Viewing files at commit c467b16393a4Commit detailsView latest

Browse files
tests/test_merge_suite.py 547 lines · 31.4 KB · python
1
"""Instrument tests for suites/merge (classification, laws, oracle parsing, vendoring).
2
3
Hermetic: no Oak binary and no network. Git-backed tests skip when git is absent.
4
"""
5
import hashlib
6
import json
7
from pathlib import Path
8
import shutil
9
import sys
10
import tempfile
11
import unittest
12
13
ROOT = Path(__file__).resolve().parents[1]
14
sys.path.insert(0, str(ROOT / "suites" / "merge"))
15
sys.path.insert(0, str(ROOT / "scripts"))
16
17
from mergesuite import classify as cl  # noqa: E402
18
from mergesuite.gitops import (Git, git_conflicts, normalize_git_conflict_type,  # noqa: E402
19
                               parse_merge_tree_z, prepare_git_home, worktree_manifest)
20
from mergesuite.oak_adapter import parse_conflict_show, summarize_prediction  # noqa: E402
21
22
VENDOR = ROOT / "suites" / "merge" / "vendor" / "gitoxide"
23
24
25
def v(tag):
26
    return ("file", hashlib.sha256(tag.encode()).hexdigest())
27
28
29
def oak(tree, conflicts=(), clean=None):
30
    return {"tree": tree, "conflicts": {p: [] for p in conflicts},
31
            "clean": (not conflicts) if clean is None else clean, "error": None, "error_kind": None}
32
33
34
def gitres(tree, conflicts=()):
35
    return {"tree": tree, "conflicts": {p: ["content"] for p in conflicts}, "clean": not conflicts}
36
37
38
class ClassifyTests(unittest.TestCase):
39
    base = {"a": v("a0"), "b": v("b0")}
40
    ours = {"a": v("a1"), "b": v("b0")}
41
    theirs = {"a": v("a0"), "b": v("b1")}
42
    merged = {"a": v("a1"), "b": v("b1")}
43
44
    def test_equal(self):
45
        r = cl.classify(gitres(self.merged), oak(self.merged), self.base, self.ours, self.theirs)
46
        self.assertEqual(r["class"], "equal-to-git")
47
48
    def test_lost_change_is_oak_wrong_even_if_git_agrees(self):
49
        # Oak drops theirs' change to b.
50
        r = cl.classify(gitres(self.merged), oak(self.ours), self.base, self.ours, self.theirs)
51
        self.assertEqual(r["class"], "oak-wrong")
52
        self.assertIn("lost_change: b", r["reasons"][0])
53
54
    def test_lost_deletion(self):
55
        theirs = {"a": v("a0")}  # theirs deletes b
56
        r = cl.classify(gitres({"a": v("a1")}), oak({"a": v("a1"), "b": v("b0")}),
57
                        self.base, self.ours, theirs)
58
        self.assertEqual(r["class"], "oak-wrong")
59
60
    def test_fabricated_content(self):
61
        r = cl.classify(gitres(self.merged), oak({"a": v("a1"), "b": v("zzz")}),
62
                        self.base, self.ours, self.theirs)
63
        self.assertEqual(r["class"], "oak-wrong")
64
        self.assertIn("fabricated", r["reasons"][0])
65
66
    def test_false_conflict_is_differs_but_valid(self):
67
        r = cl.classify(gitres(self.merged), oak(self.merged, conflicts=["a"]),
68
                        self.base, self.ours, self.theirs)
69
        self.assertEqual(r["class"], "differs-but-valid")
70
        self.assertTrue(any("oak_conflicts_where_git_clean" in x for x in r["reasons"]))
71
72
    def test_rename_not_detected_keeps_both_sides(self):
73
        # ours renames a->c; theirs edits b. git moves; Oak path-level result matches git here,
74
        # but when git conflicts and Oak keeps everything, it is differs-but-valid.
75
        base = {"a": v("a0")}
76
        ours = {"b": v("a0")}
77
        theirs = {"c": v("a0")}
78
        r = cl.classify(gitres({"b": v("a0"), "c": v("a0")}, conflicts=["a", "b", "c"]),
79
                        oak({"b": v("a0"), "c": v("a0")}), base, ours, theirs)
80
        self.assertEqual(r["class"], "differs-but-valid")
81
        self.assertTrue(any("oak_clean_where_git_conflicts" in x for x in r["reasons"]))
82
83
    def test_expectation_violation_wins(self):
84
        r = cl.classify(gitres(self.merged), oak(self.merged), self.base, self.ours, self.theirs,
85
                        expect={"paths": {"a": v("a0")[1]}})
86
        self.assertEqual(r["class"], "oak-wrong")
87
        self.assertIn("expectation_violated", r["reasons"][0])
88
89
    def test_expectation_violation_still_reports_law_reasons(self):
