Merging upstream version 2.2.0.

Signed-off-by: Daniel Baumann <daniel@debian.org>

.pre-commit-config.yaml

@@ -11,17 +11,17 @@ repos:
     -   id: requirements-txt-fixer
     -   id: trailing-whitespace
 -   repo: https://github.com/asottile/setup-cfg-fmt
-    rev: v1.16.0
+    rev: v1.17.0
     hooks:
     -   id: setup-cfg-fmt
 -   repo: https://gitlab.com/pycqa/flake8
-    rev: 3.8.4
+    rev: 3.9.0
     hooks:
     -   id: flake8
         exclude: ^identify/vendor/licenses\.py$
         additional_dependencies: [flake8-typing-imports==1.10.1]
 -   repo: https://github.com/pre-commit/mirrors-autopep8
-    rev: v1.5.4
+    rev: v1.5.5
     hooks:
     -   id: autopep8
 -   repo: https://github.com/asottile/reorder_python_imports

identify/extensions.py

@@ -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'],

identify/identify.py

@@ -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

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = identify
-version = 2.1.2
+version = 2.2.0
 description = File identification library for Python
 long_description = file: README.md
 long_description_content_type = text/markdown
@@ -24,18 +24,18 @@ classifiers =
 packages = find:
 python_requires = >=3.6.1
 
+[options.packages.find]
+exclude =
+    tests*
+    testing*
+
 [options.entry_points]
 console_scripts =
     identify-cli=identify.cli:main
 
 [options.extras_require]
 license =
-    editdistance
-
-[options.packages.find]
-exclude =
-    tests*
-    testing*
+    editdistance-s
 
 [bdist_wheel]
 universal = True

tests/identify_test.py

@@ -158,6 +158,8 @@ def test_tags_from_path_plist_text(tmpdir):
         ('Pipfile.lock', {'text', 'json'}),
         ('mod/test.py', {'text', 'python'}),
         ('mod/Dockerfile', {'text', 'dockerfile'}),
+        ('Gemfile', {'text', 'ruby'}),
+        ('Gemfile.lock', {'text'}),
 
         # does not set binary / text
         ('f.plist', {'plist'}),