afdocs 0.4.0 → 0.6.0

package/dist/helpers/detect-tabs.d.ts

@@ -0,0 +1,12 @@
+export interface TabPanel {
+    label: string | null;
+    html: string;
+}
+export interface DetectedTabGroup {
+    framework: string;
+    tabCount: number;
+    htmlSlice: string;
+    panels: TabPanel[];
+}
+export declare function detectTabGroups(html: string): DetectedTabGroup[];
+//# sourceMappingURL=detect-tabs.d.ts.map

package/dist/helpers/detect-tabs.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"detect-tabs.d.ts","sourceRoot":"","sources":["../../src/helpers/detect-tabs.ts"],"names":[],"mappings":"AAEA,MAAM,WAAW,QAAQ;IACvB,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,IAAI,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,gBAAgB;IAC/B,SAAS,EAAE,MAAM,CAAC;IAClB,QAAQ,EAAE,MAAM,CAAC;IACjB,SAAS,EAAE,MAAM,CAAC;IAClB,MAAM,EAAE,QAAQ,EAAE,CAAC;CACpB;AAqUD,wBAAgB,eAAe,CAAC,IAAI,EAAE,MAAM,GAAG,gBAAgB,EAAE,CAgBhE"}

package/dist/helpers/detect-tabs.js

@@ -0,0 +1,309 @@
+import { parse } from 'node-html-parser';
+function isDescendantOf(node, ancestor) {
+    let current = node.parentNode;
+    while (current) {
+        if (current === ancestor)
+            return true;
+        current = current.parentNode;
+    }
+    return false;
+}
+function isInsideClaimed(node, claimed) {
+    for (const container of claimed) {
+        if (container === node || isDescendantOf(node, container))
+            return true;
+    }
+    return false;
+}
+function textOf(el) {
+    // Clone to avoid mutating the original DOM, then strip <style> elements
+    // that some component libraries (e.g. LeafyGreen) embed inside tab buttons.
+    const clone = el.clone();
+    for (const style of clone.querySelectorAll('style')) {
+        style.remove();
+    }
+    return clone.textContent.trim();
+}
+/**
+ * Walk up from `start` (exclusive) looking for the nearest ancestor that
+ * contains `[role="tabpanel"]` children.  Some component libraries (e.g.
+ * LeafyGreen) place the tablist and the tab-panels as siblings rather than
+ * parent/child, so `tablist.parentNode` alone won't find the panels.
+ * Stops after `maxDepth` levels to avoid scanning the whole document.
+ */
+function findContainerWithPanels(start, claimed, maxDepth = 4) {
+    let current = start.parentNode;
+    for (let depth = 0; current && depth < maxDepth; depth++) {
+        if (isInsideClaimed(current, claimed))
+            return null;
+        const panels = current.querySelectorAll('[role="tabpanel"]');
+        if (panels.length > 0)
+            return { container: current, panels: [...panels] };
+        current = current.parentNode;
+    }
+    return null;
+}
+const docusaurus = (root, claimed) => {
+    const groups = [];
+    // Docusaurus uses role="tablist" with tabs__item class children
+    const tablists = root.querySelectorAll('[role="tablist"]');
+    for (const tablist of tablists) {
+        if (isInsideClaimed(tablist, claimed))
+            continue;
+        const tabs = tablist.querySelectorAll('.tabs__item');
+        if (tabs.length === 0)
+            continue;
+        // Find the wrapping container (parent of both tablist and tabpanels).
+        // Some component libraries put tablist and tabpanels as siblings rather
+        // than direct parent/child, so we walk up until we find panels.
