webforai 4.2.0 → 4.3.1

package/dist/cjs/adapters/sites/hackernews.js

@@ -45,11 +45,11 @@ const commentDepth = (row) => {
   return width === void 0 ? 0 : Math.round(width / 40);
 };
 const nest = (content, depth) => {
-  let current = content;
+  let current = element("div", content);
   for (let level = 0; level < Math.min(depth, MAX_QUOTE_DEPTH); level++) {
-    current = [element("blockquote", current)];
+    current = element("blockquote", [current]);
   }
-  return current[0];
+  return current;
 };
 const storyNodes = (root) => {
   const nodes = [];

package/dist/cjs/adapters/sites/youtube.js

@@ -18,6 +18,7 @@ var __copyProps = (to, from, except, desc) => {
 var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
 var youtube_exports = {};
 __export(youtube_exports, {
+  PLAYER_RESPONSE_VARIABLE: () => PLAYER_RESPONSE_VARIABLE,
   extractAssignedJson: () => extractAssignedJson,
   isYoutubeElement: () => isYoutubeElement,
   youtubeAdapter: () => youtubeAdapter,
@@ -69,12 +70,13 @@ const extractAssignedJson = (source, name) => {
 const scriptText = (root) => (0, import_hast_fast.collectElements)(root, (element2) => element2.tagName === "script").map(
   (script) => script.children.map((child) => child.type === "text" ? child.value : "").join("")
 );
+const PLAYER_RESPONSE_VARIABLE = "ytInitialPlayerResponse";
 const findPlayerResponse = (root) => {
   for (const source of scriptText(root)) {
-    if (!source.includes("ytInitialPlayerResponse")) {
+    if (!source.includes(PLAYER_RESPONSE_VARIABLE)) {
       continue;
     }
-    const parsed = extractAssignedJson(source, "ytInitialPlayerResponse");
+    const parsed = extractAssignedJson(source, PLAYER_RESPONSE_VARIABLE);
     if (parsed && typeof parsed === "object") {
       return parsed;
     }
@@ -184,6 +186,7 @@ const youtubeAdapter = {
 const isYoutubeElement = import_hast_fast.isElement;
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
+  PLAYER_RESPONSE_VARIABLE,
   extractAssignedJson,
   isYoutubeElement,
   youtubeAdapter,

package/dist/cjs/detect-client-shell.js

@@ -18,6 +18,7 @@ var __copyProps = (to, from, except, desc) => {
 var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
 var detect_client_shell_exports = {};
 __export(detect_client_shell_exports, {
+  CHALLENGE_PATTERN: () => CHALLENGE_PATTERN,
   SHELL_EMPTY_THRESHOLD: () => SHELL_EMPTY_THRESHOLD,
   SHELL_SUSPICION_THRESHOLD: () => SHELL_SUSPICION_THRESHOLD,
   detectClientShell: () => detectClientShell
@@ -72,6 +73,7 @@ const detectClientShell = (html) => {
 };
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
+  CHALLENGE_PATTERN,
   SHELL_EMPTY_THRESHOLD,
   SHELL_SUSPICION_THRESHOLD,
   detectClientShell

