@j0hanz/fetch-url-mcp 1.10.2 → 1.10.4

package/dist/lib/dom-prep.js

@@ -1,11 +1,18 @@
 import { parseHTML } from 'linkedom';
 import { config, logDebug } from './core.js';
+// ── Thresholds ──────────────────────────────────────────────────────
 const NOISE_SCAN_LIMIT = 50_000;
 const MIN_BODY_CONTENT_LENGTH = 100;
 const DIALOG_MIN_CHARS_FOR_PRESERVATION = 500;
 const NAV_FOOTER_MIN_CHARS_FOR_PRESERVATION = 500;
 const ABORT_CHECK_INTERVAL = 500;
 const NODE_FILTER_SHOW_TEXT = 4;
+const ASIDE_NAV_LINK_DENSITY_THRESHOLD = 0.5;
+const ASIDE_NAV_MIN_LINKS = 10;
+const INLINE_DEMO_INSTRUCTION_MAX_CHARS = 160;
+const REDUNDANT_PREVIEW_SEGMENT_MAX_CHARS = 60;
+const REDUNDANT_PREVIEW_MAX_SEGMENTS = 12;
+// ── Regex patterns ──────────────────────────────────────────────────
 const HTML_DOCUMENT_MARKERS = /<\s*(?:!doctype|html|head|body)\b/i;
 const HTML_FRAGMENT_MARKERS = /<\s*(?:article|main|section|div|nav|footer|header|aside|table|ul|ol)\b/i;
 const NOISE_PATTERNS = [
@@ -19,6 +26,10 @@ const HEADER_NOISE_PATTERN = /\b(site-header|masthead|topbar|navbar|nav(?:bar)?|
 const FIXED_OR_HIGH_Z_PATTERN = /\b(?:fixed|sticky|z-(?:4\d|50)|isolate)\b/;
 const HEADING_PERMALINK_TEXT_PATTERN = /^(?:#|¶|§|¤|🔗)$/u;
 const HEADING_PERMALINK_CLASS_PATTERN = /\b(?:mark|permalink|hash-link|anchor(?:js)?-?link|header-?link|heading-anchor|deep-link)\b/i;
+const HIDDEN_STYLE_REGEX = /\b(?:display\s*:\s*none|visibility\s*:\s*hidden)\b/i;
+const UTM_PARAM_REGEX = /[?&]utm_(?:source|medium|campaign)=/i;
+const NO_MATCH_REGEX = /a^/i;
+// ── URL prefixes to skip during resolution ──────────────────────────
 const SKIP_URL_PREFIXES = [
     '#',
     'javascript:',
@@ -27,6 +38,7 @@ const SKIP_URL_PREFIXES = [
     'data:',
     'blob:',
 ];
+// ── Tag / role sets ─────────────────────────────────────────────────
 const BASE_STRUCTURAL_TAGS = new Set([
     'script',
     'style',
@@ -52,9 +64,6 @@ const NAVIGATION_ROLES = new Set([
     'alertdialog',
     'search',
 ]);
-const INLINE_DEMO_INSTRUCTION_MAX_CHARS = 160;
-const REDUNDANT_PREVIEW_SEGMENT_MAX_CHARS = 60;
-const REDUNDANT_PREVIEW_MAX_SEGMENTS = 12;
 const INTERACTIVE_CONTENT_ROLES = new Set([
     'tabpanel',
     'tab',
@@ -69,6 +78,7 @@ const INTERACTIVE_CONTENT_ROLES = new Set([
     'tooltip',
     'alert',
 ]);
+// ── Promo tokens ────────────────────────────────────────────────────
 const PROMO_TOKENS_ALWAYS = [
     'banner',
     'promo',
@@ -87,20 +97,30 @@ const PROMO_TOKENS_ALWAYS = [
     'fb-post',
     'instagram-media',
     'social-embed',
+    'author-bio',
+    'byline',
+    'sharedaddy',
+    'sharing',
 ];
 const PROMO_TOKENS_AGGRESSIVE = ['ad', 'related', 'comment'];
 const PROMO_TOKENS_BY_CATEGORY = {
     'cookie-banners': ['cookie', 'consent', 'popup', 'modal', 'overlay', 'toast'],
     newsletters: ['newsletter', 'subscribe'],
-    'social-share': ['share', 'social'],
+    'social-share': ['share', 'social', 'share-button'],
+    'author-blocks': ['author-bio', 'byline', 'author-info', 'writer-profile'],
+    'related-content': [
+        'related-post',
+        'related-article',
+        'more-stories',
+        'recommended-posts',
+    ],
 };
-// Noise selector configurations
+// ── Noise selector configurations ───────────────────────────────────
 const BASE_NOISE_SELECTORS = {