90
        # Oak drops theirs' change to b AND violates the stated expectation: both are listed.
91
        r = cl.classify(gitres(self.merged), oak(self.ours), self.base, self.ours, self.theirs,
92
                        expect={"clean": False, "label": "lbl"})
93
        self.assertEqual(r["class"], "oak-wrong")
94
        self.assertEqual(r["reasons"], ["expectation_violated[lbl]: clean=True expected False", "lost_change: b"])
95
96
    def test_t6416_shape_known_violation_maps_and_new_loss_is_listed(self):
97
        from mergesuite.runner import load_known_bugs, match_known_bug
98
        bugs = load_known_bugs()
99
        # criss-cross: no unique base. Main keeps `file`, the branch deleted it; Oak picks main's side.
100
        ours = {"keep": v("k"), "ours.txt": v("o")}
101
        theirs = {"file": v("B"), "keep": v("k")}
102
        expect = {"clean": False, "label": "criss_cross_divergent_resolution"}
103
        git = gitres({"keep": v("k"), "ours.txt": v("o")}, conflicts=["file"])
104
        as_landed = {"keep": v("k"), "ours.txt": v("o"), "file": v("B")}
105
        r = cl.classify(git, oak(as_landed), None, ours, theirs, expect=expect)
106
        self.assertEqual(r["reasons"], ["expectation_violated[criss_cross_divergent_resolution]: clean=True expected False"])
107
        self.assertIsNone(r["evidence"]["law_no_lost_change"])
108
        row = {"class": r["class"], "reasons": r["reasons"], "scenario": "git-t6416-criss-cross-modify-delete"}
109
        self.assertEqual(match_known_bug(row, bugs), "fb-541")
110
        # Oak additionally loses `keep` (both sides agree on it): listed, and the row no longer maps.
111
        lossy = {"ours.txt": v("o"), "file": v("B")}
112
        r = cl.classify(git, oak(lossy), None, ours, theirs, expect=expect)
113
        self.assertEqual(r["reasons"][1:], ["lost_change: keep"])
114
        self.assertIsNone(match_known_bug(dict(row, reasons=r["reasons"]), bugs))
115
        # committed markers on a both-sides-changed file are also listed without a base
116
        ours2, theirs2 = dict(ours, t=v("t-o")), dict(theirs, t=v("t-t"))
117
        texts = {"t": {"base": b"", "ours": b"a\n", "theirs": b"b\n", "merged": b"<<<<<<< x\na\n=======\nb\n>>>>>>> y\n"}}
118
        r = cl.classify(git, oak(dict(as_landed, t=v("m"))), None, ours2, theirs2, expect=expect, texts=texts)
119
        self.assertIn("committed_conflict_markers: t", r["reasons"])
120
121
    def test_base_free_checks(self):
122
        ours = {"same": v("s"), "o_only": v("o"), "both": v("bo")}
123
        theirs = {"same": v("s"), "both": v("bt")}
124
        law = cl.base_free_checks(ours, theirs, {"both": v("bo"), "o_only": v("o")}, set())
125
        self.assertEqual(law["lost"], ["same"])
126
        law = cl.base_free_checks(ours, theirs, {"same": v("zz"), "both": v("bo")}, set())
127
        self.assertEqual((law["fabricated"], law["lost"]), (["same"], []))
128
        law = cl.base_free_checks(ours, theirs, {"same": ("exec", v("s")[1])}, {"both"})
129
        self.assertEqual(law["mode_lost"], ["same"])
130
        self.assertFalse(any(cl.base_free_checks(ours, theirs, {"same": v("s")}, {"both"}).values()))
131
132
    def test_expect_absent_and_clean(self):
133
        r = cl.classify(gitres(self.merged), oak(self.merged), self.base, self.ours, self.theirs,
134
                        expect={"clean": False})
135
        self.assertEqual(r["class"], "oak-wrong")
136
        r = cl.classify(gitres(self.merged), oak(self.merged), self.base, self.ours, self.theirs,
137
                        expect={"paths": {"gone": None}, "clean": True})
138
        self.assertEqual(r["class"], "equal-to-git")
139
140
    def test_unsupported_and_harness_and_oak_error(self):
141
        self.assertEqual(cl.classify(None, None, {}, {}, {}, unsupported="gitlink")["class"], "oak-unsupported")
142
        self.assertEqual(cl.classify(None, None, {}, {}, {}, harness_error="x")["class"], "harness-error")
143
        self.assertEqual(cl.classify(gitres({}), {"error_kind": "harness", "error": "shim"}, {}, {}, {})["class"],
144
                         "harness-error")
145
        r = cl.classify(gitres({}), {"error_kind": "oak", "error": "boom", "clean": None}, {}, {}, {})
146
        self.assertEqual(r["class"], "oak-wrong")
147
        self.assertEqual(cl.classify(gitres({}), None, {}, {}, {})["class"], "harness-error")
148
149
    def test_no_unique_base_leaves_law_unmeasured(self):
150
