    'copyright',
+    'related',
+    'recommended',
+    'recirc',
+    'trending',
+    'comments',
+    'disqus',
+]);
+/** Elements that start a new output block. */
+export const BLOCK_TAGS = new Set([
+    'address',
+    'article',
+    'blockquote',
+    'br',
+    'dd',
+    'details',
+    'div',
+    'dl',
+    'dt',
+    'fieldset',
+    'figcaption',
+    'figure',
+    'footer',
+    'form',
+    'h1',
+    'h2',
+    'h3',
+    'h4',
+    'h5',
+    'h6',
+    'header',
+    'hr',
+    'li',
+    'main',
+    'nav',
+    'ol',
+    'p',
+    'pre',
+    'section',
+    'summary',
+    'table',
+    'tbody',
+    'tfoot',
+    'thead',
+    'tr',
+    'ul',
+]);
+/** Table cells are joined into their row with ` | ` rather than broken out
+ *  into blocks of their own, so a row survives as one readable line. */
+export const CELL_TAGS = new Set(['td', 'th']);
+/** Elements whose content is raw text — `<` inside them is not markup. */
+export const RAW_TEXT_TAGS = new Set(['script', 'style', 'textarea', 'title']);
+// ── Entities ───────────────────────────────────────────────────────
+const NAMED_ENTITIES = {
+    amp: '&',
+    lt: '<',
+    gt: '>',
+    quot: '"',
+    apos: "'",
+    nbsp: ' ',
+    ensp: ' ',
+    emsp: ' ',
+    thinsp: ' ',
+    shy: '',
+    mdash: '—',
+    ndash: '–',
+    hellip: '…',
+    lsquo: '‘',
+    rsquo: '’',
+    ldquo: '“',
+    rdquo: '”',
+    bull: '•',
+    middot: '·',
+    laquo: '«',
+    raquo: '»',
+    copy: '©',
+    reg: '®',
+    trade: '™',
+    deg: '°',
+    euro: '€',
+    pound: '£',
+    yen: '¥',
+    cent: '¢',
+    sect: '§',
+    para: '¶',
+    dagger: '†',
+    times: '×',
+    divide: '÷',
+    plusmn: '±',
+    frac12: '½',
+    prime: '′',
+    larr: '←',
+    rarr: '→',
+    harr: '↔',
+};
+export function decodeEntities(s) {
+    if (!s.includes('&'))
+        return s;
+    return s.replace(/&(#[xX]?[0-9a-fA-F]+|[a-zA-Z][a-zA-Z0-9]{1,31});/g, (whole, body) => {
+        if (body[0] === '#') {
+            const hex = body[1] === 'x' || body[1] === 'X';
+            const code = Number.parseInt(hex ? body.slice(2) : body.slice(1), hex ? 16 : 10);
+            if (!Number.isFinite(code) || code <= 0 || code > 0x10ffff)
+                return whole;
+            try {
+                return String.fromCodePoint(code);
+            }
+            catch {
+                return whole;
+            }
+        }
+        const hit = NAMED_ENTITIES[body] ?? NAMED_ENTITIES[body.toLowerCase()];
+        return hit === undefined ? whole : hit;
+    });
+}
+//# sourceMappingURL=html-tags.js.map

package/dist/src/knowledge/html-tags.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"html-tags.js","sourceRoot":"","sources":["../../../src/knowledge/html-tags.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,MAAM,CAAC,MAAM,SAAS,GAAG,IAAI,GAAG,CAAC;IAC/B,MAAM;IACN,MAAM;IACN,IAAI;IACJ,KAAK;IACL,OAAO;IACP,IAAI;IACJ,KAAK;IACL,OAAO;IACP,MAAM;IACN,MAAM;IACN,OAAO;IACP,QAAQ;IACR,OAAO;IACP,KAAK;CACN,CAAC,CAAC;AAEH,4EAA4E;AAC5E,MAAM,CAAC,MAAM,aAAa,GAAG,IAAI,GAAG,CAAC;IACnC,QAAQ;IACR,OAAO;IACP,UAAU;IACV,KAAK;IACL,MAAM;IACN,UAAU;IACV,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,OAAO;IACP,OAAO;CACR,CAAC,CAAC;AAEH,qEAAqE;AACrE,MAAM,CAAC,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC;IACjC,KAAK;IACL,QAAQ;IACR,QAAQ;IACR,OAAO;IACP,MAAM;IACN,