test_symbols_v2_p1_coupling_count.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """TDD spec — Phase 1: coupling_count column on MusehubSymbolVitals. |
| 2 | |
| 3 | Problem |
| 4 | ─────── |
| 5 | The symbol list page needs a per-symbol coupling score (how many unique |
| 6 | symbols this one co-changes with) without running a COUNT(*) on |
| 7 | musehub_symbol_coupling at request time. |
| 8 | |
| 9 | Solution |
| 10 | ──────── |
| 11 | Add ``coupling_count INT DEFAULT 0`` to ``musehub_symbol_vitals``. |
| 12 | After ``_upsert_symbol_coupling`` runs, update vitals with the count derived |
| 13 | from the coupling table (SELECT COUNT(*) WHERE repo_id=? AND address=?). |
| 14 | |
| 15 | This keeps the symbol list query a single LEFT JOIN — no sub-selects, |
| 16 | no aggregations at request time. |
| 17 | |
| 18 | Tier breakdown |
| 19 | ────────────── |
| 20 | V101 Schema — coupling_count column exists on musehub_symbol_vitals |
| 21 | V102 Schema — default value is 0, not nullable |
| 22 | V103 Indexer — coupling_count populated after build_symbol_index |
| 23 | V104 Indexer — coupling_count accurate: matches distinct co_address count |
| 24 | V105 Indexer — coupling_count idempotent: re-running build_symbol_index gives same result |
| 25 | V106 Indexer — symbol with no coupling partners has coupling_count = 0 |
| 26 | V107 Schema — cascade delete removes vitals row (coupling_count included) |
| 27 | """ |
| 28 | from __future__ import annotations |
| 29 | |
| 30 | import secrets |
| 31 | from datetime import datetime, timezone |
| 32 | |
| 33 | import pytest |
| 34 | from sqlalchemy import select, text |
| 35 | from sqlalchemy.ext.asyncio import AsyncSession |
| 36 | |
| 37 | from musehub.db.musehub_intel_models import MusehubSymbolCoupling, MusehubSymbolVitals |
| 38 | from musehub.db.musehub_repo_models import MusehubCommit, MusehubCommitRef |
| 39 | from muse.core.types import blob_id, long_id |
| 40 | from musehub.services.musehub_symbol_indexer import build_symbol_index |
| 41 | from tests.factories import create_repo |
| 42 | |
| 43 | |
| 44 | # --------------------------------------------------------------------------- |
| 45 | # Helpers |
| 46 | # --------------------------------------------------------------------------- |
| 47 | |
| 48 | def _now() -> datetime: |
| 49 | return datetime.now(tz=timezone.utc) |
| 50 | |
| 51 | |
| 52 | def _cid() -> str: |
| 53 | return blob_id(secrets.token_bytes(32)) |
| 54 | |
| 55 | |
| 56 | def _lid() -> str: |
| 57 | return long_id(secrets.token_hex(32)) |
| 58 | |
| 59 | |
| 60 | async def _make_commit( |
| 61 | session: AsyncSession, |
| 62 | repo_id: str, |
| 63 | addresses: list[str], |
| 64 | *, |
| 65 | parent_id: str | None = None, |
| 66 | branch: str = "dev", |
| 67 | message: str = "feat: test", |
| 68 | op: str = "insert", |
| 69 | ) -> MusehubCommit: |
| 70 | """Create a MusehubCommit with structured_delta so the indexer processes it.""" |
| 71 | cid = _lid() |
| 72 | commit = MusehubCommit( |
| 73 | commit_id=cid, |
| 74 | branch=branch, |
| 75 | message=message, |
| 76 | author="gabriel", |
| 77 | parent_ids=[parent_id] if parent_id else [], |
| 78 | timestamp=_now(), |
| 79 | structured_delta={"ops": [ |
| 80 | {"address": addr, "op": op, "new_content_id": _cid()} |
| 81 | for addr in addresses |
| 82 | ]}, |
| 83 | ) |
| 84 | session.add(commit) |
| 85 | session.add(MusehubCommitRef(repo_id=repo_id, commit_id=cid)) |
| 86 | await session.flush() |
| 87 | return commit |
| 88 | |
| 89 | |
| 90 | async def _push_index(session: AsyncSession, repo_id: str, head_commit_id: str) -> None: |
| 91 | """Run build_symbol_index + backfill_coupling as the background job does at push time.""" |
| 92 | from musehub.services.musehub_symbol_indexer import backfill_coupling |
| 93 | await build_symbol_index(session, repo_id, head_commit_id) |
| 94 | await backfill_coupling(session, repo_id, min_shared=1) |
| 95 | await session.flush() |
| 96 | |
| 97 | |
| 98 | # --------------------------------------------------------------------------- |
| 99 | # V101 — coupling_count column exists on musehub_symbol_vitals |
| 100 | # --------------------------------------------------------------------------- |