        r = cl.classify(gitres(self.merged), oak(self.merged), None, self.ours, self.theirs)
151
        self.assertEqual(r["class"], "equal-to-git")
152
        self.assertIsNone(r["evidence"]["law_no_lost_change"])
153
154
    def test_text_merge_law(self):
155
        base = b"1\n2\n3\n4\n5\n6\n"
156
        o = b"1 ours\n2\n3\n4\n5\n6\n"
157
        t = b"1\n2\n3\n4\n5\n6 theirs\n"
158
        good = b"1 ours\n2\n3\n4\n5\n6 theirs\n"
159
        self.assertTrue(cl.text_preserves_sides(base, o, t, good))
160
        self.assertFalse(cl.text_preserves_sides(base, o, t, o))
161
        self.assertFalse(cl.text_preserves_sides(base, o, t, b"<<<<<<< a\n" + good))
162
        trees = ({"f": v("b")}, {"f": v("o")}, {"f": v("t")})
163
        texts = {"f": {"base": base, "ours": o, "theirs": t, "merged": good}}
164
        law = cl.no_lost_change(*trees, {"f": v("m")}, set(), texts)
165
        self.assertEqual(law["text_merged"], ["f"])
166
        texts["f"]["merged"] = o
167
        self.assertEqual(cl.no_lost_change(*trees, {"f": v("m")}, set(), texts)["lost"], ["f"])
168
169
    def test_one_side_taken_wholesale_is_lost(self):
170
        # base exec+empty; ours: mode-only change; theirs: content change. Oak keeping theirs'
171
        # entry (content AND exec bit) drops ours' mode change.
172
        base = {"w": ("exec", "empty")}
173
        ours = {"w": ("file", "empty")}
174
        theirs = {"w": ("exec", "123")}
175
        law = cl.no_lost_change(base, ours, theirs, {"w": ("exec", "123")}, set())
176
        self.assertEqual(law["mode_lost"], ["w"])
177
        # a content change dropped wholesale stays a plain lost change
178
        law = cl.no_lost_change({"f": v("0")}, {"f": v("1")}, {"f": v("2")}, {"f": v("1")}, set())
179
        self.assertEqual(law["lost"], ["f"])
180
        law = cl.no_lost_change(base, ours, theirs, {"w": ("file", "123")}, set(),
181
                                {"w": {"base": b"", "ours": b"", "theirs": b"1\n2\n3\n", "merged": b"1\n2\n3\n"}})
182
        self.assertEqual(law["text_merged"], ["w"])
183
184
    def test_conflicted_paths_are_exempt_from_law(self):
185
        law = cl.no_lost_change(self.base, self.ours, self.theirs, {}, {"a", "b"})
186
        self.assertEqual(law, {"lost": [], "mode_lost": [], "markers": [], "markers_no_final_newline": [], "markers_symlink": [], "fabricated": [], "text_merged": []})
187
188
    def test_text_merged_but_mode_dropped_is_oak_wrong(self):
189
        # QA shape generated/s20260928-022: ours edits lines, theirs edits other lines AND chmods.
190
        base, ours, theirs = {"f": ("file", "b")}, {"f": ("file", "o")}, {"f": ("exec", "t")}
191
        merged = b"1 ours\n2\n3\n4\n5\n6 theirs\n"
192
        texts = {"f": {"base": b"1\n2\n3\n4\n5\n6\n", "ours": b"1 ours\n2\n3\n4\n5\n6\n",
193
                       "theirs": b"1\n2\n3\n4\n5\n6 theirs\n", "merged": merged}}
194
        bad = cl.classify(gitres({"f": ("exec", "m")}), oak({"f": ("file", "m")}), base, ours, theirs, texts=texts)
195
        self.assertEqual(bad["class"], "oak-wrong")
196
        self.assertTrue(bad["reasons"][0].startswith("lost_mode_change"))
197
        good = cl.classify(gitres({"f": ("exec", "m")}), oak({"f": ("exec", "m")}), base, ours, theirs, texts=texts)
198
        self.assertEqual(good["class"], "equal-to-git")
199
200
    def test_same_content_different_modes_taken_wholesale_is_mode_loss(self):
201
        # both sides rename to the same path with different modes; Oak keeps one side silently
202
        law = cl.no_lost_change({}, {"t": ("file", "x")}, {"t": ("exec", "x")}, {"t": ("exec", "x")}, set())
203
        self.assertEqual(law["mode_lost"], ["t"])
204
205
    def test_symlink_markers_are_distinguished(self):
206
        ev = {"base": b"", "ours": b"a", "theirs": b"b", "merged": b"<<<<<<< x\na=======\nb>>>>>>> main\n"}
207
        law = cl.no_lost_change({}, {"l": ("symlink", "a")}, {"l": ("symlink", "b")}, {"l": ("symlink", "m")},
208
                                set(), {"l": ev})
209
        self.assertEqual(law["markers_symlink"], ["l"])
210
211
    def test_committed_markers_have_their_own_reason(self):
212
        base, ours, theirs = {"f": v("b")}, {"f": v("o")}, {"f": v("t")}
213
        texts = {"f": {"base": b"a\nx", "ours": b"a\ny", "theirs": b"a\nz",
214
                       "merged": b"a\n<<<<<<< br\ny||||||| original\nx=======\nz>>>>>>> main\n"}}
215
        r = cl.classify(gitres({"f": v("g")}, conflicts=["f"]), oak({"f": v("m")}), base, ours, theirs, texts=texts)
