test_hash_object_canonical.py
python
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
21 days ago
| 1 | """hash-object: canonical sha256: prefix and agent-ready JSON schema. |
| 2 | |
| 3 | Every object ID emitted by ``muse hash-object`` must carry the ``sha256:`` |
| 4 | prefix. Bare hex is only acceptable at the disk boundary (the filename |
| 5 | on disk). This test suite enforces that invariant and covers the new |
| 6 | agent-ready JSON fields. |
| 7 | |
| 8 | Test categories |
| 9 | --------------- |
| 10 | TestCanonicalPrefix — object_id always starts with 'sha256:' |
| 11 | TestStdinWriteFixed — stdin + --write was broken (bare hex bug); now fixed |
| 12 | TestAgentFields — duration_ms, exit_code, size_bytes in JSON output |
| 13 | TestTextOutputPrefix — text format also carries the prefix |
| 14 | TestCrossCheck — file and stdin produce identical canonical IDs |
| 15 | """ |
| 16 | |
| 17 | from __future__ import annotations |
| 18 | |
| 19 | import argparse |
| 20 | import json |
| 21 | import pathlib |
| 22 | |
| 23 | from muse.core.errors import ExitCode |
| 24 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 25 | from muse.core.types import blob_id, split_id |
| 26 | from muse.core.paths import muse_dir |
| 27 | from muse.core.object_store import object_path, read_object |
| 28 | |
| 29 | runner = CliRunner() |
| 30 | |
| 31 | |
| 32 | # --------------------------------------------------------------------------- |
| 33 | # Helpers |
| 34 | # --------------------------------------------------------------------------- |
| 35 | |
| 36 | def _run(*args: str, stdin: bytes | None = None, repo: pathlib.Path | None = None) -> InvokeResult: |
| 37 | from muse.cli.app import main as cli |
| 38 | env = {"MUSE_REPO_ROOT": str(repo)} if repo else {} |
| 39 | return runner.invoke(cli, ["hash-object", *args], input=stdin, env=env) |
| 40 | |
| 41 | |
| 42 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 43 | repo = tmp_path / "repo" |
| 44 | dot_muse = muse_dir(repo) |
| 45 | for sub in ("objects", "commits", "snapshots", "refs/heads"): |
| 46 | (dot_muse / sub).mkdir(parents=True) |
| 47 | (dot_muse / "HEAD").write_text("ref: refs/heads/main") |
| 48 | (dot_muse / "repo.json").write_text( |
| 49 | json.dumps({"repo_id": "test", "domain": "code"}) |
| 50 | ) |
| 51 | return repo |
| 52 | |
| 53 | |
| 54 | |
| 55 | # --------------------------------------------------------------------------- |
| 56 | # TestCanonicalPrefix |
| 57 | # --------------------------------------------------------------------------- |
| 58 | |
| 59 | class TestCanonicalPrefix: |
| 60 | """object_id in JSON output must always start with 'sha256:'.""" |
| 61 | |
| 62 | def test_file_json_object_id_has_prefix(self, tmp_path: pathlib.Path) -> None: |
| 63 | f = tmp_path / "f.txt" |
| 64 | f.write_bytes(b"hello") |
| 65 | result = _run("--json", str(f)) |
| 66 | assert result.exit_code == 0 |
| 67 | data = json.loads(result.output) |
| 68 | assert data["object_id"].startswith("sha256:"), ( |
| 69 | f"object_id must start with 'sha256:' — got {data['object_id']!r}" |
| 70 | ) |
| 71 | |
| 72 | def test_file_json_object_id_correct_length(self, tmp_path: pathlib.Path) -> None: |
| 73 | """sha256: (7 chars) + 64 hex = 71 total.""" |
| 74 | f = tmp_path / "f.txt" |
| 75 | f.write_bytes(b"hello") |
| 76 | result = _run("--json", str(f)) |
| 77 | data = json.loads(result.output) |
| 78 | assert len(data["object_id"]) == 71 |
| 79 | |
| 80 | def test_file_json_object_id_matches_canonical(self, tmp_path: pathlib.Path) -> None: |
| 81 | content = b"canonical check" |
| 82 | f = tmp_path / "f.txt" |
| 83 | f.write_bytes(content) |
| 84 | result = _run("--json", str(f)) |
| 85 | data = json.loads(result.output) |
| 86 | assert data["object_id"] == blob_id(content) |
| 87 | |
| 88 | def test_stdin_json_object_id_has_prefix(self, tmp_path: pathlib.Path) -> None: |