     navFooter: 'nav,footer,header[class*="site"],header[class*="nav"],header[class*="menu"],[role="banner"],[role="navigation"],[class*="breadcrumb"]',
     cookieBanners: '[role="dialog"]',
     hidden: '[style*="display: none"],[style*="display:none"],[style*="visibility: hidden"],[style*="visibility:hidden"],[hidden],[aria-hidden="true"]',
 };
-const NO_MATCH_REGEX = /a^/i;
 let cachedContext;
 let lastContextKey;
 function escapeRegexLiteral(value) {
@@ -111,24 +131,18 @@ function buildTokenRegex(tokens) {
         return NO_MATCH_REGEX;
     return new RegExp(`(?:^|[^a-z0-9])(?:${[...tokens].map(escapeRegexLiteral).join('|')})(?:$|[^a-z0-9])`, 'i');
 }
-function addTokens(target, tokens) {
-    for (const token of tokens)
-        target.add(token);
-}
-function getPromoMatchers(currentConfig, flags) {
+function getPromoMatchers(currentConfig, enabledCategories) {
     const baseTokens = new Set(PROMO_TOKENS_ALWAYS);
     const aggressiveTokens = new Set();
     if (currentConfig.aggressiveMode) {
-        addTokens(aggressiveTokens, PROMO_TOKENS_AGGRESSIVE);
+        for (const token of PROMO_TOKENS_AGGRESSIVE)
+            aggressiveTokens.add(token);
     }
-    if (flags.cookieBanners) {
-        addTokens(baseTokens, PROMO_TOKENS_BY_CATEGORY['cookie-banners']);
-    }
-    if (flags.newsletters) {
-        addTokens(baseTokens, PROMO_TOKENS_BY_CATEGORY['newsletters']);
-    }
-    if (flags.socialShare) {
-        addTokens(baseTokens, PROMO_TOKENS_BY_CATEGORY['social-share']);
+    for (const [category, tokens] of Object.entries(PROMO_TOKENS_BY_CATEGORY)) {
+        if (enabledCategories.has(category)) {
+            for (const token of tokens)
+                baseTokens.add(token);
+        }
     }
     for (const t of currentConfig.extraTokens) {
         const n = t.toLowerCase().trim();
@@ -149,7 +163,6 @@ function getContext() {
         extraSelectors: currentConfig.extraSelectors,
         aggressiveMode: currentConfig.aggressiveMode,
         preserveSvgCanvas: currentConfig.preserveSvgCanvas,
-        weights: currentConfig.weights,
     });
     if (cachedContext !== undefined && lastContextKey === contextKey)
         return cachedContext;
@@ -164,15 +177,13 @@ function getContext() {
     const flags = {
         navFooter: isEnabled('nav-footer'),
         cookieBanners: isEnabled('cookie-banners'),
-        newsletters: isEnabled('newsletters'),
-        socialShare: isEnabled('social-share'),
     };
     const structuralTags = new Set(BASE_STRUCTURAL_TAGS);
     if (!currentConfig.preserveSvgCanvas) {
         structuralTags.add('svg');
         structuralTags.add('canvas');
     }
-    const promoMatchers = getPromoMatchers(currentConfig, flags);
+    const promoMatchers = getPromoMatchers(currentConfig, enabled);
     const extraSelectors = currentConfig.extraSelectors
         .map((s) => s.trim())
         .filter((s) => s.length > 0);
@@ -182,7 +193,8 @@ function getContext() {
         selectors.push(BASE_NOISE_SELECTORS.navFooter);
     if (flags.cookieBanners)
         selectors.push(BASE_NOISE_SELECTORS.cookieBanners);
-    const baseSelector = selectors.join(',');
+    const noiseSelector = selectors.join(',');
+    const extraSelector = extraSelectors.length > 0 ? extraSelectors.join(',') : null;
     const candidateSelector = [
         ...structuralTags,
         ...ALWAYS_NOISE_TAGS,
