Commit 51f42da1 authored by Jan Reimes's avatar Jan Reimes
Browse files

refactor(fetch): streamline metadata fetching from whatthespec API 

* Update URL for fetching spec metadata.
* Simplify metadata extraction and return structure.
* Ensure versions are always returned as a list.
parent fb03029a

src/tdoc_crawler/specs/sources/whatthespec.py

+18 −13
Original line number Diff line number Diff line
@@ -9,20 +9,25 @@ def fetch_whatthespec_metadata(spec_number: str) -> dict[str, object]: 
    """Fetch spec metadata via whatthespec.net JSON API."""

    normalized = normalize_spec_number(spec_number)

    compact = normalized.replace(".", "")

    url = f"https://whatthespec.net/api/specs/{compact}"



    url = f"https://whatthespec.net/3gpp/spec.php?q={compact}&api=1"

    response = requests.get(url, timeout=30)

    response.raise_for_status()

    payload = response.json()

    metadata = {}



    if payload:

        payload = payload[0] if isinstance(payload, list) else payload



        versions = payload.get("versions")

        if not isinstance(versions, list):

            versions = []



    return {

        metadata = {

            "spec_number": normalized,

            "source_name": "whatthespec",

            "source_identifier": payload.get("id"),

            "metadata_payload": payload,

            "versions": versions,

        }



    return metadata
 
 No newline at end of file