+        const found = findContainerWithPanels(tablist, claimed);
+        if (!found)
+            continue;
+        const { container, panels } = found;
+        const labels = tabs.map((t) => textOf(t));
+        const panelData = panels.map((p, i) => ({
+            label: labels[i] ?? null,
+            html: p.outerHTML,
+        }));
+        claimed.add(container);
+        groups.push({
+            framework: 'docusaurus',
+            tabCount: tabs.length,
+            htmlSlice: container.outerHTML,
+            panels: panelData,
+        });
+    }
+    return groups;
+};
+const mkdocs = (root, claimed) => {
+    const groups = [];
+    const containers = root.querySelectorAll('.tabbed-set');
+    for (const container of containers) {
+        const el = container;
+        if (isInsideClaimed(el, claimed))
+            continue;
+        const labels = el.querySelectorAll('.tabbed-labels label, .tabbed-labels > *');
+        const panels = el.querySelectorAll('.tabbed-content > .tabbed-block, .tabbed-content > *');
+        if (labels.length === 0 && panels.length === 0)
+            continue;
+        const panelData = [];
+        const count = Math.max(labels.length, panels.length);
+        for (let i = 0; i < count; i++) {
+            panelData.push({
+                label: labels[i] ? textOf(labels[i]) : null,
+                html: panels[i] ? panels[i].outerHTML : '',
+            });
+        }
+        claimed.add(el);
+        groups.push({
+            framework: 'mkdocs',
+            tabCount: count,
+            htmlSlice: el.outerHTML,
+            panels: panelData,
+        });
+    }
+    return groups;
+};
+const sphinx = (root, claimed) => {
+    const groups = [];
+    const containers = root.querySelectorAll('.sphinx-tabs');
+    for (const container of containers) {
+        const el = container;
+        if (isInsideClaimed(el, claimed))
+            continue;
+        const tabs = el.querySelectorAll('.sphinx-tabs-tab');
+        const panels = el.querySelectorAll('.sphinx-tabs-panel');
+        if (tabs.length === 0 && panels.length === 0)
+            continue;
+        const panelData = [];
+        const count = Math.max(tabs.length, panels.length);
+        for (let i = 0; i < count; i++) {
+            panelData.push({
+                label: tabs[i] ? textOf(tabs[i]) : null,
+                html: panels[i] ? panels[i].outerHTML : '',
+            });
+        }
+        claimed.add(el);
+        groups.push({
+            framework: 'sphinx',
+            tabCount: count,
+            htmlSlice: el.outerHTML,
+            panels: panelData,
+        });
+    }
+    return groups;
+};
+const msLearn = (root, claimed) => {
+    const groups = [];
+    const containers = root.querySelectorAll('.tabGroup');
+    for (const container of containers) {
+        const el = container;
+        if (isInsideClaimed(el, claimed))
+            continue;
+        const tabs = el.querySelectorAll('[role="tab"][data-tab]');
+        const panels = el.querySelectorAll('[role="tabpanel"], .tabPanel');
+        if (tabs.length === 0 || panels.length === 0)
+            continue;
+        const panelData = [];
+        const count = Math.max(tabs.length, panels.length);
+        for (let i = 0; i < count; i++) {
+            panelData.push({
+                label: tabs[i] ? textOf(tabs[i]) : null,
+                html: panels[i] ? panels[i].outerHTML : '',
+            });
+        }
+        claimed.add(el);
+        groups.push({
+            framework: 'microsoft-learn',
+            tabCount: count,
+            htmlSlice: el.outerHTML,
+            panels: panelData,
+        });
+    }
+    return groups;
+};
+const genericAria = (root, claimed) => {
+    const groups = [];
+    const tablists = root.querySelectorAll('[role="tablist"]');
+    for (const tablist of tablists) {
+        const el = tablist;
+        if (isInsideClaimed(el, claimed))
+            continue;
+        const found = findContainerWithPanels(el, claimed);
+        const tabs = el.querySelectorAll('[role="tab"]');
+        if (!found) {
+            // No panels found — skip. Tabs without panels are typically site
+            // navigation, not content tab groups. The serialization check only
+            // cares about panel content, so there's nothing to measure here.
+            continue;
+        }
+        const { container, panels } = found;
+        if (tabs.length === 0 && panels.length === 0)
+            continue;
+        // Use tab count as the authority. Containers may hold panels from
+        // multiple tab groups; capping to tabs.length avoids misattributing
+        // panels from sibling groups in the same ancestor.
+        const panelData = [];
+        const count = tabs.length > 0 ? tabs.length : panels.length;
+        for (let i = 0; i < count; i++) {
+            panelData.push({
+                label: tabs[i] ? textOf(tabs[i]) : null,
+                html: panels[i] ? panels[i].outerHTML : '',
+            });
+        }
+        claimed.add(container);
+        groups.push({
+            framework: 'generic-aria',
+            tabCount: count,
+            htmlSlice: container.outerHTML,
+            panels: panelData,
+        });
+    }
+    return groups;
+};
+/**
+ * Find all `<Tabs>...</Tabs>` blocks in raw source text with proper nesting.
+ * Returns the content (including the tags) and the start offset of each
+ * top-level `<Tabs>` block.  We use regex rather than node-html-parser
+ * because the DOM parser can't reliably handle `</Tabs>` followed by
+ * markdown text followed by `<Tabs>` — it merges them into one element.