package/dist/cjs/extractors/lib/layout-tables.js

@@ -27,10 +27,51 @@ const DATA_MARKERS = /* @__PURE__ */ new Set(["caption", "thead", "tfoot", "th",
 const BLOCK_IN_CELL = /* @__PURE__ */ new Set(["h1", "h2", "h3", "h4", "h5", "h6", "table", "form"]);
 const TABLE_PARTS = /* @__PURE__ */ new Set(["tbody", "thead", "tfoot", "tr", "td", "th"]);
 const TABLE_CONTEXT = /* @__PURE__ */ new Set(["table", "tbody", "thead", "tfoot", "tr"]);
+const MAX_SMALL_TABLE_CELL = 150;
+const MAX_SMALL_TABLE_LINK_SHARE = 0.5;
+const collapsedLength = (text) => text.replace(/\s+/g, " ").trim().length;
+const cellText = (cell) => {
+  let text = "";
+  let linkText = "";
+  const visit = (node, inLink) => {
+    for (const child of node.children) {
+      if (child.type === "text") {
+        text += child.value;
+        if (inLink) {
+          linkText += child.value;
+        }
+      } else if ((0, import_hast_fast.isElement)(child) && child.tagName !== "table") {
+        visit(child, inLink || child.tagName === "a");
+      }
+    }
+  };
+  visit(cell, false);
+  return { text: collapsedLength(text), linkText: collapsedLength(linkText) };
+};
 const cellsOf = (row) => row.children.filter((cell) => (0, import_hast_fast.isElement)(cell) && (cell.tagName === "td" || cell.tagName === "th"));
 const HEADING = /^h[1-6]$/;
+const measureRow = (shape, row) => {
+  const cells = cellsOf(row);
+  shape.rows += 1;
+  shape.columns = Math.max(shape.columns, cells.length);
+  for (const cell of cells) {
+    const { text, linkText } = cellText(cell);
+    shape.longestCell = Math.max(shape.longestCell, text);
+    shape.text += text;
+    shape.linkText += linkText;
+  }
+  return cells.length === 1;
+};
 const shapeOf = (table) => {
-  const shape = { rows: 0, columns: 0, dataMarker: false, blockInCell: false };
+  const shape = {
+    rows: 0,
+    columns: 0,
+    dataMarker: false,
+    blockInCell: false,
+    longestCell: 0,
+    text: 0,
+    linkText: 0
+  };
   let captionHeadings = 0;
   const noteBlockInCell = (tag, aloneInRow) => {
     if (HEADING.test(tag) && aloneInRow) {
@@ -52,13 +93,7 @@ const shapeOf = (table) => {
       if (DATA_MARKERS.has(tag)) {
         shape.dataMarker = true;
       }
-      let alone = aloneInRow;
-      if (tag === "tr") {
-        shape.rows += 1;
-        const cells = cellsOf(child).length;
-        shape.columns = Math.max(shape.columns, cells);
-        alone = cells === 1;
-      }
+      const alone = tag === "tr" ? measureRow(shape, child) : aloneInRow;
       if (inCell && BLOCK_IN_CELL.has(tag)) {
         noteBlockInCell(tag, aloneInRow);
       }
@@ -89,7 +124,11 @@ const isLayoutTable = (table) => {
   if (shape.rows >= 10 || shape.columns > 4) {
     return false;
   }
-  return shape.rows * shape.columns <= 10;
+  if (shape.rows * shape.columns > 10) {
+    return false;
+  }
+  const linkShare = shape.text === 0 ? 1 : shape.linkText / shape.text;
+  return shape.longestCell > MAX_SMALL_TABLE_CELL || linkShare > MAX_SMALL_TABLE_LINK_SHARE;
 };
 const unwrap = (table) => {
   const cells = [];

package/dist/cjs/extractors/lib/sanitize.js

@@ -189,16 +189,52 @@ const containsMath = (element) => {
   }
   return false;
 };
+const collapsedText = (element) => {
+  let text = "";
+  (0, import_hast_fast.walk)(element, (node) => {
+    if (node.type === "text") {
+      text += node.value;
+    }
+  });
+  return text.replace(/\s+/g, " ").trim();
+};
+const visibleSiblingText = (parent, cache) => {
+  let texts = cache.get(parent);
+  if (!texts) {
+    texts = /* @__PURE__ */ new Set();
+    for (const child of parent.children) {
+      if ((0, import_hast_fast.isElement)(child) && !(0, import_hast_fast.isTruthyAttribute)(child, "aria-hidden") && !looksHidden(child)) {
+        texts.add(collapsedText(child));
+      }
+    }
+    cache.set(parent, texts);
+  }
+  return texts;
+};
+const isAriaHiddenDuplicate = (element, parent, cache) => {
+  if (!(0, import_hast_fast.isTruthyAttribute)(element, "aria-hidden") || containsMath(element)) {
+    return false;
+  }
+  const text = collapsedText(element);
+  return text.length > 0 && visibleSiblingText(parent, cache).has(text);
+};
 const stripNonContent = (tree) => {
   const tabs = new import_code_tabs.CodeTabCollector((element) => looksHidden(element) || import_constants.NON_CONTENT_TAGS.has(element.tagName));
-  hoistNoscriptImages(tree, (element, depth, parent) => tabs.visit(element, depth, parent));
+  const duplicates = /* @__PURE__ */ new Set();
+  const siblingText = /* @__PURE__ */ new WeakMap();
+  hoistNoscriptImages(tree, (element, depth, parent) => {
+    tabs.visit(element, depth, parent);
+    if (isAriaHiddenDuplicate(element, parent, siblingText)) {
+      duplicates.add(element);
+    }
+  });
   tabs.apply();
   const siblingCode = /* @__PURE__ */ new WeakMap();
   return (0, import_hast_fast.pruneInPlace)(tree, (node, parent) => {
     if (isNonContent(node)) {
       return false;
     }
-    if ((0, import_hast_fast.isElement)(node) && isHidden(node, parent, siblingCode)) {
+    if ((0, import_hast_fast.isElement)(node) && (duplicates.has(node) || isHidden(node, parent, siblingCode))) {
       return false;
     }
     return true;

