Merge pull request #1813 from codeflash-ai/fix/detect-deletion-only-diffs

fix: detect functions in deletion-only git diffs

Author: misrasaurabh1

codeflash/code_utils/git_utils.py

@@ -59,6 +59,10 @@ def get_git_diff(
 
         logger.debug(f"deleted lines: {del_line_no}")
 
+        if not add_line_no and del_line_no:
+            # Deletion-only changes: use hunk target start lines so we can still
+            # match the surrounding function in the current (target) file.
+            add_line_no = [hunk.target_start for hunk in patched_file]
         change_list[file_path] = add_line_no
     return change_list
 

tests/test_git_utils.py

@@ -3,7 +3,12 @@
 
 import git
 
-from codeflash.code_utils.git_utils import check_and_push_branch, check_running_in_git_repo, get_repo_owner_and_name
+from codeflash.code_utils.git_utils import (
+    check_and_push_branch,
+    check_running_in_git_repo,
+    get_git_diff,
+    get_repo_owner_and_name,
+)
 
 
 class TestGitUtils(unittest.TestCase):
@@ -115,5 +120,136 @@ def test_check_and_push_branch_detached_head(self, mock_repo):
         mock_origin.push.assert_not_called()
 
 
+DELETION_ONLY_DIFF = """\
+--- a/example.py
++++ b/example.py
+@@ -5,7 +5,5 @@ def foo():
+     a = 1
+     b = 2
+-    c = 3
+-    d = 4
+     e = 5
+     return a + b + e
+
+"""
+
+ADDITION_ONLY_DIFF = """\
+--- a/example.py
++++ b/example.py
+@@ -5,5 +5,7 @@ def foo():
+     a = 1
+     b = 2
++    c = 3
++    d = 4
+     e = 5
+     return a + b + e
+
+"""
+
+MIXED_DIFF = """\
+--- a/example.py
++++ b/example.py
+@@ -5,6 +5,6 @@ def foo():
+     a = 1
+     b = 2
+-    c = 3
++    c = 30
+     e = 5
+     return a + b + e
+
+"""
+
+MULTI_HUNK_DELETION_ONLY_DIFF = """\
+--- a/example.py
++++ b/example.py
+@@ -5,7 +5,5 @@ def foo():
+     a = 1
+     b = 2
+-    c = 3
+-    d = 4
+     e = 5
+     return a + b + e
+
+@@ -20,6 +18,4 @@ def bar():
+     x = 1
+     y = 2
+-    z = 3
+-    w = 4
+     return x + y
+
+"""
+
+
+class TestGetGitDiffDeletionOnly(unittest.TestCase):
+    @patch("codeflash.code_utils.git_utils.git.Repo")
+    def test_deletion_only_diff_returns_hunk_target_starts(self, mock_repo_cls):
+        repo = mock_repo_cls.return_value
+        repo.head.commit.hexsha = "abc123"
+        repo.working_dir = "/repo"
+        repo.git.diff.return_value = DELETION_ONLY_DIFF
+
+        result = get_git_diff(repo_directory=None, uncommitted_changes=True)
+
+        assert len(result) == 1
+        key = list(result.keys())[0]
+        assert str(key).endswith("example.py")
+        # The hunk target_start is 5 — this is the fix: deletion-only diffs
+        # should still report line numbers so the surrounding function is found.
+        assert result[key] == [5]
+
+    @patch("codeflash.code_utils.git_utils.git.Repo")
+    def test_addition_only_diff_returns_added_lines(self, mock_repo_cls):
+        repo = mock_repo_cls.return_value
+        repo.head.commit.hexsha = "abc123"
+        repo.working_dir = "/repo"
+        repo.git.diff.return_value = ADDITION_ONLY_DIFF
+
+        result = get_git_diff(repo_directory=None, uncommitted_changes=True)
+
+        key = list(result.keys())[0]
+        # Added lines are at target line numbers 7 and 8
+        assert result[key] == [7, 8]
+
+    @patch("codeflash.code_utils.git_utils.git.Repo")
+    def test_mixed_diff_returns_only_added_lines(self, mock_repo_cls):
+        repo = mock_repo_cls.return_value
+        repo.head.commit.hexsha = "abc123"
+        repo.working_dir = "/repo"
+        repo.git.diff.return_value = MIXED_DIFF
+
+        result = get_git_diff(repo_directory=None, uncommitted_changes=True)
+
+        key = list(result.keys())[0]
+        # Only the added line (c = 30) at target line 7
+        assert result[key] == [7]
+
+    @patch("codeflash.code_utils.git_utils.git.Repo")
+    def test_multi_hunk_deletion_only_returns_all_hunk_starts(self, mock_repo_cls):
+        repo = mock_repo_cls.return_value
+        repo.head.commit.hexsha = "abc123"
+        repo.working_dir = "/repo"
+        repo.git.diff.return_value = MULTI_HUNK_DELETION_ONLY_DIFF
+
+        result = get_git_diff(repo_directory=None, uncommitted_changes=True)
+
+        key = list(result.keys())[0]
+        # Two hunks with target_start 5 and 18
+        assert result[key] == [5, 18]
+
+    @patch("codeflash.code_utils.git_utils.git.Repo")
+    def test_deletion_only_diff_does_not_return_empty_list(self, mock_repo_cls):
+        repo = mock_repo_cls.return_value
+        repo.head.commit.hexsha = "abc123"
+        repo.working_dir = "/repo"
+        repo.git.diff.return_value = DELETION_ONLY_DIFF
+
+        result = get_git_diff(repo_directory=None, uncommitted_changes=True)
+
+        key = list(result.keys())[0]
+        # Without the fix, this would be an empty list, causing the function
+        # to be missed during discovery.
+        assert len(result[key]) > 0
+
+
 if __name__ == "__main__":
     unittest.main()