+ */
+function findTabsBlocks(source) {
+    const blocks = [];
+    const openRe = /<Tabs\b[^>]*>/gi;
+    let match;
+    while ((match = openRe.exec(source)) !== null) {
+        const startIdx = match.index;
+        // Track nesting to find the matching </Tabs>
+        let depth = 1;
+        let pos = startIdx + match[0].length;
+        while (pos < source.length && depth > 0) {
+            const nextOpen = source.indexOf('<Tabs', pos);
+            const nextClose = source.indexOf('</Tabs>', pos);
+            if (nextClose === -1)
+                break; // unclosed tag
+            if (nextOpen !== -1 && nextOpen < nextClose) {
+                depth++;
+                pos = nextOpen + 5;
+            }
+            else {
+                depth--;
+                if (depth === 0) {
+                    blocks.push(source.substring(startIdx, nextClose + 7));
+                    // Advance the outer regex past this block to avoid re-entering
+                    openRe.lastIndex = nextClose + 7;
+                }
+                pos = nextClose + 7;
+            }
+        }
+    }
+    return blocks;
+}
+const TAB_OPEN_RE = /<(Tab|TabItem)\b([^>]*)>/gi;
+const ATTR_RE = /(?:name|label|value)\s*=\s*"([^"]*)"/i;
+/**
+ * MDX-style tabs: `<Tabs>` container with `<Tab name="...">` or
+ * `<TabItem label="...">` children. Used by MongoDB docs, Docusaurus MDX, and others.
+ *
+ * Uses regex-based block finding instead of DOM parsing because
+ * node-html-parser can't reliably parse `</Tabs>` + markdown + `<Tabs>`
+ * as separate elements in mixed markdown/HTML content.
+ */
+const mdxTabs = (_root, _claimed, source) => {
+    if (!source)
+        return [];
+    const groups = [];
+    const blocks = findTabsBlocks(source);
+    for (const block of blocks) {
+        // Find direct <Tab>/<TabItem> children (depth 1 inside this block).
+        // We track nesting to skip tabs inside nested <Tabs> groups.
+        const inner = block.substring(block.indexOf('>') + 1, block.lastIndexOf('</'));
+        const panels = [];
+        TAB_OPEN_RE.lastIndex = 0;
+        let tabMatch;
+        while ((tabMatch = TAB_OPEN_RE.exec(inner)) !== null) {
+            // Check for nested <Tabs> between our position and this <Tab>
+            const before = inner.substring(0, tabMatch.index);
+            const opensInBefore = (before.match(/<Tabs\b/gi) || []).length;
+            const closesInBefore = (before.match(/<\/Tabs>/gi) || []).length;
+            const depth = opensInBefore - closesInBefore;
+            if (depth > 0)
+                continue; // This <Tab> belongs to a nested <Tabs>
+            const attrs = tabMatch[2];
+            const labelMatch = attrs.match(ATTR_RE);
+            const label = labelMatch ? labelMatch[1] : null;
+            // Find the matching closing tag for this <Tab>
+            const closeTag = `</${tabMatch[1]}>`;
+            const closeIdx = inner.indexOf(closeTag, tabMatch.index + tabMatch[0].length);
+            const tabContent = closeIdx !== -1
+                ? inner.substring(tabMatch.index, closeIdx + closeTag.length)
+                : inner.substring(tabMatch.index);
+            panels.push({ label, html: tabContent });
+        }
+        if (panels.length === 0)
+            continue;
+        groups.push({
+            framework: 'mdx',
+            tabCount: panels.length,
+            htmlSlice: block,
+            panels,
+        });
+    }
+    return groups;
+};
+const frameworkDetectors = [docusaurus, mkdocs, sphinx, msLearn, mdxTabs];
+export function detectTabGroups(html) {
+    const root = parse(html);
+    const claimed = new Set();
+    const groups = [];
+    for (const detector of frameworkDetectors) {
