test_phase2_or_set_symbol_independence.py
python
sha256:2eaa5d95f9d9383498e76947410a26e5a3ba23d182f339910c424cf88fad412b
fix: try fetch/presign before fetch/mpack to avoid Cloudfla…
Sonnet 4.6
patch
6 days ago
| 1 | """Phase 2 — OR-Set CRDT semantics + symbol-level independence-aware merge. |
| 2 | |
| 3 | Tier 1 — OR-Set for imports and variables |
| 4 | ------------------------------------------ |
| 5 | Import and variable additions from concurrent branches are always independent: |
| 6 | add-wins, never conflict. This is the OR-Set guarantee: union of additions, |
| 7 | with tombstones only when *both* sides agree on a deletion. |
| 8 | |
| 9 | Tier 2 — Symbol-level independence for functions and classes |
| 10 | ------------------------------------------------------------- |
| 11 | Two branches that add or modify *different* named symbols in the same file |
| 12 | should produce a clean merged file containing all changes. Currently they |
| 13 | produce a spurious file-level conflict because the raw blob IDs diverge. |
| 14 | |
| 15 | The fix: when merge_ops() finds that all child ops across PatchOps for the |
| 16 | same file commute (no symbol-level conflict), it reconstructs the merged |
| 17 | blob via three_way_merge_lines() and writes it to the object store. A clean |
| 18 | text merge removes the file from the conflict list and updates the manifest. |
| 19 | |
| 20 | Test categories |
| 21 | --------------- |
| 22 | TestORSetImports — concurrent import adds never conflict (Tier 1) |
| 23 | TestORSetVariables — concurrent variable adds never conflict (Tier 1) |
| 24 | TestSymbolIndependence — concurrent adds/edits of different symbols are clean (Tier 2) |
| 25 | TestSymbolConflictPreserved — genuine same-symbol conflicts still surface (Tier 2) |
| 26 | TestMergedBlobCorrectness — merged file content is complete and well-formed |
| 27 | """ |
| 28 | |
| 29 | from __future__ import annotations |
| 30 | from collections.abc import Mapping |
| 31 | |
| 32 | from typing import TYPE_CHECKING |
| 33 | import pathlib |
| 34 | import textwrap |
| 35 | |
| 36 | import pytest |
| 37 | |
| 38 | from muse.plugins.code.plugin import CodePlugin |
| 39 | from muse.core.types import blob_id |
| 40 | |
| 41 | if TYPE_CHECKING: |
| 42 | from muse.domain import MergeResult |
| 43 | |
| 44 | |
| 45 | # --------------------------------------------------------------------------- |
| 46 | # Helpers |
| 47 | # --------------------------------------------------------------------------- |
| 48 | |
| 49 | def _oid(content: bytes) -> str: |
| 50 | return blob_id(content) |
| 51 | |
| 52 | |
| 53 | def _write_blob(root: pathlib.Path, content: bytes) -> str: |
| 54 | from muse.core.object_store import write_object |
| 55 | oid = _oid(content) |
| 56 | write_object(root, oid, content) |
| 57 | return oid |
| 58 | |
| 59 | |
| 60 | def _snap(root: pathlib.Path, files: Mapping[str, bytes]) -> Mapping[str, object]: |
| 61 | return { |
| 62 | "files": {path: _write_blob(root, content) for path, content in files.items()}, |
| 63 | "domain": "code", |
| 64 | "directories": [], |
| 65 | } |
| 66 | |
| 67 | |
| 68 | def _read_merged_blob(root: pathlib.Path, result: "MergeResult", path: str) -> str: |
| 69 | from muse.core.object_store import read_object |
| 70 | oid = result.merged["files"][path] |
| 71 | raw = read_object(root, oid) |
