@thallylabs/migrate 0.2.5 → 0.2.6

package/dist/index.d.ts

@@ -66,6 +66,10 @@ interface MigrationDocsConfig {
         display?: 'tabs' | 'dropdown';
     };
     theme?: 'default' | 'maple' | 'sharp' | 'minimal';
+    /** Icon set for content `icon` names; Mintlify's `icons.library` carries through. */
+    icons?: {
+        library?: 'lucide' | 'fontawesome' | 'tabler';
+    };
     banner?: MigrationBannerConfig;
     navbar?: MigrationNavbarConfig;
     footer?: MigrationFooterConfig;

package/dist/index.js

@@ -1104,8 +1104,30 @@ function normalizeDocusaurusAdmonitions(body) {
     return line;
   }).join("\n");
 }
+var GLOBAL_DOCUSARUS_COMPONENTS = /* @__PURE__ */ new Set([
+  "Tabs",
+  "TabItem",
+  "Link",
+  "DocCardList",
+  "TOCInline"
+]);
+function isGlobalDocusaurusImport(line) {
+  const trimmed = line.trim().replace(/;$/, "").trimEnd();
+  if (!trimmed.startsWith("import ")) return false;
+  const separator = trimmed.indexOf(" from ", "import ".length);
+  if (separator < 0) return false;
+  const component = trimmed.slice("import ".length, separator).trim();
+  if (!GLOBAL_DOCUSARUS_COMPONENTS.has(component)) return false;
+  const source = trimmed.slice(separator + " from ".length);
+  if (source.length < 3) return false;
+  const quote = source[0];
+  if (quote !== "'" && quote !== '"' || source.at(-1) !== quote) return false;
+  const moduleName = source.slice(1, -1);
+  return moduleName.startsWith("@theme/") || moduleName.startsWith("@docusaurus/");
+}
 function normalizeMdx(body) {
-  return normalizeDocusaurusAdmonitions(body).replace(/^import\s+(?:Tabs|TabItem|Link|DocCardList|TOCInline)\s+from\s+['"]@(?:theme|docusaurus)\/[^'"]+['"]\s*;?\s*$/gm, "").replace(/<TabItem\b([^>]*)>/g, (_match, attributes) => {
+  const withoutGlobalImports = body.split("\n").filter((line) => !isGlobalDocusaurusImport(line)).join("\n");
+  return normalizeDocusaurusAdmonitions(withoutGlobalImports).replace(/<TabItem\b([^>]*)>/g, (_match, attributes) => {
     const title = attributes.match(/\blabel=(?:"([^"]*)"|'([^']*)')/)?.slice(1).find(Boolean) ?? attributes.match(/\bvalue=(?:"([^"]*)"|'([^']*)')/)?.slice(1).find(Boolean) ?? "Tab";
     return `<Tab title="${title.replace(/"/g, "&quot;")}">`;
   }).replace(/<\/TabItem>/g, "</Tab>").replace(/<Link\b([^>]*)\bto=(?:"([^"]*)"|'([^']*)')([^>]*)>/g, (_match, before, doubleQuoted, singleQuoted, after) => `<a${before}href="${doubleQuoted ?? singleQuoted}"${after}>`).replace(/<\/Link>/g, "</a>").replace(/<(?:DocCardList|TOCInline)\b[^>]*\/>/g, "").replace(/<!--([\s\S]*?)-->/g, (_match, content) => `{/*${content}*/}`).replace(/<Danger(\s[^>]*)?>/g, "<Error$1>").replace(/<\/Danger>/g, "</Error>").replace(/<Warn(\s[^>]*)?>/g, "<Warning$1>").replace(/<\/Warn>/g, "</Warning>").replace(/<Check(\s[^>]*)?>/g, "<Note$1>").replace(/<\/Check>/g, "</Note>").replace(/<Tree\.Folder/g, "<Folder").replace(/<\/Tree\.Folder>/g, "</Folder>").replace(/<Tree\.File/g, "<File").replace(/<\/Tree\.File>/g, "</File>");
@@ -1571,6 +1593,8 @@ function projectedCompatibleConfig(config) {
   const headingFont = projectedFont(objectValue2(config.fonts)?.heading);
   const feedback = objectValue2(config.feedback);
   const seo = objectValue2(config.seo);
+  const iconLibrary = objectValue2(config.icons)?.library;
+  const projectedIconLibrary = ["lucide", "fontawesome", "tabler"].includes(String(iconLibrary)) ? String(iconLibrary) : void 0;
   const navbar = projectedNavbar(config.navbar);