| 89 | result = _run("--json", "--stdin", stdin=b"from stdin") |
| 90 | assert result.exit_code == 0 |
| 91 | data = json.loads(result.output) |
| 92 | assert data["object_id"].startswith("sha256:") |
| 93 | |
| 94 | def test_stdin_json_object_id_matches_canonical(self, tmp_path: pathlib.Path) -> None: |
| 95 | content = b"piped data" |
| 96 | result = _run("--json", "--stdin", stdin=content) |
| 97 | data = json.loads(result.output) |
| 98 | assert data["object_id"] == blob_id(content) |
| 99 | |
| 100 | def test_empty_file_has_prefix(self, tmp_path: pathlib.Path) -> None: |
| 101 | f = tmp_path / "empty.txt" |
| 102 | f.write_bytes(b"") |
| 103 | result = _run("--json", str(f)) |
| 104 | data = json.loads(result.output) |
| 105 | assert data["object_id"] == blob_id(b"") |
| 106 | |
| 107 | def test_empty_stdin_has_prefix(self, tmp_path: pathlib.Path) -> None: |
| 108 | result = _run("--json", "--stdin", stdin=b"") |
| 109 | data = json.loads(result.output) |
| 110 | assert data["object_id"] == blob_id(b"") |
| 111 | |
| 112 | def test_no_bare_hex_in_json_output(self, tmp_path: pathlib.Path) -> None: |
| 113 | """The raw 64-char hex without prefix must not appear as object_id.""" |
| 114 | content = b"no bare hex" |
| 115 | f = tmp_path / "f.txt" |
| 116 | f.write_bytes(content) |
| 117 | result = _run("--json", str(f)) |
| 118 | data = json.loads(result.output) |
| 119 | bare_hex = split_id(blob_id(content))[1] |
| 120 | assert data["object_id"] != bare_hex, ( |
| 121 | "object_id must be 'sha256:<hex>', not bare hex" |
| 122 | ) |
| 123 | |
| 124 | |
| 125 | # --------------------------------------------------------------------------- |
| 126 | # TestStdinWriteFixed |
| 127 | # --------------------------------------------------------------------------- |
| 128 | |
| 129 | class TestStdinWriteFixed: |
| 130 | """stdin + --write was broken (passed bare hex to write_object). Now fixed.""" |
| 131 | |
| 132 | def test_stdin_write_exits_zero(self, tmp_path: pathlib.Path) -> None: |
| 133 | repo = _make_repo(tmp_path) |
| 134 | result = _run("--stdin", "--write", stdin=b"store me", repo=repo) |
| 135 | assert result.exit_code == 0, f"exit {result.exit_code}: {result.output}" |
| 136 | |
| 137 | def test_stdin_write_stored_true(self, tmp_path: pathlib.Path) -> None: |
| 138 | repo = _make_repo(tmp_path) |
| 139 | result = _run("--json", "--stdin", "--write", stdin=b"store me", repo=repo) |
| 140 | assert json.loads(result.output)["stored"] is True |
| 141 | |
| 142 | def test_stdin_write_object_file_exists(self, tmp_path: pathlib.Path) -> None: |
| 143 | repo = _make_repo(tmp_path) |
| 144 | content = b"stdin stored content" |
| 145 | result = _run("--json", "--stdin", "--write", stdin=content, repo=repo) |
| 146 | oid = json.loads(result.output)["object_id"] |
| 147 | obj_file = object_path(repo, oid) |
| 148 | assert obj_file.exists(), f"object file not found at {obj_file}" |
| 149 | assert read_object(repo, oid) == content |
| 150 | |
| 151 | def test_stdin_write_object_id_canonical(self, tmp_path: pathlib.Path) -> None: |
| 152 | repo = _make_repo(tmp_path) |
| 153 | content = b"canonical write" |
| 154 | result = _run("--json", "--stdin", "--write", stdin=content, repo=repo) |
| 155 | data = json.loads(result.output) |
| 156 | assert data["object_id"] == blob_id(content) |
| 157 | |
| 158 | def test_stdin_write_idempotent(self, tmp_path: pathlib.Path) -> None: |
| 159 | repo = _make_repo(tmp_path) |
| 160 | content = b"write twice" |
| 161 | _run("--stdin", "--write", stdin=content, repo=repo) |
| 162 | result2 = _run("--json", "--stdin", "--write", stdin=content, repo=repo) |
| 163 | assert result2.exit_code == 0 |
| 164 | assert json.loads(result2.output)["stored"] is False |
| 165 | |
| 166 | |
| 167 | # --------------------------------------------------------------------------- |
| 168 | # TestAgentFields |
