@j0hanz/fetch-url-mcp 1.10.18 → 1.10.20

package/dist/lib/dom-prep.d.ts

@@ -5,6 +5,7 @@ export declare function resolveDocumentBody(document: Document): Element;
 export declare function serializeDocumentForMarkdown(document: Document, fallback: string): string;
 /** Surface hidden tab panels, then strip unselected tab triggers. */
 export declare function normalizeTabContent(document: Document): void;
+export declare function surfaceCodeEditorContent(document: Document): void;
 export declare function prepareDocumentForMarkdown(document: Document, baseUrl?: string, signal?: AbortSignal): void;
 export declare function removeNoiseFromHtml(html: string, document?: Document, baseUrl?: string, signal?: AbortSignal): string;
 export declare function getVisibleTextLength(htmlOrDocument: string | Document): number;

package/dist/lib/dom-prep.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"dom-prep.d.ts","sourceRoot":"","sources":["../../src/lib/dom-prep.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AA8mB9D,eAAO,MAAM,sBAAsB,QAAmB,CAAC;AAyCvD,wBAAgB,qBAAqB,CAAC,QAAQ,EAAE,QAAQ,GAAG,IAAI,CA6B9D;AAuBD,wBAAgB,mBAAmB,CAAC,QAAQ,EAAE,QAAQ,GAAG,OAAO,CAY/D;AAED,wBAAgB,4BAA4B,CAC1C,QAAQ,EAAE,QAAQ,EAClB,QAAQ,EAAE,MAAM,GACf,MAAM,CASR;AA0CD,qEAAqE;AACrE,wBAAgB,mBAAmB,CAAC,QAAQ,EAAE,QAAQ,GAAG,IAAI,CAG5D;AAmUD,wBAAgB,0BAA0B,CACxC,QAAQ,EAAE,QAAQ,EAClB,OAAO,CAAC,EAAE,MAAM,EAChB,MAAM,CAAC,EAAE,WAAW,GACnB,IAAI,CAON;AA4BD,wBAAgB,mBAAmB,CACjC,IAAI,EAAE,MAAM,EACZ,QAAQ,CAAC,EAAE,QAAQ,EACnB,OAAO,CAAC,EAAE,MAAM,EAChB,MAAM,CAAC,EAAE,WAAW,GACnB,MAAM,CAcR;AA0ED,wBAAgB,oBAAoB,CAClC,cAAc,EAAE,MAAM,GAAG,QAAQ,GAChC,MAAM,CAaR;AA6ID,wBAAgB,sBAAsB,CACpC,OAAO,EAAE,gBAAgB,EACzB,QAAQ,EAAE,QAAQ,GACjB,QAAQ,GAAG,IAAI,CAsBjB"}
+{"version":3,"file":"dom-prep.d.ts","sourceRoot":"","sources":["../../src/lib/dom-prep.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AAonB9D,eAAO,MAAM,sBAAsB,QAAmB,CAAC;AAyCvD,wBAAgB,qBAAqB,CAAC,QAAQ,EAAE,QAAQ,GAAG,IAAI,CA6B9D;AAuBD,wBAAgB,mBAAmB,CAAC,QAAQ,EAAE,QAAQ,GAAG,OAAO,CAY/D;AAED,wBAAgB,4BAA4B,CAC1C,QAAQ,EAAE,QAAQ,EAClB,QAAQ,EAAE,MAAM,GACf,MAAM,CASR;AA0CD,qEAAqE;AACrE,wBAAgB,mBAAmB,CAAC,QAAQ,EAAE,QAAQ,GAAG,IAAI,CAG5D;AA0RD,wBAAgB,wBAAwB,CAAC,QAAQ,EAAE,QAAQ,GAAG,IAAI,CA2BjE;AAyDD,wBAAgB,0BAA0B,CACxC,QAAQ,EAAE,QAAQ,EAClB,OAAO,CAAC,EAAE,MAAM,EAChB,MAAM,CAAC,EAAE,WAAW,GACnB,IAAI,CAON;AA4BD,wBAAgB,mBAAmB,CACjC,IAAI,EAAE,MAAM,EACZ,QAAQ,CAAC,EAAE,QAAQ,EACnB,OAAO,CAAC,EAAE,MAAM,EAChB,MAAM,CAAC,EAAE,WAAW,GACnB,MAAM,CAcR;AA0ED,wBAAgB,oBAAoB,CAClC,cAAc,EAAE,MAAM,GAAG,QAAQ,GAChC,MAAM,CAaR;AA6ID,wBAAgB,sBAAsB,CACpC,OAAO,EAAE,gBAAgB,EACzB,QAAQ,EAAE,QAAQ,GACjB,QAAQ,GAAG,IAAI,CAsBjB"}

