Detectors in the detection package currently do both parsing and detection together:
- Parsing the source (commit message, commit trailers, git notes)
- Detecting any AI disclosures in parsing result
We need to separate these out as discussed in #49, so we can maintain detection logic (which updates frequently) in isolation from parsing logic (which changes relatively much infrequently).
Also separating them will solve the problem the logic duplication and fragmentation we have right now e.g. in case of Co-Authored-By, Assisted-By and other tool-specific trailers in message.go essentially have fragmented detection logic and duplicated parsing logic for AI trailer detection.
Detectors in the detection package currently do both parsing and detection together:
We need to separate these out as discussed in #49, so we can maintain detection logic (which updates frequently) in isolation from parsing logic (which changes relatively much infrequently).
Also separating them will solve the problem the logic duplication and fragmentation we have right now e.g. in case of Co-Authored-By, Assisted-By and other tool-specific trailers in message.go essentially have fragmented detection logic and duplicated parsing logic for AI trailer detection.