| 169 | # --------------------------------------------------------------------------- |
| 170 | |
| 171 | class TestAgentFields: |
| 172 | """JSON output must include duration_ms, exit_code, size_bytes.""" |
| 173 | |
| 174 | def test_duration_ms_present(self, tmp_path: pathlib.Path) -> None: |
| 175 | f = tmp_path / "f.txt" |
| 176 | f.write_bytes(b"timing") |
| 177 | data = json.loads(_run("--json", str(f)).output) |
| 178 | assert "duration_ms" in data, "JSON must include duration_ms" |
| 179 | |
| 180 | def test_duration_ms_non_negative(self, tmp_path: pathlib.Path) -> None: |
| 181 | f = tmp_path / "f.txt" |
| 182 | f.write_bytes(b"timing") |
| 183 | data = json.loads(_run("--json", str(f)).output) |
| 184 | assert data["duration_ms"] >= 0 |
| 185 | |
| 186 | def test_exit_code_present(self, tmp_path: pathlib.Path) -> None: |
| 187 | f = tmp_path / "f.txt" |
| 188 | f.write_bytes(b"x") |
| 189 | data = json.loads(_run("--json", str(f)).output) |
| 190 | assert "exit_code" in data, "JSON must include exit_code" |
| 191 | |
| 192 | def test_exit_code_zero_on_success(self, tmp_path: pathlib.Path) -> None: |
| 193 | f = tmp_path / "f.txt" |
| 194 | f.write_bytes(b"x") |
| 195 | data = json.loads(_run("--json", str(f)).output) |
| 196 | assert data["exit_code"] == 0 |
| 197 | |
| 198 | def test_size_bytes_present(self, tmp_path: pathlib.Path) -> None: |
| 199 | f = tmp_path / "f.txt" |
| 200 | f.write_bytes(b"twelve bytes") |
| 201 | data = json.loads(_run("--json", str(f)).output) |
| 202 | assert "size_bytes" in data, "JSON must include size_bytes" |
| 203 | |
| 204 | def test_size_bytes_correct_for_file(self, tmp_path: pathlib.Path) -> None: |
| 205 | content = b"twelve bytes" |
| 206 | f = tmp_path / "f.txt" |
| 207 | f.write_bytes(content) |
| 208 | data = json.loads(_run("--json", str(f)).output) |
| 209 | assert data["size_bytes"] == len(content) |
| 210 | |
| 211 | def test_size_bytes_correct_for_stdin(self, tmp_path: pathlib.Path) -> None: |
| 212 | content = b"stdin payload" |
| 213 | data = json.loads(_run("--json", "--stdin", stdin=content).output) |
| 214 | assert data["size_bytes"] == len(content) |
| 215 | |
| 216 | def test_size_bytes_zero_for_empty(self, tmp_path: pathlib.Path) -> None: |
| 217 | f = tmp_path / "empty.txt" |
| 218 | f.write_bytes(b"") |
| 219 | data = json.loads(_run("--json", str(f)).output) |
| 220 | assert data["size_bytes"] == 0 |
| 221 | |
| 222 | def test_stdin_duration_ms_present(self, tmp_path: pathlib.Path) -> None: |
| 223 | data = json.loads(_run("--json", "--stdin", stdin=b"x").output) |
| 224 | assert "duration_ms" in data |
| 225 | |
| 226 | def test_stdin_exit_code_present(self, tmp_path: pathlib.Path) -> None: |
| 227 | data = json.loads(_run("--json", "--stdin", stdin=b"x").output) |
| 228 | assert "exit_code" in data |
| 229 | |
| 230 | |
| 231 | # --------------------------------------------------------------------------- |
| 232 | # TestTextOutputPrefix |
| 233 | # --------------------------------------------------------------------------- |
| 234 | |
| 235 | class TestTextOutputPrefix: |
| 236 | """Text format must also emit the sha256: prefix.""" |
| 237 | |
| 238 | def test_text_file_has_prefix(self, tmp_path: pathlib.Path) -> None: |
| 239 | f = tmp_path / "f.txt" |
| 240 | f.write_bytes(b"text output") |
| 241 | result = _run(str(f)) |
| 242 | assert result.exit_code == 0 |
| 243 | assert result.output.strip().startswith("sha256:") |
| 244 | |
| 245 | def test_text_stdin_has_prefix(self, tmp_path: pathlib.Path) -> None: |
| 246 | result = _run("--stdin", stdin=b"text stdin") |
| 247 | assert result.output.strip().startswith("sha256:") |
| 248 | |
| 249 | def test_text_output_is_correct_canonical_id(self, tmp_path: pathlib.Path) -> None: |
| 250 | content = b"text canonical" |
| 251 | f = tmp_path / "f.txt" |
| 252 | f.write_bytes(content) |
| 253 | result = _run(str(f)) |