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,UAAU;CACX,CAAC,CAAC;AAEH,gEAAgE;AAChE,MAAM,CAAC,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC;IAClC,YAAY;IACZ,QAAQ;IACR,aAAa;IACb,eAAe;IACf,QAAQ;IACR,SAAS;IACT,MAAM;IACN,SAAS;IACT,QAAQ;IACR,aAAa;CACd,CAAC,CAAC;AAEH;;;;GAIG;AACH,MAAM,CAAC,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC;IAClC,QAAQ;IACR,SAAS;IACT,SAAS;IACT,MAAM;IACN,MAAM;IACN,KAAK;IACL,QAAQ;IACR,OAAO;IACP,WAAW;IACX,IAAI;IACJ,KAAK;IACL,QAAQ;IACR,eAAe;IACf,aAAa;IACb,WAAW;IACX,YAAY;IACZ,WAAW;IACX,QAAQ;IACR,SAAS;IACT,OAAO;IACP,OAAO;IACP,SAAS;IACT,UAAU;IACV,KAAK;IACL,QAAQ;IACR,YAAY;IACZ,MAAM;IACN,YAAY;IACZ,aAAa;IACb,UAAU;IACV,QAAQ;IACR,SAAS;IACT,SAAS;IACT,UAAU;IACV,YAAY;IACZ,UAAU;IACV,OAAO;IACP,OAAO;IACP,SAAS;IACT,QAAQ;IACR,SAAS;IACT,QAAQ;IACR,QAAQ;IACR,WAAW;IACX,SAAS;IACT,aAAa;IACb,QAAQ;IACR,UAAU;IACV,UAAU;IACV,QAAQ;CACT,CAAC,CAAC;AAEH,8CAA8C;AAC9C,MAAM,CAAC,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC;IAChC,SAAS;IACT,SAAS;IACT,YAAY;IACZ,IAAI;IACJ,IAAI;IACJ,SAAS;IACT,KAAK;IACL,IAAI;IACJ,IAAI;IACJ,UAAU;IACV,YAAY;IACZ,QAAQ;IACR,QAAQ;IACR,MAAM;IACN,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,QAAQ;IACR,IAAI;IACJ,IAAI;IACJ,MAAM;IACN,KAAK;IACL,IAAI;IACJ,GAAG;IACH,KAAK;IACL,SAAS;IACT,SAAS;IACT,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,IAAI;IACJ,IAAI;CACL,CAAC,CAAC;AAEH;wEACwE;AACxE,MAAM,CAAC,MAAM,SAAS,GAAG,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC;AAE/C,0EAA0E;AAC1E,MAAM,CAAC,MAAM,aAAa,GAAG,IAAI,GAAG,CAAC,CAAC,QAAQ,EAAE,OAAO,EAAE,UAAU,EAAE,OAAO,CAAC,CAAC,CAAC;AAE/E,sEAAsE;AAEtE,MAAM,cAAc,GAA2B;IAC7C,GAAG,EAAE,GAAG;IACR,EAAE,EAAE,GAAG;IACP,EAAE,EAAE,GAAG;IACP,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,MAAM,EAAE,GAAG;IACX,GAAG,EAAE,EAAE;IACP,KAAK,EAAE,GAAG;IACV,KAAK,EAAE,GAAG;IACV,MAAM,EAAE,GAAG;IACX,KAAK,EAAE,GAAG;IACV,KAAK,EAAE,GAAG;IACV,KAAK,EAAE,GAAG;IACV,KAAK,EAAE,GAAG;IACV,IAAI,EAAE,GAAG;IACT,MAAM,EAAE,GAAG;IACX,KAAK,EAAE,GAAG;IACV,KAAK,EAAE,GAAG;IACV,IAAI,EAAE,GAAG;IACT,GAAG,EAAE,GAAG;IACR,KAAK,EAAE,GAAG;IACV,GAAG,EAAE,GAAG;IACR,IAAI,EAAE,GAAG;IACT,KAAK,EAAE,GAAG;IACV,GAAG,EAAE,GAAG;IACR,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,MAAM,EAAE,GAAG;IACX,KAAK,EAAE,GAAG;IACV,MAAM,EAAE,GAAG;IACX,MAAM,EAAE,GAAG;IACX,MAAM,EAAE,GAAG;IACX,KAAK,EAAE,GAAG;IACV,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;CACV,CAAC;AAEF,MAAM,UAAU,cAAc,CAAC,CAAS;IACtC,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC,GAAG,CAAC;QAAE,OAAO,CAAC,CAAC;IAC/B,OAAO,CAAC,CAAC,OAAO,CAAC,mDAAmD,EAAE,CAAC,KAAK,EAAE,IAAY,EAAE,EAAE;QAC5F,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;YACpB,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC;YAC/C,MAAM,IAAI,GAAG,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;YACjF,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,IAAI,GAAG,QAAQ;gBAAE,OAAO,KAAK,CAAC;YACzE,IAAI,CAAC;gBACH,OAAO,MAAM,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC;YACpC,CAAC;YAAC,MAAM,CAAC;gBACP,OAAO,KAAK,CAAC;YACf,CAAC;QACH,CAAC;QACD,MAAM,GAAG,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,cAAc,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,CAAC;QACvE,OAAO,GAAG,KAAK,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC;IACzC,CAAC,CAAC,CAAC;AACL,CAAC"}

package/dist/src/knowledge/library.d.ts

@@ -0,0 +1,106 @@
+/**
+ * The capture library (RCL-09) — browsing what has been captured, rather than
+ * searching it.
