Files
dotfiles/python/eval_warnings/main.py

441 lines
13 KiB
Python

"""Detect Nix evaluation warnings and create PRs with LLM-suggested fixes."""
from __future__ import annotations
import hashlib
import logging
import re
import subprocess
from dataclasses import dataclass
from io import BytesIO
from pathlib import Path
from typing import Annotated
from zipfile import ZipFile
import typer
from httpx import HTTPError, post
from python.common import configure_logger
logger = logging.getLogger(__name__)
@dataclass(frozen=True)
class EvalWarning:
"""A single Nix evaluation warning."""
system: str
message: str
@dataclass
class FileChange:
"""A file change suggested by the LLM."""
file_path: str
original: str
fixed: str
def run_cmd(cmd: list[str], *, check: bool = True) -> subprocess.CompletedProcess[str]:
"""Run a subprocess command and return the result.
Args:
cmd: Command and arguments.
check: Whether to raise on non-zero exit.
Returns:
CompletedProcess with captured stdout/stderr.
"""
logger.debug("Running: %s", " ".join(cmd))
return subprocess.run(cmd, capture_output=True, text=True, check=check)
def download_logs(run_id: str, repo: str) -> dict[str, str]:
"""Download build logs for a GitHub Actions run.
Args:
run_id: The workflow run ID.
repo: The GitHub repository (owner/repo).
Returns:
Dict mapping zip entry names to their text content, filtered to build log files.
Raises:
RuntimeError: If log download fails.
"""
result = subprocess.run(
["gh", "api", f"repos/{repo}/actions/runs/{run_id}/logs"],
capture_output=True,
check=False,
)
if result.returncode != 0:
msg = f"Failed to download logs: {result.stderr.decode(errors='replace')}"
raise RuntimeError(msg)
logs: dict[str, str] = {}
with ZipFile(BytesIO(result.stdout)) as zip_file:
for name in zip_file.namelist():
if name.startswith("build-") and name.endswith(".txt"):
logs[name] = zip_file.read(name).decode(errors="replace")
return logs
def parse_warnings(logs: dict[str, str]) -> set[EvalWarning]:
"""Parse Nix evaluation warnings from build log contents.
Args:
logs: Dict mapping zip entry names (e.g. "build-bob/2_Build.txt") to their text.
Returns:
Deduplicated set of warnings.
"""
warnings: set[EvalWarning] = set()
warning_pattern = re.compile(r"(?:^[\d\-T:.Z]+ )?(warning:|trace: warning:)")
timestamp_prefix = re.compile(r"^[\d\-T:.Z]+ ")
for name, content in sorted(logs.items()):
system = name.split("/")[0].removeprefix("build-")
for line in content.splitlines():
if line.startswith("warning: ignoring untrusted flake configuration setting"):
logger.info("Ignoring untrusted flake configuration setting warning.")
continue
if warning_pattern.search(line):
logger.debug(f"Found warning: {line}")
message = timestamp_prefix.sub("", line).strip()
warnings.add(EvalWarning(system=system, message=message))
logger.info("Found %d unique warnings", len(warnings))
return warnings
def extract_referenced_files(warnings: set[EvalWarning]) -> dict[str, str]:
"""Extract file paths referenced in warnings and read their contents.
Args:
warnings: List of parsed warnings.
Returns:
Dict mapping repo-relative file paths to their contents.
"""
paths: set[str] = set()
warning_text = "\n".join(w.message for w in warnings)
nix_store_path = re.compile(r"/nix/store/[^/]+-source/([^:]+\.nix)")
for match in nix_store_path.finditer(warning_text):
paths.add(match.group(1))
repo_relative_path = re.compile(r"(?<![/\w])(systems|common|users|overlays)/[^:\s]+\.nix")
for match in repo_relative_path.finditer(warning_text):
paths.add(match.group(0))
files: dict[str, str] = {}
for path_str in sorted(paths):
path = Path(path_str)
if path.is_file():
files[path_str] = path.read_text()
if not files and Path("flake.nix").is_file():
files["flake.nix"] = Path("flake.nix").read_text()
logger.info("Extracted %d referenced files", len(files))
return files
def compute_warning_hash(warnings: set[EvalWarning]) -> str:
"""Compute a short hash of the warning set for deduplication.