package/dist/lib/dom-prep.js

@@ -271,6 +271,15 @@ function isNavigationAside(element) {
         return true;
     return links.length / (textLen / 100) >= ASIDE_NAV_LINK_DENSITY_THRESHOLD;
 }
+function isNavigationSidebar(element) {
+    const links = element.querySelectorAll('a[href]');
+    if (links.length < ASIDE_NAV_MIN_LINKS)
+        return false;
+    const textLen = (element.textContent || '').trim().length;
+    if (textLen === 0)
+        return true;
+    return links.length / (textLen / 100) >= ASIDE_NAV_LINK_DENSITY_THRESHOLD;
+}
 function shouldPreserve(element, tagName) {
     // Check Dialog
     const role = element.getAttribute('role');
@@ -282,12 +291,15 @@ function shouldPreserve(element, tagName) {
             return true;
         return element.querySelector('h1,h2,h3,h4,h5,h6') !== null;
     }
-    // Check Nav/Footer
     if (tagName === 'nav' || tagName === 'footer') {
         if (element.querySelector('article,main,section,[role="main"]'))
             return true;
-        return ((element.textContent || '').trim().length >=
-            NAV_FOOTER_MIN_CHARS_FOR_PRESERVATION);
+        const textLen = (element.textContent || '').trim().length;
+        if (textLen < NAV_FOOTER_MIN_CHARS_FOR_PRESERVATION)
+            return false;
+        if (isNavigationSidebar(element))
+            return false;
+        return true;
     }
     // Check Aside — preserve only if it looks like article content, not navigation
     if (tagName === 'aside') {
@@ -885,6 +897,35 @@ function separateAdjacentInlineElements(document) {
         }
     }
 }
+const CODE_EDITOR_LANG_REGEX = /\blanguage-(\S+)/;
+// Some documentation sites render code examples as highlighted, aria-hidden blocks with a textarea containing the raw code for accessibility.
+// Surface the textarea content and remove the redundant highlighted block to produce cleaner markdown output.
+export function surfaceCodeEditorContent(document) {
+    for (const pre of document.querySelectorAll('pre[aria-hidden="true"]')) {
+        const codeChild = pre.querySelector('code');
+        if (!codeChild)
+            continue;
+        const container = pre.parentElement;
+        if (!container)
+            continue;
+        const textarea = container.querySelector('textarea');
+        if (!textarea)
+            continue;
+        // Extract language from the highlighted code element
+        const langMatch = CODE_EDITOR_LANG_REGEX.exec(codeChild.getAttribute('class') ?? '');
+        const lang = langMatch?.[1] ?? '';
+        // Build a clean pre>code block from the textarea plain text
+        const newPre = document.createElement('pre');
+        const newCode = document.createElement('code');
+        if (lang)
+            newCode.setAttribute('class', `language-${lang}`);
+        newCode.textContent = textarea.textContent || '';
+        newPre.appendChild(newCode);
+        container.insertBefore(newPre, pre);
+        pre.remove();
+        textarea.remove();
+    }
+}
 function stripDocsControls(document) {
     removeNodes(document.querySelectorAll(DOCS_CONTROL_SELECTORS.join(',')));
 }
@@ -898,6 +939,7 @@ function stripAriaLiveInstructions(document) {
 }
 function runDocsControlPass(document) {
     normalizeTabContent(document);
+    surfaceCodeEditorContent(document);
     cleanHeadings(document);
     stripDocsControls(document);
     stripAriaLiveInstructions(document);
@@ -910,7 +952,15 @@ function runStructuralNoisePass(document, signal) {
 function runCodeExamplePass(document) {
     cleanCodeExamples(document);
 }
+function unwrapOrphanedTableCells(document) {
+    for (const cell of document.querySelectorAll('td, th')) {
+        if (!cell.closest('table')) {
+            cell.replaceWith(...Array.from(cell.childNodes));
+        }
+    }
+}
 function runTableNormalizationPass(document) {
+    unwrapOrphanedTableCells(document);
     normalizeTableCells(document);
     normalizeTableStructure(document);
 }

