test_blame_symlog_fast_path.py
python
sha256:87e65c3a0f38438d20b4534feb442bd274da6a5a7f0f5f95e9c4e4dfefa86c42
feat: muse#53 item 2 — muse code blame prefers symlog journal
Sonnet 5
patch
14 hours ago
| 1 | """muse#53 item 2 -- `muse code blame` prefers the symlog journal (O(1) per |
| 2 | symbol) over the full commit-history BFS scan when a journal exists. |
| 3 | |
| 4 | Correctness parity with the pre-existing BFS path is what matters here |
| 5 | (the existing test_cmd_blame*/test_blame_supercharge suites already |
| 6 | exercise that path's output and continue to pass unchanged, confirming |
| 7 | the symlog fast path produces a structurally compatible result) -- |
| 8 | this file targets the NEW decision logic: when the fast path is used, |
| 9 | when it falls back, and rename-chain correctness via the journal. |
| 10 | """ |
| 11 | |
| 12 | from __future__ import annotations |
| 13 | from collections.abc import Mapping |
| 14 | |
| 15 | import json |
| 16 | import pathlib |
| 17 | import textwrap |
| 18 | |
| 19 | import pytest |
| 20 | |
| 21 | from tests.cli_test_helper import CliRunner |
| 22 | |
| 23 | runner = CliRunner() |
| 24 | |
| 25 | |
| 26 | def _env(root: pathlib.Path) -> Mapping[str, str]: |
| 27 | return {"MUSE_REPO_ROOT": str(root)} |
| 28 | |
| 29 | |
| 30 | def _run(root: pathlib.Path, *args: str): |
| 31 | return runner.invoke(None, list(args), env=_env(root)) |
| 32 | |
| 33 | |
| 34 | @pytest.fixture() |
| 35 | def repo(tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> pathlib.Path: |
| 36 | monkeypatch.chdir(tmp_path) |
| 37 | r = _run(tmp_path, "init", "--domain", "code") |
| 38 | assert r.exit_code == 0, r.output |
| 39 | return tmp_path |
| 40 | |
| 41 | |
| 42 | def _write(repo: pathlib.Path, rel: str, content: str) -> None: |
| 43 | (repo / rel).write_text(textwrap.dedent(content)) |
| 44 | |
| 45 | |
| 46 | def _commit(repo: pathlib.Path, msg: str) -> None: |
| 47 | r1 = _run(repo, "code", "add", ".") |
| 48 | assert r1.exit_code == 0, r1.output |
| 49 | r2 = _run(repo, "commit", "-m", msg) |
| 50 | assert r2.exit_code == 0, r2.output |
| 51 | |
| 52 | |
| 53 | class TestSymlogFastPathIsActuallyUsed: |
| 54 | """Proves the symlog journal is the mechanism, not just that the final |
| 55 | result happens to be correct (the pre-existing BFS path already produces |
| 56 | correct results on its own, so output-only assertions can't distinguish |
| 57 | the two paths -- these tests break the BFS path and confirm blame still |
| 58 | works, or confirm the BFS entry point is never even called).""" |
| 59 | |
| 60 | def test_blame_still_works_when_bfs_walk_is_broken( |
| 61 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch, |
| 62 | ) -> None: |
| 63 | _write(repo, "billing.py", "def compute_total(items):\n return sum(items)\n") |
| 64 | _commit(repo, "add compute_total") |
| 65 | _write(repo, "billing.py", "def compute_total(items):\n return sum(items) * 2\n") |
| 66 | _commit(repo, "double it") |
| 67 | |
| 68 | import muse.cli.commands.blame as blame_mod |
| 69 | |
| 70 | def _broken_bfs(*args: object, **kwargs: object) -> None: |
| 71 | raise AssertionError("walk_commits_bfs must not be called on the symlog fast path") |
| 72 | |
| 73 | monkeypatch.setattr(blame_mod, "walk_commits_bfs", _broken_bfs) |
| 74 | |
| 75 | result = _run(repo, "code", "blame", "billing.py::compute_total", "--json") |
| 76 | assert result.exit_code == 0, result.output |
| 77 | data = json.loads(result.output) |
| 78 | kinds = [e["event"] for e in data["events"]] |
| 79 | assert "created" in kinds |
| 80 | assert "modified" in kinds |
| 81 | |
| 82 | def test_from_ref_falls_back_to_bfs_walk( |
| 83 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch, |
| 84 | ) -> None: |
| 85 | """--from forces the full BFS scan -- the journal has no concept of |
| 86 | 'as reachable from an older ref' and must not be used here.""" |
| 87 | _write(repo, "billing.py", "def compute_total(items):\n return sum(items)\n") |
| 88 | _commit(repo, "add compute_total") |
| 89 | _write(repo, "billing.py", "def compute_total(items):\n return sum(items) * 2\n") |
| 90 | _commit(repo, "double it") |
| 91 | |
| 92 | import muse.cli.commands.blame as blame_mod |
| 93 | |
| 94 | calls: list[bool] = [] |
| 95 | real_bfs = blame_mod.walk_commits_bfs |
| 96 | |
| 97 | def _spy_bfs(*args: object, **kwargs: object): |
| 98 | calls.append(True) |
| 99 | return real_bfs(*args, **kwargs) |
| 100 | |
| 101 | monkeypatch.setattr(blame_mod, "walk_commits_bfs", _spy_bfs) |
| 102 | |
