api / app /services /extractors /__init__.py
lljz66's picture
add extractor registry
ebccab5 verified
Raw
History Blame Contribute Delete
769 Bytes
"""Asset extractors.
Each extractor exposes a single callable:
def extract(soup: BeautifulSoup, base_url: str, include_data_uris: bool) -> list[str]
returning a list of absolute asset URLs for its category.
"""
from .audios import extract as extract_audios
from .fonts import extract as extract_fonts
from .icons_svg import extract as extract_icons
from .images import extract as extract_images
from .videos import extract as extract_videos
# Ordered registry: key -> (group label in response, extractor callable).
EXTRACTORS = {
"image": ("image", extract_images),
"video": ("video", extract_videos),
"audio": ("audio", extract_audios),
"svg_icon": ("svg_icon", extract_icons),
"font": ("font", extract_fonts),
}
__all__ = ["EXTRACTORS"]