DB
The db now includes a new table, i.e., `embeddings`. More on this below.
Similarity model
- The similarity model is fully integrated into all the interfaces of Credential Digger (both the python library, the CLI, and the UI). In order to avoid a performance bottleneck during the manual assessment of the discoveries (present in v3.3.3), we added a table in the db such that the embeddings can be computed once, at scan time.
- The UI now shows (activated by default) a flag *"update similar discoveries"* to boost the manual review operations by calling the `update_similar_snippets function`.
Scan
- The `exclude` parameter has been deprecated to make the scan process more clear: either (all) the rules of a specific category are picked for a scan, or all the rules in the db are taken. No single rules can be excluded anymore.
UI
- The re-scan button has been moved to the homepage, where the repositories are listed (instead of keeping it in the discovery views)
- The choice of how many discoveries to list per page has been made persistent
- The button *"Show on GitHub"* has been fixed
- The count of the discoveries has been fixed
AoB
- Several updates to the documentation
- The db diagram has been updated to include the new table
- Several bugfixes