package/dist/cjs/extractors/lib/terminators.js

@@ -26,6 +26,8 @@ var import_hast_fast = require("../../utils/hast-fast");
 const TERMINATOR_PATTERNS = [
   /^was this (page|article|document) helpful/i,
   /^thanks for the feedback/i,
+  // Site-generator credit closing the page ("Powered by Docsify.js", "Powered by GitBook").
+  /^powered by\b/i,
   /^(this page was )?last (modified|updated) (on|at|:)?\s/i,
   /^learn how to contribute/i,
   /^help (us )?improve\b/i,

package/dist/cjs/index.js

@@ -49,6 +49,7 @@ __export(src_exports, {
   readabilityExtractor: () => import_extractors.readabilityExtractor,
   resolveAdapter: () => import_adapters.resolveAdapter,
   runAdapters: () => import_adapters.runAdapters,
+  stripScriptBodies: () => import_parse_html.stripScriptBodies,
   takumiExtractor: () => import_extractors.takumiExtractor,
   toFrontmatter: () => import_metadata.toFrontmatter,
   youtubeCaptionTracks: () => import_adapters.youtubeCaptionTracks
@@ -64,6 +65,7 @@ var import_detect_client_shell = require("./detect-client-shell");
 var import_extract_meta_refresh = require("./extract-meta-refresh");
 var import_metadata = require("./metadata");
 var import_normalize = require("./normalize");
+var import_parse_html = require("./utils/parse-html");
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   BUILTIN_ADAPTERS,
@@ -97,6 +99,7 @@ var import_normalize = require("./normalize");
   readabilityExtractor,
   resolveAdapter,
   runAdapters,
+  stripScriptBodies,
   takumiExtractor,
   toFrontmatter,
   youtubeCaptionTracks