@@ -196,11 +208,10 @@ function getContext() {
     cachedContext = {
         flags,
         structuralTags,
-        weights: currentConfig.weights,
         promoMatchers,
-        promoEnabled: flags.cookieBanners || flags.newsletters || flags.socialShare,
-        extraSelectors,
-        baseSelector,
+        promoEnabled: Object.keys(PROMO_TOKENS_BY_CATEGORY).some((cat) => enabled.has(cat)),
+        noiseSelector,
+        extraSelector,
         candidateSelector,
     };
     lastContextKey = contextKey;
@@ -231,8 +242,6 @@ function isWithinPrimaryContent(element) {
     }
     return false;
 }
-const ASIDE_NAV_LINK_DENSITY_THRESHOLD = 0.5;
-const ASIDE_NAV_MIN_LINKS = 10;
 function isNavigationAside(element) {
     if (element.querySelector('nav'))
         return true;
@@ -278,69 +287,57 @@ function removeNodes(nodes) {
         }
     }
 }
-const HIDDEN_STYLE_REGEX = /\b(?:display\s*:\s*none|visibility\s*:\s*hidden)\b/i;
-function calculateNavFooterScore(tagName, className, id, role, weights) {
-    let score = 0;
-    if (ALWAYS_NOISE_TAGS.has(tagName))
-        score += weights.structural;
-    if (tagName === 'header') {
-        if ((role && NAVIGATION_ROLES.has(role)) ||
-            HEADER_NOISE_PATTERN.test(`${className} ${id}`)) {
-            score += weights.structural;
-        }
-    }
-    if (tagName === 'aside') {
-        score += weights.structural;
-    }
-    if (role && NAVIGATION_ROLES.has(role)) {
-        if (tagName !== 'aside' || role !== 'complementary') {
-            score += weights.structural;
-        }
-    }
-    return score;
-}
-function calculatePromoScore(element, className, id, context) {
+function isPromoMatch(className, id, element, context) {
     if (!context.promoEnabled)
-        return 0;
+        return false;
     const aggTest = context.promoMatchers.aggressive.test(className) ||
         context.promoMatchers.aggressive.test(id);
-    const isAggressiveMatch = aggTest && !isWithinPrimaryContent(element);
-    const isBaseMatch = !aggTest &&
-        (context.promoMatchers.base.test(className) ||
-            context.promoMatchers.base.test(id));
-    return isAggressiveMatch || isBaseMatch ? context.weights.promo : 0;
+    if (aggTest)
+        return !isWithinPrimaryContent(element);
+    return (context.promoMatchers.base.test(className) ||
+        context.promoMatchers.base.test(id));
 }
 function isNoiseElement(element, context) {
     const tagName = element.tagName.toLowerCase();
+    const role = element.getAttribute('role');
     const className = element.getAttribute('class') ?? '';
     const id = element.getAttribute('id') ?? '';
-    const role = element.getAttribute('role');
+    const interactive = isInteractive(element, role);
     const style = element.getAttribute('style');
-    const elIsInteractive = isInteractive(element, role);
-    const elIsHidden = element.hasAttribute('hidden') ||
+    const hidden = element.hasAttribute('hidden') ||
         element.getAttribute('aria-hidden') === 'true' ||
         (style !== null && HIDDEN_STYLE_REGEX.test(style));
-    let score = 0;
-    const { weights } = context;
-    // Structural
-    if (context.structuralTags.has(tagName) && !elIsInteractive) {
-        score += weights.structural;
-    }
-    // Nav/Footer Scoring
+    // Structural tags (script, style, form, etc.)