package/dist/transform/html-translators.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"html-translators.d.ts","sourceRoot":"","sources":["../../src/transform/html-translators.ts"],"names":[],"mappings":"AAyhBA,wBAAgB,+BAA+B,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAEpE;AAqPD,wBAAgB,4BAA4B,CAC1C,SAAS,EAAE,MAAM,GAChB,MAAM,GAAG,SAAS,CAuBpB;AAOD,wBAAgB,6BAA6B,CAC3C,SAAS,EAAE,MAAM,EACjB,QAAQ,EAAE,MAAM,GACf,MAAM,GAAG,SAAS,CAKpB;AACD,wBAAgB,sBAAsB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAKvE"}
+{"version":3,"file":"html-translators.d.ts","sourceRoot":"","sources":["../../src/transform/html-translators.ts"],"names":[],"mappings":"AAohBA,wBAAgB,+BAA+B,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAEpE;AAqPD,wBAAgB,4BAA4B,CAC1C,SAAS,EAAE,MAAM,GAChB,MAAM,GAAG,SAAS,CAuBpB;AAOD,wBAAgB,6BAA6B,CAC3C,SAAS,EAAE,MAAM,EACjB,QAAQ,EAAE,MAAM,GACf,MAAM,GAAG,SAAS,CAKpB;AACD,wBAAgB,sBAAsB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAKvE"}

package/dist/transform/html-translators.js

@@ -364,26 +364,15 @@ function buildDdTranslator() {
         postprocess: ({ content }) => content.trim() ? `: ${content.trim()}\n` : '',
     };
 }
-function buildKbdTranslator() {
+function wrapTranslator(prefix, suffix) {
     return {
-        postprocess: ({ content }) => `\`${content}\``,
-    };
-}
-function buildMarkTranslator() {
-    return {
-        postprocess: ({ content }) => `==${content}==`,
-    };
-}
-function buildSubTranslator() {
-    return {
-        postprocess: ({ content }) => `~${content}~`,
-    };
-}
-function buildSupTranslator() {
-    return {
-        postprocess: ({ content }) => `^${content}^`,
+        postprocess: ({ content }) => `${prefix}${content}${suffix}`,
     };
 }
+const buildKbdTranslator = () => wrapTranslator('`', '`');
+const buildMarkTranslator = () => wrapTranslator('==', '==');
+const buildSubTranslator = () => wrapTranslator('~', '~');
+const buildSupTranslator = () => wrapTranslator('^', '^');
 function buildDetailsTranslator() {
     return {
         postprocess: ({ content }) => {

package/dist/transform/markdown-cleanup.d.ts

@@ -4,6 +4,7 @@ interface CleanupOptions {
     url?: string;
 }
 export declare function processFencedContent(content: string, processTextSegment: (text: string) => string): string;
+export declare function finalizeMarkdownSections(content: string, options?: Pick<CleanupOptions, 'signal' | 'url'>): string;
 export declare function cleanupMarkdownArtifacts(content: string, options?: CleanupOptions): string;
 export {};
 //# sourceMappingURL=markdown-cleanup.d.ts.map

package/dist/transform/markdown-cleanup.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"markdown-cleanup.d.ts","sourceRoot":"","sources":["../../src/transform/markdown-cleanup.ts"],"names":[],"mappings":"AAqHA,UAAU,cAAc;IACtB,qBAAqB,CAAC,EAAE,OAAO,CAAC;IAChC,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB,GAAG,CAAC,EAAE,MAAM,CAAC;CACd;AAofD,wBAAgB,oBAAoB,CAClC,OAAO,EAAE,MAAM,EACf,kBAAkB,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,MAAM,GAC3C,MAAM,CAyBR;AAaD,wBAAgB,wBAAwB,CACtC,OAAO,EAAE,MAAM,EACf,OAAO,CAAC,EAAE,cAAc,GACvB,MAAM,CAqBR"}
+{"version":3,"file":"markdown-cleanup.d.ts","sourceRoot":"","sources":["../../src/transform/markdown-cleanup.ts"],"names":[],"mappings":"AAqHA,UAAU,cAAc;IACtB,qBAAqB,CAAC,EAAE,OAAO,CAAC;IAChC,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB,GAAG,CAAC,EAAE,MAAM,CAAC;CACd;AAofD,wBAAgB,oBAAoB,CAClC,OAAO,EAAE,MAAM,EACf,kBAAkB,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,MAAM,GAC3C,MAAM,CAyBR;AAaD,wBAAgB,wBAAwB,CACtC,OAAO,EAAE,MAAM,EACf,OAAO,CAAC,EAAE,IAAI,CAAC,cAAc,EAAE,QAAQ,GAAG,KAAK,CAAC,GAC/C,MAAM,CAUR;AAED,wBAAgB,wBAAwB,CACtC,OAAO,EAAE,MAAM,EACf,OAAO,CAAC,EAAE,cAAc,GACvB,MAAM,CAkBR"}