Args:
warnings: List of warnings.
Returns:
8-character hex hash.
"""
text = "\n".join(sorted(f"[{w.system}] {w.message}" for w in warnings))
return hashlib.sha256(text.encode()).hexdigest()[:8]
def check_duplicate_pr(warning_hash: str) -> bool:
"""Check if an open PR already exists for this warning hash.
Args:
warning_hash: The hash to check.
Returns:
True if a duplicate PR exists.
Raises:
RuntimeError: If the gh CLI call fails.
"""
result = run_cmd(
[
"gh",
"pr",
"list",
"--state",
"open",
"--label",
"eval-warning-fix",
"--json",
"title",
"--jq",
".[].title",
],
check=False,
)
if result.returncode != 0:
msg = f"Failed to check for duplicate PRs: {result.stderr}"
raise RuntimeError(msg)
for title in result.stdout.splitlines():
if warning_hash in title:
logger.info("Duplicate PR found for hash %s", warning_hash)
return True
return False
def query_ollama(
warnings: set[EvalWarning],
files: dict[str, str],
ollama_url: str,
) -> str | None:
"""Query Ollama for a fix suggestion.
Args:
warnings: List of warnings.
files: Referenced file contents.
ollama_url: Ollama API base URL.
Returns:
LLM response text, or None on failure.
"""
warning_text = "\n".join(f"[{w.system}] {w.message}" for w in warnings)
file_context = "\n".join(f"--- FILE: {path} ---\n{content}\n--- END FILE ---" for path, content in files.items())
prompt = f"""You are a NixOS configuration expert. \
Analyze the following Nix evaluation warnings and suggest fixes.
## Warnings
{warning_text}
## Referenced Files
{file_context}
## Instructions
- Identify the root cause of each warning
- Provide the exact file changes needed to fix the warnings
- Output your response in two clearly separated sections:
1. **REASONING**: Brief explanation of what causes each warning and how to fix it
2. **CHANGES**: For each file that needs changes, output a block like:
FILE: path/to/file.nix
<<<<<<< ORIGINAL
the original lines to replace
=======
the replacement lines
>>>>>>> FIXED
- Only suggest changes for files that exist in the repository
- Do not add unnecessary complexity
- Preserve the existing code style
- If a warning comes from upstream nixpkgs and cannot be fixed in this repo, \
say so in REASONING and do not suggest changes"""
try:
response = post(
f"{ollama_url}/api/generate",
json={
"model": "qwen3-coder:30b",
"prompt": prompt,
"stream": False,
"options": {"num_predict": 4096},
},
timeout=300,
)
response.raise_for_status()
except HTTPError:
logger.exception("Ollama request failed")
return None
return response.json().get("response")
def parse_changes(response: str) -> list[FileChange]:
"""Parse file changes from the LLM response.
Expects blocks in the format:
FILE: path/to/file.nix
<<<<<<< ORIGINAL
...
=======
...
>>>>>>> FIXED
Args:
response: Raw LLM response text.
Returns:
List of parsed file changes.
"""
changes: list[FileChange] = []
current_file = ""
section: str | None = None
original_lines: list[str] = []
fixed_lines: list[str] = []
for line in response.splitlines():
stripped = line.strip()
if stripped.startswith("FILE:"):
current_file = stripped.removeprefix("FILE:").strip()
elif stripped == "<<<<<<< ORIGINAL":
section = "original"
original_lines = []
elif stripped == "=======" and section == "original":
section = "fixed"
fixed_lines = []
elif stripped == ">>>>>>> FIXED" and section == "fixed":
section = None
if current_file:
changes.append(FileChange(current_file, "\n".join(original_lines), "\n".join(fixed_lines)))
elif section == "original":
original_lines.append(line)
elif section == "fixed":
fixed_lines.append(line)
logger.info("Parsed %d file changes", len(changes))
return changes
def apply_changes(changes: list[FileChange]) -> int:
"""Apply file changes to the working directory.