package/dist/cjs/mdast-handlers/custom-code-handler.js

@@ -29,17 +29,46 @@ var import_hast_fast = require("../utils/hast-fast");
 var import_pre_text = require("../utils/pre-text");
 var import_twoslash = require("./twoslash");
 const LANGUAGE_MATCH_REGEX = [/^language-(\S+)$/, /^highlight-source-(\S+)$/, /^CodeBlock--language-(\S+)$/];
+const LANGUAGE_TOKEN = /^[\w+#.-]+$/;
+const NO_LANGUAGE = /* @__PURE__ */ new Set(["plain", "plaintext", "text", "none", "nohighlight"]);
 const LANGUAGE_ATTRIBUTES = ["dataLanguage", "dataLang"];
 const attributeLanguage = (element) => {
   for (const name of LANGUAGE_ATTRIBUTES) {
     const value = element.properties?.[name];
-    if (typeof value === "string" && /^[\w+#.-]+$/.test(value.trim()) && value.trim() !== "plaintext") {
+    if (typeof value === "string" && LANGUAGE_TOKEN.test(value.trim()) && value.trim() !== "plaintext") {
       return value.trim();
     }
   }
   return void 0;
 };
-const codeLanguage = (...elements) => elements.flatMap(import_hast_fast.classList).map((className) => LANGUAGE_MATCH_REGEX.map((regex) => className.match(regex)).find(Boolean)?.[1]).find(Boolean) ?? elements.map(attributeLanguage).find(Boolean);
+const SANDPACK_LANGUAGES = /* @__PURE__ */ new Set([
+  "javascript",
+  "typescript",
+  "jsx",
+  "tsx",
+  "html",
+  "css",
+  "less",
+  "scss",
+  "sass",
+  "json",
+  "markdown",
+  "python",
+  "vue",
+  "svelte"
+]);
+const conventionLanguage = (element) => {
+  const classes = (0, import_hast_fast.classList)(element);
+  const brushAt = classes.findIndex((name) => name.startsWith("brush:"));
+  if (brushAt >= 0) {
+    const brush = classes[brushAt] === "brush:" ? classes[brushAt + 1] : classes[brushAt].slice("brush:".length);
+    const lang = brush?.toLowerCase();
+    return lang && LANGUAGE_TOKEN.test(lang) && !NO_LANGUAGE.has(lang) ? lang : void 0;
+  }
+  const sandpack = classes.find((name) => name.startsWith("sp-") && SANDPACK_LANGUAGES.has(name.slice(3)));
+  return sandpack?.slice(3);
+};
+const codeLanguage = (...elements) => elements.flatMap(import_hast_fast.classList).map((className) => LANGUAGE_MATCH_REGEX.map((regex) => className.match(regex)).find(Boolean)?.[1]).find(Boolean) ?? elements.map(attributeLanguage).find(Boolean) ?? elements.map(conventionLanguage).find(Boolean);
 const codeMeta = (pre) => {
   const title = pre.properties?.dataTitle;
   if (typeof title !== "string") {