+        for (const group of detector(root, claimed, html)) {
+            groups.push(group);
+        }
+    }
+    for (const group of genericAria(root, claimed)) {
+        groups.push(group);
+    }
+    return groups;
+}
+//# sourceMappingURL=detect-tabs.js.map

package/dist/helpers/detect-tabs.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"detect-tabs.js","sourceRoot":"","sources":["../../src/helpers/detect-tabs.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,EAAoB,MAAM,kBAAkB,CAAC;AAoB3D,SAAS,cAAc,CAAC,IAAiB,EAAE,QAAqB;IAC9D,IAAI,OAAO,GAAG,IAAI,CAAC,UAAU,CAAC;IAC9B,OAAO,OAAO,EAAE,CAAC;QACf,IAAI,OAAO,KAAK,QAAQ;YAAE,OAAO,IAAI,CAAC;QACtC,OAAO,GAAG,OAAO,CAAC,UAAU,CAAC;IAC/B,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,SAAS,eAAe,CAAC,IAAiB,EAAE,OAAyB;IACnE,KAAK,MAAM,SAAS,IAAI,OAAO,EAAE,CAAC;QAChC,IAAI,SAAS,KAAK,IAAI,IAAI,cAAc,CAAC,IAAI,EAAE,SAAS,CAAC;YAAE,OAAO,IAAI,CAAC;IACzE,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,SAAS,MAAM,CAAC,EAAe;IAC7B,wEAAwE;IACxE,4EAA4E;IAC5E,MAAM,KAAK,GAAG,EAAE,CAAC,KAAK,EAAiB,CAAC;IACxC,KAAK,MAAM,KAAK,IAAI,KAAK,CAAC,gBAAgB,CAAC,OAAO,CAAC,EAAE,CAAC;QACpD,KAAK,CAAC,MAAM,EAAE,CAAC;IACjB,CAAC;IACD,OAAO,KAAK,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC;AAClC,CAAC;AAED;;;;;;GAMG;AACH,SAAS,uBAAuB,CAC9B,KAAkB,EAClB,OAAyB,EACzB,QAAQ,GAAG,CAAC;IAEZ,IAAI,OAAO,GAAG,KAAK,CAAC,UAAgC,CAAC;IACrD,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,OAAO,IAAI,KAAK,GAAG,QAAQ,EAAE,KAAK,EAAE,EAAE,CAAC;QACzD,IAAI,eAAe,CAAC,OAAO,EAAE,OAAO,CAAC;YAAE,OAAO,IAAI,CAAC;QACnD,MAAM,MAAM,GAAG,OAAO,CAAC,gBAAgB,CAAC,mBAAmB,CAA6B,CAAC;QACzF,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,EAAE,SAAS,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,GAAG,MAAM,CAAC,EAAE,CAAC;QAC1E,OAAO,GAAG,OAAO,CAAC,UAAgC,CAAC;IACrD,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED,MAAM,UAAU,GAAa,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE;IAC7C,MAAM,MAAM,GAAuB,EAAE,CAAC;IACtC,gEAAgE;IAChE,MAAM,QAAQ,GAAG,IAAI,CAAC,gBAAgB,CAAC,kBAAkB,CAAC,CAAC;IAC3D,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAC/B,IAAI,eAAe,CAAC,OAAsB,EAAE,OAAO,CAAC;YAAE,SAAS;QAC/D,MAAM,IAAI,GAAG,OAAO,CAAC,gBAAgB,CAAC,aAAa,CAAC,CAAC;QACrD,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QAEhC,sEAAsE;QACtE,wEAAwE;QACxE,gEAAgE;QAChE,MAAM,KAAK,GAAG,uBAAuB,CAAC,OAAsB,EAAE,OAAO,CAAC,CAAC;QACvE,IAAI,CAAC,KAAK;YAAE,SAAS;QACrB,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,GAAG,KAAK,CAAC;QAEpC,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,CAAgB,CAAC,CAAC,CAAC;QACzD,MAAM,SAAS,GAAe,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;YAClD,KAAK,EAAE,MAAM,CAAC,CAAC,CAAC,IAAI,IAAI;YACxB,IAAI,EAAE,CAAC,CAAC,SAAS;SAClB,CAAC,CAAC,CAAC;QAEJ,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACvB,MAAM,CAAC,IAAI,CAAC;YACV,SAAS,EAAE,YAAY;YACvB,QAAQ,EAAE,IAAI,CAAC,MAAM;YACrB,SAAS,EAAE,SAAS,CAAC,SAAS;YAC9B,MAAM,EAAE,SAAS;SAClB,CAAC,CAAC;IACL,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC,CAAC;AAEF,MAAM,MAAM,GAAa,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE;IACzC,MAAM,MAAM,GAAuB,EAAE,CAAC;IACtC,MAAM,UAAU,GAAG,IAAI,CAAC,gBAAgB,CAAC,aAAa,CAAC,CAAC;IACxD,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;QACnC,MAAM,EAAE,GAAG,SAAwB,CAAC;QACpC,IAAI,eAAe,CAAC,EAAE,EAAE,OAAO,CAAC;YAAE,SAAS;QAE3C,MAAM,MAAM,GAAG,EAAE,CAAC,gBAAgB,CAAC,0CAA0C,CAAC,CAAC;QAC/E,MAAM,MAAM,GAAG,EAAE,CAAC,gBAAgB,CAAC,sDAAsD,CAAC,CAAC;QAE3F,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QAEzD,MAAM,SAAS,GAAe,EAAE,CAAC;QACjC,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC,CAAC;QACrD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE,CAAC;YAC/B,SAAS,CAAC,IAAI,CAAC;gBACb,KAAK,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAgB,CAAC,CAAC,CAAC,CAAC,IAAI;gBAC1D,