| 72 | assert raw is not None, f"merged blob for {path} not in object store" |
| 73 | return raw.decode("utf-8") |
| 74 | |
| 75 | |
| 76 | # --------------------------------------------------------------------------- |
| 77 | # Tier 1 — OR-Set for imports |
| 78 | # --------------------------------------------------------------------------- |
| 79 | |
| 80 | class TestORSetImports: |
| 81 | """Concurrent import additions from two branches must never conflict.""" |
| 82 | |
| 83 | def test_concurrent_import_adds_no_conflict(self, tmp_path: pathlib.Path) -> None: |
| 84 | """Branch A adds 'import os', branch B adds 'import sys' → clean merge.""" |
| 85 | plugin = CodePlugin() |
| 86 | |
| 87 | base_src = b"# utils.py\n\ndef process(): pass\n" |
| 88 | ours_src = b"# utils.py\nimport os\n\ndef process(): pass\n" |
| 89 | theirs_src = b"# utils.py\nimport sys\n\ndef process(): pass\n" |
| 90 | |
| 91 | base = _snap(tmp_path, {"src/utils.py": base_src}) |
| 92 | ours = _snap(tmp_path, {"src/utils.py": ours_src}) |
| 93 | theirs = _snap(tmp_path, {"src/utils.py": theirs_src}) |
| 94 | |
| 95 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 96 | |
| 97 | assert "src/utils.py" not in result.conflicts, ( |
| 98 | "Concurrent import additions must not conflict — OR-Set semantics" |
| 99 | ) |
| 100 | |
| 101 | def test_concurrent_import_adds_both_present_in_merged(self, tmp_path: pathlib.Path) -> None: |
| 102 | """The merged file must contain both imports.""" |
| 103 | plugin = CodePlugin() |
| 104 | |
| 105 | base_src = b"# utils.py\n\ndef process(): pass\n" |
| 106 | ours_src = b"# utils.py\nimport os\n\ndef process(): pass\n" |
| 107 | theirs_src = b"# utils.py\nimport sys\n\ndef process(): pass\n" |
| 108 | |
| 109 | base = _snap(tmp_path, {"src/utils.py": base_src}) |
| 110 | ours = _snap(tmp_path, {"src/utils.py": ours_src}) |
| 111 | theirs = _snap(tmp_path, {"src/utils.py": theirs_src}) |
| 112 | |
| 113 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 114 | |
| 115 | merged_text = _read_merged_blob(tmp_path, result, "src/utils.py") |
| 116 | assert "import os" in merged_text, "ours import must survive in merged blob" |
| 117 | assert "import sys" in merged_text, "theirs import must survive in merged blob" |
| 118 | |
| 119 | def test_three_concurrent_import_adds_all_survive(self, tmp_path: pathlib.Path) -> None: |
| 120 | """Even with multiple imports added per side, all survive.""" |
| 121 | plugin = CodePlugin() |
| 122 | |
| 123 | base_src = b"def fn(): pass\n" |
| 124 | ours_src = b"import os\nimport pathlib\n\ndef fn(): pass\n" |
| 125 | theirs_src = b"import sys\nimport json\n\ndef fn(): pass\n" |
| 126 | |
| 127 | base = _snap(tmp_path, {"lib.py": base_src}) |
| 128 | ours = _snap(tmp_path, {"lib.py": ours_src}) |
| 129 | theirs = _snap(tmp_path, {"lib.py": theirs_src}) |
| 130 | |
| 131 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 132 | |
| 133 | assert "lib.py" not in result.conflicts |
| 134 | merged_text = _read_merged_blob(tmp_path, result, "lib.py") |
| 135 | for imp in ("import os", "import pathlib", "import sys", "import json"): |
| 136 | assert imp in merged_text, f"{imp} missing from merged blob" |
| 137 | |
| 138 | def test_same_import_added_on_both_sides_deduplicates(self, tmp_path: pathlib.Path) -> None: |
| 139 | """Both branches adding the same import → one copy in merged file, no conflict.""" |
| 140 | plugin = CodePlugin() |
| 141 | |
