@olib-ai/owl-browser-sdk 2.0.5 → 2.0.7

package/dist/extraction/index.js

@@ -0,0 +1,19 @@
+/**
+ * Universal extraction module for Owl Browser Node.js SDK.
+ *
+ * Provides structured data extraction from any website using CSS selectors,
+ * pattern detection, table parsing, metadata extraction, and multi-page scraping.
+ */
+// Main entry point
+export { Extractor } from './extractor.js';
+// Transforms (pure functions, usable standalone)
+export { applyTransform, applyTransforms, applyPattern, coerceType, parsePrice, parseDate, resolveUrl, } from './transforms.js';
+// Selector engine (pure functions, usable standalone)
+export { extractAll, extractFirst, count, } from './selector-engine.js';
+// Table extractor (pure functions, usable standalone)
+export { extractTable, extractGrid, extractDefinitionList, detectTables, } from './table-extractor.js';
+// Meta extractor (pure functions, usable standalone)
+export { extractMeta, extractJsonLd, extractOpenGraph, extractTwitterCard, extractMicrodata, extractFeeds, extractCanonical, extractFavicon, } from './meta-extractor.js';
+// Pattern detector (pure functions, usable standalone)
+export { detect, detectAndExtract, } from './pattern-detector.js';
+//# sourceMappingURL=index.js.map

package/dist/extraction/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/extraction/index.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,mBAAmB;AACnB,OAAO,EAAE,SAAS,EAAE,MAAM,gBAAgB,CAAC;AAuB3C,iDAAiD;AACjD,OAAO,EACL,cAAc,EACd,eAAe,EACf,YAAY,EACZ,UAAU,EACV,UAAU,EACV,SAAS,EACT,UAAU,GACX,MAAM,iBAAiB,CAAC;AAEzB,sDAAsD;AACtD,OAAO,EACL,UAAU,EACV,YAAY,EACZ,KAAK,GACN,MAAM,sBAAsB,CAAC;AAE9B,sDAAsD;AACtD,OAAO,EACL,YAAY,EACZ,WAAW,EACX,qBAAqB,EACrB,YAAY,GACb,MAAM,sBAAsB,CAAC;AAE9B,qDAAqD;AACrD,OAAO,EACL,WAAW,EACX,aAAa,EACb,gBAAgB,EAChB,kBAAkB,EAClB,gBAAgB,EAChB,YAAY,EACZ,gBAAgB,EAChB,cAAc,GACf,MAAM,qBAAqB,CAAC;AAE7B,uDAAuD;AACvD,OAAO,EACL,MAAM,EACN,gBAAgB,GACjB,MAAM,uBAAuB,CAAC"}

package/dist/extraction/list-extractor.d.ts

@@ -0,0 +1,24 @@
+/**
+ * Extract from lists, cards, grids — any repeating DOM pattern with a known container.
+ *
+ * Provides auto-field detection: infers field names from child element semantics
+ * (headings → title, img → image, a → link, price-like → price, etc.).
+ */
+import type { ExtractedRecord, FieldSpec, ListOptions } from './types.js';
+/**
+ * Extract items from a container with optional auto-field detection.
+ */
+export declare function extract(html: string, containerSelector: string, options?: ListOptions): ExtractedRecord[];
+/**
+ * Analyze child elements to infer field names and selectors.
+ */
+export declare function analyzeStructure(html: string, containerSelector: string, itemSelector?: string): Record<string, FieldSpec>;
+/**
+ * Extract a simple <ul>/<ol> list as string[].
+ */
+export declare function extractList(html: string, selector?: string): string[];
+/**
+ * Extract card-like repeating structures.
+ */
+export declare function extractCards(html: string, containerSelector: string, cardSelector?: string): ExtractedRecord[];
+//# sourceMappingURL=list-extractor.d.ts.map

package/dist/extraction/list-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"list-extractor.d.ts","sourceRoot":"","sources":["../../src/extraction/list-extractor.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAIH,OAAO,KAAK,EAAE,eAAe,EAAE,SAAS,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AAG1E;;GAEG;AACH,wBAAgB,OAAO,CACrB,IAAI,EAAE,MAAM,EACZ,iBAAiB,EAAE,MAAM,EACzB,OAAO,CAAC,EAAE,WAAW,GACpB,eAAe,EAAE,CAiBnB;AAED;;GAEG;AACH,wBAAgB,gBAAgB,CAC9B,IAAI,EAAE,MAAM,EACZ,iBAAiB,EAAE,MAAM,EACzB,YAAY,CAAC,EAAE,MAAM,GACpB,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,CAY3B;AAED;;GAEG;AACH,wBAAgB,WAAW,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,MAAM,EAAE,CAWrE;AAED;;GAEG;AACH,wBAAgB,YAAY,CAC1B,IAAI,EAAE,MAAM,EACZ,iBAAiB,EAAE,MAAM,EACzB,YAAY,CAAC,EAAE,MAAM,GACpB,eAAe,EAAE,CAEnB"}

package/dist/extraction/list-extractor.js

@@ -0,0 +1,303 @@
+/**
+ * Extract from lists, cards, grids — any repeating DOM pattern with a known container.
+ *
+ * Provides auto-field detection: infers field names from child element semantics
+ * (headings → title, img → image, a → link, price-like → price, etc.).
+ */
+import { load } from 'cheerio';
+import { extractAll } from './selector-engine.js';
+/**
+ * Extract items from a container with optional auto-field detection.
+ */
+export function extract(html, containerSelector, options) {
+    const $ = load(html);
+    const container = $(containerSelector).first();
+    if (container.length === 0)
+        return [];
+    const itemSelector = options?.itemSelector ?? detectItemSelector($, container);
+    if (!itemSelector)
+        return [];
+    const fields = options?.fields ?? analyzeStructure(html, containerSelector, itemSelector);
+    let results = extractAll(html, `${containerSelector} ${itemSelector}`, fields);
+    if (options?.limit) {
+        results = results.slice(0, options.limit);
+    }
+    return results;
+}
+/**
+ * Analyze child elements to infer field names and selectors.
