test_test_selection_speedup.py
python
sha256:88ac91129873e6a496e9189515aa690eb893ae25d69c8f72af141a2be5068eb3
docs: docstring sprint contract→find-symbol — idiomatic run…
Sonnet 4.6
patch
140 days ago
| 1 | """TDD tests for muse code test speedup. |
| 2 | |
| 3 | Two root causes of slowness being fixed: |
| 4 | |
| 5 | 1. ``select_tests`` never passed ``callgraph_cache`` to ``build_forward_graph``. |
| 6 | Every invocation re-read every blob, re-parsed every AST, and re-walked |
| 7 | every function body. Fix: load ``CallGraphCache``, pass it, save it. |
| 8 | |
| 9 | 2. ``SymbolCache`` was loaded from disk twice per ``muse code test`` run: |
| 10 | once in ``changed_symbols_from_diff`` and once in ``select_tests``. |
| 11 | Fix: load once in ``test_cmd.run()``, pass to both callers. |
| 12 | |
| 13 | Coverage |
| 14 | -------- |
| 15 | - ``select_tests`` accepts ``callgraph_cache`` keyword parameter. |
| 16 | - After ``select_tests`` runs, ``.muse/callgraph_cache.msgpack`` exists. |
| 17 | - Second ``select_tests`` call with pre-populated cache skips ``parse_symbols`` |
| 18 | (warm-cache path). |
| 19 | - Results are identical whether callgraph_cache is cold, warm, or not passed. |
| 20 | - ``select_tests`` accepts a shared ``SymbolCache`` (``cache=``) without |
| 21 | double-loading from disk. |
| 22 | - ``changed_symbols_from_diff`` and ``select_tests`` can share one |
| 23 | ``SymbolCache`` instance. |
| 24 | - ``load_symbol_cache`` is called at most once when a cache is pre-supplied. |
| 25 | """ |
| 26 | |
| 27 | from __future__ import annotations |
| 28 | |
| 29 | import pathlib |
| 30 | from unittest.mock import patch, call |
| 31 | |
| 32 | import pytest |
| 33 | |
| 34 | from muse.core._types import blob_id, Manifest |
| 35 | from muse.core.object_store import write_object |
| 36 | from muse.core.test_selection import ( |
| 37 | ChangedSymbol, |
| 38 | changed_symbols_from_diff, |
| 39 | select_tests, |
| 40 | ) |
| 41 | |
| 42 | |
| 43 | # --------------------------------------------------------------------------- |
| 44 | # Shared fixture — minimal repo with prod.py + tests/test_prod.py |
| 45 | # --------------------------------------------------------------------------- |
| 46 | |
| 47 | |
| 48 | def _make_repo(tmp_path: pathlib.Path) -> tuple[pathlib.Path, Manifest]: |
| 49 | """Return (root, manifest) for a tiny repo with .muse/ initialised.""" |
| 50 | muse_dir = tmp_path / ".muse" |
| 51 | muse_dir.mkdir(exist_ok=True) |
| 52 | |
| 53 | prod_src = b"""\ |
| 54 | def compute(x: int) -> int: |
| 55 | return x * 2 |
| 56 | |
| 57 | def helper() -> int: |
| 58 | return 42 |
| 59 | """ |
| 60 | test_src = b"""\ |
| 61 | from prod import compute |
| 62 | |
| 63 | def test_compute() -> None: |
| 64 | assert compute(2) == 4 |
| 65 | """ |
| 66 | |
| 67 | prod_oid = blob_id(prod_src) |
| 68 | test_oid = blob_id(test_src) |
| 69 | |
| 70 | write_object(tmp_path, prod_oid, prod_src) |
| 71 | write_object(tmp_path, test_oid, test_src) |
| 72 | |
| 73 | (tmp_path / "prod.py").write_bytes(prod_src) |
| 74 | tests_dir = tmp_path / "tests" |
| 75 | tests_dir.mkdir(exist_ok=True) |
| 76 | (tests_dir / "test_prod.py").write_bytes(test_src) |
| 77 | |
| 78 | manifest: Manifest = { |
| 79 | "prod.py": prod_oid, |
| 80 | "tests/test_prod.py": test_oid, |
| 81 | } |
| 82 | return tmp_path, manifest |
| 83 | |
| 84 | |
| 85 | # --------------------------------------------------------------------------- |
| 86 | # 1. select_tests accepts callgraph_cache keyword |
| 87 | # --------------------------------------------------------------------------- |
| 88 | |
| 89 | |
| 90 | class TestSelectTestsAcceptsCallgraphCache: |
| 91 | def test_accepts_callgraph_cache_none(self, tmp_path: pathlib.Path) -> None: |
| 92 | root, manifest = _make_repo(tmp_path) |
| 93 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 94 | # Must not raise TypeError regardless of whether the parameter exists. |
| 95 | result = select_tests(root, changed, manifest, callgraph_cache=None) |
| 96 | assert isinstance(result, dict) |
| 97 | |
| 98 | def test_accepts_callgraph_cache_instance(self, tmp_path: pathlib.Path) -> None: |
| 99 | from muse.core.callgraph_cache import CallGraphCache |
| 100 | root, manifest = _make_repo(tmp_path) |
