learningcircuit--local-deep-research/.pre-commit-hooks/recommend-performance-tests.py

#!/usr/bin/env python3
"""Advisory reminder for subsystem fast tests + opt-in live performance tests.

When staged source files touch a subsystem listed in PATH_RULES, print a
two-tier reminder: (1) the narrow fast mocked tests that catch
logic/contract regressions in seconds, and (2) the opt-in live
performance tests under ``tests/performance/<subsystem>/`` — plus a
nudge to add one if the subsystem doesn't have a live-tests dir yet.

Always exits 0 (advisory). Mirrors recommend-tests.py style.
"""

import sys
from pathlib import Path

# Allow importing sibling module
sys.path.insert(0, str(Path(__file__).resolve().parent))

from _commit_analysis import analyze_commit

# Keep this table at the top of the file — update it alongside any
# reshuffling of tests/performance/<subsystem>/ directories or when a
# new subsystem grows a fast-test suite worth recommending.
#
# Rules are evaluated most-specific-first: a staged file is attributed
# to the FIRST matching rule only, so overlapping prefixes
# (e.g. relevance_filter.py is inside web_search_engines/) don't
# fire two reminders.
PATH_RULES = [
    {
        "name": "relevance_filter",
        "prefixes": (
            "src/local_deep_research/web_search_engines/relevance_filter.py",
        ),
        "fast_tests": [
            "tests/web_search_engines/test_relevance_filter.py",
        ],
        "perf_dir": "tests/performance/relevance_filter/",
        "perf_exists": True,
    },
    {
        "name": "search_engine_adapters",
        "prefixes": ("src/local_deep_research/web_search_engines/engines/",),
        "fast_tests": [
            "tests/web_search_engines/test_search_engine_base.py",
            "tests/web_search_engines/engines/test_full_search.py",
        ],
        "perf_dir": "tests/performance/search_engines/",
        "perf_exists": True,
    },
    {
        "name": "web_search_engines",
        "prefixes": ("src/local_deep_research/web_search_engines/",),
        "fast_tests": [
            "tests/web_search_engines/test_search_engine_base.py",
            "tests/web_search_engines/test_search_engine_factory.py",
        ],
        "perf_dir": None,
        "perf_exists": False,
    },
    {
        "name": "advanced_search_system",
        "prefixes": ("src/local_deep_research/advanced_search_system/",),
        "fast_tests": [
            "tests/advanced_search_system/constraint_checking/test_threshold_checker.py",
            "tests/advanced_search_system/constraint_checking/test_dual_confidence_checker.py",
            "tests/advanced_search_system/constraint_checking/test_base_constraint_checker.py",
        ],
        "perf_dir": "tests/performance/strategies/",
        "perf_exists": True,
    },
    {
        "name": "content_fetcher",
        "prefixes": (
            "src/local_deep_research/content_fetcher/",
            "src/local_deep_research/research_library/downloaders/",
        ),
        "fast_tests": [
            "tests/content_fetcher/test_html_content_extraction.py",
            "tests/research_library/downloaders/test_html_downloader.py",
            "tests/research_library/downloaders/test_extraction_pipeline.py",
            "tests/research_library/downloaders/test_metadata_extractor.py",
        ],
        "perf_dir": "tests/performance/content_fetcher/",
        "perf_exists": True,
    },
]


def _print_rule(rule, matched):
    print()
    print(f"  \033[33mPerformance Test Reminder — {rule['name']}\033[0m")
    print("  " + "-" * 40)
    print(f"  You touched {len(matched)} file(s) under {rule['name']}:")
    for p in matched:
        print(f"    - {p}")
    print()
    print("  Fast tests (mocked, ~5-30s) to run now:")
    print(f"    pdm run pytest {' '.join(rule['fast_tests'])}")
    print()
    if rule["perf_exists"]:
        print("  Heavier real-service tests (opt-in, live):")
        print(
            f"    LDR_TESTING_WITH_MOCKS=false pdm run pytest "
            f"{rule['perf_dir']} -m integration"
        )
        print(
            "  If you added behavior only live tests catch, consider "
            f"adding a test under {rule['perf_dir']}."
        )
    elif rule["perf_dir"]:
        print(
            f"  No live tests exist yet at {rule['perf_dir']}. If your "
            "change needs live-service coverage, consider adding one."
        )
    else:
        print(
            "  No live-tests dir for this subsystem yet — consider "
            "adding one if live coverage matters."
        )
    print()


def main():
    analysis = analyze_commit()
    if not analysis.source_files:
        return 0
    staged = [f.path for f in analysis.source_files]
    # First-match-wins assignment so overlapping prefixes
    # (relevance_filter.py is inside web_search_engines/) don't
    # double-fire. PATH_RULES is ordered most-specific-first.
    per_rule: dict[str, list[str]] = {}
    for path in staged:
        for rule in PATH_RULES:
            if path.startswith(rule["prefixes"]):
                per_rule.setdefault(rule["name"], []).append(path)
                break
    for rule in PATH_RULES:
        matched = per_rule.get(rule["name"], [])
        if matched:
            _print_rule(rule, matched)
    return 0


if __name__ == "__main__":
    sys.exit(main())