@link-assistant/web-capture 1.7.3 → 1.7.5

package/CHANGELOG.md

@@ -1,5 +1,17 @@
 # @link-assistant/web-capture
 
+## 1.7.5
+
+### Patch Changes
+
+- dfaf007: Remove the unsupported `/gdocs` HTTP API route from the JavaScript and Rust server surfaces while keeping Google Docs URL capture through normal output formats.
+
+## 1.7.4
+
+### Patch Changes
+
+- 494da4e: Move turndown-plugin-gfm into runtime dependencies so published CLI installs include the GFM plugin used by Markdown conversion.
+
 ## 1.7.3
 
 ### Patch Changes

package/bin/web-capture.js

@@ -230,7 +230,7 @@ const config = makeConfig({
         'Capture private Google Doc with API token'
       )
       .epilogue(
-        'API Endpoints (in server mode):\n  GET /html?url=<URL>&engine=<ENGINE>           Get rendered HTML\n  GET /markdown?url=<URL>                       Get Markdown conversion\n  GET /image?url=<URL>&format=png|jpeg&theme=light|dark  Screenshot\n  GET /archive?url=<URL>&localImages=true&documentFormat=markdown|html  ZIP archive\n  GET /pdf?url=<URL>&theme=light|dark           PDF with embedded images\n  GET /docx?url=<URL>                           DOCX with embedded images\n  GET /fetch?url=<URL>                          Proxy fetch\n  GET /stream?url=<URL>                         Streaming proxy\n  GET /gdocs?url=<URL>&format=markdown|html|txt  Google Docs capture'
+        'API Endpoints (in server mode):\n  GET /html?url=<URL>&engine=<ENGINE>           Get rendered HTML\n  GET /markdown?url=<URL>                       Get Markdown conversion\n  GET /image?url=<URL>&format=png|jpeg&theme=light|dark  Screenshot\n  GET /archive?url=<URL>&localImages=true&documentFormat=markdown|html  ZIP archive\n  GET /pdf?url=<URL>&theme=light|dark           PDF with embedded images\n  GET /docx?url=<URL>                           DOCX with embedded images\n  GET /fetch?url=<URL>                          Proxy fetch\n  GET /stream?url=<URL>                         Streaming proxy'
       )
       .strict(),
   lenv: {
@@ -271,7 +271,6 @@ async function startServer(port) {
       console.log(`  GET /docx?url=<URL>       - DOCX with embedded images`);
       console.log(`  GET /fetch?url=<URL>      - Proxy fetch content`);
       console.log(`  GET /stream?url=<URL>     - Stream content`);
-      console.log(`  GET /gdocs?url=<URL>      - Capture Google Docs document`);
       console.log('');
       console.log('Press Ctrl+C to stop the server');
       resolve(server);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@link-assistant/web-capture",
-  "version": "1.7.3",
+  "version": "1.7.5",
   "description": "CLI and microservice to render web pages as HTML, Markdown, or PNG",
   "main": "src/index.js",
   "bin": {
@@ -44,7 +44,8 @@
     "playwright": "^1.49.0",
     "puppeteer": "^24.8.2",
     "puppeteer-core": "^24.40.0",
-    "turndown": "^7.1.1"
+    "turndown": "^7.1.1",
+    "turndown-plugin-gfm": "^1.0.2"
   },
   "devDependencies": {
     "@babel/core": "^7.24.0",
@@ -62,8 +63,7 @@
     "lint-staged": "^16.2.6",
     "nock": "^13.5.4",
     "prettier": "^3.6.2",
-    "supertest": "^6.3.4",
-    "turndown-plugin-gfm": "^1.0.2"
+    "supertest": "^6.3.4"
   },
   "type": "module",
   "keywords": [

package/src/gdocs.js

@@ -6,8 +6,7 @@
 
 import fetch from 'node-fetch';
 import he from 'he';
-import archiver from 'archiver';
-import { convertHtmlToMarkdown, prettyPrintHtml } from './lib.js';
+import { convertHtmlToMarkdown } from './lib.js';
 
 const GDOCS_URL_PATTERN = /docs\.google\.com\/document\/d\/([a-zA-Z0-9_-]+)/;
 
