fix: skip bad data uri images on parsing

fix #36

enex2notion/note_parser/elements/media.py

@@ -50,7 +50,11 @@ def parse_img(element: Tag):
             url=src,
         )
 
-    img_resource = _parse_img_resource(src)
+    try:
+        img_resource = _parse_img_resource(src)
+    except ValueError:
+        logger.warning(f"Failed to parse image: '{src}'")
+        return None
 
     # Make SVG small by default to avoid them spreading too much
     if "svg" in img_resource.mime and not any((w, h)):

tests/test_note_parser.py

@@ -349,6 +349,21 @@ def test_embedded_inline_svg_no_size(parse_html, smallest_svg):
     )
 
 
+def test_embedded_inline_bad(parse_html, smallest_svg, caplog):
+    test_note = parse_html(
+        f"<img "
+        'src="data:image/svg+xml;utf8,'
+        "<svg xmlns='http://www.w3.org/2000/svg' width='416' height='377'>"
+        '</svg/>" />'
+    )
+
+    with caplog.at_level(logging.WARNING, logger="enex2notion"):
+        result_block = parse_note_blocks(test_note)
+
+    assert result_block == []
+    assert "Failed to parse image" in caplog.text
+
+
 def test_embedded_inline_img_url(parse_html):
     test_note = parse_html('<img src="https://google.com/image.jpg" />')