File size: 5,076 Bytes
e1f5446 | 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 | """Run the native commit_runtime pipeline over one corpus repository, locally.
This is the launch path the proposal cites (Repo2RLEnv cli.py line 355):
pipeline = pipeline_cls(gen_input, options, bootstrap=bootstrap_result)
with the prebaked BootstrapResult and the LocalSandbox adapter installed.
Usage:
python -m r2e_local.driver --repo owner/name --sha <pinned> \
--mirror assets/repos/owner__name --sandbox assets/sandboxes/owner__name \
--bootstrap assets/bootstrap/owner__name/bootstrap.json \
--out assets/heldout/owner__name \
[--llm gemini/gemini-2.5-flash --llm-key-env GEMINI_API_KEY] [--opt key=value ...]
"""
from __future__ import annotations
import argparse
import json
import logging
import sys
import time
from pathlib import Path
from repo2rlenv.spec.input import (
AuthSpec,
BootstrapSpec,
GenerationInput,
LLMSpec,
OutputSpec,
PipelineName,
PipelineSpec,
RepoSpec,
)
from repo2rlenv.spec.options import CommitRuntimeOptions
from .pipeline import LocalCommitRuntimePipeline, install_hooks, load_bootstrap
def _parse_opts(items: list[str]) -> dict:
out: dict = {}
for item in items:
k, _, v = item.partition("=")
if v.lower() in ("true", "false"):
out[k] = v.lower() == "true"
elif v.isdigit():
out[k] = int(v)
else:
try:
out[k] = float(v)
except ValueError:
out[k] = v
return out
def main(argv: list[str] | None = None) -> int:
ap = argparse.ArgumentParser()
ap.add_argument("--repo", required=True, help="owner/name")
ap.add_argument("--sha", required=True, help="pinned commit")
ap.add_argument("--mirror", required=True, type=Path)
ap.add_argument("--sandbox", required=True, type=Path)
ap.add_argument("--bootstrap", required=True, type=Path)
ap.add_argument("--out", required=True, type=Path)
ap.add_argument("--llm", default=None, help="provider/model; omit to disable synthesis")
ap.add_argument("--llm-key-env", default=None)
ap.add_argument("--llm-endpoint", default=None)
ap.add_argument("--opt", action="append", default=[], help="CommitRuntimeOptions field=value")
ap.add_argument("-v", "--verbose", action="store_true")
args = ap.parse_args(argv)
logging.basicConfig(
level=logging.DEBUG if args.verbose else logging.INFO,
format="%(asctime)s %(levelname)s %(name)s: %(message)s",
stream=sys.stderr,
)
github_url = f"https://github.com/{args.repo}"
install_hooks({github_url: args.mirror.resolve()})
llm = None
if args.llm:
provider, _, model = args.llm.partition("/")
llm = LLMSpec(
provider=provider,
model=model,
api_key_env=args.llm_key_env,
endpoint=args.llm_endpoint,
)
opts = CommitRuntimeOptions(**_parse_opts(args.opt))
if args.llm is None and opts.synthesize_with_llm:
# Shipped default is synthesize_with_llm=True; with no LLM the pipeline
# falls back to the commit message (commit_runtime.py _build_instruction).
logging.getLogger(__name__).warning("no --llm given: instructions use commit text")
gen_input = GenerationInput(
repo=RepoSpec(url=github_url, ref=args.sha, access="public"),
pipeline=PipelineSpec(name=PipelineName.COMMIT_RUNTIME, recipe="native"),
llm=llm,
output=OutputSpec(destination="local", org="commit-runtime", dataset_name="tasks"),
bootstrap=BootstrapSpec(enabled=False),
auth=AuthSpec(github_token_env="R2E_UNUSED_TOKEN", use_gh_cli=False, use_hf_cli=False),
)
bootstrap = load_bootstrap(args.bootstrap)
pipeline = LocalCommitRuntimePipeline(
gen_input, opts, bootstrap=bootstrap, sandbox_dir=args.sandbox.resolve()
)
events: list[dict] = []
pipeline.set_progress_callback(
lambda name, outcome, reason="": events.append(
{"t": time.time(), "name": name, "outcome": outcome, "reason": reason}
)
)
t0 = time.time()
result = pipeline.run(args.out.resolve())
elapsed = time.time() - t0
summary = {
"repo": args.repo,
"sha": args.sha,
"candidates": result.candidates,
"emitted": result.emitted,
"skipped": result.skipped,
"skip_reasons": result.skip_reasons,
"options": opts.model_dump() if hasattr(opts, "model_dump") else vars(opts),
"llm": llm.qualified_name if llm else None,
"llm_cost_usd": round(getattr(pipeline, "_llm_cost_usd", 0.0), 6),
"elapsed_sec": round(elapsed, 1),
"events": events,
}
args.out.mkdir(parents=True, exist_ok=True)
(args.out / "result.json").write_text(json.dumps(summary, indent=2))
print(
json.dumps(
{k: summary[k] for k in ("repo", "candidates", "emitted", "skipped", "skip_reasons", "elapsed_sec", "llm_cost_usd")}
)
)
return 0
if __name__ == "__main__":
raise SystemExit(main())
|