| 142 | base_src = b"def fn(): pass\n" |
| 143 | both_src = b"import os\n\ndef fn(): pass\n" |
| 144 | |
| 145 | base = _snap(tmp_path, {"lib.py": base_src}) |
| 146 | ours = _snap(tmp_path, {"lib.py": both_src}) |
| 147 | theirs = _snap(tmp_path, {"lib.py": both_src}) |
| 148 | |
| 149 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 150 | |
| 151 | assert "lib.py" not in result.conflicts |
| 152 | merged_text = _read_merged_blob(tmp_path, result, "lib.py") |
| 153 | assert merged_text.count("import os") == 1, "duplicate import must be deduplicated" |
| 154 | |
| 155 | |
| 156 | # --------------------------------------------------------------------------- |
| 157 | # Tier 1 — OR-Set for variables |
| 158 | # --------------------------------------------------------------------------- |
| 159 | |
| 160 | class TestORSetVariables: |
| 161 | """Concurrent top-level variable additions must never conflict.""" |
| 162 | |
| 163 | def test_concurrent_variable_adds_no_conflict(self, tmp_path: pathlib.Path) -> None: |
| 164 | """Branch A adds MAX=100, branch B adds MIN=0 → clean merge.""" |
| 165 | plugin = CodePlugin() |
| 166 | |
| 167 | base_src = b"def fn(): pass\n" |
| 168 | ours_src = b"MAX = 100\n\ndef fn(): pass\n" |
| 169 | theirs_src = b"MIN = 0\n\ndef fn(): pass\n" |
| 170 | |
| 171 | base = _snap(tmp_path, {"config.py": base_src}) |
| 172 | ours = _snap(tmp_path, {"config.py": ours_src}) |
| 173 | theirs = _snap(tmp_path, {"config.py": theirs_src}) |
| 174 | |
| 175 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 176 | |
| 177 | assert "config.py" not in result.conflicts, ( |
| 178 | "Concurrent variable additions must not conflict" |
| 179 | ) |
| 180 | |
| 181 | def test_concurrent_variable_adds_both_present(self, tmp_path: pathlib.Path) -> None: |
| 182 | """Both variables appear in the merged file.""" |
| 183 | plugin = CodePlugin() |
| 184 | |
| 185 | base_src = b"def fn(): pass\n" |
| 186 | ours_src = b"MAX = 100\n\ndef fn(): pass\n" |
| 187 | theirs_src = b"MIN = 0\n\ndef fn(): pass\n" |
| 188 | |
| 189 | base = _snap(tmp_path, {"config.py": base_src}) |
| 190 | ours = _snap(tmp_path, {"config.py": ours_src}) |
| 191 | theirs = _snap(tmp_path, {"config.py": theirs_src}) |
| 192 | |
| 193 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 194 | |
| 195 | merged_text = _read_merged_blob(tmp_path, result, "config.py") |
| 196 | assert "MAX = 100" in merged_text |
| 197 | assert "MIN = 0" in merged_text |
| 198 | |
| 199 | |
| 200 | # --------------------------------------------------------------------------- |
| 201 | # Tier 2 — Symbol independence for functions and classes |
| 202 | # --------------------------------------------------------------------------- |
| 203 | |
| 204 | class TestSymbolIndependence: |
| 205 | """Non-overlapping symbol changes in the same file must produce a clean merge.""" |
| 206 | |
| 207 | def test_concurrent_function_adds_no_conflict(self, tmp_path: pathlib.Path) -> None: |
| 208 | """Branch A adds def foo(), branch B adds def bar() → clean merge.""" |
| 209 | plugin = CodePlugin() |
| 210 | |
| 211 | base_src = textwrap.dedent("""\ |
| 212 | def existing(): |
| 213 | pass |
| 214 | """).encode() |
| 215 | |
| 216 | ours_src = textwrap.dedent("""\ |
| 217 | def existing(): |
| 218 | pass |
| 219 | |
| 220 | def foo(): |
| 221 | return 1 |
| 222 | """).encode() |
| 223 | |