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAE,MAAM,CAAC,CAAC,CAAiB,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE;aAC5D,CAAC,CAAC;QACL,CAAC;QAED,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAChB,MAAM,CAAC,IAAI,CAAC;YACV,SAAS,EAAE,QAAQ;YACnB,QAAQ,EAAE,KAAK;YACf,SAAS,EAAE,EAAE,CAAC,SAAS;YACvB,MAAM,EAAE,SAAS;SAClB,CAAC,CAAC;IACL,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC,CAAC;AAEF,MAAM,MAAM,GAAa,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE;IACzC,MAAM,MAAM,GAAuB,EAAE,CAAC;IACtC,MAAM,UAAU,GAAG,IAAI,CAAC,gBAAgB,CAAC,cAAc,CAAC,CAAC;IACzD,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;QACnC,MAAM,EAAE,GAAG,SAAwB,CAAC;QACpC,IAAI,eAAe,CAAC,EAAE,EAAE,OAAO,CAAC;YAAE,SAAS;QAE3C,MAAM,IAAI,GAAG,EAAE,CAAC,gBAAgB,CAAC,kBAAkB,CAAC,CAAC;QACrD,MAAM,MAAM,GAAG,EAAE,CAAC,gBAAgB,CAAC,oBAAoB,CAAC,CAAC;QACzD,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QAEvD,MAAM,SAAS,GAAe,EAAE,CAAC;QACjC,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC,CAAC;QACnD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE,CAAC;YAC/B,SAAS,CAAC,IAAI,CAAC;gBACb,KAAK,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAgB,CAAC,CAAC,CAAC,CAAC,IAAI;gBACtD,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAE,MAAM,CAAC,CAAC,CAAiB,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE;aAC5D,CAAC,CAAC;QACL,CAAC;QAED,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAChB,MAAM,CAAC,IAAI,CAAC;YACV,SAAS,EAAE,QAAQ;YACnB,QAAQ,EAAE,KAAK;YACf,SAAS,EAAE,EAAE,CAAC,SAAS;YACvB,MAAM,EAAE,SAAS;SAClB,CAAC,CAAC;IACL,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC,CAAC;AAEF,MAAM,OAAO,GAAa,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE;IAC1C,MAAM,MAAM,GAAuB,EAAE,CAAC;IACtC,MAAM,UAAU,GAAG,IAAI,CAAC,gBAAgB,CAAC,WAAW,CAAC,CAAC;IACtD,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;QACnC,MAAM,EAAE,GAAG,SAAwB,CAAC;QACpC,IAAI,eAAe,CAAC,EAAE,EAAE,OAAO,CAAC;YAAE,SAAS;QAE3C,MAAM,IAAI,GAAG,EAAE,CAAC,gBAAgB,CAAC,wBAAwB,CAAC,CAAC;QAC3D,MAAM,MAAM,GAAG,EAAE,CAAC,gBAAgB,CAAC,8BAA8B,CAAC,CAAC;QACnE,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QAEvD,MAAM,SAAS,GAAe,EAAE,CAAC;QACjC,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC,CAAC;QACnD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE,CAAC;YAC/B,SAAS,CAAC,IAAI,CAAC;gBACb,KAAK,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAgB,CAAC,CAAC,CAAC,CAAC,IAAI;gBACtD,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAE,MAAM,CAAC,CAAC,CAAiB,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE;aAC5D,CAAC,CAAC;QACL,CAAC;QAED,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAChB,MAAM,CAAC,IAAI,CAAC;YACV,SAAS,EAAE,iBAAiB;YAC5B,QAAQ,EAAE,KAAK;YACf,SAAS,EAAE,EAAE,CAAC,SAAS;YACvB,MAAM,EAAE,SAAS;SAClB,CAAC,CAAC;IACL,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC,CAAC;AAEF,MAAM,WAAW,GAAa,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE;IAC9C,MAAM,MAAM,GAAuB,EAAE,CAAC;IACtC,MAAM,QAAQ,GAAG,IAAI,CAAC,gBAAgB,CAAC,kBAAkB,CAAC,CAAC;IAC3D,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAC/B,MAAM,EAAE,GAAG,OAAsB,CAAC;QAClC,IAAI,eAAe,CAAC,EAAE,EAAE,OAAO,CAAC;YAAE,SAAS;QAE3C,MAAM,KAAK,GAAG,uBAAuB,CAAC,EAAE,EAAE,OAAO,CAAC,CAAC;QACnD,MAAM,IAAI,GAAG,EAAE,CAAC,gBAAgB,CAAC,cAAc,CAAC,CAAC;QAEjD,IAAI,CAAC,KAAK,EAAE,CAAC;YACX,iEAAiE;YACjE,mEAAmE;YACnE,iEAAiE;YACjE,SAAS;QACX,CAAC;QAED,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,GAAG,KAAK,CAAC;QACpC,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QAEvD,kEAAkE;QAClE,oEAAoE;QACpE,mDAAmD;QACnD,MAAM,SAAS,GAAe,EAAE,CAAC;QACjC,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC;QAC5D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE,CAAC;YAC/B,SAAS,CAAC,IAAI,CAAC;gBACb,KAAK,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAgB,CAAC,CAAC,CAAC,CAAC,IAAI;