package/dist/cjs/normalize/index.js

@@ -20,11 +20,13 @@ var normalize_exports = {};
 __export(normalize_exports, {
   bestFromSrcset: () => bestFromSrcset,
   dedupeRenderedMath: () => dedupeRenderedMath,
+  foldCodeLabels: () => foldCodeLabels,
   normalizeHast: () => normalizeHast,
   promoteAriaHeadings: () => promoteAriaHeadings,
   removeIconFonts: () => removeIconFonts,
   resolveImageSources: () => resolveImageSources,
-  separateAdjacentInlines: () => separateAdjacentInlines
+  separateAdjacentInlines: () => separateAdjacentInlines,
+  unwrapHeadingAnchors: () => unwrapHeadingAnchors
 });
 module.exports = __toCommonJS(normalize_exports);
 var import_hast_fast = require("../utils/hast-fast");
@@ -164,6 +166,82 @@ const removeIconFonts = (tree) => {
     (0, import_hast_fast.pruneInPlace)(tree, (node) => !isIcon(node));
   }
 };
+const textOfNode = (node) => node.type === "text" ? node.value : "children" in node ? node.children.map(textOfNode).join("") : "";
+const unwrapHeadingAnchors = (tree) => {
+  (0, import_hast_fast.walk)(tree, (node) => {
+    if (!((0, import_hast_fast.isElement)(node) && /^h[1-6]$/.test(node.tagName))) {
+      return;
+    }
+    const title = textOfNode(node).trim();
+    if (title === "") {
+      return;
+    }
+    const anchorAt = node.children.findIndex(
+      (child) => (0, import_hast_fast.isElement)(child) && child.tagName === "a" && (0, import_hast_fast.stringProperty)(child, "href")?.startsWith("#") && textOfNode(child).trim() === title
+    );
+    const anchor = node.children[anchorAt];
+    if ((0, import_hast_fast.isElement)(anchor)) {
+      node.children.splice(anchorAt, 1, ...anchor.children);
+    }
+  });
+};
+const LABEL_LANGUAGES = /* @__PURE__ */ new Set([
+  "js",
+  "javascript",
+  "ts",
+  "typescript",
+  "jsx",
+  "tsx",
+  "html",
+  "css",
+  "scss",
+  "json",
+  "bash",
+  "sh",
+  "shell",
+  "console",
+  "python",
+  "py",
+  "sql",
+  "xml",
+  "yaml",
+  "toml",
+  "http",
+  "rust",
+  "go",
+  "java",
+  "kotlin",
+  "swift",
+  "c",
+  "cpp",
+  "md",
+  "markdown",
+  "diff"
+]);
+const LABEL_NO_LANGUAGE = /* @__PURE__ */ new Set(["plain", "plaintext", "text"]);
+const foldCodeLabels = (tree) => {
+  (0, import_hast_fast.walk)(tree, (node) => {
+    if (!(0, import_hast_fast.isElement)(node) || node.tagName === "pre") {
+      return;
+    }
+    const elements = node.children.filter(import_hast_fast.isElement);
+    if (elements.length !== 2 || node.children.some((child) => child.type === "text" && child.value.trim() !== "")) {
+      return;
+    }
+    const [label, pre] = elements;
+    if (pre.tagName !== "pre" || /^(h[1-6]|pre)$/.test(label.tagName) || (0, import_hast_fast.findElement)(label, (el) => el.tagName === "pre")) {
+      return;
+    }
+    const name = textOfNode(label).trim().toLowerCase();
+    if (!(LABEL_LANGUAGES.has(name) || LABEL_NO_LANGUAGE.has(name))) {
+      return;
+    }
+    if (LABEL_LANGUAGES.has(name) && pre.properties.dataLanguage === void 0) {
+      pre.properties.dataLanguage = name;
+    }
+    node.children = node.children.filter((child) => child !== label);
+  });
+};
 const SPACED_INLINE = /* @__PURE__ */ new Set(["span", "time", "label"]);
 const SCRIPTS = [
   ["latin", /\p{Script=Latin}/u],
@@ -180,7 +258,6 @@ const needsSpace = (left, right) => {
   return a !== void 0 && b !== void 0 && a !== b;
 };
 const NO_SPACING = /* @__PURE__ */ new Set(["pre", "code", "kbd", "samp", "ruby", "math", "svg"]);
-const textOfNode = (node) => node.type === "text" ? node.value : "children" in node ? node.children.map(textOfNode).join("") : "";
 const separateAdjacentInlines = (tree) => {
   const visit = (node) => {
     if (!("children" in node) || (0, import_hast_fast.isElement)(node) && NO_SPACING.has(node.tagName)) {
@@ -201,7 +278,15 @@ const separateAdjacentInlines = (tree) => {
   visit(tree);
 };
 const normalizeHast = (tree, options = {}) => {
-  const { images = true, math = true, ariaHeadings = true, iconFonts = true, inlineSpacing = true } = options;
+  const {
+    images = true,
+    math = true,
+    ariaHeadings = true,
+    iconFonts = true,
+    headingAnchors = true,
+    codeLabels = true,
+    inlineSpacing = true
+  } = options;
   if (images) {
     resolveImageSources(tree);
   }
@@ -214,6 +299,12 @@ const normalizeHast = (tree, options = {}) => {
   if (iconFonts) {
     removeIconFonts(tree);
   }
+  if (headingAnchors) {
+    unwrapHeadingAnchors(tree);
+  }
+  if (codeLabels) {
+    foldCodeLabels(tree);
+  }
   if (inlineSpacing) {
     separateAdjacentInlines(tree);
   }
@@ -223,9 +314,11 @@ const normalizeHast = (tree, options = {}) => {
 0 && (module.exports = {
   bestFromSrcset,
   dedupeRenderedMath,
+  foldCodeLabels,
   normalizeHast,
   promoteAriaHeadings,
   removeIconFonts,
   resolveImageSources,
-  separateAdjacentInlines
+  separateAdjacentInlines,
+  unwrapHeadingAnchors
 });

package/dist/cjs/utils/detect-code-lang.js

@@ -62,11 +62,10 @@ const languages = [
   ["json", [/\b(true|false|null|\{})\b|\"[^"]+\":/g, 10]],
   ["yaml", [/^(\s+)?[a-z][a-z0-9]*:/gim, 10]]
 ];
-const detectLanguage = (code) => {
-  return languages.map(
-    ([lang, ...features]) => [lang, features.reduce((acc, [match, score]) => acc + [...code.matchAll(match)].length * score, 0)]
-  ).filter(([_, score]) => score > 20).sort((a, b) => b[1] - a[1])[0]?.[0] || "plain";
-};
+const MIN_CONFIDENT_SCORE = 400;
+const detectLanguage = (code) => languages.map(
+  ([lang, ...features]) => [lang, features.reduce((acc, [match, score]) => acc + [...code.matchAll(match)].length * score, 0)]
+).filter(([_, score]) => score >= MIN_CONFIDENT_SCORE).sort((a, b) => b[1] - a[1])[0]?.[0];
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   detectLanguage

