Parallelize test discovery

[KRRT7]

PR Type

Enhancement, Tests

---

Description

• Parallelize unit test discovery using processes

• Introduce caching to skip processed files

• Return total discovered count from discovery

• Update call sites and tests for new API

---

Changes walkthrough 📝

	Relevant files
Enhancement	discover_unit_tests.py Add parallel test discovery and caching                                    codeflash/discovery/discover_unit_tests.py Imported multiprocessing and ProcessPoolExecutor Added _process_single_test_file helper function Parallelized processing with futures executor Implemented caching skip logic and result counting +226/-173
Bug_fix	functions_to_optimize.py Unpack discover_unit_tests return tuple                                    codeflash/discovery/functions_to_optimize.py Unpacked discover_unit_tests tuple return Adjusted function_tests assignment +1/-1      optimizer.py Unpack discovery results in optimizer                                        codeflash/optimization/optimizer.py Unpacked discover_unit_tests tuple return Updated num_discovered_tests assignment +1/-2      concolic_testing.py Support new discovery return in concolic testing                  codeflash/verification/concolic_testing.py Unpacked discover_unit_tests return tuple Updated logging of discovered test count +1/-2
Tests	test_unit_test_discovery.py Update tests for new discover_unit_tests signature              tests/test_unit_test_discovery.py Updated tests to unpack tuple return Ensured tests consume new signature +17/-17

---

Need help?

Type /help how to ... in the comments thread for any questions about PR-Agent usage.

Check out the documentation for more information.

[github-actions]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 4 🔵🔵🔵🔵⚪

🧪 PR contains tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Missing Commit The SQLite connection does not call commit() after inserting cache entries, so cached test data may never be persisted to disk. for test_file, functions in uncached_files.items(): _, results, count, cache_entries = _process_single_test_file( test_file, functions, project_root_path, test_framework ) total_count += count file_hash = TestsCache.compute_file_hash(str(test_file)) for cache_entry in cache_entries: tests_cache.insert_test(file_path=str(test_file), file_hash=file_hash, **cache_entry) for qualified_name, function_called in results: Cache Staleness The insert_test method no longer deletes old entries for a file, which can lead to duplicate or stale records when tests are re-discovered. Consider clearing or replacing entries by file_path and file_hash. def insert_test( self, file_path: str, file_hash: str, qualified_name_with_modules_from_root: str, function_name: str, test_class: str, test_function: str, test_type: TestType, line_number: int, col_number: int, ) -> None: test_type_value = test_type.value if hasattr(test_type, "value") else test_type self.cur.execute( "INSERT INTO discovered_tests VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)", (

[github-actions]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Persist cache updates Add a DELETE statement before inserting to remove stale cache entries for the same file and hash, and call self.connection.commit() after the insert so the database updates actually persist. codeflash/discovery/discover_unit_tests.py [80-96] def insert_test( self, file_path: str, file_hash: str, qualified_name_with_modules_from_root: str, function_name: str, test_class: str, test_function: str, test_type: TestType, line_number: int, col_number: int, ) -> None: + # purge old entries for this file and hash + self.cur.execute( + "DELETE FROM discovered_tests WHERE file_path = ? AND file_hash = ?", + (file_path, file_hash), + ) test_type_value = test_type.value if hasattr(test_type, "value") else test_type self.cur.execute( "INSERT INTO discovered_tests VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)", ( file_path, file_hash, qualified_name_with_modules_from_root, function_name, test_class, test_function, test_type_value, line_number, col_number, ), ) + self.connection.commit() Suggestion importance[1-10]: 7 __ Why: Adding a DELETE before the insert prevents stale rows accumulating, and calling self.connection.commit() ensures the new entries are persisted immediately.	Medium
General	Limit worker pool correctly Bound the worker pool size to the number of uncached files rather than the total file map so you don’t spawn idle processes when most files are already cached. codeflash/discovery/discover_unit_tests.py [458] -max_workers = min(len(file_to_test_map) or 1, multiprocessing.cpu_count()) +# after segregating uncached_files +max_workers = min(len(uncached_files) or 1, multiprocessing.cpu_count()) Suggestion importance[1-10]: 6 __ Why: Bounding max_workers to len(uncached_files) avoids spawning idle processes when many files are already cached, improving resource usage.	Low
	Use serializable worker results Serializing custom FunctionCalledInTest objects across processes can fail; convert each result into plain dictionaries or tuples before returning and rebuild the objects in the main process. codeflash/discovery/discover_unit_tests.py [445] -return str(test_file), results, len(results), cache_entries +# at end of _process_single_test_file +serial_results = [ + ( + qualified_name, + { + "test_file": str(fc.tests_in_file.test_file), + "test_class": fc.tests_in_file.test_class, + "test_function": fc.tests_in_file.test_function, + "test_type": fc.tests_in_file.test_type.value if hasattr(fc.tests_in_file.test_type, "value") else fc.tests_in_file.test_type, + "line": fc.position.line_no, + "col": fc.position.col_no, + }, + ) + for qualified_name, fc in results +] +return str(test_file), serial_results, len(results), cache_entries Suggestion importance[1-10]: 5 __ Why: Converting FunctionCalledInTest objects to primitive types avoids pickling errors in ProcessPoolExecutor, enhancing robustness when deserializing results.	Low