216
        self.assertEqual(r["class"], "oak-wrong")
217
        self.assertTrue(r["reasons"][0].startswith("committed_conflict_markers_no_final_newline: f"))
218
219
    def test_failed_pull_dirty_worktree_reason(self):
220
        res = {"error_kind": "oak", "error": "IO error", "clean": None,
221
               "post_failure": {"hazard_worktree_dirty_head_unmoved": True, "worktree_changes_vs_head": 2,
222
                                "worktree_changes_sample": ["a", "b"]}}
223
        r = cl.classify(gitres({}), res, {}, {}, {})
224
        self.assertTrue(r["reasons"][1].startswith("failed_pull_left_worktree_dirty: "))
225
226
    def test_failed_pull_hazard_reason(self):
227
        res = {"error_kind": "oak", "error": "IO error", "clean": None,
228
               "post_failure": {"hazard_head_moved_worktree_stale": True, "head_before": "a" * 64,
229
                                "head_after": "b" * 64, "worktree_changes_vs_head": 2,
230
                                "repull_parent_sync": "up_to_date"}}
231
        r = cl.classify(gitres({}), res, {}, {}, {})
232
        self.assertEqual(r["class"], "oak-wrong")
233
        self.assertTrue(r["reasons"][1].startswith("failed_pull_moved_head_worktree_stale"))
234
235
    def test_unknown_class_is_rejected(self):
236
        self.assertEqual(cl.validate_class("equal-to-git"), "equal-to-git")
237
        with self.assertRaises(ValueError):
238
            cl.validate_class("pass")
239
240
    def test_symmetry(self):
241
        f = oak(self.merged, conflicts=["a"])
242
        r = oak({"a": v("other-side-markers"), "b": v("b1")}, conflicts=["a"])
243
        self.assertTrue(cl.symmetry(f, r)["holds"])
244
        self.assertFalse(cl.symmetry(f, oak(self.merged))["holds"])
245
        self.assertIsNone(cl.symmetry(f, {"clean": None}))
246
247
    def test_idempotence(self):
248
        self.assertIsNone(cl.idempotence_holds({}))
249
        good = {"idempotence": {"returncode": 0, "head_unchanged": True, "tree_unchanged": True}}
250
        self.assertTrue(cl.idempotence_holds(good))
251
        good["idempotence"]["tree_unchanged"] = False
252
        self.assertFalse(cl.idempotence_holds(good))
253
254
    def test_oak_conflict_type(self):
255
        base = {"a": ("file", "1"), "m": ("file", "1")}
256
        self.assertEqual(cl.oak_conflict_type("a", base, {"a": ("file", "2")}, {}), "modify/delete")
257
        self.assertEqual(cl.oak_conflict_type("n", base, {"n": ("file", "2")}, {"n": ("file", "3")}), "add/add")
258
        self.assertEqual(cl.oak_conflict_type("m", base, {"m": ("exec", "1")}, {"m": ("symlink", "1")}), "mode")
259
260
261
class RunnerAccountingTests(unittest.TestCase):
262
    def test_failed_pull_counts_as_prediction_disagreement(self):
263
        from mergesuite.runner import Suite
264
        oak_res = {"error_kind": "oak", "error": "x", "clean": None, "tree": {},
265
                   "prediction": {"available": True, "clean": True, "conflict_files": []},
266
                   "post_failure": {"hazard_head_moved_worktree_stale": True}}
267
        verdict = cl.classify(gitres({}), oak_res, {}, {}, {})
268
        out = Suite._result_fields(None, gitres({}), oak_res, verdict)
269
        self.assertIs(out["laws"]["prediction_agrees"], False)
270
        self.assertTrue(out["post_failure_hazard"])
271
272
    LAWS = {k: None for k in ("no_lost_change", "idempotent", "symmetric", "false_conflict",
273
                              "missed_conflict", "prediction_agrees")}
274
275
    def row(self, scenario, reasons, cls="oak-wrong", direction="A-B"):
276
        from mergesuite.runner import SCHEMA_VERSION
277
        return {"class": cls, "scenario": scenario, "direction": direction, "origin": "o",
278
                "reasons": reasons, "laws": dict(self.LAWS), "merge_suite_schema_version": SCHEMA_VERSION}
279
280
    def test_known_bug_mapping_positive(self):
281
        from mergesuite.runner import load_known_bugs, match_known_bug
282
        bugs = load_known_bugs()
283
        self.assertEqual({b["id"] for b in bugs}, {"fb-537", "fb-538", "fb-539", "fb-540", "fb-541"})
284
        cases = [
285
            (["committed_conflict_markers_symlink: link"], "fb-538"),
286
            (["committed_conflict_markers_no_final_newline: file"], "fb-537"),
287
            (["lost_mode_change: w"], "fb-539"),
288
            (["oak_error: IO error: File exists (os error 17)"], "fb-540"),