+ *
+ * `listDocuments` answers "what is indexed". A library answers a different
+ * question: everything from this site, captured last week, tagged `reading`,
+ * newest first. The facets come from the capture envelope's `meta.json`
+ * (RCL-07), which is already on every record — this module is the query layer
+ * over them, and the data layer the desktop GUI renders.
+ *
+ * Rules, once, here:
+ *  - within a facet values are OR'd, across facets AND'd (`--site a --site b
+ *    --tag x` = "from a or b, tagged x");
+ *  - `--site example.com` matches `www.example.com` and `docs.example.com`,
+ *    because nobody browsing their own library means the bare host only;
+ *  - a document with no provenance (an ordinary file on disk) sorts by its
+ *    index time and matches no capture facet — it is not a capture.
+ *
+ * @module v1/cli/knowledge/library
+ */
+import type { CaptureProvenance } from './capture-envelope.js';
+import type { DocumentMeta } from './document-pipeline.js';
+export interface LibraryFilter {
+    /** Host, or a parent domain of it. */
+    site?: string[];
+    tag?: string[];
+    collection?: string[];
+    /** `extension` | `monobrowse` | `crawl`. */
+    source?: string[];
+    /** ISO date/time, or a relative age: `7d`, `24h`, `30m`. */
+    since?: string;
+    until?: string;
+    /** Only documents that came from a capture envelope. */
+    capturedOnly?: boolean;
+    scope?: string;
+    /** Substring over title, URL and path. */
+    text?: string;
+}
+export type LibrarySort = 'captured' | 'indexed' | 'title' | 'path' | 'size';
+export interface LibraryEntry {
+    filePath: string;
+    scope: string;
+    title: string;
+    url?: string;
+    site?: string;
+    capturedAt: string;
+    indexedAt: string;
+    source?: string;
+    collection?: string;
+    tags: string[];
+    chunkCount: number;
+    size: number;
+    version?: number;
+    /** False for ordinary files on disk: no capture envelope backed this one. */
+    captured: boolean;
+}
+/** The host a document was captured from, or undefined for a local file. */
+export declare function documentSite(meta: {
+    canonicalUrl?: string;
+    provenance?: CaptureProvenance;
+}): string | undefined;
+/** When the page was captured; falls back to when it was indexed. */
+export declare function documentCapturedAt(meta: {
+    indexedAt?: string;
+    provenance?: CaptureProvenance;
+}): string;
+/**
+ * A point in time from `2026-09-01`, a full ISO stamp, or an AGE like `7d`
+ * (meaning "7 days ago"). Returns null for anything unparseable so a typo
+ * cannot silently filter the whole library away.
+ */
+export declare function parseWhen(value: string | undefined, now?: number): number | null;
+/** True when a document belongs in this view. */
+export declare function matchesLibraryFilter(meta: DocumentMeta, filter?: LibraryFilter, now?: number): boolean;
+export declare function filterDocuments(docs: DocumentMeta[], filter?: LibraryFilter, now?: number): DocumentMeta[];
+export declare function sortDocuments(docs: DocumentMeta[], sort?: LibrarySort, order?: 'asc' | 'desc'): DocumentMeta[];
+/** The flat, JSON-friendly row the GUI and `--json` consumers read. */
+export declare function libraryEntry(meta: DocumentMeta): LibraryEntry;
+export interface ListLibraryOptions extends LibraryFilter {
+    sort?: LibrarySort;
+    order?: 'asc' | 'desc';
+    limit?: number;
+    now?: number;
+}
+/** Browse the library: filter, sort, cap. Never throws on an empty store. */
+export declare function listLibrary(docs: DocumentMeta[], opts?: ListLibraryOptions): LibraryEntry[];
+/** Facet counts for the current view — what a sidebar renders. */
+export declare function libraryFacets(docs: DocumentMeta[]): {
+    sites: Array<{
+        value: string;
+        count: number;
+    }>;
+    tags: Array<{
+        value: string;
+        count: number;
+    }>;
+    collections: Array<{
+        value: string;
+        count: number;
+    }>;
+    sources: Array<{
+        value: string;
+        count: number;
+    }>;
+};
+//# sourceMappingURL=library.d.ts.map