package/dist/transform/markdown-cleanup.js

@@ -513,14 +513,22 @@ function stripLeadingBreadcrumbNoise(text) {
 function stripCopyButtonText(text) {
     return text.replace(/\[Copy\]\(#copy\)\s*/gi, '');
 }
+export function finalizeMarkdownSections(content, options) {
+    if (!content)
+        return '';
+    throwIfAborted(options?.signal, options?.url ?? '', 'markdown:cleanup:empty-headings');
+    return stripLeadingBreadcrumbNoise(stripLeadingDocsChrome(removeEmptyHeadingSections(content)));
+}
 export function cleanupMarkdownArtifacts(content, options) {
     if (!content)
         return '';
     throwIfAborted(options?.signal, options?.url ?? '', 'markdown:cleanup:begin');
     let result = stripCopyButtonText(processFencedContent(content, (text) => processTextBuffer(text.split('\n'), options)).trim());
     if (!options?.preserveEmptyHeadings) {
-        throwIfAborted(options?.signal, options?.url ?? '', 'markdown:cleanup:empty-headings');
-        result = removeEmptyHeadingSections(result);
+        result = finalizeMarkdownSections(result, options);
     }
-    return stripLeadingBreadcrumbNoise(stripLeadingDocsChrome(result));
+    else {
+        result = stripLeadingBreadcrumbNoise(stripLeadingDocsChrome(result));
+    }
+    return result;
 }

package/dist/transform/metadata.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"metadata.d.ts","sourceRoot":"","sources":["../../src/transform/metadata.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,EAAE,iBAAiB,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC;AAoCnE,wBAAgB,sBAAsB,CACpC,KAAK,EAAE,MAAM,EACb,OAAO,CAAC,EAAE,MAAM,GACf,MAAM,CAkBR;AAwID,wBAAgB,eAAe,CAC7B,QAAQ,EAAE,QAAQ,EAClB,OAAO,CAAC,EAAE,MAAM,GACf,iBAAiB,CAkBnB;AAED,wBAAgB,uBAAuB,CACrC,IAAI,EAAE,MAAM,EACZ,OAAO,CAAC,EAAE,MAAM,GACf,iBAAiB,GAAG,IAAI,CAY1B;AAED,wBAAgB,aAAa,CAC3B,KAAK,EAAE,iBAAiB,GAAG,IAAI,EAC/B,IAAI,EAAE,iBAAiB,GACtB,iBAAiB,CAmBnB;AA2GD,wBAAgB,2BAA2B,CACzC,OAAO,EAAE,MAAM,GACd,MAAM,GAAG,SAAS,CAOpB;AACD,wBAAgB,mBAAmB,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,MAAM,CAuCxE;AAmBD,wBAAgB,gBAAgB,CAAC,OAAO,EAAE,MAAM,GAAG,OAAO,CAczD;AAaD,wBAAgB,mBAAmB,CACjC,QAAQ,CAAC,EAAE,aAAa,EACxB,WAAW,CAAC,EAAE,MAAM,GACnB,MAAM,CAmBR"}
+{"version":3,"file":"metadata.d.ts","sourceRoot":"","sources":["../../src/transform/metadata.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,EAAE,iBAAiB,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC;AAoCnE,wBAAgB,sBAAsB,CACpC,KAAK,EAAE,MAAM,EACb,OAAO,CAAC,EAAE,MAAM,GACf,MAAM,CAkBR;AAgKD,wBAAgB,eAAe,CAC7B,QAAQ,EAAE,QAAQ,EAClB,OAAO,CAAC,EAAE,MAAM,GACf,iBAAiB,CAYnB;AAED,wBAAgB,uBAAuB,CACrC,IAAI,EAAE,MAAM,EACZ,OAAO,CAAC,EAAE,MAAM,GACf,iBAAiB,GAAG,IAAI,CAY1B;AAED,wBAAgB,aAAa,CAC3B,KAAK,EAAE,iBAAiB,GAAG,IAAI,EAC/B,IAAI,EAAE,iBAAiB,GACtB,iBAAiB,CAmBnB;AA2GD,wBAAgB,2BAA2B,CACzC,OAAO,EAAE,MAAM,GACd,MAAM,GAAG,SAAS,CAOpB;AACD,wBAAgB,mBAAmB,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,MAAM,CAuCxE;AAmBD,wBAAgB,gBAAgB,CAAC,OAAO,EAAE,MAAM,GAAG,OAAO,CAczD;AAaD,wBAAgB,mBAAmB,CACjC,QAAQ,CAAC,EAAE,aAAa,EACxB,WAAW,CAAC,EAAE,MAAM,GACnB,MAAM,CAmBR"}