+ */
+export function analyzeStructure(html, containerSelector, itemSelector) {
+    const $ = load(html);
+    const container = $(containerSelector).first();
+    if (container.length === 0)
+        return {};
+    const effectiveItemSelector = itemSelector ?? detectItemSelector($, container);
+    if (!effectiveItemSelector)
+        return {};
+    const firstItem = container.find(effectiveItemSelector).first();
+    if (firstItem.length === 0)
+        return {};
+    return inferFields($, firstItem);
+}
+/**
+ * Extract a simple <ul>/<ol> list as string[].
+ */
+export function extractList(html, selector) {
+    const $ = load(html);
+    const list = $(selector ?? 'ul, ol').first();
+    if (list.length === 0)
+        return [];
+    const items = [];
+    list.find('li').each((_i, el) => {
+        const text = $(el).text().trim();
+        if (text)
+            items.push(text);
+    });
+    return items;
+}
+/**
+ * Extract card-like repeating structures.
+ */
+export function extractCards(html, containerSelector, cardSelector) {
+    return extract(html, containerSelector, { itemSelector: cardSelector });
+}
+// ==================== Internal ====================
+function cssEsc(value) {
+    return value.replace(/([!"#$%&'()*+,./:;<=>?@[\\\]^`{|}~])/g, '\\$1');
+}
+/**
+ * Detect the most likely item selector within a container.
+ */
+function detectItemSelector($, container) {
+    const children = container.children();
+    if (children.length === 0)
+        return null;
+    // Count tag+class combos among direct children
+    const comboCounts = new Map();
+    children.each((_i, el) => {
+        if (el.type !== 'tag')
+            return;
+        const tag = el.tagName.toLowerCase();
+        const rawCls = $(el).attr('class')?.split(/\s+/)[0];
+        const key = rawCls ? `${tag}.${cssEsc(rawCls)}` : tag;
+        comboCounts.set(key, (comboCounts.get(key) ?? 0) + 1);
+    });
+    // Find the most common combo with at least 2 occurrences
+    let bestSelector = null;
+    let bestCount = 1;
+    for (const [selector, count] of comboCounts) {
+        if (count > bestCount) {
+            bestCount = count;
+            bestSelector = selector;
+        }
+    }
+    // Also check common list patterns
+    if (!bestSelector) {
+        const li = container.find('> li');
+        if (li.length >= 2)
+            return 'li';
+        const tr = container.find('> tr');
+        if (tr.length >= 2)
+            return 'tr';
+    }
+    return bestSelector ? `> ${bestSelector}` : null;
+}
+/**
+ * Infer field names from the structure of a representative item element.
+ */
+function inferFields($, item) {
+    const fields = {};
+    const usedNames = new Set();
+    const itemTag = item[0]?.tagName?.toLowerCase();
+    // If the item itself is an <a>, treat it as a link with title
+    if (itemTag === 'a') {
+        fields['link'] = '@href';
+        fields['title'] = ''; // empty selector = the item's own text
+        usedNames.add('link');
+        usedNames.add('title');
+    }
+    // If the item itself is a heading, treat it as a title
+    if (!usedNames.has('title') && itemTag && /^h[1-6]$/.test(itemTag)) {
+        fields['title'] = '';
+        usedNames.add('title');
+    }
+    // --- Microdata / schema.org detection (itemprop attributes) ---
+    const microdataMap = {
+        'name': 'title',
+        'headline': 'title',
+        'price': 'price',
+        'description': 'description',
+        'image': 'image',
+        'url': 'link',
+        'datePublished': 'date',
+        'dateCreated': 'date',
+    };
+    for (const [prop, fieldName] of Object.entries(microdataMap)) {
+        if (usedNames.has(fieldName))
+            continue;
+        const el = item.find(`[itemprop="${prop}"]`).first();
+        if (el.length === 0)
+            continue;
+        const tag = el[0]?.tagName?.toLowerCase() ?? '';
+        if (fieldName === 'image' && tag === 'img') {
+            fields[fieldName] = `[itemprop="${prop}"]@src`;
+        }
+        else if (fieldName === 'link' && tag === 'a') {
+            fields[fieldName] = `[itemprop="${prop}"]@href`;
+        }
+        else if (fieldName === 'date' && tag === 'time') {
+            fields[fieldName] = { selector: `[itemprop="${prop}"]`, attribute: 'datetime' };
+        }
+        else if (fieldName === 'price') {
+            const content = el.attr('content');
+            if (content) {
+                fields[fieldName] = { selector: `[itemprop="${prop}"]`, attribute: 'content' };
+            }
+            else {
+                fields[fieldName] = `[itemprop="${prop}"]`;
+            }
+        }
+        else {
+            fields[fieldName] = `[itemprop="${prop}"]`;
+        }
+        usedNames.add(fieldName);
+    }
+    // --- Standard field detection ---
+    // Check for headings → title
+    if (!usedNames.has('title')) {
+        const heading = item.find('h1, h2, h3, h4, h5, h6').first();
+        if (heading.length > 0) {
+            const tag = heading[0]?.tagName?.toLowerCase() ?? 'h2';
+            fields['title'] = tag;
+            usedNames.add('title');
+        }
+    }
+    // Check for images → image
+    if (!usedNames.has('image')) {
+        const img = item.find('img').first();
+        if (img.length > 0) {
+            fields['image'] = 'img@src';
+            usedNames.add('image');
+        }
+    }
+    // Check for links → link
+    if (!usedNames.has('link')) {
+        const link = item.find('a').first();
+        if (link.length > 0) {
+            fields['link'] = 'a@href';
+            if (!usedNames.has('title') && link.text().trim()) {
+                fields['title'] = 'a';
+                usedNames.add('title');
