test_repo_card_performance.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """ |
| 2 | Tier 7 — Performance tests for enrich_repo_cards(). |
| 3 | |
| 4 | These tests establish latency baselines that should hold on the CI database. |
| 5 | They are deliberately conservative — failing here signals a query regression, |
| 6 | not a slow machine. |
| 7 | |
| 8 | Test IDs |
| 9 | -------- |
| 10 | T700 — single-repo enrichment completes in < 100 ms |
| 11 | T701 — 10-repo batch completes in < 200 ms (sub-linear scaling) |
| 12 | T702 — p95 latency across 20 repeated single-repo calls is < 80 ms |
| 13 | T703 — enriching an empty repo (no intel) is faster than one with full data |
| 14 | """ |
| 15 | from __future__ import annotations |
| 16 | |
| 17 | import statistics |
| 18 | import time |
| 19 | from datetime import datetime, timedelta, timezone |
| 20 | |
| 21 | import pytest |
| 22 | from sqlalchemy.ext.asyncio import AsyncSession |
| 23 | |
| 24 | from musehub.db.musehub_intel_models import MusehubIntelBreakageMeta, MusehubIntelDead, MusehubSymbolIntel |
| 25 | from musehub.services.repo_card_enrichment import enrich_repo_cards |
| 26 | from tests.factories import create_commit, create_repo |
| 27 | |
| 28 | |
| 29 | def _utc_now() -> datetime: |
| 30 | return datetime.now(tz=timezone.utc) |
| 31 | |
| 32 | |
| 33 | async def _seed_full_repo(db: AsyncSession) -> str: |
| 34 | """Seed a repo with commits, symbols, dead rows, and breakage meta.""" |
| 35 | repo = await create_repo(db, visibility="public") |
| 36 | for i in range(10): |
| 37 | await create_commit(db, repo.repo_id, timestamp=_utc_now() - timedelta(days=i)) |
| 38 | for i in range(20): |
| 39 | db.add(MusehubSymbolIntel( |
| 40 | repo_id=repo.repo_id, |
| 41 | address=f"src/mod.py::fn_{i}", |
| 42 | churn_30d=i, |
| 43 | blast=i * 2, |
| 44 | )) |
| 45 | db.add(MusehubIntelDead( |
| 46 | repo_id=repo.repo_id, |
| 47 | address="src/old.py::dead_fn", |
| 48 | kind="function", |
| 49 | confidence="high", |
| 50 | ref="main", |
| 51 | )) |
| 52 | db.add(MusehubIntelBreakageMeta( |
| 53 | repo_id=repo.repo_id, |
| 54 | total_issues=1, |
| 55 | error_count=0, |
| 56 | warning_count=1, |
| 57 | file_count=1, |
| 58 | ref="main", |
| 59 | )) |
| 60 | await db.commit() |
| 61 | return repo.repo_id |
| 62 | |
| 63 | |
| 64 | # --------------------------------------------------------------------------- |
| 65 | # T700 — single-repo enrichment < 100 ms |
| 66 | # --------------------------------------------------------------------------- |
| 67 | |
| 68 | @pytest.mark.asyncio |
| 69 | async def test_t700_single_repo_under_100ms(db_session: AsyncSession) -> None: |
| 70 | """T700: enriching one fully-populated repo completes in < 100 ms.""" |
| 71 | repo_id = await _seed_full_repo(db_session) |
| 72 | |
| 73 | t0 = time.monotonic() |
| 74 | await enrich_repo_cards(db_session, [repo_id]) |
| 75 | elapsed_ms = (time.monotonic() - t0) * 1000 |
| 76 | |
| 77 | assert elapsed_ms < 100, f"Single-repo enrichment took {elapsed_ms:.1f} ms" |
| 78 | |
| 79 | |
| 80 | # --------------------------------------------------------------------------- |
| 81 | # T701 — 10-repo batch < 200 ms |
| 82 | # --------------------------------------------------------------------------- |
| 83 | |
| 84 | @pytest.mark.asyncio |
| 85 | async def test_t701_ten_repo_batch_under_200ms(db_session: AsyncSession) -> None: |
| 86 | """T701: enriching 10 repos completes in < 200 ms (sub-linear vs T700).""" |
| 87 | repo_ids = [await _seed_full_repo(db_session) for _ in range(10)] |
| 88 | |
| 89 | t0 = time.monotonic() |
| 90 | await enrich_repo_cards(db_session, repo_ids) |
| 91 | elapsed_ms = (time.monotonic() - t0) * 1000 |
| 92 | |
| 93 | assert elapsed_ms < 200, f"10-repo batch took {elapsed_ms:.1f} ms" |
| 94 | |
| 95 | |
| 96 | # --------------------------------------------------------------------------- |
| 97 | # T702 — p95 latency across 20 calls < 80 ms |
| 98 | # --------------------------------------------------------------------------- |
| 99 | |
| 100 | @pytest.mark.asyncio |
| 101 | async def test_t702_p95_single_repo_under_80ms(db_session: AsyncSession) -> None: |
| 102 | """T702: p95 latency across 20 repeated single-repo calls is < 80 ms.""" |
| 103 | repo_id = await _seed_full_repo(db_session) |
| 104 | |
| 105 | latencies = [] |
| 106 | for _ in range(20): |
| 107 | t0 = time.monotonic() |
| 108 | await enrich_repo_cards(db_session, [repo_id]) |
| 109 | latencies.append((time.monotonic() - t0) * 1000) |
| 110 | |
| 111 | p95 = statistics.quantiles(latencies, n=20)[18] # 95th percentile |
| 112 | assert p95 < 80, f"p95 latency was {p95:.1f} ms — expected < 80 ms" |
| 113 | |
| 114 | |
| 115 | # --------------------------------------------------------------------------- |
| 116 | # T703 — empty repo faster than full repo |
| 117 | # --------------------------------------------------------------------------- |
| 118 | |
| 119 | @pytest.mark.asyncio |
| 120 | async def test_t703_empty_repo_faster_than_full_repo(db_session: AsyncSession) -> None: |
| 121 | """T703: enriching an empty repo is not slower than a fully-populated one.""" |
| 122 | full_id = await _seed_full_repo(db_session) |
| 123 | empty_repo = await create_repo(db_session, visibility="public") |
| 124 | empty_id = empty_repo.repo_id |
| 125 | |
| 126 | samples_full = [] |
| 127 | samples_empty = [] |
| 128 | |
| 129 | for _ in range(10): |
| 130 | t0 = time.monotonic() |
| 131 | await enrich_repo_cards(db_session, [full_id]) |
| 132 | samples_full.append(time.monotonic() - t0) |
| 133 | |
| 134 | t0 = time.monotonic() |
| 135 | await enrich_repo_cards(db_session, [empty_id]) |
| 136 | samples_empty.append(time.monotonic() - t0) |
| 137 | |
| 138 | median_full = statistics.median(samples_full) * 1000 |
| 139 | median_empty = statistics.median(samples_empty) * 1000 |
| 140 | |
| 141 | # Empty should be no more than 2× slower than full (same 5 queries run) |
| 142 | assert median_empty < median_full * 2, ( |
| 143 | f"Empty repo ({median_empty:.1f} ms) unexpectedly slower than " |
| 144 | f"full repo ({median_full:.1f} ms) by > 2×" |
| 145 | ) |