| 224 | theirs_src = textwrap.dedent("""\ |
| 225 | def existing(): |
| 226 | pass |
| 227 | |
| 228 | def bar(): |
| 229 | return 2 |
| 230 | """).encode() |
| 231 | |
| 232 | base = _snap(tmp_path, {"module.py": base_src}) |
| 233 | ours = _snap(tmp_path, {"module.py": ours_src}) |
| 234 | theirs = _snap(tmp_path, {"module.py": theirs_src}) |
| 235 | |
| 236 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 237 | |
| 238 | assert "module.py" not in result.conflicts, ( |
| 239 | "Concurrent additions of different functions must not conflict" |
| 240 | ) |
| 241 | |
| 242 | def test_concurrent_function_adds_both_present(self, tmp_path: pathlib.Path) -> None: |
| 243 | """Both added functions appear in the merged file.""" |
| 244 | plugin = CodePlugin() |
| 245 | |
| 246 | base_src = b"def existing(): pass\n" |
| 247 | ours_src = b"def existing(): pass\n\ndef foo():\n return 1\n" |
| 248 | theirs_src = b"def existing(): pass\n\ndef bar():\n return 2\n" |
| 249 | |
| 250 | base = _snap(tmp_path, {"module.py": base_src}) |
| 251 | ours = _snap(tmp_path, {"module.py": ours_src}) |
| 252 | theirs = _snap(tmp_path, {"module.py": theirs_src}) |
| 253 | |
| 254 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 255 | |
| 256 | merged_text = _read_merged_blob(tmp_path, result, "module.py") |
| 257 | assert "def foo" in merged_text, "ours function must appear in merged file" |
| 258 | assert "def bar" in merged_text, "theirs function must appear in merged file" |
| 259 | assert "def existing" in merged_text, "base function must be preserved" |
| 260 | |
| 261 | def test_different_functions_modified_no_conflict(self, tmp_path: pathlib.Path) -> None: |
| 262 | """Branch A modifies foo(), branch B modifies bar() → clean merge.""" |
| 263 | plugin = CodePlugin() |
| 264 | |
| 265 | base_src = textwrap.dedent("""\ |
| 266 | def foo(): |
| 267 | return 0 |
| 268 | |
| 269 | def bar(): |
| 270 | return 0 |
| 271 | """).encode() |
| 272 | |
| 273 | ours_src = textwrap.dedent("""\ |
| 274 | def foo(): |
| 275 | return 1 |
| 276 | |
| 277 | def bar(): |
| 278 | return 0 |
| 279 | """).encode() |
| 280 | |
| 281 | theirs_src = textwrap.dedent("""\ |
| 282 | def foo(): |
| 283 | return 0 |
| 284 | |
| 285 | def bar(): |
| 286 | return 2 |
| 287 | """).encode() |
| 288 | |
| 289 | base = _snap(tmp_path, {"module.py": base_src}) |
| 290 | ours = _snap(tmp_path, {"module.py": ours_src}) |
| 291 | theirs = _snap(tmp_path, {"module.py": theirs_src}) |
| 292 | |
| 293 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 294 | |
| 295 | assert "module.py" not in result.conflicts, ( |
| 296 | "Modifications to different functions must not conflict" |
| 297 | ) |
| 298 | |
| 299 | def test_different_functions_modified_both_present(self, tmp_path: pathlib.Path) -> None: |
| 300 | """The merged file has ours' version of foo() and theirs' version of bar().""" |
| 301 | plugin = CodePlugin() |
| 302 | |
| 303 | base_src = b"def foo():\n return 0\n\ndef bar():\n return 0\n" |
| 304 | ours_src = b"def foo():\n return 1\n\ndef bar():\n return 0\n" |
| 305 | theirs_src = b"def foo():\n return 0\n\ndef bar():\n return 2\n" |
| 306 | |
| 307 | base = _snap(tmp_path, {"module.py": base_src}) |
| 308 | ours = _snap(tmp_path, {"module.py": ours_src}) |
| 309 | theirs = _snap(tmp_path, {"module.py": theirs_src}) |
| 310 | |
| 311 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 312 | |