| 101 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 102 | cg_cache = CallGraphCache.empty() |
| 103 | result = select_tests(root, changed, manifest, callgraph_cache=cg_cache) |
| 104 | assert isinstance(result, dict) |
| 105 | |
| 106 | def test_result_unchanged_with_or_without_cache( |
| 107 | self, tmp_path: pathlib.Path |
| 108 | ) -> None: |
| 109 | """Passing callgraph_cache does not change the selection result.""" |
| 110 | from muse.core.callgraph_cache import CallGraphCache |
| 111 | root, manifest = _make_repo(tmp_path) |
| 112 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 113 | |
| 114 | result_no_cache = select_tests(root, changed, manifest) |
| 115 | result_with_cache = select_tests( |
| 116 | root, changed, manifest, callgraph_cache=CallGraphCache.empty() |
| 117 | ) |
| 118 | |
| 119 | assert result_no_cache["changed_addresses"] == result_with_cache["changed_addresses"] |
| 120 | assert result_no_cache["fallback_used"] == result_with_cache["fallback_used"] |
| 121 | |
| 122 | |
| 123 | # --------------------------------------------------------------------------- |
| 124 | # 2. select_tests populates the callgraph cache on disk |
| 125 | # --------------------------------------------------------------------------- |
| 126 | |
| 127 | |
| 128 | class TestSelectTestsPersistsCallgraphCache: |
| 129 | def test_callgraph_cache_file_created_after_run( |
| 130 | self, tmp_path: pathlib.Path |
| 131 | ) -> None: |
| 132 | root, manifest = _make_repo(tmp_path) |
| 133 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 134 | |
| 135 | select_tests(root, changed, manifest) |
| 136 | |
| 137 | cache_file = root / ".muse" / "callgraph_cache.msgpack" |
| 138 | assert cache_file.exists(), ( |
| 139 | "select_tests should create .muse/callgraph_cache.msgpack on first run" |
| 140 | ) |
| 141 | |
| 142 | def test_callgraph_cache_nonempty_after_run(self, tmp_path: pathlib.Path) -> None: |
| 143 | from muse.core.callgraph_cache import CallGraphCache |
| 144 | root, manifest = _make_repo(tmp_path) |
| 145 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 146 | |
| 147 | select_tests(root, changed, manifest) |
| 148 | |
| 149 | loaded = CallGraphCache.load(root / ".muse") |
| 150 | assert loaded.size > 0, ( |
| 151 | "callgraph_cache should have at least one entry after select_tests" |
| 152 | ) |
| 153 | |
| 154 | def test_explicit_cache_is_populated_after_run( |
| 155 | self, tmp_path: pathlib.Path |
| 156 | ) -> None: |
| 157 | from muse.core.callgraph_cache import CallGraphCache |
| 158 | root, manifest = _make_repo(tmp_path) |
| 159 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 160 | cg_cache = CallGraphCache.empty() |
| 161 | |
| 162 | select_tests(root, changed, manifest, callgraph_cache=cg_cache) |
| 163 | |
| 164 | assert cg_cache.size > 0, ( |
| 165 | "select_tests should populate the passed callgraph_cache instance" |
| 166 | ) |
| 167 | |
| 168 | |
| 169 | # --------------------------------------------------------------------------- |
| 170 | # 3. Warm callgraph cache skips parse_symbols on second call |
| 171 | # --------------------------------------------------------------------------- |
| 172 | |
| 173 | |
| 174 | class TestWarmCallgraphCacheSkipsParse: |
| 175 | def test_second_call_skips_parse_symbols(self, tmp_path: pathlib.Path) -> None: |
| 176 | """On warm callgraph_cache, parse_symbols is never called.""" |
| 177 | from muse.core.callgraph_cache import CallGraphCache, load_callgraph_cache |
| 178 | root, manifest = _make_repo(tmp_path) |
| 179 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 180 | |
| 181 | # First call — cold cache, populates it. |
| 182 | select_tests(root, changed, manifest) |
| 183 | |
| 184 | # Second call — warm cache loaded from disk. |
| 185 | with patch("muse.plugins.code.ast_parser.parse_symbols") as mock_parse: |
| 186 | select_tests(root, changed, manifest) |
| 187 | mock_parse.assert_not_called() |
| 188 | |
| 189 | def test_warm_cache_result_matches_cold(self, tmp_path: pathlib.Path) -> None: |
| 190 | """Results are identical on cold and warm callgraph_cache.""" |
| 191 | root, manifest = _make_repo(tmp_path) |
| 192 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 193 | |
| 194 | result_cold = select_tests(root, changed, manifest) |
| 195 | result_warm = select_tests(root, changed, manifest) |
