MCPcopy Create free account
hub / github.com/CScorza/IntelOSINT / extract_urls

Method extract_urls

IntelOSINT.py:1014–1035  ·  view source on GitHub ↗
(obj)

Source from the content-addressed store, hash-verified

1012 try:
1013 extracted_media = []
1014 def extract_urls(obj):
1015 if isinstance(obj, dict):
1016 if "image_versions2" in obj:
1017 cands = obj["image_versions2"].get("candidates", [])
1018 if cands:
1019 cands_sorted = sorted(
1020 [c for c in cands if c.get("url")],
1021 key=lambda x: int(x.get("width") or 0),
1022 reverse=True
1023 )
1024 extracted_media.append(cands_sorted[0].get("url"))
1025 if "video_versions" in obj:
1026 cands = obj["video_versions"]
1027 if cands: extracted_media.append(cands[0].get("url"))
1028 if "display_url" in obj:
1029 extracted_media.append(obj["display_url"])
1030 if "video_url" in obj:
1031 extracted_media.append(obj["video_url"])
1032
1033 for k, v in obj.items(): extract_urls(v)
1034 elif isinstance(obj, list):
1035 for item in obj: extract_urls(item)
1036
1037 extract_urls(api_data)
1038

Callers

nothing calls this directly

Calls

no outgoing calls

Tested by

no test coverage detected