+            }
+            usedNames.add('link');
+        }
+    }
+    // Check for price-like content — first by class, then by text pattern
+    if (!usedNames.has('price')) {
+        const priceEl = item.find('[class*="price"], [data-price]').first();
+        if (priceEl.length > 0) {
+            const cls = priceEl.attr('class')?.split(/\s+/)[0];
+            fields['price'] = cls ? `.${cssEsc(cls)}` : '[class*="price"]';
+            usedNames.add('price');
+        }
+        else {
+            // Text-based price detection: look for currency symbols + digits
+            const priceRegex = /[$€£¥₹₩]\s*[\d,.]+|[\d,.]+\s*[$€£¥₹₩]/;
+            item.find('*').each((_i, el) => {
+                if (usedNames.has('price'))
+                    return;
+                if (el.type !== 'tag')
+                    return;
+                const $el = $(el);
+                // Only consider leaf-ish elements (no children or only text children)
+                if ($el.children().length > 2)
+                    return;
+                const text = $el.text().trim();
+                if (priceRegex.test(text) && text.length < 30) {
+                    const tag = el.tagName.toLowerCase();
+                    const cls = $el.attr('class')?.split(/\s+/)[0];
+                    fields['price'] = cls ? `.${cssEsc(cls)}` : tag;
+                    usedNames.add('price');
+                }
+            });
+        }
+    }
+    // Check for time/date elements
+    if (!usedNames.has('date')) {
+        const timeEl = item.find('time, [class*="date"], [class*="time"]').first();
+        if (timeEl.length > 0) {
+            const tag = timeEl[0]?.tagName?.toLowerCase();
+            if (tag === 'time') {
+                fields['date'] = { selector: 'time', attribute: 'datetime' };
+            }
+            else {
+                const cls = timeEl.attr('class')?.split(/\s+/)[0];
+                fields['date'] = cls ? `.${cssEsc(cls)}` : '[class*="date"]';
+            }
+            usedNames.add('date');
+        }
+    }
+    // Check for rating/star content
+    if (!usedNames.has('rating')) {
+        const ratingEl = item.find('[class*="rating"], [class*="star"], [data-rating], [aria-label*="rating" i], [aria-label*="star" i]').first();
+        if (ratingEl.length > 0) {
+            const dataRating = ratingEl.attr('data-rating');
+            if (dataRating) {
+                fields['rating'] = { selector: '[data-rating]', attribute: 'data-rating' };
+            }
+            else {
+                const ariaLabel = ratingEl.attr('aria-label');
+                if (ariaLabel) {
+                    fields['rating'] = { selector: '[aria-label]', attribute: 'aria-label' };
+                }
+                else {
+                    const cls = ratingEl.attr('class')?.split(/\s+/)[0];
+                    fields['rating'] = cls ? `.${cssEsc(cls)}` : '[class*="rating"]';
+                }
+            }
+            usedNames.add('rating');
+        }
+    }
+    // Check for badges/status/tags
+    if (!usedNames.has('badge')) {
+        const badgeEl = item.find('[class*="badge"]:not(a), [class*="status"]:not(a), [class*="label"]:not(label):not(a), [class*="chip"]:not(a)').first();
+        if (badgeEl.length > 0) {
+            const cls = badgeEl.attr('class')?.split(/\s+/)[0];
+            fields['badge'] = cls ? `.${cssEsc(cls)}` : '[class*="badge"]';
+            usedNames.add('badge');
+        }
+    }
+    // Check for description/body text (longer text blocks or <p> elements)
+    if (!usedNames.has('description')) {
+        // First, look for <p> elements anywhere in the item (up to depth 3)
+        const pEl = item.find('p').first();
+        if (pEl.length > 0 && pEl.text().trim().length > 10) {
+            fields['description'] = 'p';
+            usedNames.add('description');
+        }
+        // Fallback: check all child elements for longer text (threshold lowered to 20)
+        if (!usedNames.has('description')) {
+            item.find('*').each((_i, el) => {
+                if (usedNames.has('description'))
+                    return;
+                if (el.type !== 'tag')
+                    return;
+                const $el = $(el);
+                const tag = el.tagName.toLowerCase();
+                if (['h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'img', 'a', 'time', 'p'].includes(tag))
+                    return;
+                if ($el.find('h1, h2, h3, h4, h5, h6, img').length > 0)
+                    return; // Skip containers of other fields
+                const text = $el.text().trim();
+                if (text.length > 20) {
+                    const cls = $el.attr('class')?.split(/\s+/)[0];
+                    fields['description'] = cls ? `.${cssEsc(cls)}` : tag;
+                    usedNames.add('description');
+                }
+            });
+        }
+    }
+    return fields;
+}
+//# sourceMappingURL=list-extractor.js.map