   const globalLinks = projectedGlobalNavigationLinks(objectValue2(config.navigation)?.global);
   const navbarLinks = [...new Map(
@@ -1579,6 +1603,7 @@ function projectedCompatibleConfig(config) {
   const projectedNavigation = navbarLinks.length > 0 || navbar?.primary ? { ...navbarLinks.length > 0 ? { links: navbarLinks } : {}, ...navbar?.primary ? { primary: navbar.primary } : {} } : void 0;
   return {
     ...projectedTheme(config.theme) ? { theme: projectedTheme(config.theme) } : {},
+    ...projectedIconLibrary ? { icons: { library: projectedIconLibrary } } : {},
     ...bannerContent ? {
       banner: {
         content: bannerContent,
@@ -2670,7 +2695,10 @@ function renderMigrationFiles(bundle, options = {}) {
 
 // src/url.ts
 import { load } from "cheerio";
+import remarkMdx3 from "remark-mdx";
+import remarkParse3 from "remark-parse";
 import TurndownService from "turndown";
+import { unified as unified3 } from "unified";
 import { parse as parseYaml3, stringify as stringifyYaml } from "yaml";
 var DEFAULT_MAX_PAGES = 1e3;
 var DEFAULT_MAX_TOTAL_BYTES = 1e8;
@@ -2773,6 +2801,8 @@ var LOCALE_CODES = /* @__PURE__ */ new Set([
   "zh-hant"
 ]);
 var PORTABLE_URL_PROPS = "href|src|img|primaryHref|secondaryHref";
+var PORTABLE_URL_PROP_NAMES = new Set(PORTABLE_URL_PROPS.split("|"));
+var remoteMarkdownParser = unified3().use(remarkParse3).use(remarkMdx3);
 function validateMigrationUrl(value) {
   let url;
   try {
@@ -2902,7 +2932,7 @@ function detectUrlPlatform(document) {
   const value = `${document.body.slice(0, 2e5)} ${headers}`.toLowerCase();
   if (value.includes("__mintlify") || value.includes("/mintlify-assets/") || value.includes("x-mintlify-") || value.includes("/_mintlify/")) return "mintlify";
   if (value.includes("docusaurus") || value.includes("__docusaurus")) return "docusaurus";
-  if (value.includes("gitbook") || value.includes("gitbook.io")) return "gitbook";
+  if (value.includes("gitbook")) return "gitbook";
   if (value.includes("nextra")) return "nextra";
   if (value.includes("vitepress")) return "vitepress";
   if (value.includes("starlight")) return "starlight";
@@ -3041,19 +3071,184 @@ function stripEventHandlerProps(line) {
   }
   return cursor === 0 ? line : `${result}${line.slice(cursor)}`;
 }
-function hasUnsafeUrlScheme(value) {
-  const normalized = value.trim().replace(/[\u0000-\u0020]/g, "").replace(/&(?:#x0*3a|#0*58|colon);/gi, ":");
-  return /^(?:javascript|data|vbscript):/i.test(normalized);
+function isSafePortableUrl(value) {
+  try {
+    const parsed = new URL(value, "https://relative.thally.invalid");
+    return ["http:", "https:", "mailto:", "tel:"].includes(parsed.protocol);
+  } catch {
+    return false;
+  }
+}
+function decodeHtmlAttribute(value) {
+  const document = load(`<i data-thally-value="${value.replace(/"/g, "&quot;")}"></i>`);
+  return document("i").attr("data-thally-value") ?? "";
+}
+function escapeMarkdownText(value, escapeTableSeparator = false) {
+  let escaped = "";
+  for (const character of value) {
+    if (character === "\\") escaped += "\\\\";
+    else if (character === "[" || character === "]") escaped += `\\${character}`;
+    else if (character === "<") escaped += "&lt;";
+    else if (character === ">") escaped += "&gt;";
+    else if (character === "{") escaped += "&#123;";
+    else if (character === "}") escaped += "&#125;";
+    else if (character === "|" && escapeTableSeparator) escaped += "\\|";
+    else escaped += character;
+  }
+  return escaped;
+}
+function markdownLinkDestination(value) {