289
            (["oak_error: IO error: Not a directory (os error 20)",
290
              "failed_pull_left_worktree_dirty: HEAD unchanged, 2 worktree path(s) differ from HEAD (e.g. a)"], "fb-540"),
291
            (['oak_error: cannot build tree with both file "e" and child path "e/e"'], "fb-540"),
292
            (["oak_error: refusing to replace non-empty directory '/x/a' with a tracked file",
293
              "failed_pull_moved_head_worktree_stale: head a -> b"], "fb-540"),
294
            (["expectation_violated[criss_cross_divergent_resolution]: clean=True expected False"], "fb-541"),
295
        ]
296
        for reasons, bug in cases:
297
            self.assertEqual(match_known_bug(self.row("gix-x", reasons), bugs), bug, reasons)
298
299
    def test_known_bug_escape_routes_stay_unmapped(self):
300
        from mergesuite.runner import load_known_bugs, match_known_bug
301
        bugs = load_known_bugs()
302
        escapes = [
303
            # any other Oak error (e.g. a panic or server 500) is not fb-540
304
            ("gix-tree/a", ["oak_error: Server error: HTTP 500"]),
305
            ("gix-tree/a", ["oak_error: thread 'main' panicked at src/x.rs"]),
306
            # t6416 rows with another reason are not fb-541
307
            ("git-t6416-criss-cross-modify-delete", ["lost_change: file"]),
308
            ("git-t6416-criss-cross-modify-delete", ["expectation_violated: clean=True expected False"]),
309
            # one explained reason plus one unexplained reason stays unmapped
310
            ("gix-text/x", ["committed_conflict_markers_no_final_newline: f", "lost_change: g"]),
311
            ("gix-tree/x", ["oak_error: IO error: File exists (os error 17)", "fabricated_content: q"]),
312
            # markers in files that DO end with a newline are a different root cause
313
            ("gix-text/x", ["committed_conflict_markers: f"]),
314
            ("gix-text/x", []),
315
        ]
316
        for scenario, reasons in escapes:
317
            self.assertIsNone(match_known_bug(self.row(scenario, reasons), bugs), (scenario, reasons))
318
        # only oak-wrong rows are ever mapped
319
        self.assertIsNone(match_known_bug(self.row("gix-x", ["lost_mode_change: w"], cls="differs-but-valid"), bugs))
320
321
    def test_setup_failure_is_not_an_oak_verdict(self):
322
        res = {"error_kind": "oak-setup", "error": "oak clone exited 101: panicked", "clean": None}
323
        r = cl.classify(gitres({}), res, {}, {}, {})
324
        self.assertEqual(r["class"], "harness-error")
325
        self.assertTrue(r["reasons"][0].startswith("setup_failed: "))
326
327
    def test_markers_split_by_final_newline(self):
328
        base, ours, theirs = {"f": v("b")}, {"f": v("o")}, {"f": v("t")}
329
        glued = {"base": b"a\nx", "ours": b"a\ny", "theirs": b"a\nz",
330
                 "merged": b"a\n<<<<<<< br\ny||||||| original\nx=======\nz>>>>>>> main\n"}
331
        law = cl.no_lost_change(base, ours, theirs, {"f": v("m")}, set(), {"f": glued})
332
        self.assertEqual(law["markers_no_final_newline"], ["f"])
333
        clean_eol = {"base": b"a\nx\n", "ours": b"a\ny\n", "theirs": b"a\nz\n",
334
                     "merged": b"a\n<<<<<<< br\ny\n=======\nz\n>>>>>>> main\n"}
335
        law = cl.no_lost_change(base, ours, theirs, {"f": v("m")}, set(), {"f": clean_eol})
336
        self.assertEqual((law["markers"], law["markers_no_final_newline"]), (["f"], []))
337
338
    def test_pinned_rows_report_new_and_vanished(self):
339
        from mergesuite.runner import summarize
340
        bugs = [{"id": "fb-x", "family": "F", "reason_patterns": ["^lost_mode_change: "],
341
                 "expected_rows": ["gix-a|A-B", "gix-b|A-B"]},
342
                {"id": "fb-y", "family": "G", "reason_patterns": ["^nothing"], "expected_rows": ["z|A-B"]}]
343
        rows = [self.row("gix-a", ["lost_mode_change: w"]), self.row("gix-c", ["lost_mode_change: w"]),
344
                self.row("gix-d", ["lost_change: q"]), self.row("ok", [], cls="equal-to-git"),
345
                self.row("gix-b", [], cls="equal-to-git"), self.row("z", [], cls="equal-to-git")]
346
        summary = summarize(rows, bugs)
347
        fx = summary["known_bugs"]["fb-x"]
348
        self.assertEqual((fx["new_rows"], fx["vanished_rows"], fx["status"]),
349
                         (["gix-c|A-B"], ["gix-b|A-B"], "changed"))
350
        self.assertEqual(summary["known_bugs"]["fb-y"]["status"], "NOT reproduced (fixed?)")