package/dist/cjs/utils/parse-html.js

@@ -26,6 +26,8 @@ __export(parse_html_exports, {
 module.exports = __toCommonJS(parse_html_exports);
 var import_hast_util_from_parse5 = require("hast-util-from-parse5");
 var import_parse5 = require("parse5");
+var import_youtube = require("../adapters/sites/youtube");
+var import_detect_client_shell = require("../detect-client-shell");
 const MAX_ELEMENT_DEPTH = 256;
 const isParent = (node) => "childNodes" in node;
 const LINE_ELEMENTS = /* @__PURE__ */ new Set([
@@ -92,12 +94,78 @@ const flattenDeep = (root, maxDepth) => {
     }
   }
 };
-const SCRIPT_OR_STYLE = /<(script|style)\b([^>]*)>[\s\S]*?<\/\1\s*>/gi;
+const SCRIPT_OR_STYLE_START = /<!--|<(script|style)(?=[\s/>])/gi;
+const CLOSING_TAG = {
+  script: /<\/script[\s/>]/gi,
+  style: /<\/style[\s/>]/gi
+};
+const startTagEnd = (html, from) => {
+  let index = from;
+  while (index < html.length) {
+    const char = html[index];
+    if (char === ">") {
+      return index + 1;
+    }
+    if (char === "=") {
+      let value = index + 1;
+      while (value < html.length && /\s/.test(html[value] ?? "")) {
+        value++;
+      }
+      const quote = html[value];
+      if (quote === '"' || quote === "'") {
+        const close = html.indexOf(quote, value + 1);
+        if (close < 0) {
+          return -1;
+        }
+        index = close + 1;
+        continue;
+      }
+      index = value;
+      continue;
+    }
+    index++;
+  }
+  return -1;
+};
+const isReadScript = (attributes, element) => /application\/ld\+json/i.test(attributes) || element.includes(import_youtube.PLAYER_RESPONSE_VARIABLE) || import_detect_client_shell.CHALLENGE_PATTERN.test(element);
 const LARGE_DOCUMENT_CHARS = 2e6;
-const stripScriptBodies = (html) => html.replace(
-  SCRIPT_OR_STYLE,
-  (match, tag, attributes) => /application\/ld\+json/i.test(attributes) ? match : `<${tag}${attributes}></${tag}>`
-);
+const stripScriptBodies = (html) => {
+  const out = [];
+  let copied = 0;
+  const start = new RegExp(SCRIPT_OR_STYLE_START);
+  for (let match = start.exec(html); match; match = start.exec(html)) {
+    if (match[1] === void 0) {
+      const commentEnd = html.indexOf("-->", match.index + 4);
+      if (commentEnd < 0) {
+        break;
+      }
+      start.lastIndex = commentEnd + 3;
+      continue;
+    }
+    const tag = match[1].toLowerCase();
+    const bodyStart = startTagEnd(html, match.index + match[0].length);
+    if (bodyStart < 0) {
+      break;
+    }
+    const closing = CLOSING_TAG[tag];
+    closing.lastIndex = bodyStart;
+    const close = closing.exec(html);
+    if (!close) {
+      break;
+    }
+    const tagEnd = html.indexOf(">", close.index);
+    const elementEnd = tagEnd < 0 ? html.length : tagEnd + 1;
+    const element = html.slice(match.index, elementEnd);
+    const openingTag = html.slice(match.index, bodyStart);
+    if (!(tag === "script" && isReadScript(openingTag, element))) {
+      out.push(html.slice(copied, bodyStart), html.slice(close.index, elementEnd));
+      copied = elementEnd;
+    }
+    start.lastIndex = elementEnd;
+  }
+  out.push(html.slice(copied));
+  return out.join("");
+};
 const parseHtml = (html, options = {}) => {
   const source = html.length > LARGE_DOCUMENT_CHARS ? stripScriptBodies(html) : html;
   const document = options.fragment ? (0, import_parse5.parseFragment)(source, { scriptingEnabled: false }) : (0, import_parse5.parse)(source, { scriptingEnabled: false });