| 101 | |
| 102 | @pytest.mark.asyncio |
| 103 | async def test_v101_coupling_count_column_exists(db_session: AsyncSession) -> None: |
| 104 | """musehub_symbol_vitals must have a coupling_count column.""" |
| 105 | result = await db_session.execute( |
| 106 | text( |
| 107 | "SELECT column_name FROM information_schema.columns " |
| 108 | "WHERE table_name = 'musehub_symbol_vitals' AND column_name = 'coupling_count'" |
| 109 | ) |
| 110 | ) |
| 111 | assert result.fetchone() is not None, "coupling_count column not found on musehub_symbol_vitals" |
| 112 | |
| 113 | |
| 114 | # --------------------------------------------------------------------------- |
| 115 | # V102 — default is 0, not nullable |
| 116 | # --------------------------------------------------------------------------- |
| 117 | |
| 118 | @pytest.mark.asyncio |
| 119 | async def test_v102_coupling_count_default_zero_not_nullable(db_session: AsyncSession) -> None: |
| 120 | """coupling_count must default to 0 and be NOT NULL.""" |
| 121 | result = await db_session.execute( |
| 122 | text( |
| 123 | "SELECT is_nullable, column_default " |
| 124 | "FROM information_schema.columns " |
| 125 | "WHERE table_name = 'musehub_symbol_vitals' AND column_name = 'coupling_count'" |
| 126 | ) |
| 127 | ) |
| 128 | row = result.fetchone() |
| 129 | assert row is not None |
| 130 | is_nullable, column_default = row |
| 131 | assert is_nullable == "NO", "coupling_count must be NOT NULL" |
| 132 | assert column_default is not None and "0" in str(column_default), \ |
| 133 | f"coupling_count must default to 0, got: {column_default}" |
| 134 | |
| 135 | |
| 136 | # --------------------------------------------------------------------------- |
| 137 | # V103 — coupling_count populated after build_symbol_index |
| 138 | # --------------------------------------------------------------------------- |
| 139 | |
| 140 | @pytest.mark.asyncio |
| 141 | async def test_v103_coupling_count_populated_after_index(db_session: AsyncSession) -> None: |
| 142 | """After build_symbol_index, symbols with coupling partners have coupling_count > 0.""" |
| 143 | repo = await create_repo(db_session) |
| 144 | repo_id = repo.repo_id |
| 145 | |
| 146 | # Two symbols changed in the same commit → they are coupled |
| 147 | commit = await _make_commit( |
| 148 | db_session, repo_id, |
| 149 | ["src/foo.py::alpha", "src/foo.py::beta"], |
| 150 | ) |
| 151 | await _push_index(db_session, repo_id, commit.commit_id) |
| 152 | |
| 153 | vitals_alpha = (await db_session.execute( |
| 154 | select(MusehubSymbolVitals).where( |
| 155 | MusehubSymbolVitals.repo_id == repo_id, |
| 156 | MusehubSymbolVitals.address == "src/foo.py::alpha", |
| 157 | ) |
| 158 | )).scalar_one_or_none() |
| 159 | |
| 160 | assert vitals_alpha is not None |
| 161 | assert vitals_alpha.coupling_count == 1, \ |
| 162 | f"alpha coupled to beta → coupling_count should be 1, got {vitals_alpha.coupling_count}" |
| 163 | |
| 164 | |
| 165 | # --------------------------------------------------------------------------- |
| 166 | # V104 — coupling_count accurate: matches distinct co_address count |
| 167 | # --------------------------------------------------------------------------- |
| 168 | |
| 169 | @pytest.mark.asyncio |
| 170 | async def test_v104_coupling_count_matches_distinct_co_address_count(db_session: AsyncSession) -> None: |
| 171 | """coupling_count must equal the number of distinct partners in musehub_symbol_coupling.""" |
| 172 | repo = await create_repo(db_session) |
| 173 | repo_id = repo.repo_id |
| 174 | |
| 175 | # Commit 1: alpha + beta + gamma change together |
| 176 | c1 = await _make_commit( |
| 177 | db_session, repo_id, |
| 178 | ["src/a.py::alpha", "src/a.py::beta", "src/a.py::gamma"], |
| 179 | message="feat: first", |
| 180 | ) |
| 181 | # Commit 2: alpha + delta change together (another partner for alpha) |
| 182 | c2 = await _make_commit( |
| 183 | db_session, repo_id, |
| 184 | ["src/a.py::alpha", "src/a.py::delta"], |
| 185 | parent_id=c1.commit_id, |
| 186 | message="feat: second", |
| 187 | op="replace", |
| 188 | ) |
| 189 | await _push_index(db_session, repo_id, c2.commit_id) |
| 190 | |
