@mcp-b/smart-dom-reader-server 5.1.0 → 6.0.0

package/lib/smart-dom-reader.bundle.js

@@ -128,30 +128,20 @@ var ContentDetection = class ContentDetection {
    * Detect page landmarks
    */
   static detectLandmarks(doc) {
-    const landmarks = {
-      navigation: [],
-      main: [],
-      complementary: [],
-      contentinfo: [],
-      banner: [],
-      search: [],
-      form: [],
-      region: [],
+    return {
+      navigation: Array.from(doc.querySelectorAll('nav, [role="navigation"]')),
+      main: Array.from(doc.querySelectorAll('main, [role="main"]')),
+      complementary: Array.from(doc.querySelectorAll('aside, [role="complementary"]')),
+      contentinfo: Array.from(doc.querySelectorAll('footer, [role="contentinfo"]')),
+      banner: Array.from(doc.querySelectorAll('header, [role="banner"]')),
+      search: Array.from(doc.querySelectorAll('[role="search"]')),
+      form: Array.from(
+        doc.querySelectorAll('form[aria-label], form[aria-labelledby], [role="form"]')
+      ),
+      region: Array.from(
+        doc.querySelectorAll('section[aria-label], section[aria-labelledby], [role="region"]')
+      ),
     };
-    for (const [landmark, selector] of Object.entries({
-      navigation: 'nav, [role="navigation"]',
-      main: 'main, [role="main"]',
-      complementary: 'aside, [role="complementary"]',
-      contentinfo: 'footer, [role="contentinfo"]',
-      banner: 'header, [role="banner"]',
-      search: '[role="search"]',
-      form: 'form[aria-label], form[aria-labelledby], [role="form"]',
-      region: 'section[aria-label], section[aria-labelledby], [role="region"]',
-    })) {
-      const elements = doc.querySelectorAll(selector);
-      landmarks[landmark] = Array.from(elements);
-    }
-    return landmarks;
   }
 };
 //#endregion
@@ -243,8 +233,7 @@ function selectorQualitySummary(inter) {
 }
 function renderInteractive(inter, opts) {
   const parts = [];
-  const limit = (arr) =>
-    typeof opts?.maxElements === 'number' ? arr.slice(0, opts.maxElements) : arr;
+  const limit = (arr) => arr.slice(0, opts?.maxElements ?? arr.length);
   if (inter.buttons.length) {
     parts.push('Buttons:');
     for (const el of limit(inter.buttons)) parts.push(elementLine(el, opts));
@@ -377,8 +366,7 @@ var MarkdownFormatter = class {
       lines.push('');
     }
     if (content.text.paragraphs?.length) {
-      const limit =
-        typeof opts.maxElements === 'number' ? opts.maxElements : content.text.paragraphs.length;
+      const limit = opts.maxElements ?? content.text.paragraphs.length;
       lines.push('Paragraphs:');
       for (const p of content.text.paragraphs.slice(0, limit))
         lines.push(`- ${truncate(p, opts.maxTextLength ?? 200)}`);
@@ -388,7 +376,7 @@ var MarkdownFormatter = class {
       lines.push('Lists:');
       for (const list of content.text.lists) {
         lines.push(`- ${list.type.toUpperCase()}:`);
-        const limit = typeof opts.maxElements === 'number' ? opts.maxElements : list.items.length;
+        const limit = opts.maxElements ?? list.items.length;
         for (const item of list.items.slice(0, limit))
           lines.push(`  - ${truncate(item, opts.maxTextLength ?? 120)}`);
       }
@@ -398,14 +386,14 @@ var MarkdownFormatter = class {
       lines.push('Tables:');
       for (const t of content.tables) {
         lines.push(`- Headers: ${t.headers.join(' | ')}`);
-        const limit = typeof opts.maxElements === 'number' ? opts.maxElements : t.rows.length;
+        const limit = opts.maxElements ?? t.rows.length;
         for (const row of t.rows.slice(0, limit)) lines.push(`  - ${row.join(' | ')}`);
       }
       lines.push('');
     }
     if (content.media?.length) {
       lines.push('Media:');
-      const limit = typeof opts.maxElements === 'number' ? opts.maxElements : content.media.length;
+      const limit = opts.maxElements ?? content.media.length;
       for (const m of content.media.slice(0, limit))
         lines.push(`- ${m.type.toUpperCase()}: ${m.alt ?? ''} ${m.src ? `→ ${m.src}` : ''}`.trim());
       lines.push('');
@@ -647,8 +635,8 @@ var SelectorGenerator = class SelectorGenerator {
       const tag = current.nodeName.toLowerCase();
       let descriptor = tag;
       if (current.id) descriptor = `${tag}#${current.id}`;
-      else if (current.className && typeof current.className === 'string') {
-        const firstClass = current.className.split(' ')[0];
+      else {
+        const firstClass = current.getAttribute('class')?.split(' ')[0];
         if (firstClass) descriptor = `${tag}.${firstClass}`;
       }
       const role = current.getAttribute('role');
@@ -743,11 +731,9 @@ var DOMTraversal = class DOMTraversal {
       for (const [attr, value] of Object.entries(filter.attributeValues)) {
         const attrValue = element.getAttribute(attr);
         if (!attrValue) return false;
