CM-55207-Fix tests and lint

Author: mateusz-sterczewski

cycode/cli/files_collector/commit_range_documents.py

@@ -67,11 +67,12 @@ def collect_commit_range_diff_documents(
     commit_documents_to_scan = []
 
     repo = git_proxy.get_repo(path)
-    
+
     normalized_commit_range = normalize_commit_range(commit_range, path)
-    
+
     total_commits_count = int(repo.git.rev_list('--count', normalized_commit_range))
-    logger.debug('Calculating diffs for %s commits in the commit range %s', total_commits_count, normalized_commit_range)
+    logger.debug('Calculating diffs for %s commits in the commit range %s',
+        total_commits_count, normalized_commit_range)
 
     progress_bar.set_section_length(ScanProgressBarSection.PREPARE_LOCAL_FILES, total_commits_count)
 
@@ -100,9 +101,9 @@ def collect_commit_range_diff_documents(
         logger.debug(
             'Found all relevant files in commit %s',
             {
-                'path': path, 
-                'commit_range': commit_range, 
-                'normalized_commit_range': normalized_commit_range, 
+                'path': path,
+                'commit_range': commit_range,
+                'normalized_commit_range': normalized_commit_range,
                 'commit_id': commit_id,
             },
         )
@@ -437,7 +438,8 @@ def get_pre_commit_modified_documents(
 
 
 def parse_commit_range(commit_range: str, path: str) -> tuple[Optional[str], Optional[str], Optional[str]]:
-    """Parses a git commit range string and returns the full SHAs for the 'from' and 'to' commits as well as the separator.
+    """Parses a git commit range string and returns the full SHAs for the 'from' and 'to' commits.
+    Also, it returns the separator in the commit range.
 
     Supports:
     - 'from..to'
@@ -462,8 +464,10 @@ def parse_commit_range(commit_range: str, path: str) -> tuple[Optional[str], Opt
     # If a spec is empty (e.g., from '..master'), default it to 'HEAD'
     if not from_spec:
         from_spec = consts.GIT_HEAD_COMMIT_REV
+        separator = '..'
     if not to_spec:
         to_spec = consts.GIT_HEAD_COMMIT_REV
+        separator = '..'
 
     try:
         # Use rev_parse to resolve each specifier to its full commit SHA
@@ -476,7 +480,7 @@ def parse_commit_range(commit_range: str, path: str) -> tuple[Optional[str], Opt
 
 def normalize_commit_range(commit_range: str, path: str) -> str:
     """Normalize a commit range string to handle various formats consistently with all scan types.
-        
+
     Returns:
         A normalized commit range string suitable for Git operations (e.g., 'full_sha1..full_sha2')
     """
@@ -488,7 +492,7 @@ def normalize_commit_range(commit_range: str, path: str) -> str:
         )
         # Fall back to using the raw commit_range string
         return commit_range
-    
+
     # Construct a normalized range string using the original separator for iter_commits
     # This preserves the semantics of two-dot vs three-dot syntax
     normalized_commit_range = f'{from_commit_rev}{separator}{to_commit_rev}'
@@ -497,4 +501,4 @@ def normalize_commit_range(commit_range: str, path: str) -> str:
         commit_range,
         normalized_commit_range,
     )
-    return normalized_commit_range
+    return normalized_commit_range

tests/cli/files_collector/test_commit_range_documents.py

@@ -847,40 +847,40 @@ def test_two_dot_linear_history(self) -> None:
         with temporary_git_repository() as (temp_dir, repo):
             a, b, c = self._make_linear_history(repo, temp_dir)
 
-            parsed_from, parsed_to = parse_commit_range(f'{a}..{c}', temp_dir)
-            assert (parsed_from, parsed_to) == (a, c)
+            parsed_from, parsed_to, separator = parse_commit_range(f'{a}..{c}', temp_dir)
+            assert (parsed_from, parsed_to, separator) == (a, c, '..')
 
     def test_three_dot_linear_history(self) -> None:
         """For 'A...C' in linear history, expect (A,C)."""
         with temporary_git_repository() as (temp_dir, repo):
             a, b, c = self._make_linear_history(repo, temp_dir)
 
-            parsed_from, parsed_to = parse_commit_range(f'{a}...{c}', temp_dir)
-            assert (parsed_from, parsed_to) == (a, c)
+            parsed_from, parsed_to, separator = parse_commit_range(f'{a}...{c}', temp_dir)
+            assert (parsed_from, parsed_to, separator) == (a, c, '...')
 
     def test_open_right_linear_history(self) -> None:
         """For 'A..', expect (A,HEAD=C)."""
         with temporary_git_repository() as (temp_dir, repo):
             a, b, c = self._make_linear_history(repo, temp_dir)
 
-            parsed_from, parsed_to = parse_commit_range(f'{a}..', temp_dir)
-            assert (parsed_from, parsed_to) == (a, c)
+            parsed_from, parsed_to, separator = parse_commit_range(f'{a}..', temp_dir)
+            assert (parsed_from, parsed_to, separator) == (a, c, '..')
 
     def test_open_left_linear_history(self) -> None:
         """For '..C' where HEAD==C, expect (HEAD=C,C)."""
         with temporary_git_repository() as (temp_dir, repo):
             a, b, c = self._make_linear_history(repo, temp_dir)
 
-            parsed_from, parsed_to = parse_commit_range(f'..{c}', temp_dir)
-            assert (parsed_from, parsed_to) == (c, c)
+            parsed_from, parsed_to, separator = parse_commit_range(f'..{c}', temp_dir)
+            assert (parsed_from, parsed_to, separator) == (c, c, '..')
 
     def test_single_commit_spec(self) -> None:
         """For 'A', expect (A,HEAD=C)."""
         with temporary_git_repository() as (temp_dir, repo):
             a, b, c = self._make_linear_history(repo, temp_dir)
 
-            parsed_from, parsed_to = parse_commit_range(a, temp_dir)
-            assert (parsed_from, parsed_to) == (a, c)
+            parsed_from, parsed_to, separator = parse_commit_range(a, temp_dir)
+            assert (parsed_from, parsed_to, separator) == (a, c, '..')
 
 
 class TestParsePreReceiveInput:
@@ -1085,7 +1085,8 @@ def test_collect_with_various_commit_range_formats(self) -> None:
             documents = collect_commit_range_diff_documents(mock_ctx, temp_dir, commit_range)
             assert len(documents) == 2, f'Expected 2 documents from range A..C, got {len(documents)}'
             commit_ids_in_documents = {doc.unique_id for doc in documents if doc.unique_id}
-            assert b_commit.hexsha in commit_ids_in_documents and c_commit.hexsha in commit_ids_in_documents
+            assert b_commit.hexsha in commit_ids_in_documents
+            assert c_commit.hexsha in commit_ids_in_documents
 
             # Test three-dot range - should collect documents from commits B and C (2 commits, 2 documents)
             commit_range = f'{a_commit.hexsha}...{c_commit.hexsha}'
@@ -1101,4 +1102,4 @@ def test_collect_with_various_commit_range_formats(self) -> None:
             # Test single commit spec - should be interpreted as A..HEAD (commits B and C, 2 documents)
             commit_range = a_commit.hexsha
             documents = collect_commit_range_diff_documents(mock_ctx, temp_dir, commit_range)
-            assert len(documents) == 2, f'Expected 2 documents from single commit A (interpreted as A..HEAD), got {len(documents)}'
+            assert len(documents) == 2, f'Expected 2 documents from single commit A, got {len(documents)}'