requirements_txt_fixer.py: Included an option to fail if no version is specified for a requirement

Author: binka-dev

.pre-commit-hooks.yaml

@@ -192,7 +192,7 @@
     always_run: true
 -   id: requirements-txt-fixer
     name: fix requirements.txt
-    description: sorts entries in requirements.txt.
+    description: sorts entries in requirements.txt and checks whether a version is specified (parameterized).
     entry: requirements-txt-fixer
     language: python
     files: (requirements|constraints).*\.txt$

README.md

@@ -186,7 +186,10 @@ the following commandline options:
   - `--top-keys comma,separated,keys` - Keys to keep at the top of mappings.
 
 #### `requirements-txt-fixer`
-Sorts entries in requirements.txt and constraints.txt and removes incorrect entry for `pkg-resources==0.0.0`
+Sorts entries in requirements.txt and constraints.txt and removes incorrect entry for `pkg-resources==0.0.0`  
+Provides also an optional check if a version is specified for each requirement. You can configure this with
+the following commandline options:
+  - `--fail-without-version` - Fails when no version is specified for a requirement
 
 #### `sort-simple-yaml`
 Sorts simple YAML files which consist only of top-level

pre_commit_hooks/requirements_txt_fixer.py

@@ -13,6 +13,7 @@
 class Requirement:
     UNTIL_COMPARISON = re.compile(b'={2,3}|!=|~=|>=?|<=?')
     UNTIL_SEP = re.compile(rb'[^;\s]+')
+    VERSION_SPECIFIED = re.compile(b'.+(={2,3}|!=|~=|>=?|<=?).+')
 
     def __init__(self) -> None:
         self.value: bytes | None = None
@@ -58,14 +59,17 @@ def is_complete(self) -> bool:
             not self.value.rstrip(b'\r\n').endswith(b'\\')
         )
 
+    def contains_version_specifier(self) -> bool:
+        return bool(self.VERSION_SPECIFIED.match(self.value))
+
     def append_value(self, value: bytes) -> None:
         if self.value is not None:
             self.value += value
         else:
             self.value = value
 
 
-def fix_requirements(f: IO[bytes]) -> int:
+def fix_requirements(f: IO[bytes], fail_without_version: bool) -> int:
     requirements: list[Requirement] = []
     before = list(f)
     after: list[bytes] = []
@@ -121,6 +125,17 @@ def fix_requirements(f: IO[bytes]) -> int:
         ]
     ]
 
+    # check for requirements without a version specified
+    if fail_without_version:
+        missing_requirement_found = False
+        for req in requirements:
+            if not req.contains_version_specifier():
+                print(f'Missing version for requirement {req.name.decode()}')
+                missing_requirement_found = True
+
+        if missing_requirement_found:
+            return FAIL
+
     # sort the requirements and remove duplicates
     prev = None
     for requirement in sorted(requirements):
@@ -145,13 +160,15 @@ def fix_requirements(f: IO[bytes]) -> int:
 def main(argv: Sequence[str] | None = None) -> int:
     parser = argparse.ArgumentParser()
     parser.add_argument('filenames', nargs='*', help='Filenames to fix')
+    parser.add_argument("--fail-without-version", action="store_true",
+                        help="Fail if a requirement is missing a version")
     args = parser.parse_args(argv)
 
     retv = PASS
 
     for arg in args.filenames:
         with open(arg, 'rb+') as file_obj:
-            ret_for_file = fix_requirements(file_obj)
+            ret_for_file = fix_requirements(file_obj, args.fail_without_version)
 
             if ret_for_file:
                 print(f'Sorting {arg}')

