@link-assistant/web-capture 1.7.25 → 1.7.26

package/CHANGELOG.md

@@ -1,5 +1,11 @@
 # @link-assistant/web-capture
 
+## 1.7.26
+
+### Patch Changes
+
+- 112f2f9: Fix `stripBase64Images` (used by `--keep-original-links`) dropping every base64 image with empty alt text instead of leaving a visible placeholder. Google Docs HTML exports emit `<img alt="" src="data:image/png;base64,...">` for every image, so the previous behaviour silently deleted all images from the rendered markdown. Empty-alt now renders as `![]()` (a valid empty image reference); non-empty alt continues to render as `*[image: <alt>]*`.
+
 ## 1.7.25
 
 ### Patch Changes

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@link-assistant/web-capture",
-  "version": "1.7.25",
+  "version": "1.7.26",
   "description": "CLI and microservice to render web pages as HTML, Markdown, or PNG",
   "main": "src/index.js",
   "bin": {

package/src/extract-images.js

@@ -95,9 +95,16 @@ export function extractBase64ToBuffers(markdown, imagesDir = 'images') {
 }
 
 /**
- * Strip base64 data URI images from markdown, leaving only the alt text
- * as a placeholder. Used when keepOriginalLinks is enabled — base64 images
- * have no original URL to restore, so we remove the heavy data URI.
+ * Strip base64 data URI images from markdown, leaving a visible placeholder.
+ * Used when keepOriginalLinks is enabled — base64 images have no original URL
+ * to restore, so we remove the heavy data URI but still leave a marker so the
+ * reader can see that an image was here.
+ *
+ * Non-empty alt becomes `*[image: <alt>]*`. Empty alt — common for Google Docs
+ * HTML exports, which emit `<img alt="" src="data:...">` for every image —
+ * becomes `![]()`, an empty markdown image reference that renderers still
+ * surface as a slot. Emitting `''` for empty-alt would silently delete every
+ * image in the document (see issue #117).
  *
  * @param {string} markdown - Markdown content with data URI images
  * @returns {{markdown: string, stripped: number}}
@@ -108,7 +115,7 @@ export function stripBase64Images(markdown) {
     BASE64_IMAGE_REGEX,
     (_match, altText) => {
       stripped++;
-      return altText ? `*[image: ${altText}]*` : '';
+      return altText ? `*[image: ${altText}]*` : '![]()';
     }
   );
   return { markdown: updatedMarkdown, stripped };

package/tests/unit/extract-images.test.js

@@ -204,12 +204,18 @@ describe('extract-images module', () => {
       expect(result.markdown).not.toContain('data:image');
     });
 
-    it('strips base64 images with empty alt text', () => {
-      const md = `![](data:image/png;base64,${TINY_PNG})`;
+    it('leaves a visible placeholder when stripping a base64 image with empty alt', () => {
+      // Google Docs HTML exports emit `<img alt="" src="data:image/png;base64,...">`,
+      // which renders as `![](data:...)`. If stripping drops the markdown image
+      // entirely, every image in the doc vanishes from the output with no
+      // indication that anything was lost. Leave a visible placeholder so the
+      // reader can see that an image was here.
+      const md = `Hi.\n\n![](data:image/png;base64,${TINY_PNG})\n\nBye.\n`;
       const result = stripBase64Images(md);
 
       expect(result.stripped).toBe(1);
-      expect(result.markdown).toBe('');
+      expect(result.markdown).not.toMatch(/data:image/);
+      expect(result.markdown).toMatch(/!\[|\[image/);
     });
 
     it('preserves remote image URLs', () => {