package/dist/extraction/list-extractor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"list-extractor.js","sourceRoot":"","sources":["../../src/extraction/list-extractor.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,IAAI,EAAiC,MAAM,SAAS,CAAC;AAG9D,OAAO,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAC;AAElD;;GAEG;AACH,MAAM,UAAU,OAAO,CACrB,IAAY,EACZ,iBAAyB,EACzB,OAAqB;IAErB,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,MAAM,SAAS,GAAG,CAAC,CAAC,iBAAiB,CAAC,CAAC,KAAK,EAAE,CAAC;IAC/C,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IAEtC,MAAM,YAAY,GAAG,OAAO,EAAE,YAAY,IAAI,kBAAkB,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC;IAC/E,IAAI,CAAC,YAAY;QAAE,OAAO,EAAE,CAAC;IAE7B,MAAM,MAAM,GAAG,OAAO,EAAE,MAAM,IAAI,gBAAgB,CAAC,IAAI,EAAE,iBAAiB,EAAE,YAAY,CAAC,CAAC;IAE1F,IAAI,OAAO,GAAG,UAAU,CAAC,IAAI,EAAE,GAAG,iBAAiB,IAAI,YAAY,EAAE,EAAE,MAAM,CAAC,CAAC;IAE/E,IAAI,OAAO,EAAE,KAAK,EAAE,CAAC;QACnB,OAAO,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC;IAC5C,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,gBAAgB,CAC9B,IAAY,EACZ,iBAAyB,EACzB,YAAqB;IAErB,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,MAAM,SAAS,GAAG,CAAC,CAAC,iBAAiB,CAAC,CAAC,KAAK,EAAE,CAAC;IAC/C,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IAEtC,MAAM,qBAAqB,GAAG,YAAY,IAAI,kBAAkB,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC;IAC/E,IAAI,CAAC,qBAAqB;QAAE,OAAO,EAAE,CAAC;IAEtC,MAAM,SAAS,GAAG,SAAS,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAC,KAAK,EAAE,CAAC;IAChE,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IAEtC,OAAO,WAAW,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC;AACnC,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,WAAW,CAAC,IAAY,EAAE,QAAiB;IACzD,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,MAAM,IAAI,GAAG,CAAC,CAAC,QAAQ,IAAI,QAAQ,CAAC,CAAC,KAAK,EAAE,CAAC;IAC7C,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IAEjC,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;QAC9B,MAAM,IAAI,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC;QACjC,IAAI,IAAI;YAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC7B,CAAC,CAAC,CAAC;IACH,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,YAAY,CAC1B,IAAY,EACZ,iBAAyB,EACzB,YAAqB;IAErB,OAAO,OAAO,CAAC,IAAI,EAAE,iBAAiB,EAAE,EAAE,YAAY,EAAE,YAAY,EAAE,CAAC,CAAC;AAC1E,CAAC;AAED,qDAAqD;AAErD,SAAS,MAAM,CAAC,KAAa;IAC3B,OAAO,KAAK,CAAC,OAAO,CAAC,uCAAuC,EAAE,MAAM,CAAC,CAAC;AACxE,CAAC;AAED;;GAEG;AACH,SAAS,kBAAkB,CAAC,CAAa,EAAE,SAA2B;IACpE,MAAM,QAAQ,GAAG,SAAS,CAAC,QAAQ,EAAE,CAAC;IACtC,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IAEvC,+CAA+C;IAC/C,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC9C,QAAQ,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;QACvB,IAAI,EAAE,CAAC,IAAI,KAAK,KAAK;YAAE,OAAO;QAC9B,MAAM,GAAG,GAAI,EAA0B,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC;QAC9D,MAAM,MAAM,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QACpD,MAAM,GAAG,GAAG,MAAM,CAAC,CAAC,CAAC,GAAG,GAAG,IAAI,MAAM,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC;QACtD,WAAW,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,WAAW,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;IACxD,CAAC,CAAC,CAAC;IAEH,yDAAyD;IACzD,IAAI,YAAY,GAAkB,IAAI,CAAC;IACvC,IAAI,SAAS,GAAG,CAAC,CAAC;IAClB,KAAK,MAAM,CAAC,QAAQ,EAAE,KAAK,CAAC,IAAI,WAAW,EAAE,CAAC;QAC5C,IAAI,KAAK,GAAG,SAAS,EAAE,CAAC;YACtB,SAAS,GAAG,KAAK,CAAC;YAClB,YAAY,GAAG,QAAQ,CAAC;QAC1B,CAAC;IACH,CAAC;IAED,kCAAkC;IAClC,IAAI,CAAC,YAAY,EAAE,CAAC;QAClB,MAAM,EAAE,GAAG,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAClC,IAAI,EAAE,CAAC,MAAM,IAAI,CAAC;YAAE,OAAO,IAAI,CAAC;QAChC,MAAM,EAAE,GAAG,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAClC,IAAI,EAAE,CAAC,MAAM,IAAI,CAAC;YAAE,OAAO,IAAI,CAAC;IAClC,CAAC;IAED,OAAO,YAAY,CAAC,CAAC,CAAC,KAAK,YAAY,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;AACnD,CAAC;AAED;;GAEG;AACH,SAAS,WAAW,CAClB,CAAa,EACb,IAAsB;IAEtB,MAAM,MAAM,GAA8B,EAAE,CAAC;IAC7C,MAAM,SAAS,GAAG,IAAI,GAAG,EAAU,CAAC;IAEpC,MAAM,OAAO,GAAI,IAAI,CAAC,CAAC,CAA0B,EAAE,OAAO,EAAE,WAAW,EAAE,CAAC;IAE1E,8DAA8D;IAC9D,IAAI,OAAO,KAAK,GAAG,EAAE,CAAC;QACpB,MAAM,CAAC,MAAM,CAAC,GAAG,OAAO,CAAC;QACzB,MAAM,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC,CAAE,uCAAuC;QAC9D,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;QACtB,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;IACzB,CAAC;IAED,uDAAuD;IACvD,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,OAAO,IAAI,UAAU,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;QACnE,MAAM,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC;QACrB,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;IACzB,CAAC;IAED,iEAAiE;IACjE,MAAM,YAAY,GAA2B;QAC3C,MAAM,EAAE,OAAO;QACf,UAAU,EAAE,OAAO;QACnB,OAAO,EAAE,OAAO;QAChB,aAAa,EAAE,aAAa;QAC5B,OAAO,EAAE,OAAO;QAChB,KAAK,EAAE,MAAM;QACb,eAAe,EAAE,MAAM;QACvB,aAAa,EAAE,MAAM;KACtB,CAAC;IAEF,KAAK,MAAM,CAAC,IAAI,EAAE,SAAS,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,YAAY,CAAC,EAAE,CAAC;QAC7D,IAAI,SAAS,CAAC,GAAG,CAAC,SAAS,CAAC;YAAE,SAAS;QACvC,MAAM,EAAE,GAAG,IAAI,CAAC,IAAI,CAAC,cAAc,IAAI,IAAI,CAAC,CAAC,KAAK,EAAE,CAAC;QACrD,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QAE9B,MAAM,GAAG,GAAI,EAAE,CAAC,CAAC,CAA0B,EAAE,OAAO,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC;QAC1E,IAAI,SAAS,KAAK,OAAO,IAAI,GAAG,KAAK,KAAK,EAAE,CAAC;YAC3C,MAAM,CAAC,SAAS,CAAC,GAAG,cAAc,IAAI,QAAQ,CAAC;QACjD,CAAC;aAAM,IAAI,SAAS,KAAK,MAAM,IAAI,GAAG,KAAK,GAAG,EAAE,CAAC;YAC/C,MAAM,CAAC,SAAS,CAAC,GAAG,cAAc,IAAI,SAAS,CAAC;QAClD,CAAC;aAAM,IAAI,SAAS,KAAK,MAAM,IAAI,GAAG,KAAK,MAAM,EAAE,CAAC;YAClD,MAAM,CAAC,SAAS,CAAC,GAAG,EAAE,QAAQ,EAAE,cAAc,IAAI,IAAI,EAAE,SAAS,EAAE,UAAU,EAAE,CAAC;QAClF,CAAC;aAAM,IAAI,SAAS,KAAK,OAAO,EAAE,CAAC;YACjC,MAAM,OAAO,GAAG,EAAE,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;YACnC,IAAI,OAAO,EAAE,CAAC;gBACZ,MAAM,CAAC,SAAS,CAAC,GAAG,EAAE,QAAQ,EAAE,cAAc,IAAI,IAAI,EAAE,SAAS,EAAE,SAAS,EAAE,CAAC;YACjF,CAAC;iBAAM,CAAC;gBACN,MAAM,CAAC,SAAS,CAAC,GAAG,cAAc,IAAI,IAAI,CAAC;YAC7C,CAAC;QACH,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,SAAS,CAAC,GAAG,cAAc,IAAI,IAAI,CAAC;QAC7C,CAAC;QACD,SAAS,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;IAC3B,CAAC;IAED,mCAAmC;IAEnC,6BAA6B;IAC7B,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;QAC5B,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,wBAAwB,CAAC,CAAC,KAAK,EAAE,CAAC;QAC5D,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACvB,MAAM,GAAG,GAAI,OAAO,CAAC,CAAC,CAA0B,EAAE,OAAO,EAAE,WAAW,EAAE,IAAI,IAAI,CAAC;YACjF,MAAM,CAAC,OAAO,CAAC,GAAG,GAAG,CAAC;YACtB,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACzB,CAAC;IACH,CAAC;IAED,2BAA2B;IAC3B,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;QAC5B,MAAM,GAAG,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,KAAK,EAAE,CAAC;QACrC,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACnB,MAAM,CAAC,OAAO,CAAC,GAAG,SAAS,CAAC;YAC5B,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACzB,CAAC;IACH,CAAC;IAED,yBAAyB;IACzB,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC;QAC3B,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,CAAC;QACpC,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACpB,MAAM,CAAC,MAAM,CAAC,GAAG,QAAQ,CAAC;YAC1B,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC;gBAClD,MAAM,CAAC,OAAO,CAAC,GAAG,GAAG,CAAC;gBACtB,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;YACzB,CAAC;YACD,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;QACxB,CAAC;IACH,CAAC;IAED,sEAAsE;IACtE,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;QAC5B,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,gCAAgC,CAAC,CAAC,KAAK,EAAE,CAAC;QACpE,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACvB,MAAM,GAAG,GAAG,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;YACnD,MAAM,CAAC,OAAO,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,kBAAkB,CAAC;YAC/D,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACzB,CAAC;aAAM,CAAC;YACN,iEAAiE;YACjE,MAAM,UAAU,GAAG,uCAAuC,CAAC;YAC3D,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;gBAC7B,IAAI,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC;oBAAE,OAAO;gBACnC,IAAI,EAAE,CAAC,IAAI,KAAK,KAAK;oBAAE,OAAO;gBAC9B,MAAM,GAAG,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC;gBAClB,sEAAsE;gBACtE,IAAI,GAAG,CAAC,QAAQ,EAAE,CAAC,MAAM,GAAG,CAAC;oBAAE,OAAO;gBACtC,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC;gBAC/B,IAAI,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,MAAM,GAAG,EAAE,EAAE,CAAC;oBAC9C,MAAM,GAAG,GAAI,EAA0B,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC;oBAC9D,MAAM,GAAG,GAAG,GAAG,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;oBAC/C,MAAM,CAAC,OAAO,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC;oBAChD,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;gBACzB,CAAC;YACH,CAAC,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,+BAA+B;IAC/B,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC;QAC3B,MAAM,MAAM,GAAG,IAAI,CAAC,IAAI,CAAC,wCAAwC,CAAC,CAAC,KAAK,EAAE,CAAC;QAC3E,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACtB,MAAM,GAAG,GAAI,MAAM,CAAC,CAAC,CAA0B,EAAE,OAAO,EAAE,WAAW,EAAE,CAAC;YACxE,IAAI,GAAG,KAAK,MAAM,EAAE,CAAC;gBACnB,MAAM,CAAC,MAAM,CAAC,GAAG,EAAE,QAAQ,EAAE,MAAM,EAAE,SAAS,EAAE,UAAU,EAAE,CAAC;YAC/D,CAAC;iBAAM,CAAC;gBACN,MAAM,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;gBAClD,MAAM,CAAC,MAAM,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,iBAAiB,CAAC;YAC/D,CAAC;YACD,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;QACxB,CAAC;IACH,CAAC;IAED,gCAAgC;IAChC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC7B,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,qGAAqG,CAAC,CAAC,KAAK,EAAE,CAAC;QAC1I,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACxB,MAAM,UAAU,GAAG,QAAQ,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;YAChD,IAAI,UAAU,EAAE,CAAC;gBACf,MAAM,CAAC,QAAQ,CAAC,GAAG,EAAE,QAAQ,EAAE,eAAe,EAAE,SAAS,EAAE,aAAa,EAAE,CAAC;YAC7E,CAAC;iBAAM,CAAC;gBACN,MAAM,SAAS,GAAG,QAAQ,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;gBAC9C,IAAI,SAAS,EAAE,CAAC;oBACd,MAAM,CAAC,QAAQ,CAAC,GAAG,EAAE,QAAQ,EAAE,cAAc,EAAE,SAAS,EAAE,YAAY,EAAE,CAAC;gBAC3E,CAAC;qBAAM,CAAC;oBACN,MAAM,GAAG,GAAG,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;oBACpD,MAAM,CAAC,QAAQ,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,mBAAmB,CAAC;gBACnE,CAAC;YACH,CAAC;YACD,SAAS,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;QAC1B,CAAC;IACH,CAAC;IAED,+BAA+B;IAC/B,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;QAC5B,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,+GAA+G,CAAC,CAAC,KAAK,EAAE,CAAC;QACnJ,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACvB,MAAM,GAAG,GAAG,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;YACnD,MAAM,CAAC,OAAO,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,kBAAkB,CAAC;YAC/D,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACzB,CAAC;IACH,CAAC;IAED,uEAAuE;IACvE,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,aAAa,CAAC,EAAE,CAAC;QAClC,oEAAoE;QACpE,MAAM,GAAG,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,CAAC;QACnC,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,GAAG,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,EAAE,EAAE,CAAC;YACpD,MAAM,CAAC,aAAa,CAAC,GAAG,GAAG,CAAC;YAC5B,SAAS,CAAC,GAAG,CAAC,aAAa,CAAC,CAAC;QAC/B,CAAC;QAED,+EAA+E;QAC/E,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,aAAa,CAAC,EAAE,CAAC;YAClC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;gBAC7B,IAAI,SAAS,CAAC,GAAG,CAAC,aAAa,CAAC;oBAAE,OAAO;gBACzC,IAAI,EAAE,CAAC,IAAI,KAAK,KAAK;oBAAE,OAAO;gBAC9B,MAAM,GAAG,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC;gBAClB,MAAM,GAAG,GAAI,EAA0B,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC;gBAC9D,IAAI,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,GAAG,CAAC,CAAC,QAAQ,CAAC,GAAG,CAAC;oBAAE,OAAO;gBACxF,IAAI,GAAG,CAAC,IAAI,CAAC,6BAA6B,CAAC,CAAC,MAAM,GAAG,CAAC;oBAAE,OAAO,CAAC,kCAAkC;gBAClG,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC;gBAC/B,IAAI,IAAI,CAAC,MAAM,GAAG,EAAE,EAAE,CAAC;oBACrB,MAAM,GAAG,GAAG,GAAG,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;oBAC/C,MAAM,CAAC,aAAa,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC;oBACtD,SAAS,CAAC,GAAG,CAAC,aAAa,CAAC,CAAC;gBAC/B,CAAC;YACH,CAAC,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/dist/extraction/meta-extractor.d.ts

@@ -0,0 +1,40 @@
+/**
+ * Structured metadata extraction from HTML.
