Adds unix shell-style wildcard matching to /learn
#989
Conversation
andrewfulton9
force-pushed
the
learn_wildcard_matching
branch
from
7b9b695 to
40f3ea8
Compare
|
@andrewfulton9 Thank you for this change! We plan to get this reviewed by tomorrow afternoon. |
There was a problem hiding this comment.
@andrewfulton9 Nice enhancement! Makes a huge difference, TY. I tried it with multiple nested folders with files of the same extension and it works as expected (inserted print statements to see all collected file_paths).
Thanks also for additions to the documentation.
And yes, the branch needs to be rebased.
Would it be possible to update the tests also in test_directory.py to test for the extended case in this PR? Check that the **/*.ipy* case (for example) is handled correctly through nested directories and subdirectories.
Co-authored-by: Michał Krassowski <[email protected]>
dlqqq
force-pushed
the
learn_wildcard_matching
branch
from
c883f5b to
03cca3d
Compare
|
Rebased to run |
…pyter-ai into learn_wildcard_matching
There was a problem hiding this comment.
@andrewfulton9 Thank you for this contribution! Left a comment on the indexing logic below.
There was a problem hiding this comment.
@andrewfulton9 Awesome work, thanks for this new feature! This will be included in the next release.
* adds wildcard matching to /learn * Add documentation * improve docs * cleanup * adds wildcard matching to /learn * Add documentation * improve docs * Update docs/source/users/index.md Co-authored-by: Michał Krassowski <[email protected]> * update for test * improve test * improve directory handling * remove dir only logic --------- Co-authored-by: Michał Krassowski <[email protected]>
4 participants
Closes #988
This PR updates the
LearnChatHandlerto accept paths in the form of unix filename matching patterns to allow more fine-grained selection of files to learn on. It also updates the documentation to reflect the changes.