| 313 | merged_text = _read_merged_blob(tmp_path, result, "module.py") |
| 314 | assert "return 1" in merged_text, "ours change to foo() must be in merged file" |
| 315 | assert "return 2" in merged_text, "theirs change to bar() must be in merged file" |
| 316 | |
| 317 | def test_concurrent_class_adds_no_conflict(self, tmp_path: pathlib.Path) -> None: |
| 318 | """Branch A adds class Foo, branch B adds class Bar → clean merge.""" |
| 319 | plugin = CodePlugin() |
| 320 | |
| 321 | base_src = b"# module\n" |
| 322 | ours_src = b"# module\n\nclass Foo:\n pass\n" |
| 323 | theirs_src = b"# module\n\nclass Bar:\n pass\n" |
| 324 | |
| 325 | base = _snap(tmp_path, {"module.py": base_src}) |
| 326 | ours = _snap(tmp_path, {"module.py": ours_src}) |
| 327 | theirs = _snap(tmp_path, {"module.py": theirs_src}) |
| 328 | |
| 329 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 330 | |
| 331 | assert "module.py" not in result.conflicts |
| 332 | |
| 333 | def test_independent_changes_in_multiple_files_all_clean(self, tmp_path: pathlib.Path) -> None: |
| 334 | """Multiple files with independent changes all merge cleanly.""" |
| 335 | plugin = CodePlugin() |
| 336 | |
| 337 | base = _snap(tmp_path, { |
| 338 | "a.py": b"def fa(): pass\n", |
| 339 | "b.py": b"def fb(): pass\n", |
| 340 | }) |
| 341 | ours = _snap(tmp_path, { |
| 342 | "a.py": b"def fa(): pass\n\ndef fa2(): pass\n", |
| 343 | "b.py": b"def fb(): pass\n", |
| 344 | }) |
| 345 | theirs = _snap(tmp_path, { |
| 346 | "a.py": b"def fa(): pass\n", |
| 347 | "b.py": b"def fb(): pass\n\ndef fb2(): pass\n", |
| 348 | }) |
| 349 | |
| 350 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 351 | |
| 352 | assert result.conflicts == [], f"Expected no conflicts, got: {result.conflicts}" |
| 353 | |
| 354 | |
| 355 | # --------------------------------------------------------------------------- |
| 356 | # Tier 2 — Genuine conflicts still surface |
| 357 | # --------------------------------------------------------------------------- |
| 358 | |
| 359 | class TestSymbolConflictPreserved: |
| 360 | """Genuine same-symbol conflicts must still be detected and reported.""" |
| 361 | |
| 362 | def test_same_function_modified_both_sides_conflicts(self, tmp_path: pathlib.Path) -> None: |
| 363 | """Both branches modified the same function body → real conflict.""" |
| 364 | plugin = CodePlugin() |
| 365 | |
| 366 | base_src = b"def compute():\n return 0\n" |
| 367 | ours_src = b"def compute():\n return 1\n" |
| 368 | theirs_src = b"def compute():\n return 2\n" |
| 369 | |
| 370 | base = _snap(tmp_path, {"ops.py": base_src}) |
| 371 | ours = _snap(tmp_path, {"ops.py": ours_src}) |
| 372 | theirs = _snap(tmp_path, {"ops.py": theirs_src}) |
| 373 | |
| 374 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 375 | |
| 376 | assert any("ops.py" in c for c in result.conflicts), ( |
| 377 | "Same-function conflict must still be detected" |
| 378 | ) |
| 379 | |
| 380 | def test_mixed_file_some_symbols_conflict_some_independent( |
| 381 | self, tmp_path: pathlib.Path |
| 382 | ) -> None: |
| 383 | """When one symbol conflicts, the file is in conflicts — independent ones don't suppress it.""" |
| 384 | plugin = CodePlugin() |
| 385 | |
| 386 | base_src = textwrap.dedent("""\ |
| 387 | def shared(): |
| 388 | return 0 |
| 389 | |
| 390 | def independent_a(): |
| 391 | pass |
| 392 | """).encode() |
| 393 | |