+    if (context.structuralTags.has(tagName) && !interactive)
+        return true;
     if (context.flags.navFooter) {
-        score += calculateNavFooterScore(tagName, className, id, role, weights);
-    }
-    // Hidden
-    if (elIsHidden && !elIsInteractive) {
-        score += weights.hidden;
-    }
-    // Sticky/Fixed
-    if (FIXED_OR_HIGH_Z_PATTERN.test(className)) {
-        score += weights.stickyFixed;
+        // Always-noise tags (nav, footer)
+        if (ALWAYS_NOISE_TAGS.has(tagName))
+            return true;
+        // Header with navigation role or noise class/id
+        if (tagName === 'header' &&
+            ((role !== null && NAVIGATION_ROLES.has(role)) ||
+                HEADER_NOISE_PATTERN.test(`${className} ${id}`)))
+            return true;
+        // Aside elements
+        if (tagName === 'aside')
+            return true;
+        // Navigation roles (except aside+complementary)
+        if (role !== null &&
+            NAVIGATION_ROLES.has(role) &&
+            (tagName !== 'aside' || role !== 'complementary'))
+            return true;
     }
-    // Promo
-    score += calculatePromoScore(element, className, id, context);
-    return score >= weights.threshold;
+    // Hidden elements
+    if (hidden && !interactive)
+        return true;
+    // Sticky/fixed positioned elements
+    if (FIXED_OR_HIGH_Z_PATTERN.test(className))
+        return true;
+    // Promotional/noise content
+    if (isPromoMatch(className, id, element, context))
+        return true;
+    return false;
 }
 function cleanHeadings(document) {
     const headings = document.querySelectorAll('h1,h2,h3,h4,h5,h6');
@@ -421,13 +418,18 @@ function hoistNestedRows(table) {
         }
     }
 }
-function stripNoise(document, context, signal) {
-    cleanHeadings(document);
+function stripNoise(document, signal) {
+    const context = getContext();
+    if (config.noiseRemoval.debug) {
+        logDebug('Noise removal audit enabled', {
+            categories: [...(context.flags.navFooter ? ['nav-footer'] : [])],
+        });
+    }
     // Structural Removal
-    const { baseSelector, extraSelectors } = context;
-    removeNodes(document.querySelectorAll(baseSelector));
-    if (extraSelectors.length > 0) {
-        removeNodes(document.querySelectorAll(extraSelectors.join(',')));
+    removeNodes(document.querySelectorAll(context.noiseSelector));
+    // Extra selectors (evaluated after base removal so DOM state is updated)
+    if (context.extraSelector) {
+        removeNodes(document.querySelectorAll(context.extraSelector));
     }
     // Candidates (conditional removal)
     const candidates = document.querySelectorAll(context.candidateSelector);
@@ -531,20 +533,54 @@ function surfaceHiddenTabPanels(document) {
     }
 }
 function stripTabTriggers(document) {
-    surfaceHiddenTabPanels(document);
-    const tabs = document.querySelectorAll('button[role="tab"]');
+    const tabs = document.querySelectorAll('[role="tab"]');
     for (let i = tabs.length - 1; i >= 0; i--) {
-        tabs[i]?.remove();
+        const tab = tabs[i];
+        if (!tab)
+            continue;
+        const isSelected = tab.getAttribute('aria-selected') === 'true' ||