package/dist/detect-client-shell.js

@@ -47,6 +47,7 @@ var detectClientShell = (html) => {
   return { isShell: false, visibleTextLength };
 };
 export {
+  CHALLENGE_PATTERN,
   SHELL_EMPTY_THRESHOLD,
   SHELL_SUSPICION_THRESHOLD,
   detectClientShell

package/dist/extractors/lib/layout-tables.js

@@ -4,10 +4,51 @@ var DATA_MARKERS = /* @__PURE__ */ new Set(["caption", "thead", "tfoot", "th", "
 var BLOCK_IN_CELL = /* @__PURE__ */ new Set(["h1", "h2", "h3", "h4", "h5", "h6", "table", "form"]);
 var TABLE_PARTS = /* @__PURE__ */ new Set(["tbody", "thead", "tfoot", "tr", "td", "th"]);
 var TABLE_CONTEXT = /* @__PURE__ */ new Set(["table", "tbody", "thead", "tfoot", "tr"]);
+var MAX_SMALL_TABLE_CELL = 150;
+var MAX_SMALL_TABLE_LINK_SHARE = 0.5;
+var collapsedLength = (text) => text.replace(/\s+/g, " ").trim().length;
+var cellText = (cell) => {
+  let text = "";
+  let linkText = "";
+  const visit = (node, inLink) => {
+    for (const child of node.children) {
+      if (child.type === "text") {
+        text += child.value;
+        if (inLink) {
+          linkText += child.value;
+        }
+      } else if (isElement(child) && child.tagName !== "table") {
+        visit(child, inLink || child.tagName === "a");
+      }
+    }
+  };
+  visit(cell, false);
+  return { text: collapsedLength(text), linkText: collapsedLength(linkText) };
+};
 var cellsOf = (row) => row.children.filter((cell) => isElement(cell) && (cell.tagName === "td" || cell.tagName === "th"));
 var HEADING = /^h[1-6]$/;
+var measureRow = (shape, row) => {
+  const cells = cellsOf(row);
+  shape.rows += 1;
+  shape.columns = Math.max(shape.columns, cells.length);
+  for (const cell of cells) {
+    const { text, linkText } = cellText(cell);
+    shape.longestCell = Math.max(shape.longestCell, text);
+    shape.text += text;
+    shape.linkText += linkText;
+  }
+  return cells.length === 1;
+};
 var shapeOf = (table) => {
-  const shape = { rows: 0, columns: 0, dataMarker: false, blockInCell: false };
+  const shape = {
+    rows: 0,
+    columns: 0,
+    dataMarker: false,
+    blockInCell: false,
+    longestCell: 0,
+    text: 0,
+    linkText: 0
+  };
   let captionHeadings = 0;
   const noteBlockInCell = (tag, aloneInRow) => {
     if (HEADING.test(tag) && aloneInRow) {
@@ -29,13 +70,7 @@ var shapeOf = (table) => {
       if (DATA_MARKERS.has(tag)) {
         shape.dataMarker = true;
       }
-      let alone = aloneInRow;
-      if (tag === "tr") {
-        shape.rows += 1;
-        const cells = cellsOf(child).length;
-        shape.columns = Math.max(shape.columns, cells);
-        alone = cells === 1;
-      }
+      const alone = tag === "tr" ? measureRow(shape, child) : aloneInRow;
       if (inCell && BLOCK_IN_CELL.has(tag)) {
         noteBlockInCell(tag, aloneInRow);
       }
@@ -66,7 +101,11 @@ var isLayoutTable = (table) => {
   if (shape.rows >= 10 || shape.columns > 4) {
     return false;
   }
-  return shape.rows * shape.columns <= 10;
+  if (shape.rows * shape.columns > 10) {
+    return false;
+  }
+  const linkShare = shape.text === 0 ? 1 : shape.linkText / shape.text;
+  return shape.longestCell > MAX_SMALL_TABLE_CELL || linkShare > MAX_SMALL_TABLE_LINK_SHARE;
 };
 var unwrap = (table) => {
   const cells = [];