gBACtD,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAE,MAAM,CAAC,CAAC,CAAiB,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE;aAC5D,CAAC,CAAC;QACL,CAAC;QAED,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACvB,MAAM,CAAC,IAAI,CAAC;YACV,SAAS,EAAE,cAAc;YACzB,QAAQ,EAAE,KAAK;YACf,SAAS,EAAE,SAAS,CAAC,SAAS;YAC9B,MAAM,EAAE,SAAS;SAClB,CAAC,CAAC;IACL,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC,CAAC;AAEF;;;;;;GAMG;AACH,SAAS,cAAc,CAAC,MAAc;IACpC,MAAM,MAAM,GAAa,EAAE,CAAC;IAC5B,MAAM,MAAM,GAAG,iBAAiB,CAAC;IACjC,IAAI,KAA6B,CAAC;IAClC,OAAO,CAAC,KAAK,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;QAC9C,MAAM,QAAQ,GAAG,KAAK,CAAC,KAAK,CAAC;QAC7B,6CAA6C;QAC7C,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,IAAI,GAAG,GAAG,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC;QACrC,OAAO,GAAG,GAAG,MAAM,CAAC,MAAM,IAAI,KAAK,GAAG,CAAC,EAAE,CAAC;YACxC,MAAM,QAAQ,GAAG,MAAM,CAAC,OAAO,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC;YAC9C,MAAM,SAAS,GAAG,MAAM,CAAC,OAAO,CAAC,SAAS,EAAE,GAAG,CAAC,CAAC;YACjD,IAAI,SAAS,KAAK,CAAC,CAAC;gBAAE,MAAM,CAAC,eAAe;YAC5C,IAAI,QAAQ,KAAK,CAAC,CAAC,IAAI,QAAQ,GAAG,SAAS,EAAE,CAAC;gBAC5C,KAAK,EAAE,CAAC;gBACR,GAAG,GAAG,QAAQ,GAAG,CAAC,CAAC;YACrB,CAAC;iBAAM,CAAC;gBACN,KAAK,EAAE,CAAC;gBACR,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;oBAChB,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,QAAQ,EAAE,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC;oBACvD,+DAA+D;oBAC/D,MAAM,CAAC,SAAS,GAAG,SAAS,GAAG,CAAC,CAAC;gBACnC,CAAC;gBACD,GAAG,GAAG,SAAS,GAAG,CAAC,CAAC;YACtB,CAAC;QACH,CAAC;IACH,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,MAAM,WAAW,GAAG,4BAA4B,CAAC;AACjD,MAAM,OAAO,GAAG,uCAAuC,CAAC;AAExD;;;;;;;GAOG;AACH,MAAM,OAAO,GAAa,CAAC,KAAK,EAAE,QAAQ,EAAE,MAAe,EAAE,EAAE;IAC7D,IAAI,CAAC,MAAM;QAAE,OAAO,EAAE,CAAC;IACvB,MAAM,MAAM,GAAuB,EAAE,CAAC;IACtC,MAAM,MAAM,GAAG,cAAc,CAAC,MAAM,CAAC,CAAC;IAEtC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,oEAAoE;QACpE,6DAA6D;QAC7D,MAAM,KAAK,GAAG,KAAK,CAAC,SAAS,CAAC,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,KAAK,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC;QAC/E,MAAM,MAAM,GAAe,EAAE,CAAC;QAE9B,WAAW,CAAC,SAAS,GAAG,CAAC,CAAC;QAC1B,IAAI,QAAgC,CAAC;QACrC,OAAO,CAAC,QAAQ,GAAG,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;YACrD,8DAA8D;YAC9D,MAAM,MAAM,GAAG,KAAK,CAAC,SAAS,CAAC,CAAC,EAAE,QAAQ,CAAC,KAAK,CAAC,CAAC;YAClD,MAAM,aAAa,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC;YAC/D,MAAM,cAAc,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC;YACjE,MAAM,KAAK,GAAG,aAAa,GAAG,cAAc,CAAC;YAE7C,IAAI,KAAK,GAAG,CAAC;gBAAE,SAAS,CAAC,wCAAwC;YAEjE,MAAM,KAAK,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;YAC1B,MAAM,UAAU,GAAG,KAAK,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;YACxC,MAAM,KAAK,GAAG,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;YAEhD,+CAA+C;YAC/C,MAAM,QAAQ,GAAG,KAAK,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC;YACrC,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,QAAQ,EAAE,QAAQ,CAAC,KAAK,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC;YAC9E,MAAM,UAAU,GACd,QAAQ,KAAK,CAAC,CAAC;gBACb,CAAC,CAAC,KAAK,CAAC,SAAS,CAAC,QAAQ,CAAC,KAAK,EAAE,QAAQ,GAAG,QAAQ,CAAC,MAAM,CAAC;gBAC7D,CAAC,CAAC,KAAK,CAAC,SAAS,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;YAEtC,MAAM,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,IAAI,EAAE,UAAU,EAAE,CAAC,CAAC;QAC3C,CAAC;QAED,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QAElC,MAAM,CAAC,IAAI,CAAC;YACV,SAAS,EAAE,KAAK;YAChB,QAAQ,EAAE,MAAM,CAAC,MAAM;YACvB,SAAS,EAAE,KAAK;YAChB,MAAM;SACP,CAAC,CAAC;IACL,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC,CAAC;AAEF,MAAM,kBAAkB,GAAe,CAAC,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,OAAO,CAAC,CAAC;AAEtF,MAAM,UAAU,eAAe,CAAC,IAAY;IAC1C,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,CAAC;IACzB,MAAM,OAAO,GAAG,IAAI,GAAG,EAAe,CAAC;IACvC,MAAM,MAAM,GAAuB,EAAE,CAAC;IAEtC,KAAK,MAAM,QAAQ,IAAI,kBAAkB,EAAE,CAAC;QAC1C,KAAK,MAAM,KAAK,IAAI,QAAQ,CAAC,IAAI,EAAE,OAAO,EAAE,IAAI,CAAC,EAAE,CAAC;YAClD,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QACrB,CAAC;IACH,CAAC;IAED,KAAK,MAAM,KAAK,IAAI,WAAW,CAAC,IAAI,EAAE,OAAO,CAAC,EAAE,CAAC;QAC/C,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IACrB,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/dist/helpers/fetch-page.d.ts