+            tab.getAttribute('data-state') === 'active' ||
+            tab.hasAttribute('data-selected');
+        if (!isSelected) {
+            tab.remove();
+        }
     }
 }
-function escapeTableCellPipes(document) {
+/** Surface hidden tab panels, then strip unselected tab triggers. */
+export function normalizeTabContent(document) {
+    surfaceHiddenTabPanels(document);
+    stripTabTriggers(document);
+}
+function normalizeTableCells(document) {
     const cells = document.querySelectorAll('td, th');
     for (const cell of cells) {
+        const brs = cell.querySelectorAll('br');
+        for (const br of brs) {
+            br.replaceWith(' ');
+        }
+        const blocks = Array.from(cell.querySelectorAll('div, p, ul, li, h1, h2, h3, h4, h5, h6'));
+        for (const block of blocks) {
+            if (!block.parentNode)
+                continue;
+            const span = document.createElement('span');
+            span.appendChild(document.createTextNode(' '));
+            while (block.firstChild) {
+                span.appendChild(block.firstChild);
+            }
+            span.appendChild(document.createTextNode(' '));
+            for (const attr of Array.from(block.attributes)) {
+                span.setAttribute(attr.name, attr.value);
+            }
+            block.replaceWith(span);
+        }
         const walker = document.createTreeWalker(cell, NODE_FILTER_SHOW_TEXT);
         let node;
         while ((node = walker.nextNode())) {
-            if (node.textContent?.includes('|')) {
-                node.textContent = node.textContent.replace(/\|/g, '\\|');
+            if (node.nodeValue) {
+                node.nodeValue = node.nodeValue.replace(/\r?\n/g, ' ');
+                if (node.nodeValue.includes('|')) {
+                    node.nodeValue = node.nodeValue.replace(/\|/g, '\\|');
+                }
             }
         }
     }
@@ -673,8 +709,20 @@ function cleanCodeExamples(document) {
     pruneDemoInstructionBlocks(document);
     normalizeHighlightedCodeLines(document);
 }
+function stripPromoLinks(document) {
+    const links = document.querySelectorAll('a[href]');
+    for (let i = links.length - 1; i >= 0; i--) {
+        const link = links[i];
+        if (!link)
+            continue;
+        const href = link.getAttribute('href');
+        if (href && UTM_PARAM_REGEX.test(href)) {
+            link.remove();
+        }
+    }
+}
 function separateAdjacentInlineElements(document) {
-    const badges = document.querySelectorAll('span.chakra-badge, [data-scope="badge"], [class*="badge"]');
+    const badges = document.querySelectorAll('span.chakra-badge, [data-scope="badge"], [class*="badge"], [data-slot="label"], [slot="label"]');
     for (const badge of badges) {
         const next = badge.nextSibling;
         if (next?.nodeType === 1) {
@@ -682,19 +730,17 @@ function separateAdjacentInlineElements(document) {
         }
     }
 }
+// Called on both raw documents (pre-article path) and article fragments
+// (post-Readability). Some passes (stripTabTriggers, etc.) are no-ops
+// on Readability output since tabs are already stripped or absent.