-        if (typeof value === 'string') {
-          if (attrValue !== value) return false;
-        } else if (value instanceof RegExp) {
+        if (value instanceof RegExp) {
           if (!value.test(attrValue)) return false;
-        }
+        } else if (attrValue !== value) return false;
       }
     if (filter.withinSelectors?.length) {
       let isWithin = false;
@@ -884,21 +870,25 @@ var DOMTraversal = class DOMTraversal {
    * Get interaction information for an element (compact format)
    */
   static getInteractionInfo(element) {
-    const htmlElement = element;
     const interaction = {};
+    const view = element.ownerDocument.defaultView ?? window;
     if (
-      htmlElement.onclick ||
+      ('onclick' in element && element.onclick) ||
       element.getAttribute('onclick') ||
       element.matches('button, a[href], [role="button"], [tabindex]:not([tabindex="-1"])')
     )
       interaction.click = true;
     if (
-      htmlElement.onchange ||
+      ('onchange' in element && element.onchange) ||
       element.getAttribute('onchange') ||
       element.matches('input, select, textarea')
     )
       interaction.change = true;
-    if (htmlElement.onsubmit || element.getAttribute('onsubmit') || element.matches('form'))
+    if (
+      ('onsubmit' in element && element.onsubmit) ||
+      element.getAttribute('onsubmit') ||
+      element.matches('form')
+    )
       interaction.submit = true;
     if (element.matches('a[href], button[type="submit"]')) interaction.nav = true;
     if (element.hasAttribute('disabled') || element.getAttribute('aria-disabled') === 'true')
@@ -907,7 +897,13 @@ var DOMTraversal = class DOMTraversal {
     const ariaRole = element.getAttribute('role');
     if (ariaRole) interaction.role = ariaRole;
     if (element.matches('input, textarea, select, button')) {
-      const form = element.form || element.closest('form');
+      const form =
+        (element instanceof view.HTMLInputElement ||
+        element instanceof view.HTMLTextAreaElement ||
+        element instanceof view.HTMLSelectElement ||
+        element instanceof view.HTMLButtonElement
+          ? element.form
+          : null) || element.closest('form');
       if (form) interaction.form = SelectorGenerator.generateSelectors(form).css;
     }
     return interaction;
@@ -916,11 +912,13 @@ var DOMTraversal = class DOMTraversal {
    * Get text content of an element (limited length)
    */
   static getElementText(element, options) {
-    if (element.matches('input, textarea')) {
-      const input = element;
-      return input.value || input.placeholder || '';
-    }
-    if (element.matches('img')) return element.alt || '';
+    const view = element.ownerDocument.defaultView ?? window;
+    if (element.matches('input, textarea'))
+      return element instanceof view.HTMLInputElement || element instanceof view.HTMLTextAreaElement
+        ? element.value || element.placeholder || ''
+        : '';
+    if (element.matches('img'))
+      return element instanceof view.HTMLImageElement ? element.alt || '' : '';
     const text = element.textContent?.trim() || '';
     const maxLength = options?.textTruncateLength;
     if (maxLength && text.length > maxLength) return `${text.substring(0, maxLength)}...`;
@@ -1031,13 +1029,13 @@ var SmartDOMReader = class SmartDOMReader {
   extractLandmarks(doc) {
     const detected = ContentDetection.detectLandmarks(doc);
     return {
-      navigation: this.elementsToSelectors(detected.navigation || []),
-      main: this.elementsToSelectors(detected.main || []),
-      forms: this.elementsToSelectors(detected.form || []),
-      headers: this.elementsToSelectors(detected.banner || []),
-      footers: this.elementsToSelectors(detected.contentinfo || []),
-      articles: this.elementsToSelectors(detected.region || []),
-      sections: this.elementsToSelectors(detected.region || []),
+      navigation: this.elementsToSelectors(detected.navigation),
+      main: this.elementsToSelectors(detected.main),
+      forms: this.elementsToSelectors(detected.form),
+      headers: this.elementsToSelectors(detected.banner),
+      footers: this.elementsToSelectors(detected.contentinfo),
+      articles: this.elementsToSelectors(detected.region),
+      sections: this.elementsToSelectors(detected.region),
     };
   }
   /**
@@ -1321,7 +1319,7 @@ var ProgressiveExtractor = class ProgressiveExtractor {
     if (options.includeLists !== false) {
       const lists = element.querySelectorAll('ul, ol');
       result.text.lists = Array.from(lists).map((list) => ({
-        type: list.tagName.toLowerCase(),
+        type: list.tagName.toLowerCase() === 'ul' ? 'ul' : 'ol',
         items: Array.from(list.querySelectorAll('li')).map((li) =>
           ProgressiveExtractor.getTextContent(li, options.maxTextLength)
         ),

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mcp-b/smart-dom-reader-server",
-  "version": "5.1.0",
+  "version": "6.0.0",
   "description": "MCP server for smart-dom-reader backed by Playwright",
   "license": "MIT",
   "repository": {
@@ -26,7 +26,7 @@
   "devDependencies": {
     "@types/node": "22.20.1",
     "typescript": "^5.8.3",
-    "@mcp-b/smart-dom-reader": "5.1.0"
+    "@mcp-b/smart-dom-reader": "6.0.0"
   },
   "engines": {
     "node": ">=20"

package/src/index.ts

@@ -19,8 +19,9 @@ const { F_OK } = fsConstants;
 
 const __filename = fileURLToPath(import.meta.url);
 const __dirname = dirname(__filename);
-const packageVersion = (createRequire(import.meta.url)('../package.json') as { version: string })
-  .version;
+const packageVersion = z
+  .object({ version: z.string() })
+  .parse(createRequire(import.meta.url)('../package.json')).version;
 
 const EMBEDDED_LIBRARY_RELATIVE_PATH = join('..', 'lib', 'smart-dom-reader.bundle.js');
 const DEFAULT_LAUNCH_ARGS = ['--no-sandbox', '--disable-setuid-sandbox'] as const;
@@ -61,7 +62,11 @@ type ScreenshotArgs = {
   fullPage: boolean;
 };
 
-type LibraryOperation = 'structure' | 'region' | 'content' | 'interactive';
+type LibraryRequest =
+  | [operation: 'structure', args: StructureOperationArgs & { format: FormatOptions }]
+  | [operation: 'region', args: RegionOperationArgs & { format: FormatOptions }]
+  | [operation: 'content', args: ContentOperationArgs & { format: FormatOptions }]
+  | [operation: 'interactive', args: InteractiveOperationArgs & { format: FormatOptions }];
 
 type StructureOperationArgs = {
   selector: string | null;
@@ -84,32 +89,41 @@ type InteractiveOperationArgs = {
 
 const CHROME_ENV_VARS = ['CHROME_PATH', 'GOOGLE_CHROME_SHIM', 'BROWSER_PATH'] as const;
 