tests/requirements_txt_fixer_test.py

@@ -9,130 +9,155 @@
 
 
 @pytest.mark.parametrize(
-    ('input_s', 'expected_retval', 'output'),
+    ('input_s', 'argv', 'expected_retval', 'output'),
     (
-        (b'', PASS, b''),
-        (b'\n', PASS, b'\n'),
-        (b'# intentionally empty\n', PASS, b'# intentionally empty\n'),
-        (b'foo\n# comment at end\n', PASS, b'foo\n# comment at end\n'),
-        (b'foo\nbar\n', FAIL, b'bar\nfoo\n'),
-        (b'bar\nfoo\n', PASS, b'bar\nfoo\n'),
-        (b'a\nc\nb\n', FAIL, b'a\nb\nc\n'),
-        (b'a\nc\nb', FAIL, b'a\nb\nc\n'),
-        (b'a\nb\nc', FAIL, b'a\nb\nc\n'),
-        (
-            b'#comment1\nfoo\n#comment2\nbar\n',
-            FAIL,
-            b'#comment2\nbar\n#comment1\nfoo\n',
-        ),
-        (
-            b'#comment1\nbar\n#comment2\nfoo\n',
-            PASS,
-            b'#comment1\nbar\n#comment2\nfoo\n',
-        ),
-        (b'#comment\n\nfoo\nbar\n', FAIL, b'#comment\n\nbar\nfoo\n'),
-        (b'#comment\n\nbar\nfoo\n', PASS, b'#comment\n\nbar\nfoo\n'),
-        (
-            b'foo\n\t#comment with indent\nbar\n',
-            FAIL,
-            b'\t#comment with indent\nbar\nfoo\n',
-        ),
-        (
-            b'bar\n\t#comment with indent\nfoo\n',
-            PASS,
-            b'bar\n\t#comment with indent\nfoo\n',
-        ),
-        (b'\nfoo\nbar\n', FAIL, b'bar\n\nfoo\n'),
-        (b'\nbar\nfoo\n', PASS, b'\nbar\nfoo\n'),
-        (
-            b'pyramid-foo==1\npyramid>=2\n',
-            FAIL,
-            b'pyramid>=2\npyramid-foo==1\n',
-        ),
-        (
-            b'a==1\n'
-            b'c>=1\n'
-            b'bbbb!=1\n'
-            b'c-a>=1;python_version>="3.6"\n'
-            b'e>=2\n'
-            b'd>2\n'
-            b'g<2\n'
-            b'f<=2\n',
-            FAIL,
-            b'a==1\n'
-            b'bbbb!=1\n'
-            b'c>=1\n'
-            b'c-a>=1;python_version>="3.6"\n'
-            b'd>2\n'
-            b'e>=2\n'
-            b'f<=2\n'
-            b'g<2\n',
-        ),
-        (b'a==1\nb==1\na==1\n', FAIL, b'a==1\nb==1\n'),
-        (
-            b'a==1\nb==1\n#comment about a\na==1\n',
-            FAIL,
-            b'#comment about a\na==1\nb==1\n',
-        ),
-        (b'ocflib\nDjango\nPyMySQL\n', FAIL, b'Django\nocflib\nPyMySQL\n'),
-        (
-            b'-e git+ssh://git_url@tag#egg=ocflib\nDjango\nPyMySQL\n',
-            FAIL,
-            b'Django\n-e git+ssh://git_url@tag#egg=ocflib\nPyMySQL\n',
-        ),
-        (b'bar\npkg-resources==0.0.0\nfoo\n', FAIL, b'bar\nfoo\n'),
-        (b'foo\npkg-resources==0.0.0\nbar\n', FAIL, b'bar\nfoo\n'),
-        (b'bar\npkg_resources==0.0.0\nfoo\n', FAIL, b'bar\nfoo\n'),
-        (b'foo\npkg_resources==0.0.0\nbar\n', FAIL, b'bar\nfoo\n'),
-        (
-            b'git+ssh://git_url@tag#egg=ocflib\nDjango\nijk\n',
-            FAIL,
-            b'Django\nijk\ngit+ssh://git_url@tag#egg=ocflib\n',
-        ),
-        (
-            b'b==1.0.0\n'
-            b'c=2.0.0 \\\n'
-            b' --hash=sha256:abcd\n'
-            b'a=3.0.0 \\\n'
-            b'  --hash=sha256:a1b1c1d1',
-            FAIL,
-            b'a=3.0.0 \\\n'
-            b'  --hash=sha256:a1b1c1d1\n'
-            b'b==1.0.0\n'
-            b'c=2.0.0 \\\n'
-            b' --hash=sha256:abcd\n',
-        ),
-        (
-            b'a=2.0.0 \\\n --hash=sha256:abcd\nb==1.0.0\n',
-            PASS,
-            b'a=2.0.0 \\\n --hash=sha256:abcd\nb==1.0.0\n',
-        ),
+
+            (b'', [], PASS, b''),
+            (b'\n', [], PASS, b'\n'),
+            (b'# intentionally empty\n', [], PASS, b'# intentionally empty\n'),
+            (b'foo\n# comment at end\n', [], PASS, b'foo\n# comment at end\n'),
+            (b'foo\nbar\n', [], FAIL, b'bar\nfoo\n'),
+            (b'bar\nfoo\n', [], PASS, b'bar\nfoo\n'),
+            (b'a\nc\nb\n', [], FAIL, b'a\nb\nc\n'),
+            (b'a\nc\nb', [], FAIL, b'a\nb\nc\n'),
+            (b'a\nb\nc', [], FAIL, b'a\nb\nc\n'),
+            (
+                    b'#comment1\nfoo\n#comment2\nbar\n',
+                    [],
+                    FAIL,
+                    b'#comment2\nbar\n#comment1\nfoo\n',
+            ),
+            (
