Implement approximate file matching #5 #6
Conversation
Signed-off-by: Jono Yang <jyang@nexb.com>
* Compute resource fingerprints in fingerprint_codebase Signed-off-by: Jono Yang <jyang@nexb.com>
Signed-off-by: Jono Yang <jyang@nexb.com>
Signed-off-by: Jono Yang <jyang@nexb.com>
Signed-off-by: Jono Yang <jyang@nexb.com>
Signed-off-by: Jono Yang <jyang@nexb.com>
JonoYang
force-pushed
the
5-approximate-file-matching
branch
from
9293abc
to
287f8ec
Compare
* Use open instead of codecs.open Signed-off-by: Jono Yang <jyang@nexb.com>
| file_fingerprint = BitAverageHaloHash(ngs_bytes) if ngs_bytes else None | ||
|
|
||
| return dict( | ||
| halo1=file_fingerprint.hexdigest().decode('utf-8') if file_fingerprint else '' |
There was a problem hiding this comment.
@pombredanne What should the file fingerprint be called? Currently it is halo1 but I don't think that's a great label. Also, if a fingerprint could not be generated for a file, should we still populate that resource's extra_data field with {'halo1': ''} or just avoid returning anything at all?
There was a problem hiding this comment.
What should the file fingerprint be called? Currently it is
halo1but I don't think that's a great label.
This is OK for now IMHO.
Also, if a fingerprint could not be generated for a file, should we still populate that resource's extra_data field with
{'halo1': ''}or just avoid returning anything at all?
IMHO, always provide a value, null or empty (the same we do for other hashes)
There was a problem hiding this comment.
@pombredanne I have a question regarding the file fingerprint name
pombredanne
changed the title
|
@pombredanne What additional tests do you suggest? |
|
@JonoYang what about a test that would be using two real code files where the second one has been modified a little? |
Signed-off-by: Jono Yang <jyang@nexb.com>
No description provided.