351
        self.assertEqual(summary["oak_wrong_unmapped_rows"], ["gix-d|A-B"])
352
        # a subset run (--groups/--only) that never executed a pinned row reports nothing vanished
353
        # a pinned row that became a harness error is indeterminate, never "fixed?"
354
        broken = [self.row("gix-a", ["setup_failed: x"], cls="harness-error"), self.row("gix-b", [], cls="harness-error")]
355
        fx = summarize(broken, bugs)["known_bugs"]["fb-x"]
356
        self.assertEqual((fx["status"], fx["vanished_rows"], fx["indeterminate_rows"]),
357
                         ("harness-error (indeterminate)", [], ["gix-a|A-B", "gix-b|A-B"]))
358
        subset = summarize(rows[:4], bugs)
359
        self.assertEqual(subset["known_bugs"]["fb-x"]["vanished_rows"], [])
360
        self.assertEqual(subset["known_bugs"]["fb-x"]["expected_rows_total"], 2)
361
        self.assertTrue(subset["known_bugs"]["fb-y"]["status"].startswith("not in this run"))
362
363
    def test_stale_schema_rows_are_indeterminate_not_fixed(self):
364
        from mergesuite.runner import stale_schema_rows, summarize
365
        bugs = [{"id": "fb-y", "family": "G", "reason_patterns": ["^nothing"], "expected_rows": ["z|A-B"]}]
366
        rows = [self.row("gix-a", ["lost_mode_change: w"]), self.row("z", [], cls="equal-to-git")]
367
        self.assertEqual(stale_schema_rows(rows), {})
368
        self.assertEqual(summarize(rows, bugs)["known_bugs"]["fb-y"]["status"], "NOT reproduced (fixed?)")
369
        rows[0].pop("merge_suite_schema_version")
370
        rows[1]["merge_suite_schema_version"] = 1
371
        self.assertEqual(stale_schema_rows(rows), {"missing": 1, "1": 1})
372
        summary = summarize(rows, bugs)
373
        self.assertTrue(summary["known_bugs"]["fb-y"]["status"].startswith("indeterminate"))
374
        self.assertEqual(summary["stale_schema_rows"], {"missing": 1, "1": 1})
375
376
    def test_resummarize_refuses_pre_schema_runs(self):
377
        import contextlib
378
        import io
379
        sys.path.insert(0, str(ROOT / "suites" / "merge"))
380
        import run_merge_suite
381
        with tempfile.TemporaryDirectory() as tmp:
382
            run = Path(tmp) / "run.jsonl"
383
            fb541 = self.row("git-t6416-criss-cross-modify-delete",
384
                             ["expectation_violated[criss_cross_divergent_resolution]: clean=True expected False"],
385
                             direction="branch-into-main-sync")
386
            fb541["repro"] = "x"
387
            old = dict(fb541, merge_suite_schema_version=1)
388
            run.write_text(json.dumps(old) + "\n")
389
            err = io.StringIO()
390
            with contextlib.redirect_stderr(err):
391
                self.assertEqual(run_merge_suite.main(["--workdir", tmp, "--resummarize", str(run)]), 2)
392
            self.assertIn("predate merge-suite row/reason schema", err.getvalue())
393
            # a newer-schema row is not called "predates"
394
            run.write_text(json.dumps(dict(fb541, merge_suite_schema_version=99)) + "\n")
395
            err = io.StringIO()
396
            with contextlib.redirect_stderr(err):
397
                self.assertEqual(run_merge_suite.main(["--workdir", tmp, "--resummarize", str(run)]), 2)
398
            self.assertIn("NEWER merge suite", err.getvalue())
399
            self.assertNotIn("predate", err.getvalue())
400
            # an empty run is an error, not an all-clear
401
            run.write_text("\n")
402
            with contextlib.redirect_stderr(io.StringIO()):
403
                self.assertNotEqual(run_merge_suite.main(["--workdir", tmp, "--resummarize", str(run)]), 0)
404
            run.write_text(json.dumps(old) + "\n")
405
            summary_path = Path(tmp) / "s.json"
406
            with contextlib.redirect_stderr(io.StringIO()):
407
                run_merge_suite.main(["--workdir", tmp, "--resummarize", str(run), "--allow-stale-schema",
408
                                      "--summary", str(summary_path)])
409