Args:
changes: List of changes to apply.
Returns:
Number of changes successfully applied.
"""
applied = 0
for change in changes:
path = Path(change.file_path)
if not path.is_file():
logger.warning("File not found: %s", change.file_path)
continue
content = path.read_text()
if change.original not in content:
logger.warning("Original text not found in %s", change.file_path)
continue
path.write_text(content.replace(change.original, change.fixed, 1))
logger.info("Applied fix to %s", change.file_path)
applied += 1
return applied
def create_pr(
warning_hash: str,
warnings: set[EvalWarning],
llm_response: str,
run_url: str,
) -> None:
"""Create a git branch and PR with the applied fixes.
Args:
warning_hash: Short hash for branch naming and deduplication.
warnings: Original warnings for the PR body.
llm_response: Full LLM response for extracting reasoning.
run_url: URL to the triggering build run.
"""
branch = f"fix/eval-warning-{warning_hash}"
warning_text = "\n".join(f"[{w.system}] {w.message}" for w in warnings)
if "**REASONING**" not in llm_response:
logger.warning("LLM response missing **REASONING** section")
reasoning = ""
else:
_, after = llm_response.split("**REASONING**", 1)
reasoning = "\n".join(after.split("**CHANGES**", 1)[0].strip().splitlines()[:50])
run_cmd(["git", "config", "user.name", "github-actions[bot]"])
run_cmd(["git", "config", "user.email", "github-actions[bot]@users.noreply.github.com"])
run_cmd(["git", "checkout", "-b", branch])
run_cmd(["git", "add", "-A"])
diff_result = run_cmd(["git", "diff", "--cached", "--quiet"], check=False)
if diff_result.returncode == 0:
logger.info("No file changes to commit")
return
run_cmd(["git", "commit", "-m", f"fix: resolve nix evaluation warnings ({warning_hash})"])
run_cmd(["git", "push", "origin", branch, "--force"])
body = f"""## Nix Evaluation Warnings
Detected in [build_systems run]({run_url}):
```
{warning_text}
```
## LLM Analysis (qwen3-coder:30b)
{reasoning}
---
*Auto-generated by fix_eval_warnings. Review carefully before merging.*"""
run_cmd(
[
"gh",
"pr",
"create",
"--title",
f"fix: resolve nix eval warnings ({warning_hash})",
"--label",
"automated",
"--label",
"eval-warning-fix",
"--body",
body,
]
)
logger.info("PR created on branch %s", branch)
def main(
run_id: Annotated[str, typer.Option("--run-id", help="GitHub Actions run ID")],
repo: Annotated[str, typer.Option("--repo", help="GitHub repository (owner/repo)")],
ollama_url: Annotated[str, typer.Option("--ollama-url", help="Ollama API base URL")],
run_url: Annotated[str, typer.Option("--run-url", help="URL to the triggering build run")],
log_level: Annotated[str, typer.Option("--log-level", "-l", help="Log level")] = "INFO",
) -> None:
"""Detect Nix evaluation warnings and create PRs with LLM-suggested fixes."""
configure_logger(log_level)
logs = download_logs(run_id, repo)
warnings = parse_warnings(logs)
if not warnings:
return
warning_hash = compute_warning_hash(warnings)
if check_duplicate_pr(warning_hash):
return
files = extract_referenced_files(warnings)
llm_response = query_ollama(warnings, files, ollama_url)
if not llm_response:
return
changes = parse_changes(llm_response)
applied = apply_changes(changes)
if applied == 0:
logger.info("No changes could be applied")
return
create_pr(warning_hash, warnings, llm_response, run_url)
if __name__ == "__main__":
typer.run(main)