Commit ebf5dbef authored by Jan Reimes's avatar Jan Reimes
Browse files

Fix: _get_tdoc now correctly retrieves TDocs from database 

Fixed bug in TDocDatabase._get_tdoc that caused TDocs to not be
found in the local database, leading to unnecessary external API calls
to whatthespec.net.

Issue:
- _get_tdoc used self.connection.model_from_table("tdocs", tdoc_id.upper())
- This method was throwing KeyError, causing all lookups to fail
- As a result, checkout_tdoc_to_workspace fell back to whatthespec.net
  even for TDocs that existed in the local database

Fix:
- Changed _get_tdoc to use direct SQL query instead of model_from_table
- Query retrieves all required fields from actual database schema
- Returns properly constructed TDocMetadata object

Impact:
- TDocs with URLs in database are now correctly found
- whatthespec.net is only contacted for TDocs NOT in database
- Commands like `ai workspace add-members` work efficiently

Tested with:
- S4AQ200164: Now correctly retrieved from database with URL
- Non-existent TDocs: Return None as expected
parent 67486cd8

src/tdoc_crawler/database/tdocs.py

+25 −3
Original line number Diff line number Diff line
@@ -287,11 +287,33 @@ class TDocDatabase(MeetingDatabase): 


    def _get_tdoc(self, tdoc_id: str) -> TDocMetadata | None:

        """Get a TDoc by ID."""

        import sqlite3



        try:

            return self.connection.model_from_table("tdocs", tdoc_id.upper())

        except KeyError:

            # Use direct SQL query to retrieve TDoc by ID

            # Query all required fields from actual database schema

            with sqlite3.connect(self.connection.filename) as conn:

                cursor = conn.cursor()

                cursor.execute(

                    "SELECT tdoc_id, meeting_id, title, url, source, contact, agenda_item_nbr "

                    "FROM tdocs WHERE tdoc_id = ?",

                    (tdoc_id.upper(),)

                )

                row = cursor.fetchone()

                if row:

                    from tdoc_crawler.tdocs.models import TDocMetadata

                    return TDocMetadata(

                        tdoc_id=row[0],

                        meeting_id=row[1],

                        title=row[2],

                        url=row[3],

                        source=row[4],

                        contact=row[5],

                        agenda_item_nbr=row[6],

                    )

                return None

        except sqlite3.Error:

            return None



    # ------------------------------------------------------------------

    # Normalisation helpers

    # ------------------------------------------------------------------