            summary = json.loads(summary_path.read_text())
410
            self.assertTrue(all(b["status"].startswith("indeterminate") for b in summary["known_bugs"].values()))
411
            self.assertNotIn("fixed?", json.dumps(summary["known_bugs"]))
412
            run.write_text(json.dumps(fb541) + "\n")
413
            with contextlib.redirect_stderr(io.StringIO()):
414
                run_merge_suite.main(["--workdir", tmp, "--resummarize", str(run), "--summary", str(summary_path)])
415
            summary = json.loads(summary_path.read_text())
416
            self.assertEqual(summary["known_bugs"]["fb-541"]["rows"], 1)
417
            self.assertEqual(summary["stale_schema_rows"], {})
418
419
    def test_committed_known_bugs_are_pinned(self):
420
        from mergesuite.runner import load_known_bugs
421
        for bug in load_known_bugs():
422
            self.assertTrue(bug["expected_rows"], bug["id"])
423
            self.assertEqual(len(bug["expected_rows"]), len(set(bug["expected_rows"])), bug["id"])
424
425
426
class OracleParsingTests(unittest.TestCase):
427
    def test_parse_merge_tree_z(self):
428
        out = (b"tree123\0"
429
               b"100644 aaa 1\ta\x00100644 bbb 2\ta\x00100644 ccc 3\ta\x00"
430
               b"\0"
431
               b"1\0a\0Auto-merging\0Auto-merging a\n\0"
432
               b"1\0a\0CONFLICT (contents)\0CONFLICT (content): Merge conflict in a\n\0")
433
        parsed = parse_merge_tree_z(out)
434
        self.assertEqual(parsed["tree_oid"], "tree123")
435
        self.assertEqual(parsed["conflict_stages"], {"a": [1, 2, 3]})
436
        self.assertEqual(git_conflicts(parsed), {"a": ["content"]})
437
438
    def test_parse_clean(self):
439
        parsed = parse_merge_tree_z(b"t\0\0")
440
        self.assertEqual(parsed["conflict_stages"], {})
441
        self.assertEqual(parsed["messages"], [])
442
443
    def test_type_vocabulary(self):
444
        self.assertIsNone(normalize_git_conflict_type("Auto-merging"))
445
        self.assertEqual(normalize_git_conflict_type("CONFLICT (modify/delete)"), "modify/delete")
446
        self.assertEqual(normalize_git_conflict_type("CONFLICT (rename/delete)"), "rename/delete")
447
        self.assertEqual(normalize_git_conflict_type("CONFLICT (directory rename split)"), "directory-rename")
448
449
    def test_conflict_show_and_prediction_parsing(self):
450
        doc = {"conflicts": [{"path": "f", "has_conflict_markers": False, "exists": True}]}
451
        self.assertEqual(parse_conflict_show(doc, {}), {"f": ["no-markers"]})
452
        self.assertEqual(parse_conflict_show(None, {"conflict_paths": ["x"]}), {"x": []})
453
        self.assertEqual(summarize_prediction(None, 6), {"available": False, "returncode": 6})
454
        pred = summarize_prediction({"prediction_available": True, "clean": False,
455
                                     "conflict_files": [{"path": "b"}, "a"]}, 0)
456
        self.assertEqual(pred["conflict_files"], ["a", "b"])
457
458
459
@unittest.skipUnless(shutil.which("git"), "git not installed")
460
class GitBackedTests(unittest.TestCase):
461
    def setUp(self):
462
        self.tmp = Path(tempfile.mkdtemp(prefix="mergesuite-test-"))
463
        self.home = self.tmp / "home"
464
        from oakbench.subjects import which_measured_git
465
        self.git_bin = which_measured_git() or shutil.which("git")
466
        prepare_git_home(self.home, self.git_bin)
467
468
    def tearDown(self):
469
        shutil.rmtree(self.tmp, ignore_errors=True)
470
471
    def test_builder_oracle_and_materialize_roundtrip(self):
472
        from mergesuite.builder import RepoBuilder, lines
473
        b = RepoBuilder(self.tmp / "r", self.git_bin, self.home)
474
        base = b.commit({"f": lines("1", "2", "3"), "x": ("exec", b"#!/bin/sh\n"), "l": ("link", "f")})
475
        a = b.commit({"f": lines("1 a", "2", "3"), "x": ("exec", b"#!/bin/sh\n"), "l": ("link", "f")}, [base])
476
        c = b.commit({"f": lines("1", "2", "3 c"), "l": ("link", "f")}, [base], message="unique")
