git/identify
1
0
Fork 0
Code Issues Activity

Merging upstream version 2.2.0.

Browse source 
Signed-off-by: Daniel Baumann <daniel@debian.org>
This commit is contained in:
Daniel Baumann 2025-02-12 21:46:10 +01:00
parent 13a62a0f54
commit e5b299766a
Signed by: daniel
GPG key ID: FBB4F0E80A80222F
5 changed files with 16 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

2
identify/extensions.py
View file

@ -226,6 +226,7 @@ NAMES = {
'.bash_aliases': EXTENSIONS['bash'],
'.bash_profile': EXTENSIONS['bash'],
'.bashrc': EXTENSIONS['bash'],
'.bazelrc': {'text', 'bazelrc'},
'.bowerrc': EXTENSIONS['json'] | {'bowerrc'},
'.browserslistrc': {'text', 'browserslistrc'},
'.clang-format': EXTENSIONS['yaml'],
@ -261,6 +262,7 @@ NAMES = {
'COPYING': EXTENSIONS['txt'],
'Dockerfile': {'text', 'dockerfile'},
'Gemfile': EXTENSIONS['rb'],
'Gemfile.lock': {'text'},
'Jenkinsfile': {'text', 'groovy'},
'LICENSE': EXTENSIONS['txt'],
'MAINTAINERS': EXTENSIONS['txt'],

4
identify/identify.py
View file

@ -240,7 +240,7 @@ def license_id(filename: str) -> Optional[str]:
3. check exact text match with existing licenses
4. failing that use edit distance
"""
import editdistance # `pip install identify[license]`
import editdistance_s # `pip install identify[license]`
with open(filename, encoding='UTF-8') as f:
contents = f.read()
@ -260,7 +260,7 @@ def license_id(filename: str) -> Optional[str]:
if norm and abs(len(norm) - len(norm_license)) / len(norm) > .05:
continue
edit_dist = editdistance.eval(norm, norm_license)
edit_dist = editdistance_s.distance(norm, norm_license)
if edit_dist < min_edit_dist:
min_edit_dist = edit_dist
min_edit_dist_spdx = spdx