package/dist/transform/metadata.js

@@ -136,6 +136,13 @@ function resolveMetadataFromContext(ctx) {
 // ---------------------------------------------------------------------------
 // Favicon resolution
 // ---------------------------------------------------------------------------
+/** Ordered by preference: exact 32×32, SVG, any generic icon, legacy shortcut. */
+const FAVICON_SELECTORS = [
+    'link[rel="icon"][sizes="32x32"]',
+    'link[rel="icon"][type="image/svg+xml"]',
+    'link[rel="icon"]',
+    'link[rel="shortcut icon"]',
+];
 function resolveFaviconUrl(href, baseUrl) {
     const trimmed = href.trim();
     if (!trimmed)
@@ -151,6 +158,19 @@ function resolveFaviconUrl(href, baseUrl) {
     }
     return resolved.toString();
 }
+function extractFavicon(document, baseUrl) {
+    for (const selector of FAVICON_SELECTORS) {
+        for (const el of document.querySelectorAll(selector)) {
+            const href = el.getAttribute('href');
+            if (href) {
+                const resolved = resolveFaviconUrl(href, baseUrl);
+                if (resolved)
+                    return resolved;
+            }
+        }
+    }
+    return undefined;
+}
 // ---------------------------------------------------------------------------
 // Public interface
 // ---------------------------------------------------------------------------
@@ -161,13 +181,9 @@ export function extractMetadata(document, baseUrl) {
         metadata.title = normalizeDocumentTitle(metadata.title, baseUrl);
     }
     if (baseUrl) {
-        const icon32 = document.querySelector('link[rel="icon"][sizes="32x32"]');
-        const href = icon32?.getAttribute('href');
-        if (href) {
-            const resolved = resolveFaviconUrl(href, baseUrl);
-            if (resolved)
-                metadata.favicon = resolved;
-        }
+        const favicon = extractFavicon(document, baseUrl);
+        if (favicon)
+            metadata.favicon = favicon;
     }
     return metadata;
 }

package/dist/transform/transform.d.ts

@@ -1,4 +1,4 @@
-import { cleanupMarkdownArtifacts, processFencedContent } from './markdown-cleanup.js';
+import { cleanupMarkdownArtifacts, finalizeMarkdownSections, processFencedContent } from './markdown-cleanup.js';
 import type { ExtractedArticle, ExtractedMetadata, ExtractionResult, MarkdownTransformResult, MetadataBlock, TransformOptions, TransformStageContext } from './types.js';
 interface StageBudget {
     totalBudgetMs: number;
@@ -34,5 +34,5 @@ type TransformExecutionOptions = TransformOptions & {
 };
 export declare function transformHtmlToMarkdown(html: string, url: string, options: TransformOptions): Promise<MarkdownTransformResult>;
 export declare function transformBufferToMarkdown(htmlBuffer: Uint8Array, url: string, options: TransformExecutionOptions): Promise<MarkdownTransformResult>;
-export { cleanupMarkdownArtifacts, processFencedContent };
+export { cleanupMarkdownArtifacts, finalizeMarkdownSections, processFencedContent, };
 //# sourceMappingURL=transform.d.ts.map

