Spaces:
Sleeping
Sleeping
File size: 7,785 Bytes
7a658b7 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 | """RealBugGenerator — scenario family: real_bug."""
from __future__ import annotations
import random
from ci_triage_env.data.clustering.archetypes import Archetype
from ci_triage_env.data.generators._helpers import (
ArchetypedGenerator,
build_base_outputs,
fake_short_sha,
fake_timestamp,
fill_template,
make_failure_summary,
pick_test_name,
scenario_id_for,
)
from ci_triage_env.schemas.diagnosis import DiagnosisLabel
from ci_triage_env.schemas.scenario import (
GroundTruth,
Scenario,
ScenarioMetadata,
TerminalActionSpec,
ToolOutput,
)
_DEFAULT_LOG_TEMPLATE = (
"FAILED {TEST_MODULE}::{TEST_FUNC} - AssertionError\n"
" assert result == expected\n"
" where result = {ACTUAL}\n"
" and expected = {EXPECTED}\n"
"E AssertionError: assertion failed at line {LINENO}\n"
"short test summary info\n"
"FAILED {TEST_MODULE}::{TEST_FUNC}"
)
_DEFAULT_BUGGY_CODE = (
"def {TEST_FUNC}(self):\n"
" result = self.service.compute({INPUT})\n"
" assert result == {EXPECTED} # broke after {COMMIT_MSG}\n"
)
class RealBugGenerator(ArchetypedGenerator):
family_name = "real_bug"
label = DiagnosisLabel.REAL_BUG
def informative_tools(self) -> list[str]:
return ["read_logs", "inspect_test_code", "recent_commits", "rerun_test"]
def minimal_evidence_set(self) -> list[str]:
return ["recent_commits", "inspect_test_code"]
def _default_archetypes(self) -> list[Archetype]:
return [
Archetype(
archetype_id="real_bug_001",
family="real_bug",
pattern_summary="AssertionError after recent commit changed return value",
log_template=_DEFAULT_LOG_TEMPLATE,
slot_distributions={
"TEST_MODULE": ["tests/unit/test_core", "tests/unit/test_api"],
"TEST_FUNC": ["test_compute", "test_process", "test_validate"],
"ACTUAL": ["None", "0", "-1", "[]"],
"EXPECTED": ["42", "True", "{'ok': True}"],
"LINENO": ["42", "87", "115", "203"],
},
informative_tools_hint=["read_logs:full", "inspect_test_code", "recent_commits"],
minimal_evidence_hint=["recent_commits", "inspect_test_code"],
),
Archetype(
archetype_id="real_bug_002",
family="real_bug",
pattern_summary="AttributeError / NullPointerException in core logic",
log_template=(
"AttributeError: 'NoneType' object has no attribute '{ATTR}'\n"
" File \"{TEST_MODULE}.py\", line {LINENO}, in {TEST_FUNC}\n"
" return obj.{ATTR}\n"
"FAILED {TEST_MODULE}::{TEST_FUNC}"
),
slot_distributions={
"ATTR": ["name", "id", "value", "data", "result"],
"TEST_MODULE": ["tests/unit/test_models", "tests/unit/test_service"],
"TEST_FUNC": ["test_create", "test_update", "test_fetch"],
"LINENO": ["33", "67", "91", "144"],
},
informative_tools_hint=["read_logs:full", "inspect_test_code", "recent_commits"],
minimal_evidence_hint=["inspect_test_code"],
),
]
def generate(self, seed: int, source_log_hash: str | None = None) -> Scenario:
rng = random.Random(seed)
archetype = self._pick_archetype(rng)
log_text = fill_template(archetype.log_template, archetype.slot_distributions, rng)
test_name = pick_test_name(rng)
summary = make_failure_summary(
self.family_name, rng, test_name=test_name, log_excerpt=log_text
)
branch = summary.branch
outputs = build_base_outputs(
test_name, branch, rng,
log_lines=log_text.splitlines(),
rerun_passes=False,
)
# --- informative overrides ---
breaking_author = rng.choice(["@alice", "@bob", "@carol"])
breaking_sha = fake_short_sha(rng)
breaking_commit = {
"sha": breaking_sha,
"author": breaking_author,
"msg": rng.choice([
f"fix: update {test_name.split('::')[-1].replace('test_', '')} logic",
"refactor: change return contract of compute()",
f"feat: extend {test_name.split('::')[-1].split('_')[1]} API",
]),
"files": [
f"src/{test_name.split('/')[1].replace('test_', '')}.py",
test_name.rsplit("::", 1)[0],
],
}
outputs[f"recent_commits:{branch}"] = ToolOutput(
tool_name="recent_commits",
payload={"commits": [breaking_commit, {
"sha": fake_short_sha(rng),
"author": rng.choice(["@dave", "@eve"]),
"msg": "chore: update lockfile",
"files": ["pyproject.toml"],
}]},
cost_units=0.002,
)
buggy_code = self._pick_buggy_code(rng)
outputs[f"inspect_test_code:{test_name}"] = ToolOutput(
tool_name="inspect_test_code",
payload={"source": buggy_code, "fixtures": []},
cost_units=0.002,
)
# Rerun also fails — it's a real bug, not a flake
outputs["rerun_test"] = ToolOutput(
tool_name="rerun_test",
payload={"results": [{"passed": False, "duration_s": round(rng.uniform(5, 30), 2),
"log_excerpt": log_text.splitlines()[:3]}]},
cost_units=0.01,
)
# Flake history is clean (test was stable before the bad commit)
outputs[f"query_flake_history:{test_name}"] = ToolOutput(
tool_name="query_flake_history",
payload={"failure_count": 0, "pass_count": 50, "recent_failures": []},
cost_units=0.002,
)
difficulty = rng.choice(["easy", "medium", "hard"])
rationale = (
f"The commit {breaking_sha} by {breaking_author} changed the return contract "
f"of the production code exercised by {test_name}. "
f"inspect_test_code shows the assertion that now fails; "
f"recent_commits:{branch} shows the introducing commit. "
f"query_flake_history shows no prior failures — not a flake. "
f"rerun_test fails again — confirms deterministic breakage."
)
return Scenario(
schema_version="1.0",
scenario_id=scenario_id_for(self.family_name, seed),
family=self.family_name,
seed=seed,
ground_truth=GroundTruth(
label=self.label,
rationale=rationale,
is_ambiguous=False,
confidence_target=1.0,
),
failure_summary=summary,
tool_outputs=outputs,
informative_tools=self.informative_tools(),
minimal_evidence_set=self.minimal_evidence_set(),
correct_terminal_action=TerminalActionSpec(
primary="submit_diagnosis",
args={
"diagnosis": self.label.value,
"confidence": 1.0,
"secondary_actions": [{"name": "file_bug", "owner": breaking_author}],
},
acceptable_alternatives=[],
),
metadata=ScenarioMetadata(
generator_version="1.0",
generated_at=fake_timestamp(rng),
source_log_hash=source_log_hash,
difficulty=difficulty,
),
)
|