lljz66 commited on
Commit
ebccab5
·
verified ·
1 Parent(s): 0082cb1

add extractor registry

Browse files
Files changed (1) hide show
  1. app/services/extractors/__init__.py +25 -0
app/services/extractors/__init__.py ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """Asset extractors.
2
+
3
+ Each extractor exposes a single callable:
4
+
5
+ def extract(soup: BeautifulSoup, base_url: str, include_data_uris: bool) -> list[str]
6
+
7
+ returning a list of absolute asset URLs for its category.
8
+ """
9
+
10
+ from .audios import extract as extract_audios
11
+ from .fonts import extract as extract_fonts
12
+ from .icons_svg import extract as extract_icons
13
+ from .images import extract as extract_images
14
+ from .videos import extract as extract_videos
15
+
16
+ # Ordered registry: key -> (group label in response, extractor callable).
17
+ EXTRACTORS = {
18
+ "image": ("image", extract_images),
19
+ "video": ("video", extract_videos),
20
+ "audio": ("audio", extract_audios),
21
+ "svg_icon": ("svg_icon", extract_icons),
22
+ "font": ("font", extract_fonts),
23
+ }
24
+
25
+ __all__ = ["EXTRACTORS"]