-const DEFAULT_CHROME_LOCATIONS: Partial<Record<NodeJS.Platform, readonly string[]>> = {
-  darwin: [
-    '/Applications/Google Chrome.app/Contents/MacOS/Google Chrome',
-    '/Applications/Google Chrome Beta.app/Contents/MacOS/Google Chrome Beta',
-    '/Applications/Google Chrome Canary.app/Contents/MacOS/Google Chrome Canary',
-    '/Applications/Chromium.app/Contents/MacOS/Chromium',
+const DEFAULT_CHROME_LOCATIONS = new Map<NodeJS.Platform, readonly string[]>([
+  [
+    'darwin',
+    [
+      '/Applications/Google Chrome.app/Contents/MacOS/Google Chrome',
+      '/Applications/Google Chrome Beta.app/Contents/MacOS/Google Chrome Beta',
+      '/Applications/Google Chrome Canary.app/Contents/MacOS/Google Chrome Canary',
+      '/Applications/Chromium.app/Contents/MacOS/Chromium',
+    ],
   ],
-  linux: [
-    '/usr/bin/google-chrome',
-    '/usr/bin/google-chrome-stable',
-    '/usr/bin/chromium-browser',
-    '/usr/bin/chromium',
-    '/snap/bin/chromium',
+  [
+    'linux',
+    [
+      '/usr/bin/google-chrome',
+      '/usr/bin/google-chrome-stable',
+      '/usr/bin/chromium-browser',
+      '/usr/bin/chromium',
+      '/snap/bin/chromium',
+    ],
   ],
-  win32: [
-    ...(process.env.PROGRAMFILES
-      ? [join(process.env.PROGRAMFILES, 'Google/Chrome/Application/chrome.exe')]
-      : []),
-    ...(process.env['ProgramFiles(x86)']
-      ? [join(process.env['ProgramFiles(x86)'], 'Google/Chrome/Application/chrome.exe')]
-      : []),
-    ...(process.env.LOCALAPPDATA
-      ? [join(process.env.LOCALAPPDATA, 'Google/Chrome/Application/chrome.exe')]
-      : []),
+  [
+    'win32',
+    [
+      ...(process.env.PROGRAMFILES
+        ? [join(process.env.PROGRAMFILES, 'Google/Chrome/Application/chrome.exe')]
+        : []),
+      ...(process.env['ProgramFiles(x86)']
+        ? [join(process.env['ProgramFiles(x86)'], 'Google/Chrome/Application/chrome.exe')]
+        : []),
+      ...(process.env.LOCALAPPDATA
+        ? [join(process.env.LOCALAPPDATA, 'Google/Chrome/Application/chrome.exe')]
+        : []),
+    ],
   ],
-};
+]);
 
 interface ProgressiveExtractorRegionConfig {
   mode?: 'interactive' | 'full' | undefined;
@@ -135,7 +149,7 @@ interface FormatOptions {
   maxElements?: number | undefined;
 }
 
-const COMMON_TOOL_OPTIONS_SHAPE = {
+const COMMON_TOOL_OPTIONS = {
   maxDepth: z
     .number()
     .int()
@@ -167,11 +181,7 @@ type LibraryCache = {
   mtimeMs: number;
 };
 
-async function pathExists(candidate: string | undefined): Promise<boolean> {
-  if (!candidate) {
-    return false;
-  }
-
+async function pathExists(candidate: string): Promise<boolean> {
   try {
     await access(candidate, F_OK);
     return true;
@@ -325,7 +335,7 @@ class SmartDomReaderServer {
                 .boolean()
                 .describe('Include elements that are hidden/offscreen. Default: false.')
                 .optional(),
-              ...COMMON_TOOL_OPTIONS_SHAPE,
+              ...COMMON_TOOL_OPTIONS,
             })
             .optional(),
         }),
@@ -402,7 +412,7 @@ class SmartDomReaderServer {
                 .boolean()
                 .default(false)
                 .describe('Only include elements currently within the viewport. Default: false.'),
-              ...COMMON_TOOL_OPTIONS_SHAPE,
+              ...COMMON_TOOL_OPTIONS,
             })
             .optional(),
         }),
@@ -500,12 +510,12 @@ class SmartDomReaderServer {
 
     for (const envVar of CHROME_ENV_VARS) {
       const envPath = process.env[envVar];
-      if (await pathExists(envPath)) {
-        return { path: resolve(envPath as string), source: `env:${envVar}` };
+      if (envPath && (await pathExists(envPath))) {
+        return { path: resolve(envPath), source: `env:${envVar}` };
       }
     }
 