+  try {
+    return encodeURI(value).replace(/\(/g, "%28").replace(/\)/g, "%29").replace(/&/g, "&amp;");
+  } catch {
+    return null;
+  }
+}
+function htmlOpeningTagEnd(value, start) {
+  let quote = "";
+  for (let index = start; index < value.length; index += 1) {
+    const character = value[index];
+    if (quote) {
+      if (character === quote) quote = "";
+      continue;
+    }
+    if (character === '"' || character === "'") {
+      quote = character;
+      continue;
+    }
+    if (character === ">") return index + 1;
+  }
+  return null;
+}
+function replacePortableHtmlImages(value) {
+  const matcher = /<img\b/gi;
+  let cursor = 0;
+  let output = "";
+  for (const match of value.matchAll(matcher)) {
+    const start = match.index;
+    if (start < cursor) continue;
+    const end = htmlOpeningTagEnd(value, start + match[0].length);
+    if (end === null) continue;
+    const fragment = load(value.slice(start, end));
+    const image = fragment("img").first();
+    const source = image.attr("src");
+    const destination = source && isSafePortableUrl(source) ? markdownLinkDestination(source) : null;
+    const alternative = escapeMarkdownText(image.attr("alt") ?? "");
+    output += value.slice(cursor, start);
+    output += destination ? `![${alternative}](${destination})` : alternative;
+    cursor = end;
+  }
+  return cursor === 0 ? value : `${output}${value.slice(cursor)}`;
+}
+function replacePortableHtmlAnchors(value) {
+  const matcher = /<a\b/gi;
+  let cursor = 0;
+  let output = "";
+  for (const match of value.matchAll(matcher)) {
+    const start = match.index;
+    if (start < cursor) continue;
+    const openingEnd = htmlOpeningTagEnd(value, start + match[0].length);
+    if (openingEnd === null) continue;
+    const closingMatch = /<\/a\s*>/gi;
+    closingMatch.lastIndex = openingEnd;
+    const closing = closingMatch.exec(value);
+    if (!closing) continue;
+    const end = closing.index + closing[0].length;
+    const fragment = load(value.slice(start, end));
+    const anchor = fragment("a").first();
+    const href = anchor.attr("href");
+    const label = escapeMarkdownText(anchor.text().replace(/\s+/g, " ").trim());
+    const destination = href && isSafePortableUrl(href) ? markdownLinkDestination(href) : null;
+    output += value.slice(cursor, start);
+    output += destination ? `[${label}](${destination})` : label;
+    cursor = end;
+  }
+  return cursor === 0 ? value : `${output}${value.slice(cursor)}`;
+}
+function markdownNodeText(node) {
+  if (node.type === "text" || node.type === "inlineCode") return node.value ?? "";
+  if (node.type === "image" || node.type === "imageReference") return node.alt ?? "";
+  return (node.children ?? []).map(markdownNodeText).join("");
+}
+function sanitizeMarkdownUrls(value) {
+  let root;
+  try {
+    root = remoteMarkdownParser.parse(value);
+  } catch {
+    return null;
+  }
+  let hasUnsafeMdxSyntax = false;
+  const validateMdxSyntax = (node) => {
+    if (node.type === "mdxTextExpression" || node.type === "mdxFlowExpression" || node.type === "mdxjsEsm") {
+      hasUnsafeMdxSyntax = true;
+      return;
+    }
+    if (node.type === "mdxJsxTextElement" || node.type === "mdxJsxFlowElement") {
+      for (const attribute of node.attributes ?? []) {
+        if (attribute.type !== "mdxJsxAttribute" || !attribute.name || /^on[A-Z]/.test(attribute.name)) {
+          hasUnsafeMdxSyntax = true;
+          return;
+        }
