@link-assistant/web-capture 1.7.15 → 1.7.16

package/CHANGELOG.md

@@ -1,5 +1,11 @@
 # @link-assistant/web-capture
 
+## 1.7.16
+
+### Patch Changes
+
+- bf7eaec: Keep Google Docs browser-model soft line breaks outside inline mark tags and preserve image dimensions in HTML output.
+
 ## 1.7.15
 
 ### Patch Changes

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@link-assistant/web-capture",
-  "version": "1.7.15",
+  "version": "1.7.16",
   "description": "CLI and microservice to render web pages as HTML, Markdown, or PNG",
   "main": "src/index.js",
   "bin": {

package/src/gdocs-render-markdown.js

@@ -138,9 +138,27 @@ function renderTextRunsMarkdown(nodes = []) {
       italic: Boolean(node.italic),
       strike: Boolean(node.strike),
     };
-    output += markdownStyleTransition(active, next);
-    output += node.text || '';
-    active = next;
+    const text = node.text || '';
+    let start = 0;
+    for (let idx = 0; idx < text.length; idx++) {
+      if (text[idx] !== '\n') {
+        continue;
+      }
+      if (idx > start) {
+        output += markdownStyleTransition(active, next);
+        output += text.slice(start, idx);
+        active = next;
+      }
+      output += markdownStyleTransition(active, inactive);
+      output += '\n';
+      active = inactive;
+      start = idx + 1;
+    }
+    if (start < text.length) {
+      output += markdownStyleTransition(active, next);
+      output += text.slice(start);
+      active = next;
+    }
   }
   output += markdownStyleTransition(active, inactive);
   return output;

package/src/gdocs.js

@@ -1219,7 +1219,17 @@ function renderContentHtml(content = []) {
 }
 
 function renderMarkedHtml(node) {
-  let output = escapeHtml(node.text || '').replace(/\n/g, '<br>');
+  return String(node.text || '')
+    .split('\n')
+    .map((segment) => renderMarkedHtmlSegment(node, segment))
+    .join('<br>');
+}
+
+function renderMarkedHtmlSegment(node, text) {
+  if (!text) {
+    return '';
+  }
+  let output = escapeHtml(text);
   if (node.bold) {
     output = `<strong>${output}</strong>`;
   }

package/tests/unit/gdocs.test.js

@@ -26,10 +26,26 @@ import {
 import fs from 'fs';
 import os from 'os';
 import path from 'path';
+import { fileURLToPath } from 'url';
 import nock from 'nock';
 
 jest.setTimeout(30000);
 
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = path.dirname(__filename);
+const REPO_ROOT = path.resolve(__dirname, '..', '..', '..');
+const ISSUE_104_FIXTURE_DIR = path.join(
+  REPO_ROOT,
+  'docs',
+  'case-studies',
+  'issue-104',
+  'fixtures'
+);
+
+function readIssue104Fixture(filename) {
+  return fs.readFileSync(path.join(ISSUE_104_FIXTURE_DIR, filename), 'utf-8');
+}
+
 describe('gdocs', () => {
   describe('isGoogleDocsUrl', () => {
     it('returns true for standard Google Docs edit URL', () => {
@@ -648,6 +664,74 @@ describe('gdocs', () => {
       );
     });
 
+    it('renders soft breaks outside marks and preserves image dimensions (issue #104)', () => {
+      const fixture = JSON.parse(
+        readIssue104Fixture('multiline-marked-inline-image-model.json')
+      );
+      const capture = parseGoogleDocsModelChunks(
+        fixture.chunks,
+        fixture.cidUrlMap
+      );
+
+      expect(capture.blocks[0].content).toEqual([
+        expect.objectContaining({
+          type: 'text',
+          text: 'Line one of bold text.',
+          bold: true,
+        }),
+        expect.objectContaining({ type: 'text', text: '\n', bold: false }),
+        expect.objectContaining({
+          type: 'text',
+          text: 'Line two of bold text.',
+          bold: true,
+        }),
+        expect.objectContaining({
+          type: 'image',
+          alt: 'Inline diagram',
+          width: 320,
+          height: 180,
+        }),
+        expect.objectContaining({ type: 'text', text: '\n\n', bold: false }),
+        expect.objectContaining({
+          type: 'text',
+          text: 'Line three of bold text.',
+          bold: true,
+        }),
+      ]);
+      expect(renderGoogleDocsCapture(capture, 'html')).toBe(
+        readIssue104Fixture(
+          'multiline-marked-inline-image.expected.html'
+        ).trimEnd()
+      );
+      expect(renderGoogleDocsCapture(capture, 'markdown')).toBe(
+        readIssue104Fixture('multiline-marked-inline-image.expected.md')
+      );
+    });
+
+    it('does not keep inline marks open across embedded newlines (issue #104)', () => {
+      const capture = {
+        blocks: [
+          {
+            type: 'paragraph',
+            content: [
+              {
+                type: 'text',
+                text: 'Alpha\nBeta',
+                bold: true,
+              },
+            ],
+          },
+        ],
+      };
+
+      expect(renderGoogleDocsCapture(capture, 'html')).toBe(
+        '<!doctype html><html><body><p><strong>Alpha</strong><br><strong>Beta</strong></p></body></html>'
+      );
+      expect(renderGoogleDocsCapture(capture, 'markdown')).toBe(
+        '**Alpha**\n**Beta**\n'
+      );
+    });
+
     it('keeps nested bold and italic markers balanced (issue #96)', () => {
       const text = 'Bold text with italic inside and back to bold\n';
       const chunks = [