+                    b'#comment1\nbar\n#comment2\nfoo\n',
+                    [],
+                    PASS,
+                    b'#comment1\nbar\n#comment2\nfoo\n',
+            ),
+            (b'#comment\n\nfoo\nbar\n', [], FAIL, b'#comment\n\nbar\nfoo\n'),
+            (b'#comment\n\nbar\nfoo\n', [], PASS, b'#comment\n\nbar\nfoo\n'),
+            (
+                    b'foo\n\t#comment with indent\nbar\n',
+                    [],
+                    FAIL,
+                    b'\t#comment with indent\nbar\nfoo\n',
+            ),
+            (
+                    b'bar\n\t#comment with indent\nfoo\n',
+                    [],
+                    PASS,
+                    b'bar\n\t#comment with indent\nfoo\n',
+            ),
+            (b'\nfoo\nbar\n', [], FAIL, b'bar\n\nfoo\n'),
+            (b'\nbar\nfoo\n', [], PASS, b'\nbar\nfoo\n'),
+            (
+                    b'pyramid-foo==1\npyramid>=2\n',
+                    [],
+                    FAIL,
+                    b'pyramid>=2\npyramid-foo==1\n',
+            ),
+            (
+                    b'a==1\n'
+                    b'c>=1\n'
+                    b'bbbb!=1\n'
+                    b'c-a>=1;python_version>="3.6"\n'
+                    b'e>=2\n'
+                    b'd>2\n'
+                    b'g<2\n'
+                    b'f<=2\n',
+                    [],
+                    FAIL,
+                    b'a==1\n'
+                    b'bbbb!=1\n'
+                    b'c>=1\n'
+                    b'c-a>=1;python_version>="3.6"\n'
+                    b'd>2\n'
+                    b'e>=2\n'
+                    b'f<=2\n'
+                    b'g<2\n',
+            ),
+            (b'a==1\nb==1\na==1\n', [], FAIL, b'a==1\nb==1\n'),
+            (
+                    b'a==1\nb==1\n#comment about a\na==1\n',
+                    [],
+                    FAIL,
+                    b'#comment about a\na==1\nb==1\n',
+            ),
+            (b'ocflib\nDjango\nPyMySQL\n', [], FAIL, b'Django\nocflib\nPyMySQL\n'),
+            (
+                    b'-e git+ssh://git_url@tag#egg=ocflib\nDjango\nPyMySQL\n',
+                    [],
+                    FAIL,
+                    b'Django\n-e git+ssh://git_url@tag#egg=ocflib\nPyMySQL\n',
+            ),
+            (b'bar\npkg-resources==0.0.0\nfoo\n', [], FAIL, b'bar\nfoo\n'),
+            (b'foo\npkg-resources==0.0.0\nbar\n', [], FAIL, b'bar\nfoo\n'),
+            (b'bar\npkg_resources==0.0.0\nfoo\n', [], FAIL, b'bar\nfoo\n'),
+            (b'foo\npkg_resources==0.0.0\nbar\n', [], FAIL, b'bar\nfoo\n'),
+            (
+                    b'git+ssh://git_url@tag#egg=ocflib\nDjango\nijk\n',
+                    [],
+                    FAIL,
+                    b'Django\nijk\ngit+ssh://git_url@tag#egg=ocflib\n',
+            ),
+            (
+                    b'b==1.0.0\n'
+                    b'c=2.0.0 \\\n'
+                    b' --hash=sha256:abcd\n'
+                    b'a=3.0.0 \\\n'
+                    b'  --hash=sha256:a1b1c1d1',
+                    [],
+                    FAIL,
+                    b'a=3.0.0 \\\n'
+                    b'  --hash=sha256:a1b1c1d1\n'
+                    b'b==1.0.0\n'
+                    b'c=2.0.0 \\\n'
+                    b' --hash=sha256:abcd\n',
+            ),
+            (
+                    b'a=2.0.0 \\\n --hash=sha256:abcd\nb==1.0.0\n',
+                    [],
+                    PASS,
+                    b'a=2.0.0 \\\n --hash=sha256:abcd\nb==1.0.0\n',
+            ),
+            (b'bar\nfoo\n', ["--fail-without-version"], FAIL, b'bar\nfoo\n'),
+            (b'bar==1.0\nfoo==1.1a\n', ["--fail-without-version"], PASS, b'bar==1.0\nfoo==1.1a\n'),
+            (b'#test\nbar==1.0\nfoo==1.1a\n', ["--fail-without-version"], PASS, b'#test\nbar==1.0\nfoo==1.1a\n'),
+            (b'bar==1.0\n#test\nfoo==1.1a\n', ["--fail-without-version"], PASS, b'bar==1.0\n#test\nfoo==1.1a\n'),
     ),
 )
-def test_integration(input_s, expected_retval, output, tmpdir):
-    path = tmpdir.join('file.txt')
-    path.write_binary(input_s)
+def test_integration(input_s, argv, expected_retval, output, tmpdir):
+        path = tmpdir.join('file.txt')
+        path.write_binary(input_s)
 