| 196 | |
| 197 | assert result_cold["changed_addresses"] == result_warm["changed_addresses"] |
| 198 | assert result_cold["fallback_used"] == result_warm["fallback_used"] |
| 199 | assert set(t["node_id"] for t in result_cold["test_targets"]) == set( |
| 200 | t["node_id"] for t in result_warm["test_targets"] |
| 201 | ) |
| 202 | |
| 203 | |
| 204 | # --------------------------------------------------------------------------- |
| 205 | # 4. Shared SymbolCache — load_symbol_cache not called when cache= is supplied |
| 206 | # --------------------------------------------------------------------------- |
| 207 | |
| 208 | |
| 209 | class TestSelectTestsSharedSymbolCache: |
| 210 | def test_load_symbol_cache_not_called_when_cache_supplied( |
| 211 | self, tmp_path: pathlib.Path |
| 212 | ) -> None: |
| 213 | """When ``cache=`` is passed, select_tests must not load it from disk again.""" |
| 214 | from muse.core.symbol_cache import SymbolCache |
| 215 | root, manifest = _make_repo(tmp_path) |
| 216 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 217 | shared_cache = SymbolCache.load(root / ".muse") |
| 218 | |
| 219 | with patch("muse.core.test_selection.load_symbol_cache") as mock_load: |
| 220 | select_tests(root, changed, manifest, cache=shared_cache) |
| 221 | mock_load.assert_not_called() |
| 222 | |
| 223 | def test_load_symbol_cache_called_once_when_none( |
| 224 | self, tmp_path: pathlib.Path |
| 225 | ) -> None: |
| 226 | """Without cache=, select_tests loads from disk exactly once.""" |
| 227 | root, manifest = _make_repo(tmp_path) |
| 228 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 229 | from muse.core.symbol_cache import SymbolCache |
| 230 | |
| 231 | real_load = __import__( |
| 232 | "muse.core.symbol_cache", fromlist=["load_symbol_cache"] |
| 233 | ).load_symbol_cache |
| 234 | call_count = [] |
| 235 | |
| 236 | def counting_load(root_arg): |
| 237 | call_count.append(1) |
| 238 | return real_load(root_arg) |
| 239 | |
| 240 | with patch("muse.core.test_selection.load_symbol_cache", side_effect=counting_load): |
| 241 | select_tests(root, changed, manifest) |
| 242 | |
| 243 | assert sum(call_count) == 1, ( |
| 244 | f"load_symbol_cache called {sum(call_count)} times — expected 1" |
| 245 | ) |
| 246 | |
| 247 | |
| 248 | # --------------------------------------------------------------------------- |
| 249 | # 5. changed_symbols_from_diff + select_tests can share one SymbolCache |
| 250 | # --------------------------------------------------------------------------- |
| 251 | |
| 252 | |
| 253 | class TestSharedCacheAcrossBothCalls: |
| 254 | def test_shared_cache_produces_same_diff(self, tmp_path: pathlib.Path) -> None: |
| 255 | """changed_symbols_from_diff with a shared SymbolCache returns same result.""" |
| 256 | from muse.core.symbol_cache import load_symbol_cache |
| 257 | root, manifest = _make_repo(tmp_path) |
| 258 | |
| 259 | # Edit a file on disk so there's something to diff. |
| 260 | (root / "prod.py").write_bytes(b"def compute(x: int) -> int:\n return x * 3\n") |
| 261 | |
| 262 | shared_cache = load_symbol_cache(root) |
| 263 | result_shared = changed_symbols_from_diff(root, manifest, cache=shared_cache) |
| 264 | result_own = changed_symbols_from_diff(root, manifest) |
| 265 | |
| 266 | assert {c["address"] for c in result_shared} == {c["address"] for c in result_own} |
| 267 | |
| 268 | def test_select_tests_with_prepopulated_cache( |
| 269 | self, tmp_path: pathlib.Path |
| 270 | ) -> None: |
| 271 | """select_tests with a pre-warmed SymbolCache returns correct results.""" |
| 272 | from muse.core.symbol_cache import load_symbol_cache |
| 273 | root, manifest = _make_repo(tmp_path) |
| 274 | changed = [ChangedSymbol(address="prod.py::compute", change_kind="modified")] |
| 275 | |
| 276 | # Warm the symbol cache by running select_tests once. |
| 277 | select_tests(root, changed, manifest) |
| 278 | shared_cache = load_symbol_cache(root) |
| 279 | |
| 280 | result = select_tests(root, changed, manifest, cache=shared_cache) |
| 281 | assert "prod.py::compute" in result["changed_addresses"] |
File History
1 commit
sha256:88ac91129873e6a496e9189515aa690eb893ae25d69c8f72af141a2be5068eb3
docs: docstring sprint contract→find-symbol — idiomatic run…
Sonnet 4.6
patch
140 days ago