 export function prepareDocumentForMarkdown(document, baseUrl, signal) {
-    const context = getContext();
-    if (config.noiseRemoval.debug) {
-        logDebug('Noise removal audit enabled', {
-            categories: [...(context.flags.navFooter ? ['nav-footer'] : [])],
-        });
-    }
-    stripNoise(document, context, signal);
-    stripTabTriggers(document);
+    normalizeTabContent(document);
+    cleanHeadings(document);
+    stripNoise(document, signal);
+    stripPromoLinks(document);
     cleanCodeExamples(document);
     separateAdjacentInlineElements(document);
-    flattenTableCellBreaks(document);
-    escapeTableCellPipes(document);
+    normalizeTableCells(document);
     normalizeTableStructure(document);
     if (baseUrl)
         resolveUrls(document, baseUrl);
@@ -723,41 +769,6 @@ function normalizeTableStructure(document) {
         hoistNestedRows(table);
     }
 }
-function flattenTableCellBreaks(document) {
-    const cells = document.querySelectorAll('td, th');
-    for (const cell of cells) {
-        const brs = cell.querySelectorAll('br');
-        for (const br of brs) {
-            br.replaceWith(' ');
-        }
-        const blocks = Array.from(cell.querySelectorAll('div, p, ul, li, h1, h2, h3, h4, h5, h6'));
-        for (const block of blocks) {
-            if (!block.parentNode)
-                continue;
-            const span = document.createElement('span');
-            span.appendChild(document.createTextNode(' '));
-            while (block.firstChild) {
-                span.appendChild(block.firstChild);
-            }
-            span.appendChild(document.createTextNode(' '));
-            for (const attr of Array.from(block.attributes)) {
-                span.setAttribute(attr.name, attr.value);
-            }
-            block.replaceWith(span);
-        }
-        const filterNewlines = (node) => {
-            if (node.nodeType === 3 && node.nodeValue) {
-                node.nodeValue = node.nodeValue.replace(/\r?\n/g, ' ');
-            }
-            else {
-                for (const child of Array.from(node.childNodes)) {
-                    filterNewlines(child);
-                }
-            }
-        };
-        filterNewlines(cell);
-    }
-}
 export function removeNoiseFromHtml(html, document, baseUrl, signal) {
     const shouldParse = isFullDocumentHtml(html) ||
         mayContainNoise(html) ||

package/dist/lib/md-cleanup.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"md-cleanup.d.ts","sourceRoot":"","sources":["../../src/lib/md-cleanup.ts"],"names":[],"mappings":"AAwEA,UAAU,cAAc;IACtB,qBAAqB,CAAC,EAAE,OAAO,CAAC;IAChC,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB,GAAG,CAAC,EAAE,MAAM,CAAC;CACd;AA6cD;;;;GAIG;AACH,wBAAgB,oBAAoB,CAClC,OAAO,EAAE,MAAM,EACf,kBAAkB,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,MAAM,GAC3C,MAAM,CAuCR;AAWD,wBAAgB,wBAAwB,CACtC,OAAO,EAAE,MAAM,EACf,OAAO,CAAC,EAAE,cAAc,GACvB,MAAM,CAmBR"}
+{"version":3,"file":"md-cleanup.d.ts","sourceRoot":"","sources":["../../src/lib/md-cleanup.ts"],"names":[],"mappings":"AAiGA,UAAU,cAAc;IACtB,qBAAqB,CAAC,EAAE,OAAO,CAAC;IAChC,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB,GAAG,CAAC,EAAE,MAAM,CAAC;CACd;AAkeD;;;;GAIG;AACH,wBAAgB,oBAAoB,CAClC,OAAO,EAAE,MAAM,EACf,kBAAkB,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,MAAM,GAC3C,MAAM,CAuCR;AAWD,wBAAgB,wBAAwB,CACtC,OAAO,EAAE,MAAM,EACf,OAAO,CAAC,EAAE,cAAc,GACvB,MAAM,CAmBR"}