@@ -223,113 +222,3 @@ export async function fetchGoogleDocAsArchive(url, options = {}) {
     images,
   };
 }
-
-/**
- * Express handler for the /gdocs API endpoint.
- *
- * Query parameters:
- * - url (required): Google Docs URL
- * - format (optional): Export format (html, markdown, md, txt, pdf, docx)
- * - apiToken (optional): API token for private documents
- *
- * The API token can also be provided via:
- * - Authorization header: Bearer <token>
- * - X-Api-Token header: <token>
- *
- * @param {import('express').Request} req
- * @param {import('express').Response} res
- */
-export async function gdocsHandler(req, res) {
-  const url = req.query.url;
-  if (!url) {
-    return res.status(400).send('Missing `url` parameter');
-  }
-
-  if (!isGoogleDocsUrl(url)) {
-    return res.status(400).send('URL is not a Google Docs document URL');
-  }
-
-  // Resolve API token from query, headers, or environment
-  const apiToken =
-    req.query.apiToken ||
-    extractBearerToken(req.headers.authorization) ||
-    req.headers['x-api-token'] ||
-    undefined;
-
-  const format = (req.query.format || 'markdown').toLowerCase();
-
-  try {
-    if (format === 'archive' || format === 'zip') {
-      return await sendGDocsArchive(res, url, apiToken);
-    }
-
-    if (format === 'markdown' || format === 'md') {
-      const result = await fetchGoogleDocAsMarkdown(url, { apiToken });
-      return res.type('text/markdown').send(result.markdown);
-    }
-
-    return await sendGDocsExport(res, url, format, apiToken);
-  } catch (err) {
-    console.error('Google Docs capture error:', err.message);
-    return res.status(500).send(`Error capturing Google Doc: ${err.message}`);
-  }
-}
-
-async function sendGDocsArchive(res, url, apiToken) {
-  const archiveResult = await fetchGoogleDocAsArchive(url, { apiToken });
-
-  res.set('Content-Type', 'application/zip');
-  res.set(
-    'Content-Disposition',
-    `attachment; filename="gdoc-${archiveResult.documentId}.zip"`
-  );
-
-  const archive = archiver('zip', { zlib: { level: 9 } });
-  archive.pipe(res);
-  archive.append(archiveResult.markdown, { name: 'document.md' });
-  archive.append(prettyPrintHtml(archiveResult.html), {
-    name: 'document.html',
-  });
-  for (const img of archiveResult.images) {
-    archive.append(img.data, { name: `images/${img.filename}` });
-  }
-  await archive.finalize();
-}
-
-const FORMAT_CONTENT_TYPES = {
-  html: 'text/html',
-  txt: 'text/plain',
-  pdf: 'application/pdf',
-  docx: 'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
-};
-
-const FORMAT_DISPOSITIONS = {
-  pdf: 'inline; filename="document.pdf"',
-  docx: 'inline; filename="document.docx"',
-};
-
-async function sendGDocsExport(res, url, format, apiToken) {
-  const result = await fetchGoogleDoc(url, { format, apiToken });
-  const contentType = FORMAT_CONTENT_TYPES[format] || 'text/html';
-  const disposition = FORMAT_DISPOSITIONS[format];
-
-  res.type(contentType);
-  if (disposition) {
-    res.set('Content-Disposition', disposition);
-  }
-  return res.send(result.content);
-}
-
-/**
- * Extract Bearer token from Authorization header value.
- *
- * @param {string|undefined} authHeader - Authorization header value
- * @returns {string|undefined} Token or undefined
- */
-function extractBearerToken(authHeader) {
-  if (!authHeader || typeof authHeader !== 'string') {
-    return undefined;
-  }
-  const match = authHeader.match(/^Bearer\s+(.+)$/i);
-  return match ? match[1] : undefined;
-}

package/src/index.js

@@ -11,7 +11,6 @@ import { docxHandler } from './docx.js';
 import { animationHandler } from './animation.js';
 import { figuresHandler } from './figures.js';
 import { themedImageHandler } from './themed-image.js';
-import { gdocsHandler } from './gdocs.js';
 
 const app = express();
 const port = process.env.PORT || 3000;
@@ -28,7 +27,6 @@ app.get('/docx', docxHandler);
 app.get('/animation', animationHandler);
 app.get('/figures', figuresHandler);
 app.get('/themed-image', themedImageHandler);
-app.get('/gdocs', gdocsHandler);
 
 // Start the server if this is the main module
 const isMainModule =

package/tests/integration/api-endpoints.test.js

@@ -28,6 +28,17 @@ afterEach(() => {
 });
 
 describe('API Endpoint Tests', () => {
+  describe('unsupported routes', () => {
+    it('does not expose /gdocs as an output-format endpoint', async () => {
+      await request(app)
+        .get('/gdocs')
+        .query({
+          url: 'https://docs.google.com/document/d/test-doc/edit',
+        })
+        .expect(404);
+    });
+  });
+
   describe('GET /image', () => {
     it('rejects invalid format', async () => {
       await request(app)

package/tests/unit/cli.test.js

@@ -222,6 +222,7 @@ describe('CLI', () => {
       expect(result.stdout).toContain('/html');
       expect(result.stdout).toContain('/markdown');
       expect(result.stdout).toContain('/image');
+      expect(result.stdout).not.toContain('/gdocs');
     });
   });
 

package/tests/unit/package-runtime-dependencies.test.js

@@ -0,0 +1,48 @@
+import { readFileSync } from 'fs';
+import { dirname, resolve } from 'path';
+import { fileURLToPath } from 'url';
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = dirname(__filename);
+const jsRoot = resolve(__dirname, '../..');
+
+function readJson(relativePath) {
+  return JSON.parse(readFileSync(resolve(jsRoot, relativePath), 'utf8'));
+}
+
+const runtimeImports = [
+  {
+    packageName: 'turndown-plugin-gfm',
+    sourceFile: 'src/lib.js',
+  },
+];
+
+describe('runtime package dependencies', () => {
+  test.each(runtimeImports)(
+    '$packageName is declared as a production dependency',
+    ({ packageName, sourceFile }) => {
+      const source = readFileSync(resolve(jsRoot, sourceFile), 'utf8');
+      const packageJson = readJson('package.json');
+
+      expect(source).toContain(`from '${packageName}'`);
+      expect(packageJson.dependencies).toHaveProperty(packageName);
+      expect(packageJson.devDependencies || {}).not.toHaveProperty(packageName);
+    }
+  );
+
+  test.each(runtimeImports)(
+    '$packageName is not marked dev-only in package-lock.json',
+    ({ packageName }) => {
+      const packageLock = readJson('package-lock.json');
+
+      expect(packageLock.packages[''].dependencies).toHaveProperty(packageName);
+      expect(packageLock.packages[''].devDependencies || {}).not.toHaveProperty(
+        packageName
+      );
+      expect(packageLock.packages[`node_modules/${packageName}`]).toBeDefined();
+      expect(packageLock.packages[`node_modules/${packageName}`].dev).not.toBe(
+        true
+      );
+    }
+  );
+});