+        let staticValue = attribute.value;
+        if (attribute.value && typeof attribute.value === "object") {
+          if (attribute.value.type !== "mdxJsxAttributeValueExpression") {
+            hasUnsafeMdxSyntax = true;
+            return;
+          }
+          try {
+            staticValue = JSON.parse(attribute.value.value);
+          } catch {
+            hasUnsafeMdxSyntax = true;
+            return;
+          }
+        }
+        if (PORTABLE_URL_PROP_NAMES.has(attribute.name) && (typeof staticValue !== "string" || !isSafePortableUrl(decodeHtmlAttribute(staticValue)))) {
+          hasUnsafeMdxSyntax = true;
+          return;
+        }
+      }
+    }
+    for (const child of node.children ?? []) validateMdxSyntax(child);
+  };
+  validateMdxSyntax(root);
+  if (hasUnsafeMdxSyntax) return null;
+  const unsafeDefinitions = /* @__PURE__ */ new Set();
+  const visitDefinitions = (node) => {
+    if (node.type === "definition" && node.identifier && node.url && !isSafePortableUrl(node.url)) unsafeDefinitions.add(node.identifier.toLowerCase());
+    for (const child of node.children ?? []) visitDefinitions(child);
+  };
+  visitDefinitions(root);
+  const replacements = [];
+  const replaceNode = (node, replacement) => {
+    const start = node.position?.start.offset;
+    const end = node.position?.end.offset;
+    if (start === void 0 || end === void 0) return false;
+    replacements.push({ start, end, value: replacement });
+    return true;
+  };
+  const visitUrls = (node) => {
+    const hasUnsafeDirectUrl = (node.type === "link" || node.type === "image") && typeof node.url === "string" && !isSafePortableUrl(node.url);
+    const hasUnsafeReference = (node.type === "linkReference" || node.type === "imageReference") && typeof node.identifier === "string" && unsafeDefinitions.has(node.identifier.toLowerCase());
+    if (hasUnsafeDirectUrl || hasUnsafeReference) {
+      if (replaceNode(node, escapeMarkdownText(markdownNodeText(node)))) return;
+    }
+    if (node.type === "definition" && node.identifier && unsafeDefinitions.has(node.identifier.toLowerCase())) {
+      if (replaceNode(node, "")) return;
+    }
+    for (const child of node.children ?? []) visitUrls(child);
+  };
+  visitUrls(root);
+  return replacements.sort((left, right) => right.start - left.start).reduce((result, replacement) => `${result.slice(0, replacement.start)}${replacement.value}${result.slice(replacement.end)}`, value);
 }
 function hasUnsafePortableUrlProp(value) {
   const quotedProps = new RegExp(`\\b(?:${PORTABLE_URL_PROPS})\\s*=\\s*(['"])([^'"\\r\\n]*)\\1`, "gi");
   for (const match of value.matchAll(quotedProps)) {
-    if (hasUnsafeUrlScheme(match[2])) return true;
+    if (!isSafePortableUrl(decodeHtmlAttribute(match[2]))) return true;
   }
   const expressionProps = new RegExp(`\\b(?:${PORTABLE_URL_PROPS})\\s*=\\s*\\{\\s*("(?:\\\\.|[^"\\\\])*")\\s*\\}`, "gi");
   for (const match of value.matchAll(expressionProps)) {
     try {
-      if (hasUnsafeUrlScheme(JSON.parse(match[1]))) return true;
+      if (!isSafePortableUrl(decodeHtmlAttribute(JSON.parse(match[1])))) return true;
     } catch {
       return true;
     }
@@ -3061,12 +3256,7 @@ function hasUnsafePortableUrlProp(value) {
   return false;
 }
 function portableHtmlToMarkdown(value) {
-  return value.replace(/<span\b[^>]*>|<\/span>/gi, "").replace(/<a\b([^>]*)>([\s\S]*?)<\/a>/gi, (_original, attributes, label) => {
-    const href = attributes.match(/\bhref\s*=\s*(['"])(.*?)\1/i)?.[2];