package/dist/lib/md-cleanup.js

@@ -1,5 +1,6 @@
 import { config } from './core.js';
 import { throwIfAborted } from './utils.js';
+// ── ASCII code constants ────────────────────────────────────────────
 const ASCII_HASH = 35;
 const ASCII_ASTERISK = 42;
 const ASCII_PLUS = 43;
@@ -10,13 +11,21 @@ const ASCII_DIGIT_9 = 57;
 const ASCII_EXCLAMATION = 33;
 const ASCII_QUESTION = 63;
 const ASCII_BRACKET_OPEN = 91;
+// ── Title heuristic thresholds ──────────────────────────────────────
 const TITLE_MIN_WORDS = 2;
 const TITLE_MAX_WORDS = 10;
 const TITLE_MIN_CAPITALIZED = 2;
+// ── Processing limits ───────────────────────────────────────────────
 const HAS_FOLLOWING_LOOKAHEAD = 10;
 const PROPERTY_FIX_MAX_PASSES = 5;
 const MAX_LINE_LENGTH = 80;
+// ── TOC thresholds ──────────────────────────────────────────────────
+const TOC_SCAN_LIMIT = 20;
+const TOC_MAX_NON_EMPTY = 12;
+const TOC_LINK_RATIO_THRESHOLD = 0.8;
+// ── Fence pattern ───────────────────────────────────────────────────
 const FENCE_PATTERN = /^\s*(`{3,}|~{3,})/;
+// ── Regex collection ────────────────────────────────────────────────
 const REGEX = {
     HEADING_MARKER: /^#{1,6}\s/m,
     HEADING_STRICT: /^#{1,6}\s+/m,
@@ -47,11 +56,11 @@ const REGEX = {
     NESTED_LIST_INDENT: /^( +)((?:[-*+])|\d+\.)\s/gm,
     TYPEDOC_COMMENT: /(`+)(?:(?!\1)[\s\S])*?\1|\s?\/\\?\*[\s\S]*?\\?\*\//g,
 };
+// ── Heading keywords (config-driven) ────────────────────────────────
 const HEADING_KEYWORDS = new Set(config.markdownCleanup.headingKeywords.map((value) => value.toLocaleLowerCase(config.i18n.locale)));
+// ── Prefix patterns ─────────────────────────────────────────────────
 const SPECIAL_PREFIXES = /^(?:example|note|tip|warning|important|caution):\s+\S/i;
-const TOC_SCAN_LIMIT = 20;
-const TOC_MAX_NON_EMPTY = 12;
-const TOC_LINK_RATIO_THRESHOLD = 0.8;
+// ── TypeDoc prefixes ────────────────────────────────────────────────
 const TYPEDOC_PREFIXES = [
     'Defined in:',
     'Returns:',
@@ -399,24 +408,48 @@ function fixConcatenatedProperties(text) {
 }
 function applyGlobalRegexes(text, options) {
     const checkAbort = createAbortChecker(options);
-    let result = text.replace(/\u00A0/g, ' ');
-    checkAbort('markdown:cleanup:headings');
-    result = result
-        .replace(REGEX.HEADING_SPACING, '$1\n\n$2')
-        .replace(REGEX.HEADING_CODE_BLOCK, '$1\n\n```');
-    if (config.markdownCleanup.removeTypeDocComments) {
-        checkAbort('markdown:cleanup:typedoc');
-        result = removeTypeDocArtifacts(result);
-    }
-    if (config.markdownCleanup.removeSkipLinks) {
-        checkAbort('markdown:cleanup:skip-links');
-        result = removeSkipLinks(result);
+    const passes = [
+        {
+            stage: 'markdown:cleanup:nbsp',
+            apply: (t) => t.replace(/\u00A0/g, ' '),
+        },
+        {
+            stage: 'markdown:cleanup:headings',
+            apply: (t) => t
+                .replace(REGEX.HEADING_SPACING, '$1\n\n$2')
+                .replace(REGEX.HEADING_CODE_BLOCK, '$1\n\n```'),
+        },
+        {
+            stage: 'markdown:cleanup:typedoc',
+            enabled: () => config.markdownCleanup.removeTypeDocComments,
+            apply: removeTypeDocArtifacts,
+        },
+        {
+            stage: 'markdown:cleanup:skip-links',
+            enabled: () => config.markdownCleanup.removeSkipLinks,