@@ -0,0 +1,8 @@
+import type { CheckContext, FetchedPage } from '../types.js';
+/**
+ * Fetch a page URL, returning the body and content-type metadata.
+ * Results are cached on `ctx.htmlCache` so that multiple checks
+ * within the same run avoid redundant HTTP requests.
+ */
+export declare function fetchPage(ctx: CheckContext, url: string): Promise<FetchedPage>;
+//# sourceMappingURL=fetch-page.d.ts.map

package/dist/helpers/fetch-page.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"fetch-page.d.ts","sourceRoot":"","sources":["../../src/helpers/fetch-page.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,YAAY,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAE7D;;;;GAIG;AACH,wBAAsB,SAAS,CAAC,GAAG,EAAE,YAAY,EAAE,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC,WAAW,CAAC,CAcpF"}

package/dist/helpers/fetch-page.js

@@ -0,0 +1,20 @@
+import { looksLikeHtml } from './detect-markdown.js';
+/**
+ * Fetch a page URL, returning the body and content-type metadata.
+ * Results are cached on `ctx.htmlCache` so that multiple checks
+ * within the same run avoid redundant HTTP requests.
+ */
+export async function fetchPage(ctx, url) {
+    const cached = ctx.htmlCache.get(url);
+    if (cached)
+        return cached;
+    const response = await ctx.http.fetch(url);
+    const body = await response.text();
+    const contentType = response.headers.get('content-type') ?? '';
+    const isMarkdownType = contentType.includes('text/markdown') || contentType.includes('text/plain');
+    const isHtml = !isMarkdownType && (contentType.includes('text/html') || looksLikeHtml(body));
+    const result = { url, status: response.status, body, contentType, isHtml };
+    ctx.htmlCache.set(url, result);
+    return result;
+}
+//# sourceMappingURL=fetch-page.js.map

package/dist/helpers/fetch-page.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"fetch-page.js","sourceRoot":"","sources":["../../src/helpers/fetch-page.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAC;AAGrD;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,GAAiB,EAAE,GAAW;IAC5D,MAAM,MAAM,GAAG,GAAG,CAAC,SAAS,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;IACtC,IAAI,MAAM;QAAE,OAAO,MAAM,CAAC;IAE1B,MAAM,QAAQ,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IAC3C,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAC,IAAI,EAAE,CAAC;IACnC,MAAM,WAAW,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,cAAc,CAAC,IAAI,EAAE,CAAC;IAC/D,MAAM,cAAc,GAClB,WAAW,CAAC,QAAQ,CAAC,eAAe,CAAC,IAAI,WAAW,CAAC,QAAQ,CAAC,YAAY,CAAC,CAAC;IAC9E,MAAM,MAAM,GAAG,CAAC,cAAc,IAAI,CAAC,WAAW,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC;IAE7F,MAAM,MAAM,GAAgB,EAAE,GAAG,EAAE,MAAM,EAAE,QAAQ,CAAC,MAAM,EAAE,IAAI,EAAE,WAAW,EAAE,MAAM,EAAE,CAAC;IACxF,GAAG,CAAC,SAAS,CAAC,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC;IAC/B,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/dist/helpers/get-page-urls.d.ts

@@ -7,6 +7,7 @@ export declare function parseSitemapUrls(xml: string): {
     urls: string[];
     sitemapIndexUrls: string[];
 };
+export declare function getUrlsFromCachedLlmsTxt(ctx: CheckContext): Promise<string[]>;
 /**
  * Parse `Sitemap:` directives from a robots.txt body.
