Fix: Include generated tests in PR summaries

codeflash/result/create_pr.py

@@ -61,7 +61,10 @@ def existing_tests_source_for(
     instrumented_to_original: dict[Path, Path] = {}
     if test_files_registry:
         for registry_tf in test_files_registry.test_files:
+            # For existing tests, map instrumented → original
+            # For generated tests (original_file_path=None), map instrumented → instrumented (self)
             if registry_tf.original_file_path:
+                # Existing test: map to original file
                 if registry_tf.instrumented_behavior_file_path:
                     instrumented_to_original[registry_tf.instrumented_behavior_file_path.resolve()] = (
                         registry_tf.original_file_path.resolve()
@@ -76,13 +79,46 @@ def existing_tests_source_for(
                     logger.debug(
                         f"[PR-DEBUG] Mapping (perf): {registry_tf.benchmarking_file_path.name} -> {registry_tf.original_file_path.name}"
                     )
+            else:
+                # Generated test (no original file): map to itself
+                if registry_tf.instrumented_behavior_file_path:
+                    behavior_resolved = registry_tf.instrumented_behavior_file_path.resolve()
+                    instrumented_to_original[behavior_resolved] = behavior_resolved
+                    logger.debug(
+                        f"[PR-DEBUG] Mapping (generated behavior): {registry_tf.instrumented_behavior_file_path.name} -> {registry_tf.instrumented_behavior_file_path.name}"
+                    )
+                if registry_tf.benchmarking_file_path:
+                    perf_resolved = registry_tf.benchmarking_file_path.resolve()
+                    instrumented_to_original[perf_resolved] = perf_resolved
+                    logger.debug(
+                        f"[PR-DEBUG] Mapping (generated perf): {registry_tf.benchmarking_file_path.name} -> {registry_tf.benchmarking_file_path.name}"
+                    )
 
     # Resolve all paths to absolute for consistent comparison
+    # Include both existing tests (from function_to_tests) and generated tests (from registry)
     non_generated_tests: set[Path] = set()
     for test_file in test_files:
         resolved = test_file.tests_in_file.test_file.resolve()
         non_generated_tests.add(resolved)
         logger.debug(f"[PR-DEBUG] Added to non_generated_tests: {resolved}")
+
+    # Also add generated tests from registry to the set
+    # Generated tests have original_file_path=None and should be included by their instrumented paths
+    if test_files_registry:
+        for registry_tf in test_files_registry.test_files:
+            if registry_tf.original_file_path is None:  # Generated test
+                if registry_tf.instrumented_behavior_file_path:
+                    generated_resolved = registry_tf.instrumented_behavior_file_path.resolve()
+                    non_generated_tests.add(generated_resolved)
+                    logger.debug(
+                        f"[PR-DEBUG] Added generated test (behavior) to non_generated_tests: {generated_resolved}"
+                    )
+                if registry_tf.benchmarking_file_path:
+                    generated_perf_resolved = registry_tf.benchmarking_file_path.resolve()
+                    non_generated_tests.add(generated_perf_resolved)
+                    logger.debug(
+                        f"[PR-DEBUG] Added generated test (perf) to non_generated_tests: {generated_perf_resolved}"
+                    )
     # TODO confirm that original and optimized have the same keys
     all_invocation_ids = original_runtimes_all.keys() | optimized_runtimes_all.keys()
     logger.debug(f"[PR-DEBUG] Processing {len(all_invocation_ids)} invocation_ids")

tests/result/test_pr_generated_tests.py

@@ -0,0 +1,93 @@
+"""Unit tests for PR creation with generated tests.
+
+Bug: Generated tests with original_file_path=None are excluded from PR summaries.
+Location: codeflash/result/create_pr.py:62-85
+"""
+from pathlib import Path
+
+import pytest
+
+from codeflash.models.models import TestFile, TestFiles
+from codeflash.models.test_type import TestType
+
+
+@pytest.fixture
+def test_files_registry() -> TestFiles:
+    """Create a test files registry with both generated and existing tests."""
+    registry = TestFiles(test_files=[])
+
+    # Generated test (has None for original_file_path)
+    generated_behavior = Path("/workspace/target/src/test_saveCronStore__unit_test_0.test.ts")
+    generated_perf = Path("/workspace/target/src/test_saveCronStore__perf_test_0.test.ts")
+
+    generated_test = TestFile(
+        instrumented_behavior_file_path=generated_behavior,
+        benchmarking_file_path=generated_perf,
+        original_file_path=None,  # Generated tests have no original
+        original_source="test code",
+        test_type=TestType.GENERATED_REGRESSION,
+        tests_in_file=None,
+    )
+    registry.add(generated_test)
+
+    # Existing instrumented test (has original_file_path)
+    existing_behavior = Path("/workspace/target/src/store__perfinstrumented.test.ts")
+    existing_perf = Path("/workspace/target/src/store__perfonlyinstrumented.test.ts")
+    existing_original = Path("/workspace/target/src/store.test.ts")
+
+    existing_test = TestFile(
+        instrumented_behavior_file_path=existing_behavior,
+        benchmarking_file_path=existing_perf,
+        original_file_path=existing_original,
+        original_source=None,
+        test_type=TestType.EXISTING_UNIT_TEST,
+        tests_in_file=None,
+    )
+    registry.add(existing_test)
+
+    return registry
+
+
+def test_instrumented_to_original_mapping_includes_generated_tests(test_files_registry: TestFiles) -> None:
+    """Test that the instrumented_to_original mapping includes generated tests.
+
+    This is the direct test of the bug at lines 62-78 in create_pr.py.
+    """
+    # Build the mapping as create_pr.py does (lines 62-95) - with fix
+    instrumented_to_original = {}
+    for registry_tf in test_files_registry.test_files:
+        # For existing tests, map instrumented → original
+        # For generated tests (original_file_path=None), map instrumented → instrumented (self)
+        if registry_tf.original_file_path:
+            # Existing test: map to original file
+            if registry_tf.instrumented_behavior_file_path:
+                instrumented_to_original[registry_tf.instrumented_behavior_file_path.resolve()] = (
+                    registry_tf.original_file_path.resolve()
+                )
+            if registry_tf.benchmarking_file_path:
+                instrumented_to_original[registry_tf.benchmarking_file_path.resolve()] = (
+                    registry_tf.original_file_path.resolve()
+                )
+        else:
+            # Generated test (no original file): map to itself
+            if registry_tf.instrumented_behavior_file_path:
+                behavior_resolved = registry_tf.instrumented_behavior_file_path.resolve()
+                instrumented_to_original[behavior_resolved] = behavior_resolved
+            if registry_tf.benchmarking_file_path:
+                perf_resolved = registry_tf.benchmarking_file_path.resolve()
+                instrumented_to_original[perf_resolved] = perf_resolved
+
+    # Check the bug
+    generated_behavior_path = Path("/workspace/target/src/test_saveCronStore__unit_test_0.test.ts").resolve()
+    existing_behavior_path = Path("/workspace/target/src/store__perfinstrumented.test.ts").resolve()
+
+    # Existing test should be in mapping
+    assert existing_behavior_path in instrumented_to_original, \
+        "Existing test should be in mapping"
+
+    # After fix: Generated test SHOULD be in mapping (mapped to itself)
+    assert generated_behavior_path in instrumented_to_original, \
+        "Generated test should be in mapping"
+    # Verify it's mapped to itself
+    assert instrumented_to_original[generated_behavior_path] == generated_behavior_path, \
+        "Generated test should map to itself"