477
        self.assertEqual(b.merge_bases(a, c), [base])
478
        merge = b.merge_tree(a, c)
479
        self.assertTrue(merge["clean"])
480
        self.assertEqual(sorted(merge["tree"]), ["f", "l"])
481
        self.assertEqual(merge["tree"]["l"][0], "symlink")
482
        out = self.tmp / "mat"
483
        out.mkdir()
484
        b.materialize(base, out)
485
        self.assertEqual(worktree_manifest(out), b.tree(base))
486
        # first parent order is preserved
487
        m = b.commit({"f": lines("z")}, [a, c])
488
        self.assertEqual(b.out("rev-parse", m + "^1"), a)
489
490
    def test_oracle_reports_conflict_types(self):
491
        from mergesuite.builder import RepoBuilder, lines
492
        b = RepoBuilder(self.tmp / "r2", self.git_bin, self.home)
493
        base = b.commit({"f": lines("1", "2"), "g": lines("g")})
494
        a = b.commit({"f": lines("1", "A")}, [base])
495
        c = b.commit({"f": lines("1", "C"), "g": lines("g2")}, [base])
496
        merge = b.merge_tree(a, c)
497
        self.assertFalse(merge["clean"])
498
        self.assertEqual(git_conflicts(merge), {"f": ["content"], "g": ["modify/delete"]})
499
500
    def test_sources_build_without_oak(self):
501
        from mergesuite import sources
502
        scen = sources.git_t64xx(self.tmp / "t64", self.git_bin, self.home)
503
        ids = [s["id"] for s in scen]
504
        self.assertIn("git-t6423-1a", ids)
505
        self.assertTrue(all(len(s["directions"]) == 2 and s["unsupported"] is None for s in scen))
506
        native = sources.native_incidents(self.tmp / "nat", self.git_bin, self.home)
507
        by_id = {s["id"]: s for s in native}
508
        cc = by_id["incident/criss-cross-tie-break-side-newer"]["directions"][0]
509
        self.assertEqual(len(cc["merge_bases"]), 2)  # two best common ancestors by construction
510
        self.assertIsNone(cc["trees"]["base"])
511
        skew = by_id["incident/clock-skew-merge-base-skewed"]["directions"][0]
512
        self.assertEqual(len(skew["merge_bases"]), 1)
513
        # self-merges produce a single direction (no duplicate rows sharing a case dir)
514
        rev = by_id["incident/revert-after-client-sync"]
515
        tip = rev["directions"][0]["oracle"]["theirs"]
516
        one = sources.symmetric("self", "o", {}, Git(rev["repo"], self.git_bin, self.home), tip, tip)
517
        self.assertEqual(len(one["directions"]), 1)
518
        gen = sources.generated(self.tmp / "gen", self.git_bin, self.home, 3, 7)
519
        gen2 = sources.generated(self.tmp / "gen2", self.git_bin, self.home, 3, 7)
520
        self.assertEqual([s["notes"] for s in gen], [s["notes"] for s in gen2])  # seeded
521
522
523
class VendoringTests(unittest.TestCase):
524
    def test_vendored_scripts_are_byte_identical_to_pinned_upstream(self):
525
        meta = json.loads((VENDOR / "UPSTREAM.json").read_text())
526
        self.assertEqual(meta["license"], "MIT OR Apache-2.0")
527
        for name, info in meta["files"].items():
528
            path = VENDOR / name
529
            self.assertTrue(path.is_file(), name)
530
            if "sha256" in info:
531
                self.assertEqual(hashlib.sha256(path.read_bytes()).hexdigest(), info["sha256"], name)
532
        notice = (VENDOR / "NOTICE").read_text()
533
        self.assertIn(meta["commit"], notice)
534
        self.assertIn("Permission is hereby granted", (VENDOR / "LICENSE-MIT").read_text())
535
        self.assertIn("Apache License", (VENDOR / "LICENSE-APACHE").read_text())
536
537
    def test_no_git_test_suite_text_is_vendored(self):
538
        merge_root = ROOT / "suites" / "merge"
539
        for path in merge_root.rglob("*"):
540
            if path.is_file() and path.suffix in (".py", ".sh", ".md", ".json"):
541
                text = path.read_text(errors="replace")
542
                self.assertNotIn("test_expect_success", text, path)
543
                self.assertNotIn(". ./test-lib.sh", text, path)
544
545
546
if __name__ == "__main__":
547
    unittest.main()