+ *
+ * Extracts JSON-LD, OpenGraph, Twitter Card, microdata (itemprop/itemscope),
+ * canonical URLs, favicons, and RSS/Atom feeds.
+ */
+import type { MetaData, FeedLink, ExtractedRecord } from './types.js';
+/**
+ * Extract all structured metadata from HTML.
+ */
+export declare function extractMeta(html: string): MetaData;
+/**
+ * Extract JSON-LD structured data.
+ */
+export declare function extractJsonLd(html: string): object[];
+/**
+ * Extract OpenGraph meta tags.
+ */
+export declare function extractOpenGraph(html: string): Record<string, string>;
+/**
+ * Extract Twitter Card meta tags.
+ */
+export declare function extractTwitterCard(html: string): Record<string, string>;
+/**
+ * Extract microdata (itemprop/itemscope) from HTML.
+ */
+export declare function extractMicrodata(html: string): ExtractedRecord[];
+/**
+ * Extract RSS/Atom feed links.
+ */
+export declare function extractFeeds(html: string): FeedLink[];
+/**
+ * Extract the canonical URL.
+ */
+export declare function extractCanonical(html: string): string | null;
+/**
+ * Extract the favicon URL.
+ */
+export declare function extractFavicon(html: string): string | null;
+//# sourceMappingURL=meta-extractor.d.ts.map

package/dist/extraction/meta-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"meta-extractor.d.ts","sourceRoot":"","sources":["../../src/extraction/meta-extractor.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAGH,OAAO,KAAK,EAAE,QAAQ,EAAE,QAAQ,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAEtE;;GAEG;AACH,wBAAgB,WAAW,CAAC,IAAI,EAAE,MAAM,GAAG,QAAQ,CA8DlD;AAED;;GAEG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,CAoBpD;AAED;;GAEG;AACH,wBAAgB,gBAAgB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAarE;AAED;;GAEG;AACH,wBAAgB,kBAAkB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAavE;AAED;;GAEG;AACH,wBAAgB,gBAAgB,CAAC,IAAI,EAAE,MAAM,GAAG,eAAe,EAAE,CA8ChE;AAED;;GAEG;AACH,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,GAAG,QAAQ,EAAE,CAuBrD;AAED;;GAEG;AACH,wBAAgB,gBAAgB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAI5D;AAED;;GAEG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAI1D"}

package/dist/extraction/meta-extractor.js

@@ -0,0 +1,216 @@
+/**
+ * Structured metadata extraction from HTML.
+ *
+ * Extracts JSON-LD, OpenGraph, Twitter Card, microdata (itemprop/itemscope),
+ * canonical URLs, favicons, and RSS/Atom feeds.
+ */
+import { load } from 'cheerio';
+/**
+ * Extract all structured metadata from HTML.
+ */
+export function extractMeta(html) {
+    const $ = load(html);
+    const result = {
+        title: null,
+        description: null,
+        canonical: null,
+        og: {},
+        twitter: {},
+        jsonLd: [],
+        microdata: [],
+        feeds: [],
+        favicon: null,
+        other: {},
+    };
+    // Title
+    const titleEl = $('title').first();
+    if (titleEl.length > 0) {
+        result.title = titleEl.text().trim() || null;
+    }
+    // Canonical
+    const canonicalEl = $('link[rel="canonical"]').first();
+    if (canonicalEl.length > 0) {
+        result.canonical = canonicalEl.attr('href') ?? null;
+    }
+    // Favicon
+    const faviconEl = $('link[rel="icon"], link[rel="shortcut icon"]').first();
+    if (faviconEl.length > 0) {
+        result.favicon = faviconEl.attr('href') ?? null;
+    }
+    // Meta tags
+    $('meta').each((_i, el) => {
+        const $el = $(el);
+        const name = $el.attr('name') ?? $el.attr('property') ?? '';
+        const content = $el.attr('content') ?? '';
+        if (!name || !content)
+            return;
+        if (name === 'description') {
+            result.description = content;
+        }
+        else if (name.startsWith('og:')) {
+            result.og[name.slice(3)] = content;
+        }
+        else if (name.startsWith('twitter:')) {
+            result.twitter[name.slice(8)] = content;
+        }
+        else {
+            result.other[name] = content;
+        }
+    });
+    // JSON-LD
+    result.jsonLd = extractJsonLd(html);
+    // Microdata
+    result.microdata = extractMicrodata(html);
+    // Feeds
+    result.feeds = extractFeeds(html);
+    return result;
+}
+/**
+ * Extract JSON-LD structured data.
+ */
+export function extractJsonLd(html) {
+    const $ = load(html);
+    const results = [];
+    $('script[type="application/ld+json"]').each((_i, el) => {
+        const text = $(el).html();
+        if (!text)
+            return;
+        try {
+            const parsed = JSON.parse(text);
+            if (Array.isArray(parsed)) {
+                results.push(...parsed);
+            }
+            else if (typeof parsed === 'object' && parsed !== null) {
+                results.push(parsed);
+            }
+        }
+        catch {
+            // Skip malformed JSON-LD
+        }
+    });
+    return results;
+}
+/**
+ * Extract OpenGraph meta tags.