| 103 | result = _run(repo, "code", "blame", "billing.py::compute_total", "--from", "HEAD~1", "--json") |
| 104 | assert result.exit_code == 0, result.output |
| 105 | assert calls, "walk_commits_bfs should have been called for --from" |
| 106 | |
| 107 | def test_kind_filter_for_unsupported_kind_falls_back_to_bfs( |
| 108 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch, |
| 109 | ) -> None: |
| 110 | """A --kind moved/signature query can never be answered from the |
| 111 | journal (it doesn't track those categories) -- must fall back.""" |
| 112 | _write(repo, "billing.py", "def compute_total(items):\n return sum(items)\n") |
| 113 | _commit(repo, "add compute_total") |
| 114 | |
| 115 | import muse.cli.commands.blame as blame_mod |
| 116 | |
| 117 | calls: list[bool] = [] |
| 118 | real_bfs = blame_mod.walk_commits_bfs |
| 119 | |
| 120 | def _spy_bfs(*args: object, **kwargs: object): |
| 121 | calls.append(True) |
| 122 | return real_bfs(*args, **kwargs) |
| 123 | |
| 124 | monkeypatch.setattr(blame_mod, "walk_commits_bfs", _spy_bfs) |
| 125 | |
| 126 | result = _run(repo, "code", "blame", "billing.py::compute_total", "--kind", "signature", "--json") |
| 127 | assert result.exit_code == 0, result.output |
| 128 | assert calls, "walk_commits_bfs should have been called for --kind signature" |
| 129 | |
| 130 | |
| 131 | class TestEventsFromSymlogDirect: |
| 132 | """Direct unit tests against _events_from_symlog() -- exercises the new |
| 133 | function in isolation rather than through blame's path-selection logic.""" |
| 134 | |
| 135 | def test_returns_none_when_no_journal_exists(self, repo: pathlib.Path) -> None: |
| 136 | from muse.cli.commands.blame import _events_from_symlog |
| 137 | |
| 138 | _write(repo, "billing.py", "def compute_total(items):\n return sum(items)\n") |
| 139 | _commit(repo, "add compute_total") |
| 140 | |
| 141 | result = _events_from_symlog(repo, "billing.py::nonexistent", 500) |
| 142 | assert result is None |
| 143 | |
| 144 | def test_builds_created_and_modified_events(self, repo: pathlib.Path) -> None: |
| 145 | from muse.cli.commands.blame import _events_from_symlog |
| 146 | |
| 147 | _write(repo, "billing.py", "def compute_total(items):\n return sum(items)\n") |
| 148 | _commit(repo, "add compute_total") |
| 149 | _write(repo, "billing.py", "def compute_total(items):\n return sum(items) * 2\n") |
| 150 | _commit(repo, "double it") |
| 151 | |
| 152 | events = _events_from_symlog(repo, "billing.py::compute_total", 500) |
| 153 | assert events is not None |
| 154 | kinds = [e.kind for e in events] |
| 155 | assert kinds == ["modified", "created"] # newest-first |
| 156 | assert events[0].commit.message == "double it" |
| 157 | |
| 158 | def test_deleted_symbol(self, repo: pathlib.Path) -> None: |
| 159 | from muse.cli.commands.blame import _events_from_symlog |
| 160 | |
| 161 | _write(repo, "billing.py", ( |
| 162 | "def compute_total(items):\n return sum(items)\n\n" |
| 163 | "def apply_discount(total, pct):\n return total * (1 - pct)\n" |
| 164 | )) |
| 165 | _commit(repo, "add two functions") |
| 166 | _write(repo, "billing.py", "def compute_total(items):\n return sum(items)\n") |
| 167 | _commit(repo, "remove apply_discount") |
| 168 | |
| 169 | events = _events_from_symlog(repo, "billing.py::apply_discount", 500) |
| 170 | assert events is not None |
| 171 | assert events[0].kind == "deleted" |
| 172 | |
| 173 | def test_rename_chain_crosses_address_boundary(self, repo: pathlib.Path) -> None: |
| 174 | from muse.cli.commands.blame import _events_from_symlog |
| 175 | |
| 176 | _write(repo, "billing.py", "def compute_total(items):\n return sum(items)\n") |
| 177 | _commit(repo, "add compute_total") |
| 178 | _write(repo, "billing.py", "def compute_invoice_total(items):\n return sum(items)\n") |
| 179 | _commit(repo, "rename to compute_invoice_total") |
| 180 | |
| 181 | events = _events_from_symlog(repo, "billing.py::compute_invoice_total", 500) |
| 182 | assert events is not None |
| 183 | kinds = [e.kind for e in events] |
| 184 | assert "created" in kinds |
| 185 | # The creation event must be attributed to the OLD address, since |
| 186 | # that's what the symbol was called at the time it was created. |
| 187 | created_event = next(e for e in events if e.kind == "created") |
| 188 | assert created_event.address == "billing.py::compute_total" |
File History
1 commit
sha256:87e65c3a0f38438d20b4534feb442bd274da6a5a7f0f5f95e9c4e4dfefa86c42
feat: muse#53 item 2 — muse code blame prefers symlog journal
Sonnet 5
patch
14 hours ago