fix: update image regex to handle URLs with parentheses (#704)

* fix: update image regex to handle URLs with parentheses

Co-Authored-By: Max Prilutskiy <maks.prilutskiy@gmail.com>

* chore: add changeset for image regex fix

Co-Authored-By: Max Prilutskiy <maks.prilutskiy@gmail.com>

---------

Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Co-authored-by: Max Prilutskiy <maks.prilutskiy@gmail.com>

Author: devin-ai-integration[bot]

.changeset/angry-chairs-shave.md

@@ -0,0 +1,5 @@
+---
+"lingo.dev": patch
+---
+
+Fix image regex in MDX2 loader to handle URLs with parentheses

packages/cli/src/cli/loaders/mdx2/code-placeholder.spec.ts

@@ -450,4 +450,64 @@ describe("MDX Code Placeholder Loader", () => {
       expect(ruPushed).toBe("Примените `бар` функцию.");
     });
   });
+
+  describe("Image URLs with Parentheses", () => {
+    it("should handle image URLs with parentheses", async () => {
+      const md = dedent`
+        Text above.
+
+        ![](https://example.com/image(with)parentheses.jpg)
+
+        Text below.
+      `;
+
+      const pulled = await loader.pull("en", md);
+      const pushed = await loader.push("es", pulled);
+      expect(pushed).toBe(md);
+    });
+
+    it("should handle image URLs with nested parentheses", async () => {
+      const md = dedent`
+        Text above.
+
+        ![Alt text](https://example.com/image(with(nested)parentheses).jpg)
+
+        Text below.
+      `;
+
+      const pulled = await loader.pull("en", md);
+      const pushed = await loader.push("es", pulled);
+      expect(pushed).toBe(md);
+    });
+
+    it("should handle image URLs with parentheses in blockquotes", async () => {
+      const md = dedent`
+        > ![Blockquote image](https://example.com/image(in)blockquote.jpg)
+      `;
+
+      const pulled = await loader.pull("en", md);
+      const pushed = await loader.push("es", pulled);
+      expect(pushed).toBe(md);
+    });
+
+    it("should handle image URLs with parentheses in JSX components", async () => {
+      const md = dedent`
+        <Component>
+        ![Component image](https://example.com/image(in)component.jpg)
+        </Component>
+      `;
+
+      const expected = dedent`
+        <Component>
+
+        ![Component image](https://example.com/image(in)component.jpg)
+
+        </Component>
+      `;
+
+      const pulled = await loader.pull("en", md);
+      const pushed = await loader.push("es", pulled);
+      expect(pushed).toBe(expected);
+    });
+  });
 });

packages/cli/src/cli/loaders/mdx2/code-placeholder.ts

@@ -8,7 +8,7 @@ const inlineCodeRegex = /(?<!`)`([^`\r\n]+?)`(?!`)/g;
 
 // Matches markdown image tags, with optional alt text & parenthesis URL, possibly inside blockquotes
 // Captures patterns like ![](url) or ![alt](url), with optional leading '> ' for blockquotes
-const imageRegex = /([ \t]*)(^>\s*)?!\[[^\]]*?\]\([^\n\r]*?\)/gm;
+const imageRegex = /([ \t]*)(^>\s*)?!\[[^\]]*?\]\(([^()]*(\([^()]*\)[^()]*)*)\)/gm;
 
 /**
  * Ensures that markdown image tags are surrounded by blank lines (\n\n) so that they are properly