package/dist/transform/transform.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"transform.d.ts","sourceRoot":"","sources":["../../src/transform/transform.ts"],"names":[],"mappings":"AAwCA,OAAO,EACL,wBAAwB,EACxB,oBAAoB,EACrB,MAAM,uBAAuB,CAAC;AAqB/B,OAAO,KAAK,EACV,gBAAgB,EAChB,iBAAiB,EACjB,gBAAgB,EAChB,uBAAuB,EACvB,aAAa,EACb,gBAAgB,EAChB,qBAAqB,EAEtB,MAAM,YAAY,CAAC;AA+BpB,UAAU,WAAW;IACnB,aAAa,EAAE,MAAM,CAAC;IACtB,SAAS,EAAE,MAAM,CAAC;CACnB;AAgJD,wBAAgB,mBAAmB,CACjC,GAAG,EAAE,MAAM,EACX,KAAK,EAAE,MAAM,EACb,MAAM,CAAC,EAAE,WAAW,GACnB,qBAAqB,GAAG,IAAI,CAE9B;AAED,wBAAgB,iBAAiB,CAC/B,OAAO,EAAE,qBAAqB,GAAG,IAAI,EACrC,OAAO,CAAC,EAAE;IAAE,SAAS,CAAC,EAAE,OAAO,CAAA;CAAE,GAChC,MAAM,CAER;AA6XD,wBAAgB,cAAc,CAC5B,IAAI,EAAE,MAAM,EACZ,GAAG,EAAE,MAAM,EACX,OAAO,GAAE;IAAE,cAAc,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,WAAW,CAAA;CAExD,GACA,gBAAgB,CAGlB;AAyKD,wBAAgB,cAAc,CAC5B,IAAI,EAAE,MAAM,EACZ,QAAQ,CAAC,EAAE,aAAa,EACxB,OAAO,CAAC,EAAE;IACR,GAAG,CAAC,EAAE,MAAM,CAAC;IACb,MAAM,CAAC,EAAE,WAAW,GAAG,SAAS,CAAC;IACjC,QAAQ,CAAC,EAAE,QAAQ,GAAG,SAAS,CAAC;IAChC,gBAAgB,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;CACxC,GACA,MAAM,CAyBR;AA2DD,wBAAgB,sBAAsB,CACpC,OAAO,EAAE,gBAAgB,GAAG,IAAI,EAChC,sBAAsB,EAAE,MAAM,GAAG,QAAQ,GACxC,OAAO,CAQT;AAKD,wBAAgB,gCAAgC,CAC9C,OAAO,EAAE,gBAAgB,GAAG,IAAI,GAC/B,OAAO,IAAI,gBAAgB,CAE7B;AAED,wBAAgB,0BAA0B,CACxC,GAAG,EAAE,MAAM,EACX,OAAO,EAAE,gBAAgB,GAAG,IAAI,EAChC,aAAa,EAAE,iBAAiB,EAChC,wBAAwB,EAAE,OAAO,EACjC,eAAe,EAAE,OAAO,GACvB,aAAa,GAAG,SAAS,CAuB3B;AA4bD,wBAAgB,gCAAgC,CAC9C,IAAI,EAAE,MAAM,EACZ,GAAG,EAAE,MAAM,EACX,OAAO,EAAE,gBAAgB,GACxB,uBAAuB,CAgBzB;AAaD,UAAU,kBAAkB;IAC1B,UAAU,EAAE,MAAM,CAAC;IACnB,aAAa,EAAE,MAAM,CAAC;IACtB,QAAQ,EAAE,MAAM,CAAC;CAClB;AAED,wBAAgB,qBAAqB,IAAI,kBAAkB,GAAG,IAAI,CAEjE;AAED,wBAAsB,2BAA2B,IAAI,OAAO,CAAC,IAAI,CAAC,CAEjE;AAED,KAAK,yBAAyB,GAAG,gBAAgB,GAAG;IAAE,QAAQ,CAAC,EAAE,MAAM,CAAA;CAAE,CAAC;AAgH1E,wBAAsB,uBAAuB,CAC3C,IAAI,EAAE,MAAM,EACZ,GAAG,EAAE,MAAM,EACX,OAAO,EAAE,gBAAgB,GACxB,OAAO,CAAC,uBAAuB,CAAC,CAElC;AAED,wBAAsB,yBAAyB,CAC7C,UAAU,EAAE,UAAU,EACtB,GAAG,EAAE,MAAM,EACX,OAAO,EAAE,yBAAyB,GACjC,OAAO,CAAC,uBAAuB,CAAC,CAElC;AAED,OAAO,EAAE,wBAAwB,EAAE,oBAAoB,EAAE,CAAC"}