| 191 | # alpha is coupled to beta, gamma, delta → coupling_count = 3 |
| 192 | vitals = (await db_session.execute( |
| 193 | select(MusehubSymbolVitals).where( |
| 194 | MusehubSymbolVitals.repo_id == repo_id, |
| 195 | MusehubSymbolVitals.address == "src/a.py::alpha", |
| 196 | ) |
| 197 | )).scalar_one() |
| 198 | |
| 199 | coupling_rows = (await db_session.execute( |
| 200 | select(MusehubSymbolCoupling).where( |
| 201 | MusehubSymbolCoupling.repo_id == repo_id, |
| 202 | MusehubSymbolCoupling.address == "src/a.py::alpha", |
| 203 | ) |
| 204 | )).scalars().all() |
| 205 | |
| 206 | assert vitals.coupling_count == len(coupling_rows), ( |
| 207 | f"coupling_count {vitals.coupling_count} != distinct coupling rows {len(coupling_rows)}" |
| 208 | ) |
| 209 | assert vitals.coupling_count == 3 |
| 210 | |
| 211 | |
| 212 | # --------------------------------------------------------------------------- |
| 213 | # V105 — idempotent: re-running build_symbol_index gives same result |
| 214 | # --------------------------------------------------------------------------- |
| 215 | |
| 216 | @pytest.mark.asyncio |
| 217 | async def test_v105_coupling_count_idempotent(db_session: AsyncSession) -> None: |
| 218 | """Running build_symbol_index twice produces the same coupling_count.""" |
| 219 | repo = await create_repo(db_session) |
| 220 | repo_id = repo.repo_id |
| 221 | |
| 222 | c1 = await _make_commit(db_session, repo_id, ["src/b.py::x", "src/b.py::y"]) |
| 223 | |
| 224 | await _push_index(db_session, repo_id, c1.commit_id) |
| 225 | await _push_index(db_session, repo_id, c1.commit_id) |
| 226 | |
| 227 | vitals = (await db_session.execute( |
| 228 | select(MusehubSymbolVitals).where( |
| 229 | MusehubSymbolVitals.repo_id == repo_id, |
| 230 | MusehubSymbolVitals.address == "src/b.py::x", |
| 231 | ) |
| 232 | )).scalar_one() |
| 233 | |
| 234 | assert vitals.coupling_count == 1 |
| 235 | |
| 236 | |
| 237 | # --------------------------------------------------------------------------- |
| 238 | # V106 — isolated symbol has coupling_count = 0 |
| 239 | # --------------------------------------------------------------------------- |
| 240 | |
| 241 | @pytest.mark.asyncio |
| 242 | async def test_v106_isolated_symbol_has_zero_coupling_count(db_session: AsyncSession) -> None: |
| 243 | """A symbol that never changes with others must have coupling_count = 0.""" |
| 244 | repo = await create_repo(db_session) |
| 245 | repo_id = repo.repo_id |
| 246 | |
| 247 | # Single symbol in the commit → no coupling partners |
| 248 | c1 = await _make_commit(db_session, repo_id, ["src/solo.py::lone_wolf"]) |
| 249 | await _push_index(db_session, repo_id, c1.commit_id) |
| 250 | |
| 251 | vitals = (await db_session.execute( |
| 252 | select(MusehubSymbolVitals).where( |
| 253 | MusehubSymbolVitals.repo_id == repo_id, |
| 254 | MusehubSymbolVitals.address == "src/solo.py::lone_wolf", |
| 255 | ) |
| 256 | )).scalar_one() |
| 257 | |
| 258 | assert vitals.coupling_count == 0 |
| 259 | |
| 260 | |
| 261 | # --------------------------------------------------------------------------- |
| 262 | # V107 — cascade delete removes vitals row (coupling_count included) |
| 263 | # --------------------------------------------------------------------------- |
| 264 | |
| 265 | @pytest.mark.asyncio |
| 266 | async def test_v107_cascade_delete_removes_vitals(db_session: AsyncSession) -> None: |
| 267 | """Deleting the repo must cascade-delete all musehub_symbol_vitals rows.""" |
| 268 | repo = await create_repo(db_session) |
| 269 | repo_id = repo.repo_id |
| 270 | |
| 271 | row = MusehubSymbolVitals( |
| 272 | repo_id=repo_id, |
| 273 | address="src/c.py::fn", |
| 274 | first_introduced=_now(), |
| 275 | change_count=1, |
| 276 | version_count=1, |
| 277 | op_add=1, |
| 278 | op_modify=0, |
| 279 | op_delete=0, |
| 280 | op_move=0, |
| 281 | coupling_count=3, |
| 282 | ) |
| 283 | db_session.add(row) |
| 284 | await db_session.flush() |
| 285 | |
| 286 | await db_session.delete(repo) |
| 287 | await db_session.flush() |
| 288 | |
| 289 | remaining = (await db_session.execute( |
| 290 | select(MusehubSymbolVitals).where(MusehubSymbolVitals.repo_id == repo_id) |
| 291 | )).scalars().all() |
| 292 | assert remaining == [] |