Requirements parser handles trailing comments (#315)

requirements parser handles trailing comments

Author: clavedeluna

src/codemodder/project_analysis/file_parsers/requirements_txt_file_parser.py

@@ -28,7 +28,9 @@ def _parse_file(self, file: Path) -> PackageStore | None:
             logger.debug("Unknown encoding for file: %s", file)
             return None
 
-        dependencies = set(line.strip() for line in lines if not line.startswith("#"))
+        dependencies = set(
+            line.split("#")[0].strip() for line in lines if not line.startswith("#")
+        )
 
         return PackageStore(
             type=self.file_type,

tests/conftest.py

@@ -44,6 +44,15 @@ def pkg_with_reqs_txt(tmp_path_factory):
     return base_dir
 
 
+@pytest.fixture(scope="module")
+def pkg_with_reqs_txt_and_comments(tmp_path_factory):
+    base_dir = tmp_path_factory.mktemp("foo")
+    req_file = base_dir / "requirements.txt"
+    reqs = "# comment\nrequests==2.31.0\nblack==23.7.*\nmypy~=1.4 # comment\npylint>1\n"
+    req_file.write_text(reqs)
+    return base_dir
+
+
 @pytest.fixture(scope="module")
 def pkg_with_reqs_txt_utf_16(tmp_path_factory):
     base_dir = tmp_path_factory.mktemp("foo")

tests/project_analysis/file_parsers/test_requirements_txt_file_parser.py

@@ -37,3 +37,13 @@ def test_open_error(self, pkg_with_reqs_txt, mocker):
         parser = RequirementsTxtParser(pkg_with_reqs_txt)
         found = parser.parse()
         assert len(found) == 0
+
+    def test_trailing_comments(self, pkg_with_reqs_txt_and_comments):
+        parser = RequirementsTxtParser(pkg_with_reqs_txt_and_comments)
+        found = parser.parse()
+        assert len(found) == 1
+        store = found[0]
+        assert store.type.value == "requirements.txt"
+        assert store.file == pkg_with_reqs_txt_and_comments / parser.file_type.value
+        assert store.py_versions == []
+        assert len(store.dependencies) == 4