+            apply: removeSkipLinks,
+        },
+        {
+            stage: 'markdown:cleanup:spacing',
+            apply: normalizeMarkdownSpacing,
+        },
+        {
+            stage: 'markdown:cleanup:properties',
+            apply: fixConcatenatedProperties,
+        },
+        {
+            stage: 'markdown:cleanup:permalinks',
+            apply: stripTrailingHeadingPermalinks,
+        },
+    ];
+    let result = text;
+    for (const pass of passes) {
+        if (pass.enabled !== undefined && !pass.enabled())
+            continue;
+        checkAbort(pass.stage);
+        result = pass.apply(result);
     }
-    checkAbort('markdown:cleanup:spacing');
-    result = normalizeMarkdownSpacing(result);
-    checkAbort('markdown:cleanup:properties');
-    result = fixConcatenatedProperties(result);
-    return stripTrailingHeadingPermalinks(result);
+    return result;
 }
 function normalizeNestedListIndentation(text) {
     return text.replace(REGEX.NESTED_LIST_INDENT, (match, spaces, marker) => {

package/dist/lib/progress.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"progress.d.ts","sourceRoot":"","sources":["../../src/lib/progress.ts"],"names":[],"mappings":"AAOA,KAAK,aAAa,GAAG,MAAM,GAAG,MAAM,CAAC;AACrC,UAAU,WAAW;IACnB,aAAa,CAAC,EAAE,aAAa,GAAG,SAAS,CAAC;IAC1C,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC;CACxB;AACD,MAAM,WAAW,0BAA0B;IACzC,aAAa,EAAE,aAAa,CAAC;IAC7B,QAAQ,EAAE,MAAM,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CACjC;AACD,MAAM,WAAW,oBAAoB;IACnC,MAAM,EAAE,wBAAwB,CAAC;IACjC,MAAM,EAAE,0BAA0B,CAAC;CACpC;AACD,MAAM,WAAW,gBAAgB;IAC/B,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB,SAAS,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;IAC5B,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,WAAW,CAAC,EAAE,OAAO,CAAC;IACtB,KAAK,CAAC,EAAE,WAAW,CAAC;IACpB,gBAAgB,CAAC,EAAE,CAAC,YAAY,EAAE,oBAAoB,KAAK,OAAO,CAAC,IAAI,CAAC,CAAC;IACzE,UAAU,CAAC,EAAE,CAAC,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,KAAK,IAAI,CAAC;IACzD,iBAAiB,CAAC,EAAE,MAAM,OAAO,CAAC;CACnC;AACD,MAAM,WAAW,gBAAgB;IAC/B,MAAM,EAAE,CAAC,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,KAAK,IAAI,CAAC;CACrD;AAkJD,wBAAgB,sBAAsB,CACpC,KAAK,CAAC,EAAE,gBAAgB,GACvB,gBAAgB,CAElB"}
+{"version":3,"file":"progress.d.ts","sourceRoot":"","sources":["../../src/lib/progress.ts"],"names":[],"mappings":"AAOA,KAAK,aAAa,GAAG,MAAM,GAAG,MAAM,CAAC;AACrC,UAAU,WAAW;IACnB,aAAa,CAAC,EAAE,aAAa,GAAG,SAAS,CAAC;IAC1C,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC;CACxB;AACD,MAAM,WAAW,0BAA0B;IACzC,aAAa,EAAE,aAAa,CAAC;IAC7B,QAAQ,EAAE,MAAM,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CACjC;AACD,MAAM,WAAW,oBAAoB;IACnC,MAAM,EAAE,wBAAwB,CAAC;IACjC,MAAM,EAAE,0BAA0B,CAAC;CACpC;AACD,MAAM,WAAW,gBAAgB;IAC/B,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB,SAAS,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;IAC5B,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,WAAW,CAAC,EAAE,OAAO,CAAC;IACtB,KAAK,CAAC,EAAE,WAAW,CAAC;IACpB,gBAAgB,CAAC,EAAE,CAAC,YAAY,EAAE,oBAAoB,KAAK,OAAO,CAAC,IAAI,CAAC,CAAC;IACzE,UAAU,CAAC,EAAE,CAAC,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,KAAK,IAAI,CAAC;IACzD,iBAAiB,CAAC,EAAE,MAAM,OAAO,CAAC;CACnC;AACD,MAAM,WAAW,gBAAgB;IAC/B,MAAM,EAAE,CAAC,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,KAAK,IAAI,CAAC;CACrD;AAiLD,wBAAgB,sBAAsB,CACpC,KAAK,CAAC,EAAE,gBAAgB,GACvB,gBAAgB,CAElB"}