| 394 | ours_src = textwrap.dedent("""\ |
| 395 | def shared(): |
| 396 | return 1 |
| 397 | |
| 398 | def independent_a(): |
| 399 | pass |
| 400 | |
| 401 | def only_on_ours(): |
| 402 | pass |
| 403 | """).encode() |
| 404 | |
| 405 | theirs_src = textwrap.dedent("""\ |
| 406 | def shared(): |
| 407 | return 2 |
| 408 | |
| 409 | def independent_a(): |
| 410 | pass |
| 411 | |
| 412 | def only_on_theirs(): |
| 413 | pass |
| 414 | """).encode() |
| 415 | |
| 416 | base = _snap(tmp_path, {"mixed.py": base_src}) |
| 417 | ours = _snap(tmp_path, {"mixed.py": ours_src}) |
| 418 | theirs = _snap(tmp_path, {"mixed.py": theirs_src}) |
| 419 | |
| 420 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 421 | |
| 422 | # The shared() conflict must surface — even though independent symbols exist |
| 423 | assert any("mixed.py" in c for c in result.conflicts), ( |
| 424 | "File with a genuine symbol conflict must still appear in conflicts" |
| 425 | ) |
| 426 | |
| 427 | |
| 428 | # --------------------------------------------------------------------------- |
| 429 | # Merged blob correctness |
| 430 | # --------------------------------------------------------------------------- |
| 431 | |
| 432 | class TestMergedBlobCorrectness: |
| 433 | """Reconstructed blobs must be syntactically valid and not contain conflict markers.""" |
| 434 | |
| 435 | def test_merged_blob_has_no_conflict_markers(self, tmp_path: pathlib.Path) -> None: |
| 436 | """Blobs auto-resolved via independence must not contain <<<<<<< markers.""" |
| 437 | plugin = CodePlugin() |
| 438 | |
| 439 | base_src = b"def existing(): pass\n" |
| 440 | ours_src = b"def existing(): pass\n\ndef foo(): return 1\n" |
| 441 | theirs_src = b"def existing(): pass\n\ndef bar(): return 2\n" |
| 442 | |
| 443 | base = _snap(tmp_path, {"m.py": base_src}) |
| 444 | ours = _snap(tmp_path, {"m.py": ours_src}) |
| 445 | theirs = _snap(tmp_path, {"m.py": theirs_src}) |
| 446 | |
| 447 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 448 | |
| 449 | merged_text = _read_merged_blob(tmp_path, result, "m.py") |
| 450 | assert "<<<<<<<" not in merged_text, "auto-resolved blob must not contain conflict markers" |
| 451 | assert "=======" not in merged_text |
| 452 | assert ">>>>>>>" not in merged_text |
| 453 | |
| 454 | def test_merged_blob_is_valid_python(self, tmp_path: pathlib.Path) -> None: |
| 455 | """Reconstructed blob must parse without SyntaxError.""" |
| 456 | import ast |
| 457 | |
| 458 | plugin = CodePlugin() |
| 459 | |
| 460 | base_src = b"def existing(): pass\n" |
| 461 | ours_src = b"def existing(): pass\n\ndef foo():\n return 1\n" |
| 462 | theirs_src = b"def existing(): pass\n\ndef bar():\n return 2\n" |
| 463 | |
| 464 | base = _snap(tmp_path, {"m.py": base_src}) |
| 465 | ours = _snap(tmp_path, {"m.py": ours_src}) |
| 466 | theirs = _snap(tmp_path, {"m.py": theirs_src}) |
| 467 | |
| 468 | result = plugin.merge_ops(base, ours, theirs, [], [], repo_root=tmp_path) |
| 469 | |
| 470 | merged_text = _read_merged_blob(tmp_path, result, "m.py") |
| 471 | try: |
| 472 | ast.parse(merged_text) |
| 473 | except SyntaxError as exc: |
| 474 | pytest.fail(f"Merged blob is not valid Python: {exc}\n\n{merged_text}") |
File History
1 commit
sha256:2eaa5d95f9d9383498e76947410a26e5a3ba23d182f339910c424cf88fad412b
fix: try fetch/presign before fetch/mpack to avoid Cloudfla…
Sonnet 4.6
patch
6 days ago