Merge pull request #24 from aboutcode-org/23-fix-mutation-ending-without-newline

Fix mutation in files ending without a newline

Author: JonoYang

src/matchcode_toolkit/stemming.py

@@ -149,7 +149,8 @@ def get_stem_code(location):
 
     # Apply mutations bottom-up
     mutations = dict(sorted(mutations.items(), reverse=True))
-    text = source.decode()
+    # Ensure the text always ends with a newline character.
+    text = source.decode().rstrip("\n") + "\n"
     cur_count = 0
     lines = text.splitlines(keepends=True)
     successive_line_count = [cur_count := cur_count + len(line) for line in lines]

tests/test_stemming.py

@@ -78,3 +78,9 @@ def test_rust_code_stemming(self):
         expected_file_location = self.test_data_dir / "rust/metrics-stemmeds.rs"
         results = stemming.get_stem_code(location=str(file_location))
         check_against_expected_code_file(results, expected_file_location)
+
+    def test_javascript_code_stemming_endwith_no_newline(self):
+        file_location = self.test_data_dir / "javascript/main.js"
+        expected_file_location = self.test_data_dir / "javascript/main-stemmed.js"
+        results = stemming.get_stem_code(location=str(file_location))
+        check_against_expected_code_file(results, expected_file_location)