| 254 | assert result.output.strip() == blob_id(content) |
| 255 | |
| 256 | def test_text_length_is_71(self, tmp_path: pathlib.Path) -> None: |
| 257 | """sha256: (7) + 64 hex = 71 characters.""" |
| 258 | f = tmp_path / "f.txt" |
| 259 | f.write_bytes(b"length check") |
| 260 | result = _run(str(f)) |
| 261 | assert len(result.output.strip()) == 71 |
| 262 | |
| 263 | |
| 264 | # --------------------------------------------------------------------------- |
| 265 | # TestCrossCheck |
| 266 | # --------------------------------------------------------------------------- |
| 267 | |
| 268 | class TestCrossCheck: |
| 269 | """File and stdin paths produce identical canonical IDs for the same bytes.""" |
| 270 | |
| 271 | def test_file_and_stdin_same_id(self, tmp_path: pathlib.Path) -> None: |
| 272 | content = b"cross check content" |
| 273 | f = tmp_path / "f.txt" |
| 274 | f.write_bytes(content) |
| 275 | file_id = json.loads(_run("--json", str(f)).output)["object_id"] |
| 276 | stdin_id = json.loads(_run("--json", "--stdin", stdin=content).output)["object_id"] |
| 277 | assert file_id == stdin_id |
| 278 | |
| 279 | def test_write_file_and_stdin_same_id(self, tmp_path: pathlib.Path) -> None: |
| 280 | repo = _make_repo(tmp_path) |
| 281 | content = b"write cross check" |
| 282 | f = repo / "f.txt" |
| 283 | f.write_bytes(content) |
| 284 | file_id = json.loads(_run("--json", "--write", str(f), repo=repo).output)["object_id"] |
| 285 | stdin_id = json.loads( |
| 286 | _run("--json", "--stdin", "--write", stdin=content, repo=repo).output |
| 287 | )["object_id"] |
| 288 | assert file_id == stdin_id |
| 289 | |
| 290 | def test_hash_bytes_returns_canonical(self) -> None: |
| 291 | """_hash_bytes must return sha256:-prefixed ID, not bare hex.""" |
| 292 | from muse.cli.commands.hash_object import _hash_bytes |
| 293 | result = _hash_bytes(b"test data") |
| 294 | assert result.startswith("sha256:"), ( |
| 295 | f"_hash_bytes must return 'sha256:<hex>', got {result!r}" |
| 296 | ) |
| 297 | assert len(result) == 71 |
| 298 | |
| 299 | |
| 300 | # --------------------------------------------------------------------------- |
| 301 | # TestRegisterFlags — argparse-level verification |
| 302 | # --------------------------------------------------------------------------- |
| 303 | |
| 304 | |
| 305 | class TestRegisterFlags: |
| 306 | """Verify that register() wires --json / -j correctly.""" |
| 307 | |
| 308 | def _make_parser(self) -> "argparse.ArgumentParser": |
| 309 | import argparse |
| 310 | from muse.cli.commands.hash_object import register |
| 311 | ap = argparse.ArgumentParser() |
| 312 | subs = ap.add_subparsers() |
| 313 | register(subs) |
| 314 | return ap |
| 315 | |
| 316 | def test_json_flag_long(self) -> None: |
| 317 | ns = self._make_parser().parse_args(["hash-object", "--stdin", "--json"]) |
| 318 | assert ns.json_out is True |
| 319 | |
| 320 | def test_j_alias(self) -> None: |
| 321 | ns = self._make_parser().parse_args(["hash-object", "--stdin", "-j"]) |
| 322 | assert ns.json_out is True |
| 323 | |
| 324 | def test_default_is_text(self) -> None: |
| 325 | ns = self._make_parser().parse_args(["hash-object", "--stdin"]) |
| 326 | assert ns.json_out is False |
| 327 | |
| 328 | def test_dest_is_json_out(self) -> None: |
| 329 | ns = self._make_parser().parse_args(["hash-object", "--stdin", "-j"]) |
| 330 | assert hasattr(ns, "json_out") |
| 331 | assert not hasattr(ns, "fmt") |
File History
4 commits
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
21 days ago
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
22 days ago
sha256:c06a9b9b9fee26c68ea725b44d54b2c0a171301ce9de746d5b656617b4463a9a
fix: repair four test failures from post-migration audit
Sonnet 4.6
patch
28 days ago
sha256:1900655993c83c4107067375548a7be823e471d2515830842f1a12cba4bd3cdf
fix: unified object store migration — idempotent writes, JS…
Sonnet 4.6
minor
⚠
29 days ago