-    const candidates = DEFAULT_CHROME_LOCATIONS[process.platform] ?? [];
+    const candidates = DEFAULT_CHROME_LOCATIONS.get(process.platform) ?? [];
     for (const candidate of candidates) {
       if (await pathExists(candidate)) {
         return { path: resolve(candidate), source: 'default' };
@@ -529,10 +539,7 @@ class SmartDomReaderServer {
   private async extractStructure(args: OptionalSelectorArgs): Promise<CallToolResult> {
     const started = Date.now();
     try {
-      const text = await this.runLibraryOperation<
-        string,
-        StructureOperationArgs & { format: FormatOptions }
-      >('structure', {
+      const text = await this.runLibraryOperation('structure', {
         selector: args.selector && args.selector.trim().length > 0 ? args.selector : null,
         format: {
           detail: args.detail ?? 'summary',
@@ -550,10 +557,7 @@ class SmartDomReaderServer {
   private async extractRegion(args: RegionArgs): Promise<CallToolResult> {
     const started = Date.now();
     try {
-      const text = await this.runLibraryOperation<
-        string,
-        RegionOperationArgs & { format: FormatOptions }
-      >('region', {
+      const text = await this.runLibraryOperation('region', {
         selector: args.selector,
         options: args.options ?? {},
         format: {
@@ -574,10 +578,7 @@ class SmartDomReaderServer {
   private async extractContent(args: ContentArgs): Promise<CallToolResult> {
     const started = Date.now();
     try {
-      const text = await this.runLibraryOperation<
-        string,
-        ContentOperationArgs & { format: FormatOptions }
-      >('content', {
+      const text = await this.runLibraryOperation('content', {
         selector: args.selector,
         options: args.options ?? {},
         format: {
@@ -598,10 +599,7 @@ class SmartDomReaderServer {
   private async extractInteractive(args: InteractiveArgs): Promise<CallToolResult> {
     const started = Date.now();
     try {
-      const text = await this.runLibraryOperation<
-        string,
-        InteractiveOperationArgs & { format: FormatOptions }
-      >('interactive', {
+      const text = await this.runLibraryOperation('interactive', {
         selector: args.selector && args.selector.trim().length > 0 ? args.selector : null,
         options: args.options ?? {},
         format: {
@@ -661,15 +659,12 @@ class SmartDomReaderServer {
    * All formatting happens in-page via the bundled library; the server never
    * imports it, so there are no server-side dynamic imports.
    */
-  private async runLibraryOperation<
-    TResult,
-    TArgs extends Record<string, unknown> & { format: FormatOptions },
-  >(operation: LibraryOperation, args: TArgs): Promise<TResult> {
+  private async runLibraryOperation(...request: LibraryRequest): Promise<string> {
     const page = this.getActivePage();
     const code = await this.readLibraryFile(resolve(__dirname, EMBEDDED_LIBRARY_RELATIVE_PATH));
 
-    return page.evaluate<TResult, { code: string; operation: LibraryOperation; args: TArgs }>(
-      async ({ code, operation, args }) => {
+    const markdown = await page.evaluate<unknown, { code: string; request: LibraryRequest }>(
+      async ({ code, request: [operation, args] }) => {
         const blob = new Blob([code], { type: 'text/javascript' });
         const url = URL.createObjectURL(blob);
 
@@ -687,12 +682,12 @@ class SmartDomReaderServer {
                 throw new Error('ProgressiveExtractor export is unavailable.');
               }
 
-              const { selector } = args as StructureOperationArgs;
+              const { selector } = args;
               const target = selector ? document.querySelector(selector) : document;
               if (!target) throw new Error(`No matching element for selector ${selector}`);
               const overview = ProgressiveExtractor.extractStructure(target);
               if (!MarkdownFormatter) throw new Error('MarkdownFormatter export is unavailable.');
-              return MarkdownFormatter.structure(overview, fmt, meta) as TResult;
+              return MarkdownFormatter.structure(overview, fmt, meta);
             }
 
             case 'region': {
@@ -700,12 +695,11 @@ class SmartDomReaderServer {
                 throw new Error('ProgressiveExtractor export is unavailable.');
               }
 
