fix(ai): broaden SSRF pre-scan regex to cover srcset, poster, formaction, @import

packages/ai/python/doc_html_pdf.py

@@ -10,11 +10,11 @@ import re
 import sys
 
 _REMOTE_REF_RE = re.compile(
-    r'(?:src|href|action)\s*=\s*["\']?\s*(https?:/{1,2}[^\s"\'>\)]{1,200})',
+    r'(?:src|href|action|srcset|poster|formaction)\s*=\s*["\']?\s*(https?:/{1,2}[^\s"\'>\)]{1,200})',
     re.IGNORECASE,
 )
 _REMOTE_CSS_URL_RE = re.compile(
-    r'url\s*\(\s*["\']?\s*(https?:/{1,2}[^\s"\'>\)]{1,200})',
+    r'(?:url\s*\(|@import)\s*["\']?\s*(https?:/{1,2}[^\s"\'>\)]{1,200})',
     re.IGNORECASE,
 )
 

tests/unit/ai/ssrf-prescan-regex.test.ts

@@ -49,6 +49,18 @@ describe.skipIf(!hasPython)("SSRF pre-scan regexes (doc_html_pdf.py)", () => {
     it("does NOT match relative paths", () => {
       expect(testRegex('<a href="page2.xhtml">', "_REMOTE_REF_RE")).toBe(false);
     });
+
+    it("matches srcset attribute", () => {
+      expect(testRegex('<img srcset="https://h/x.png 2x">', "_REMOTE_REF_RE")).toBe(true);
+    });
+
+    it("matches poster attribute", () => {
+      expect(testRegex('<video poster="https://h/thumb.jpg">', "_REMOTE_REF_RE")).toBe(true);
+    });
+
+    it("matches formaction attribute", () => {
+      expect(testRegex('<button formaction="https://h/submit">', "_REMOTE_REF_RE")).toBe(true);
+    });
   });
 
   describe("_REMOTE_CSS_URL_RE", () => {
@@ -63,5 +75,17 @@ describe.skipIf(!hasPython)("SSRF pre-scan regexes (doc_html_pdf.py)", () => {
     it("does NOT match data: CSS url()", () => {
       expect(testRegex("url(data:image/png;base64,AA==)", "_REMOTE_CSS_URL_RE")).toBe(false);
     });
+
+    it("matches @import with double-quoted URL", () => {
+      expect(testRegex('@import "https://h/style.css";', "_REMOTE_CSS_URL_RE")).toBe(true);
+    });
+
+    it("matches @import with single-quoted URL", () => {
+      expect(testRegex("@import 'https://h/style.css';", "_REMOTE_CSS_URL_RE")).toBe(true);
+    });
+
+    it("matches @import with bare URL", () => {
+      expect(testRegex("@import https://h/style.css;", "_REMOTE_CSS_URL_RE")).toBe(true);
+    });
   });
 });