+ */
+export function extractOpenGraph(html) {
+    const $ = load(html);
+    const og = {};
+    $('meta[property^="og:"]').each((_i, el) => {
+        const prop = $(el).attr('property') ?? '';
+        const content = $(el).attr('content') ?? '';
+        if (prop && content) {
+            og[prop.slice(3)] = content;
+        }
+    });
+    return og;
+}
+/**
+ * Extract Twitter Card meta tags.
+ */
+export function extractTwitterCard(html) {
+    const $ = load(html);
+    const twitter = {};
+    $('meta[name^="twitter:"]').each((_i, el) => {
+        const name = $(el).attr('name') ?? '';
+        const content = $(el).attr('content') ?? '';
+        if (name && content) {
+            twitter[name.slice(8)] = content;
+        }
+    });
+    return twitter;
+}
+/**
+ * Extract microdata (itemprop/itemscope) from HTML.
+ */
+export function extractMicrodata(html) {
+    const $ = load(html);
+    const results = [];
+    $('[itemscope]').each((_i, el) => {
+        const $el = $(el);
+        // Skip nested itemscopes (only extract top-level)
+        if ($el.parents('[itemscope]').length > 0)
+            return;
+        const record = {};
+        const itemType = $el.attr('itemtype');
+        if (itemType)
+            record['@type'] = itemType;
+        $el.find('[itemprop]').each((_j, prop) => {
+            const $prop = $(prop);
+            // Skip props that belong to a nested itemscope
+            const closestScope = $prop.closest('[itemscope]');
+            if (closestScope.length > 0 && closestScope[0] !== el)
+                return;
+            const name = $prop.attr('itemprop') ?? '';
+            if (!name)
+                return;
+            // Extract value based on element type
+            let value = null;
+            const tag = prop.tagName?.toLowerCase() ?? '';
+            if (tag === 'meta') {
+                value = $prop.attr('content') ?? null;
+            }
+            else if (tag === 'a' || tag === 'link') {
+                value = $prop.attr('href') ?? null;
+            }
+            else if (tag === 'img') {
+                value = $prop.attr('src') ?? null;
+            }
+            else if (tag === 'time') {
+                value = $prop.attr('datetime') ?? ($prop.text().trim() || null);
+            }
+            else {
+                value = $prop.text().trim() || null;
+            }
+            record[name] = value;
+        });
+        if (Object.keys(record).length > 0) {
+            results.push(record);
+        }
+    });
+    return results;
+}
+/**
+ * Extract RSS/Atom feed links.
+ */
+export function extractFeeds(html) {
+    const $ = load(html);
+    const feeds = [];
+    $('link[rel="alternate"]').each((_i, el) => {
+        const $el = $(el);
+        const type = $el.attr('type') ?? '';
+        const href = $el.attr('href');
+        if (!href)
+            return;
+        let feedType = 'unknown';
+        if (type.includes('rss'))
+            feedType = 'rss';
+        else if (type.includes('atom'))
+            feedType = 'atom';
+        else
+            return; // Skip non-feed alternate links
+        feeds.push({
+            type: feedType,
+            href,
+            title: $el.attr('title') ?? null,
+        });
+    });
+    return feeds;
+}
+/**
+ * Extract the canonical URL.
+ */
+export function extractCanonical(html) {
+    const $ = load(html);
+    const el = $('link[rel="canonical"]').first();
+    return el.length > 0 ? (el.attr('href') ?? null) : null;
+}
+/**
+ * Extract the favicon URL.
+ */
+export function extractFavicon(html) {
+    const $ = load(html);
+    const el = $('link[rel="icon"], link[rel="shortcut icon"]').first();
+    return el.length > 0 ? (el.attr('href') ?? null) : null;
+}
+//# sourceMappingURL=meta-extractor.js.map

package/dist/extraction/meta-extractor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"meta-extractor.js","sourceRoot":"","sources":["../../src/extraction/meta-extractor.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,IAAI,EAAE,MAAM,SAAS,CAAC;AAG/B;;GAEG;AACH,MAAM,UAAU,WAAW,CAAC,IAAY;IACtC,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IAErB,MAAM,MAAM,GAAa;QACvB,KAAK,EAAE,IAAI;QACX,WAAW,EAAE,IAAI;QACjB,SAAS,EAAE,IAAI;QACf,EAAE,EAAE,EAAE;QACN,OAAO,EAAE,EAAE;QACX,MAAM,EAAE,EAAE;QACV,SAAS,EAAE,EAAE;QACb,KAAK,EAAE,EAAE;QACT,OAAO,EAAE,IAAI;QACb,KAAK,EAAE,EAAE;KACV,CAAC;IAEF,QAAQ;IACR,MAAM,OAAO,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,CAAC;IACnC,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACvB,MAAM,CAAC,KAAK,GAAG,OAAO,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,IAAI,IAAI,CAAC;IAC/C,CAAC;IAED,YAAY;IACZ,MAAM,WAAW,GAAG,CAAC,CAAC,uBAAuB,CAAC,CAAC,KAAK,EAAE,CAAC;IACvD,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC3B,MAAM,CAAC,SAAS,GAAG,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC;IACtD,CAAC;IAED,UAAU;IACV,MAAM,SAAS,GAAG,CAAC,CAAC,6CAA6C,CAAC,CAAC,KAAK,EAAE,CAAC;IAC3E,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACzB,MAAM,CAAC,OAAO,GAAG,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC;IAClD,CAAC;IAED,YAAY;IACZ,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;QACxB,MAAM,GAAG,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC;QAClB,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC;QAC5D,MAAM,OAAO,GAAG,GAAG,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;QAC1C,IAAI,CAAC,IAAI,IAAI,CAAC,OAAO;YAAE,OAAO;QAE9B,IAAI,IAAI,KAAK,aAAa,EAAE,CAAC;YAC3B,MAAM,CAAC,WAAW,GAAG,OAAO,CAAC;QAC/B,CAAC;aAAM,IAAI,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;YAClC,MAAM,CAAC,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC;QACrC,CAAC;aAAM,IAAI,IAAI,CAAC,UAAU,CAAC,UAAU,CAAC,EAAE,CAAC;YACvC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC;QAC1C,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,OAAO,CAAC;QAC/B,CAAC;IACH,CAAC,CAAC,CAAC;IAEH,UAAU;IACV,MAAM,CAAC,MAAM,GAAG,aAAa,CAAC,IAAI,CAAC,CAAC;IAEpC,YAAY;IACZ,MAAM,CAAC,SAAS,GAAG,gBAAgB,CAAC,IAAI,CAAC,CAAC;IAE1C,QAAQ;IACR,MAAM,CAAC,KAAK,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IAElC,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,aAAa,CAAC,IAAY;IACxC,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,MAAM,OAAO,GAAa,EAAE,CAAC;IAE7B,CAAC,CAAC,oCAAoC,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;QACtD,MAAM,IAAI,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;QAC1B,IAAI,CAAC,IAAI;YAAE,OAAO;QAClB,IAAI,CAAC;YACH,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;YAChC,IAAI,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC1B,OAAO,CAAC,IAAI,CAAC,GAAG,MAAM,CAAC,CAAC;YAC1B,CAAC;iBAAM,IAAI,OAAO,MAAM,KAAK,QAAQ,IAAI,MAAM,KAAK,IAAI,EAAE,CAAC;gBACzD,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;YACvB,CAAC;QACH,CAAC;QAAC,MAAM,CAAC;YACP,yBAAyB;QAC3B,CAAC;IACH,CAAC,CAAC,CAAC;IAEH,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,gBAAgB,CAAC,IAAY;IAC3C,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,MAAM,EAAE,GAA2B,EAAE,CAAC;IAEtC,CAAC,CAAC,uBAAuB,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;QACzC,MAAM,IAAI,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC;QAC1C,MAAM,OAAO,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;QAC5C,IAAI,IAAI,IAAI,OAAO,EAAE,CAAC;YACpB,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC;QAC9B,CAAC;IACH,CAAC,CAAC,CAAC;IAEH,OAAO,EAAE,CAAC;AACZ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAAC,IAAY;IAC7C,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,MAAM,OAAO,GAA2B,EAAE,CAAC;IAE3C,CAAC,CAAC,wBAAwB,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;QAC1C,MAAM,IAAI,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;QACtC,MAAM,OAAO,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;QAC5C,IAAI,IAAI,IAAI,OAAO,EAAE,CAAC;YACpB,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC;QACnC,CAAC;IACH,CAAC,CAAC,CAAC;IAEH,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,gBAAgB,CAAC,IAAY;IAC3C,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,MAAM,OAAO,GAAsB,EAAE,CAAC;IAEtC,CAAC,CAAC,aAAa,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;QAC/B,MAAM,GAAG,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC;QAClB,kDAAkD;QAClD,IAAI,GAAG,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO;QAElD,MAAM,MAAM,GAAoB,EAAE,CAAC;QACnC,MAAM,QAAQ,GAAG,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QACtC,IAAI,QAAQ;YAAE,MAAM,CAAC,OAAO,CAAC,GAAG,QAAQ,CAAC;QAEzC,GAAG,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,IAAI,EAAE,EAAE;YACvC,MAAM,KAAK,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC;YACtB,+CAA+C;YAC/C,MAAM,YAAY,GAAG,KAAK,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC;YAClD,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,IAAI,YAAY,CAAC,CAAC,CAAC,KAAK,EAAE;gBAAE,OAAO;YAE9D,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC;YAC1C,IAAI,CAAC,IAAI;gBAAE,OAAO;YAElB,sCAAsC;YACtC,IAAI,KAAK,GAAkB,IAAI,CAAC;YAChC,MAAM,GAAG,GAAI,IAA6B,CAAC,OAAO,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC;YACxE,IAAI,GAAG,KAAK,MAAM,EAAE,CAAC;gBACnB,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,IAAI,CAAC;YACxC,CAAC;iBAAM,IAAI,GAAG,KAAK,GAAG,IAAI,GAAG,KAAK,MAAM,EAAE,CAAC;gBACzC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC;YACrC,CAAC;iBAAM,IAAI,GAAG,KAAK,KAAK,EAAE,CAAC;gBACzB,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC;YACpC,CAAC;iBAAM,IAAI,GAAG,KAAK,MAAM,EAAE,CAAC;gBAC1B,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,IAAI,IAAI,CAAC,CAAC;YAClE,CAAC;iBAAM,CAAC;gBACN,KAAK,GAAG,KAAK,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,IAAI,IAAI,CAAC;YACtC,CAAC;YAED,MAAM,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC;QACvB,CAAC,CAAC,CAAC;QAEH,IAAI,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACnC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACvB,CAAC;IACH,CAAC,CAAC,CAAC;IAEH,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,YAAY,CAAC,IAAY;IACvC,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,MAAM,KAAK,GAAe,EAAE,CAAC;IAE7B,CAAC,CAAC,uBAAuB,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE;QACzC,MAAM,GAAG,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC;QAClB,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;QACpC,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAC9B,IAAI,CAAC,IAAI;YAAE,OAAO;QAElB,IAAI,QAAQ,GAAqB,SAAS,CAAC;QAC3C,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC;YAAE,QAAQ,GAAG,KAAK,CAAC;aACtC,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC;YAAE,QAAQ,GAAG,MAAM,CAAC;;YAC7C,OAAO,CAAC,gCAAgC;QAE7C,KAAK,CAAC,IAAI,CAAC;YACT,IAAI,EAAE,QAAQ;YACd,IAAI;YACJ,KAAK,EAAE,GAAG,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,IAAI;SACjC,CAAC,CAAC;IACL,CAAC,CAAC,CAAC;IAEH,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,gBAAgB,CAAC,IAAY;IAC3C,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,MAAM,EAAE,GAAG,CAAC,CAAC,uBAAuB,CAAC,CAAC,KAAK,EAAE,CAAC;IAC9C,OAAO,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;AAC1D,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,cAAc,CAAC,IAAY;IACzC,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;IACrB,MAAM,EAAE,GAAG,CAAC,CAAC,6CAA6C,CAAC,CAAC,KAAK,EAAE,CAAC;IACpE,OAAO,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;AAC1D,CAAC"}