exclude svg from image group (belongs to svg_icon)

app/services/extractors/images.py

@@ -1,10 +1,15 @@
-"""Extract image assets from HTML."""
+"""Extract image assets from HTML.
+
+SVG files are intentionally excluded — they belong to the ``svg_icon``
+category, not ``image``.
+"""
 
 import re
 
 from ..resolver import resolve
 
 _IMAGE_EXT = r"\.(?:jpe?g|png|gif|webp|avif|bmp|apng)(?:$|[\?#])"
+_SVG_EXT = r"\.svg(?:$|[\?#])"
 
 
 def _parse_srcset(srcset: str) -> list[str]:
@@ -34,6 +39,9 @@ def extract(soup, base_url: str, include_data_uris: bool = False) -> list[str]:
             return
         if ref.startswith("data:") and not include_data_uris:
             return
+        # SVG belongs to svg_icon, never image.
+        if re.search(_SVG_EXT, resolved, re.IGNORECASE):
+            return
         if resolved not in found:
             found.append(resolved)