-              const { selector, options } = args as RegionOperationArgs;
+              const { selector, options } = args;
               const result = ProgressiveExtractor.extractRegion(selector, document, options ?? {});
-              if (!result)
-                return `No matching region for selector ${selector}` as unknown as TResult;
+              if (!result) return `No matching region for selector ${selector}`;
               if (!MarkdownFormatter) throw new Error('MarkdownFormatter export is unavailable.');
-              return MarkdownFormatter.region(result, fmt, meta) as TResult;
+              return MarkdownFormatter.region(result, fmt, meta);
             }
 
             case 'content': {
@@ -713,15 +707,15 @@ class SmartDomReaderServer {
                 throw new Error('ProgressiveExtractor export is unavailable.');
               }
 
-              const { selector, options } = args as ContentOperationArgs;
+              const { selector, options } = args;
               const content = ProgressiveExtractor.extractContent(
                 selector,
                 document,
                 options ?? {}
               );
-              if (!content) return `No content for selector ${selector}` as unknown as TResult;
+              if (!content) return `No content for selector ${selector}`;
               if (!MarkdownFormatter) throw new Error('MarkdownFormatter export is unavailable.');
-              return MarkdownFormatter.content(content, fmt, meta) as TResult;
+              return MarkdownFormatter.content(content, fmt, meta);
             }
 
             case 'interactive': {
@@ -729,20 +723,20 @@ class SmartDomReaderServer {
                 throw new Error('SmartDOMReader export is unavailable.');
               }
 
-              const { selector, options } = args as InteractiveOperationArgs;
+              const { selector, options } = args;
               const target = selector ? document.querySelector(selector) : document;
               if (!target) throw new Error(`No matching element for selector ${selector}`);
 
               if (typeof SmartDOMReader.extractInteractive === 'function') {
                 const result = SmartDOMReader.extractInteractive(target, options ?? {});
                 if (!MarkdownFormatter) throw new Error('MarkdownFormatter export is unavailable.');
-                return MarkdownFormatter.region(result, fmt, meta) as TResult;
+                return MarkdownFormatter.region(result, fmt, meta);
               }
 
               const reader = new SmartDOMReader({ ...options, mode: 'interactive' });
               const result = reader.extract(target, options ?? {});
               if (!MarkdownFormatter) throw new Error('MarkdownFormatter export is unavailable.');
-              return MarkdownFormatter.region(result, fmt, meta) as TResult;
+              return MarkdownFormatter.region(result, fmt, meta);
             }
 
             default: {
@@ -753,8 +747,12 @@ class SmartDomReaderServer {
           URL.revokeObjectURL(url);
         }
       },
-      { code, operation, args }
+      { code, request }
     );
+    if (typeof markdown !== 'string') {
+      throw new TypeError('The embedded library returned a non-string result.');
+    }
+    return markdown;
   }
 
   private getActivePage(): Page {
@@ -773,12 +771,7 @@ class SmartDomReaderServer {
     try {
       file = await open(resolvedPath, 'r');
     } catch (error) {
-      if (
-        typeof error === 'object' &&
-        error !== null &&
-        'code' in error &&
-        error.code === 'ENOENT'
-      ) {
+      if (error instanceof Error && 'code' in error && error.code === 'ENOENT') {
         throw new ProtocolError(
           ProtocolErrorCode.InvalidParams,
           `Embedded library file not found at ${resolvedPath}. Ensure the bundled file exists.`
@@ -802,14 +795,14 @@ class SmartDomReaderServer {
     }
   }
 
-  private handleToolError(error: unknown, fallbackMessage: string): never {
-    if (error instanceof ProtocolError) {
-      throw error;
+  private handleToolError(cause: unknown, fallbackMessage: string): never {
+    if (cause instanceof ProtocolError) {
+      throw cause;
     }
 
-    if (error instanceof Error) {
-      console.error(fallbackMessage, error);
-      throw new ProtocolError(ProtocolErrorCode.InternalError, error.message);
+    if (cause instanceof Error) {
+      console.error(fallbackMessage, cause);
+      throw new ProtocolError(ProtocolErrorCode.InternalError, cause.message);
     }
 
     throw new ProtocolError(ProtocolErrorCode.InternalError, fallbackMessage);
@@ -817,7 +810,7 @@ class SmartDomReaderServer {
 }
 
 const server = new SmartDomReaderServer();
-server.run().catch((error) => {
-  console.error('Smart DOM Reader server failed to start', error);
+server.run().catch((cause) => {
+  console.error('Smart DOM Reader server failed to start', cause);
   process.exitCode = 1;
 });