-    const plainLabel = label.replace(/<[^>]+>/g, "").trim();
-    if (!href || hasUnsafeUrlScheme(href)) return plainLabel;
-    return `[${plainLabel.replace(/]/g, "\\]")}](${href})`;
-  }).replace(/<h([1-6])\b[^>]*>([\s\S]*?)<\/h\1>/gi, (_original, level, content) => {
+  return replacePortableHtmlAnchors(replacePortableHtmlImages(value)).replace(/<span\b[^>]*>|<\/span>/gi, "").replace(/<h([1-6])\b[^>]*>([\s\S]*?)<\/h\1>/gi, (_original, level, content) => {
     return `
 ${"#".repeat(Number(level))} ${content.trim()}
 `;
@@ -3123,16 +3313,11 @@ function sanitizeRemoteMarkdown(body) {
       return line;
     }
     if (fenceCharacter) return line;
-    return portableHtmlToMarkdown(stripEventHandlerProps(line.replace(/!\[([^\]]*)\]\(\s*([^)]+)\)/g, (original, alternative, destination) => {
-      return hasUnsafeUrlScheme(destination.replace(/^<|>$/g, "")) ? alternative : original;
-    }).replace(/<img\s+([^>]*?)\/?\s*>/gi, (original, attributes) => {
-      const source = attributes.match(/\bsrc\s*=\s*(['"])(.*?)\1/i)?.[2];
-      const alternative = attributes.match(/\balt\s*=\s*(['"])(.*?)\1/i)?.[2] ?? "";
-      if (!source || hasUnsafeUrlScheme(source)) return "";
-      return `![${alternative.replace(/]/g, "\\]")}](${source})`;
-    })));
+    return portableHtmlToMarkdown(stripEventHandlerProps(line.replace(/!\[([^\]]*)\]\(\s*([^)]+)\)/g, (original, alternative, destination) => isSafePortableUrl(decodeHtmlAttribute(destination.replace(/^<|>$/g, ""))) ? original : escapeMarkdownText(alternative))));
   }).join("\n");
-  const executable = withoutFencedCode(sanitized).replace(/\{\/\*[\s\S]*?\*\/\}/g, "").replace(/<(?:https?:\/\/|mailto:)[^>]+>/gi, "").replace(/(`+)[\s\S]*?\1/g, "");
+  const urlSafeMarkdown = sanitizeMarkdownUrls(sanitized);
+  if (urlSafeMarkdown === null) return null;
+  const executable = withoutFencedCode(urlSafeMarkdown).replace(/\{\/\*[\s\S]*?\*\/\}/g, "").replace(/<(?:https?:\/\/|mailto:)[^>]+>/gi, "").replace(/(`+)[\s\S]*?\1/g, "");
   if (/^\s*(?:import|export)\s/m.test(executable)) return null;
   let hasUnsafeExpression = false;
   const withoutStaticProps = executable.replace(/=\{([^{}\n]*)\}/g, (original, value) => {
@@ -3149,7 +3334,7 @@ function sanitizeRemoteMarkdown(body) {
     const component = match[1].split(".", 1)[0];
     if (!PORTABLE_MDX_COMPONENTS.has(component)) return null;
   }
-  return sanitized;
+  return urlSafeMarkdown;
 }
 function migratedHref(href, currentUrl, source, scopePath, importedIds, sourceHomePageId) {
   if (/^(?:javascript|data|vbscript):/i.test(href)) return "#";
@@ -3436,7 +3621,10 @@ ${tabs.join("\n\n")}
     filter: "table",
     replacement: (_content, node) => {
       const fragment = load(node.outerHTML);
-      const rows = fragment("tr").toArray().map((row) => fragment(row).children("th,td").toArray().map((cell) => fragment(cell).text().replace(/\s+/g, " ").replace(/\|/g, "\\|").replace(/</g, "&lt;").replace(/>/g, "&gt;").replace(/\{/g, "&#123;").replace(/\}/g, "&#125;").trim())).filter((row) => row.length > 0);
+      const rows = fragment("tr").toArray().map((row) => fragment(row).children("th,td").toArray().map((cell) => escapeMarkdownText(
+        fragment(cell).text().replace(/\s+/g, " ").trim(),
+        true
+      ))).filter((row) => row.length > 0);
       if (rows.length === 0) return "";
       const width = Math.max(...rows.map((row) => row.length));
       const normalized = rows.map((row) => [...row, ...Array(width - row.length).fill("")]);