+{"version":3,"file":"transform.d.ts","sourceRoot":"","sources":["../../src/transform/transform.ts"],"names":[],"mappings":"AAyCA,OAAO,EACL,wBAAwB,EACxB,wBAAwB,EACxB,oBAAoB,EACrB,MAAM,uBAAuB,CAAC;AAqB/B,OAAO,KAAK,EACV,gBAAgB,EAChB,iBAAiB,EACjB,gBAAgB,EAChB,uBAAuB,EACvB,aAAa,EACb,gBAAgB,EAChB,qBAAqB,EAEtB,MAAM,YAAY,CAAC;AA+BpB,UAAU,WAAW;IACnB,aAAa,EAAE,MAAM,CAAC;IACtB,SAAS,EAAE,MAAM,CAAC;CACnB;AAgJD,wBAAgB,mBAAmB,CACjC,GAAG,EAAE,MAAM,EACX,KAAK,EAAE,MAAM,EACb,MAAM,CAAC,EAAE,WAAW,GACnB,qBAAqB,GAAG,IAAI,CAE9B;AAED,wBAAgB,iBAAiB,CAC/B,OAAO,EAAE,qBAAqB,GAAG,IAAI,EACrC,OAAO,CAAC,EAAE;IAAE,SAAS,CAAC,EAAE,OAAO,CAAA;CAAE,GAChC,MAAM,CAER;AA8XD,wBAAgB,cAAc,CAC5B,IAAI,EAAE,MAAM,EACZ,GAAG,EAAE,MAAM,EACX,OAAO,GAAE;IAAE,cAAc,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,WAAW,CAAA;CAExD,GACA,gBAAgB,CAGlB;AAyKD,wBAAgB,cAAc,CAC5B,IAAI,EAAE,MAAM,EACZ,QAAQ,CAAC,EAAE,aAAa,EACxB,OAAO,CAAC,EAAE;IACR,GAAG,CAAC,EAAE,MAAM,CAAC;IACb,MAAM,CAAC,EAAE,WAAW,GAAG,SAAS,CAAC;IACjC,QAAQ,CAAC,EAAE,QAAQ,GAAG,SAAS,CAAC;IAChC,gBAAgB,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;CACxC,GACA,MAAM,CAyBR;AA2DD,wBAAgB,sBAAsB,CACpC,OAAO,EAAE,gBAAgB,GAAG,IAAI,EAChC,sBAAsB,EAAE,MAAM,GAAG,QAAQ,GACxC,OAAO,CAQT;AAKD,wBAAgB,gCAAgC,CAC9C,OAAO,EAAE,gBAAgB,GAAG,IAAI,GAC/B,OAAO,IAAI,gBAAgB,CAE7B;AAED,wBAAgB,0BAA0B,CACxC,GAAG,EAAE,MAAM,EACX,OAAO,EAAE,gBAAgB,GAAG,IAAI,EAChC,aAAa,EAAE,iBAAiB,EAChC,wBAAwB,EAAE,OAAO,EACjC,eAAe,EAAE,OAAO,GACvB,aAAa,GAAG,SAAS,CAuB3B;AA4bD,wBAAgB,gCAAgC,CAC9C,IAAI,EAAE,MAAM,EACZ,GAAG,EAAE,MAAM,EACX,OAAO,EAAE,gBAAgB,GACxB,uBAAuB,CAgBzB;AAaD,UAAU,kBAAkB;IAC1B,UAAU,EAAE,MAAM,CAAC;IACnB,aAAa,EAAE,MAAM,CAAC;IACtB,QAAQ,EAAE,MAAM,CAAC;CAClB;AAED,wBAAgB,qBAAqB,IAAI,kBAAkB,GAAG,IAAI,CAEjE;AAED,wBAAsB,2BAA2B,IAAI,OAAO,CAAC,IAAI,CAAC,CAEjE;AAED,KAAK,yBAAyB,GAAG,gBAAgB,GAAG;IAAE,QAAQ,CAAC,EAAE,MAAM,CAAA;CAAE,CAAC;AAgH1E,wBAAsB,uBAAuB,CAC3C,IAAI,EAAE,MAAM,EACZ,GAAG,EAAE,MAAM,EACX,OAAO,EAAE,gBAAgB,GACxB,OAAO,CAAC,uBAAuB,CAAC,CAElC;AAED,wBAAsB,yBAAyB,CAC7C,UAAU,EAAE,UAAU,EACtB,GAAG,EAAE,MAAM,EACX,OAAO,EAAE,yBAAyB,GACjC,OAAO,CAAC,uBAAuB,CAAC,CAElC;AAED,OAAO,EACL,wBAAwB,EACxB,wBAAwB,EACxB,oBAAoB,GACrB,CAAC"}