-    output_retval = main([str(path)])
+        output_retval = main([str(path)] + argv)
 
-    assert path.read_binary() == output
-    assert output_retval == expected_retval
+        assert path.read_binary() == output
+        assert output_retval == expected_retval
 
 
 def test_requirement_object():
-    top_of_file = Requirement()
-    top_of_file.comments.append(b'#foo')
-    top_of_file.value = b'\n'
+        top_of_file = Requirement()
+        top_of_file.comments.append(b'#foo')
+        top_of_file.value = b'\n'
+
+        requirement_foo = Requirement()
+        requirement_foo.value = b'foo'
+
+        requirement_bar = Requirement()
+        requirement_bar.value = b'bar'
 
-    requirement_foo = Requirement()
-    requirement_foo.value = b'foo'
+        requirements_bar_versioned = Requirement()
+        requirements_bar_versioned.value = b'bar==1.0'
 
-    requirement_bar = Requirement()
-    requirement_bar.value = b'bar'
+        # check for version specification
+        assert top_of_file.contains_version_specifier() is False
+        assert requirement_foo.contains_version_specifier() is False
+        assert requirement_bar.contains_version_specifier() is False
+        assert requirements_bar_versioned.contains_version_specifier() is True
 
-    # This may look redundant, but we need to test both foo.__lt__(bar) and
-    # bar.__lt__(foo)
-    assert requirement_foo > top_of_file
-    assert top_of_file < requirement_foo
-    assert requirement_foo > requirement_bar
-    assert requirement_bar < requirement_foo
+        # This may look redundant, but we need to test both foo.__lt__(bar) and
+        # bar.__lt__(foo)
+        assert requirement_foo > top_of_file
+        assert top_of_file < requirement_foo
+        assert requirement_foo > requirement_bar
+        assert requirement_bar < requirement_foo