  * Returns an array of sitemap URLs found.
@@ -16,6 +17,7 @@ export interface PageUrlResult {
     urls: string[];
     warnings: string[];
 }
+export declare function getUrlsFromSitemap(ctx: CheckContext, warnings: string[], maxUrls?: number, originOverride?: string): Promise<string[]>;
 /**
  * Discover page URLs from llms.txt links, sitemap, or fall back to baseUrl.
  *
@@ -37,6 +39,13 @@ export interface SampledPages {
  *
  * The result is cached on ctx so that all checks within a single run
  * share the same sampled page list, avoiding inconsistent results.
+ *
+ * Sampling strategies:
+ * - `random`: Fisher-Yates shuffle, then take the first maxLinksToTest. (Default.)
+ * - `deterministic`: Sort URLs lexicographically, then pick every Nth URL
+ *   so that the result is reproducible across runs (as long as the discovered
+ *   URL set is stable).
+ * - `none`: Skip discovery entirely; return only the baseUrl.
  */
 export declare function discoverAndSamplePages(ctx: CheckContext): Promise<SampledPages>;
 //# sourceMappingURL=get-page-urls.d.ts.map

package/dist/helpers/get-page-urls.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"get-page-urls.d.ts","sourceRoot":"","sources":["../../src/helpers/get-page-urls.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,YAAY,EAAkB,MAAM,aAAa,CAAC;AAEhE;;;GAGG;AACH,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,MAAM,GAAG;IAAE,IAAI,EAAE,MAAM,EAAE,CAAC;IAAC,gBAAgB,EAAE,MAAM,EAAE,CAAA;CAAE,CA2B5F;AAyDD;;;GAGG;AACH,wBAAgB,sBAAsB,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,EAAE,CASlE;AAsBD,MAAM,WAAW,aAAa;IAC5B,IAAI,EAAE,MAAM,EAAE,CAAC;IACf,QAAQ,EAAE,MAAM,EAAE,CAAC;CACpB;AAyED;;;;;;;GAOG;AACH,wBAAsB,WAAW,CAAC,GAAG,EAAE,YAAY,GAAG,OAAO,CAAC,aAAa,CAAC,CAmB3E;AAED,MAAM,WAAW,YAAY;IAC3B,IAAI,EAAE,MAAM,EAAE,CAAC;IACf,UAAU,EAAE,MAAM,CAAC;IACnB,OAAO,EAAE,OAAO,CAAC;IACjB,QAAQ,EAAE,MAAM,EAAE,CAAC;CACpB;AAED;;;;;;GAMG;AACH,wBAAsB,sBAAsB,CAAC,GAAG,EAAE,YAAY,GAAG,OAAO,CAAC,YAAY,CAAC,CAkBrF"}
+{"version":3,"file":"get-page-urls.d.ts","sourceRoot":"","sources":["../../src/helpers/get-page-urls.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAE,YAAY,EAAkB,MAAM,aAAa,CAAC;AAEhE;;;GAGG;AACH,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,MAAM,GAAG;IAAE,IAAI,EAAE,MAAM,EAAE,CAAC;IAAC,gBAAgB,EAAE,MAAM,EAAE,CAAA;CAAE,CA2B5F;AAED,wBAAsB,wBAAwB,CAAC,GAAG,EAAE,YAAY,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,CAMnF;AAiID;;;GAGG;AACH,wBAAgB,sBAAsB,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,EAAE,CASlE;AAoCD,MAAM,WAAW,aAAa;IAC5B,IAAI,EAAE,MAAM,EAAE,CAAC;IACf,QAAQ,EAAE,MAAM,EAAE,CAAC;CACpB;AA0BD,wBAAsB,kBAAkB,CACtC,GAAG,EAAE,YAAY,EACjB,QAAQ,EAAE,MAAM,EAAE,EAClB,OAAO,GAAE,MAAyB,EAClC,cAAc,CAAC,EAAE,MAAM,GACtB,OAAO,CAAC,MAAM,EAAE,CAAC,CA8CnB;AAED;;;;;;;GAOG;AACH,wBAAsB,WAAW,CAAC,GAAG,EAAE,YAAY,GAAG,OAAO,CAAC,aAAa,CAAC,CAmB3E;AAED,MAAM,WAAW,YAAY;IAC3B,IAAI,EAAE,MAAM,EAAE,CAAC;IACf,UAAU,EAAE,MAAM,CAAC;IACnB,OAAO,EAAE,OAAO,CAAC;IACjB,QAAQ,EAAE,MAAM,EAAE,CAAC;CACpB;AAED;;;;;;;;;;;;;GAaG;AACH,wBAAsB,sBAAsB,CAAC,GAAG,EAAE,YAAY,GAAG,OAAO,CAAC,YAAY,CAAC,CA2CrF"}