package/dist/transform/transform.js

@@ -3,12 +3,12 @@ import { isProbablyReaderable, Readability } from '@mozilla/readability';
 import { parseHTML } from 'linkedom';
 import { config } from '../lib/core.js';
 import { getOperationId, getRequestId, logDebug, logError, logInfo, logWarn, redactUrl, } from '../lib/core.js';
-import { evaluateArticleContent, extractNoscriptImages, getVisibleTextLength, normalizeTabContent, prepareDocumentForMarkdown, removeNoiseFromHtml, serializeDocumentForMarkdown, } from '../lib/dom-prep.js';
+import { evaluateArticleContent, extractNoscriptImages, getVisibleTextLength, normalizeTabContent, prepareDocumentForMarkdown, removeNoiseFromHtml, serializeDocumentForMarkdown, surfaceCodeEditorContent, } from '../lib/dom-prep.js';
 import { isRawTextContentUrl } from '../lib/http.js';
 import { composeAbortSignal, FetchError, getErrorMessage, getUtf8ByteLength, isAsciiOnly, isObject, throwIfAborted, toError, trimDanglingTagFragment, truncateToUtf8Boundary, } from '../lib/utils.js';
 import { extractLanguageFromClassName } from './html-translators.js';
 import { translateHtmlFragmentToMarkdown } from './html-translators.js';
-import { cleanupMarkdownArtifacts, processFencedContent, } from './markdown-cleanup.js';
+import { cleanupMarkdownArtifacts, finalizeMarkdownSections, processFencedContent, } from './markdown-cleanup.js';
 import { addSourceToMarkdown, buildMetadataFooter, extractTitleFromRawMarkdown, isRawTextContent, } from './metadata.js';
 import { extractMetadata, extractMetadataFromHead, mergeMetadata, normalizeDocumentTitle, } from './metadata.js';
 import { supplementMarkdownFromNextFlight } from './next-flight.js';
@@ -249,6 +249,7 @@ function prepareReadabilityDocument(readabilityDoc) {
     preserveAlertElements(readabilityDoc);
     preserveCodeLanguageAttributes(readabilityDoc);
     normalizeTabContent(readabilityDoc);
+    surfaceCodeEditorContent(readabilityDoc);
     for (const el of readabilityDoc.querySelectorAll('[class*="breadcrumb"],[class*="pagination"]')) {
         if (el.tagName === 'HTML' || el.tagName === 'BODY')
             continue;
@@ -825,7 +826,7 @@ function postprocessMarkdownStage({ context, url, signal }, markdown) {
     let content = maybeStripGithubPrimaryHeading(markdown, context.primaryHeading, url);
     content = maybePrependSyntheticTitle(content, context, url);
     content = supplementMarkdownFromNextFlight(content, context.originalHtml);
-    content = cleanupMarkdownArtifacts(content, signal ? { signal, url } : { url });
+    content = finalizeMarkdownSections(content, signal ? { signal, url } : { url });
     return {
         markdown: content,
         title: context.title,
@@ -972,4 +973,4 @@ export async function transformHtmlToMarkdown(html, url, options) {
 export async function transformBufferToMarkdown(htmlBuffer, url, options) {
     return transformInputToMarkdown(htmlBuffer, url, options);
 }
-export { cleanupMarkdownArtifacts, processFencedContent };
+export { cleanupMarkdownArtifacts, finalizeMarkdownSections, processFencedContent, };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@j0hanz/fetch-url-mcp",
-  "version": "1.10.18",
+  "version": "1.10.20",
   "mcpName": "io.github.j0hanz/fetch-url-mcp",
   "description": "A web content fetcher MCP server that converts HTML to clean, AI and human readable markdown.",
   "type": "module",