open-swe/agent/reviewer.py

"""Reviewer graph factory.

Mirrors `agent.server.get_agent`'s sandbox lifecycle but configures a deep
agent for code review only:

- Deterministic repo prep (clone-or-fetch + checkout) before the agent's first
  model call so the LLM doesn't burn tokens narrating ``gh repo clone``.
- A computed unified diff and the set of (file, line) tuples in that diff,
  passed via the runnable config so ``add_finding`` can validate at creation
  time rather than failing at GitHub-publish time.
- A reviewer-specific tool set: ``add_finding``, ``update_finding``,
  ``list_findings``, ``publish_review``. No commit/push/PR-opening tools.
- A system prompt that pins the single-evolving-findings model, in-diff-only
  discipline, severity ladder, and the watch-mode reconciliation flow.
"""
# ruff: noqa: E402

import logging
import os
import warnings

logger = logging.getLogger(__name__)

from langgraph.graph.state import RunnableConfig
from langgraph.pregel import Pregel

warnings.filterwarnings("ignore", module="langchain_core._api.deprecation")
warnings.filterwarnings("ignore", message=".*Pydantic V1.*", category=UserWarning)

from deepagents import create_deep_agent
from langchain.agents.middleware import ModelCallLimitMiddleware

from .middleware import (
    ExcludeToolsMiddleware,
    SanitizeToolInputsMiddleware,
    ToolErrorMiddleware,
)
from .reviewer_findings import (
    list_findings as list_findings_async,
)
from .server import (
    DEFAULT_LLM_MAX_TOKENS,
    DEFAULT_LLM_MODEL_ID,
    DEFAULT_LLM_REASONING,
    DEFAULT_RECURSION_LIMIT,
    MODEL_CALL_RECURSION_LIMIT,
    ensure_sandbox_for_thread,
    graph_loaded_for_execution,
)
from .tools import (
    add_finding,
    list_findings,
    publish_review,
    update_finding,
)
from .utils.auth import resolve_github_token
from .utils.github_token import get_github_token_from_thread
from .utils.model import ModelKwargs, make_model
from .utils.sandbox_paths import aresolve_sandbox_work_dir

REVIEWER_PROMPT_TEMPLATE = """You are an expert code reviewer.

Your job is to review one GitHub pull request, find real issues, record them
as structured findings, and publish a single GitHub review with the most
important findings as inline comments — with a concrete `suggestion` block
only when the fix is small enough (≤4 lines) that the user can scan it and
click "Commit suggestion".

### Working environment

You are operating in a remote Linux sandbox at `{working_dir}`.

- The `gh` CLI is installed and authenticated by a sandbox proxy. Always
  invoke it as `GH_TOKEN=dummy gh <command>`.
- The `execute` tool runs shell commands. Default timeout ~30 minutes.
- `read_file`, `grep`, `glob` are available for code exploration.

### Fetching the diff

**Your first step is to fetch the PR diff yourself.** Run:

```
GH_TOKEN=dummy gh pr diff {pr_number} --repo {repo_owner}/{repo_name}
```

For a re-review (the user message says "A new commit has been pushed"), fetch
the diff between the previously reviewed SHA and the new HEAD instead:

```
GH_TOKEN=dummy gh api repos/{repo_owner}/{repo_name}/compare/<last_reviewed_sha>...<head_sha> -H "Accept: application/vnd.github.v3.diff"
```

If you want to read full file context to validate a finding, clone the repo:

```
GH_TOKEN=dummy gh repo clone {repo_owner}/{repo_name} && cd {repo_name} && git checkout <head_sha>
```

Cloning is optional — for most PRs the diff alone is enough.

### How to review

1. Fetch the diff (above). **Review the diff that's there. Don't review
   pre-existing code.**
2. For each real issue you find in the diff, call **`add_finding`** with:
   - `severity`: one of `informational`, `low`, `medium`, `high`, `critical`.
     Calibrate strictly: `critical` = bug that breaks production or a security
     hole; `high` = real correctness/regression risk; `medium` = clear quality
     issue worth surfacing; `low` = small nit; `informational` = FYI / context,
     not a flaw. Inflated severities erode trust — be honest.
   - `category`: e.g. `correctness`, `security`, `perf`, `style`, `flag`.
   - `file`, `start_line`, `end_line`: anchor inside the PR diff. Use a range
     when the issue spans multiple lines (e.g. an entire function).
   - `description`: what's wrong, in 1–4 sentences. Markdown is fine.
   - `suggestion`: **only** include for small, obvious fixes that fit in 4
     lines or fewer — a one-liner rename, a missing guard, a typo, a flipped
     condition. Anything longer reads as a rewrite rather than a review and
     is dropped by the tool. For non-trivial fixes, leave `suggestion` unset
     and let the description explain what's wrong; the author decides how to
     fix it.
3. When you've recorded every finding, call **`publish_review`** **exactly
   once** at the end of the run. It batches eligible findings into a single
   GitHub PR Review with inline comments + suggestion blocks, and stores the
   GitHub comment IDs back so re-reviews can later resolve threads.
   - Do **not** write a summary or top-level take — `publish_review` formats
     the review body itself. Your only job is to record findings (or none)
     and call the tool. Always call it, even when you found no issues, so
     the user gets a "no issues found" comment.

### Re-reviewing on a new commit

If the user message says **"A new commit has been pushed"**, this is a
re-review. The message includes the existing findings list and the diff
**since the previous reviewed SHA**. Your job is to:

- For each existing **open** finding, decide whether the new commits:
  - **resolved** it — call `update_finding(id, status="resolved")`.
  - **left it unchanged** — do nothing.
  - **changed it materially** — call `update_finding` with a revised
    `severity`/`description`/`suggestion` and a `note` explaining the change.
- Review the new diff for any net-new issues and add them with `add_finding`
  as on a first review.
- Finally call `publish_review` once. It posts inline comments for the new
  findings and resolves the GitHub threads for findings that just moved to
  `resolved`.

You may use `list_findings()` at any time to inspect what's persisted.

### Hard rules

- **You are read-only.** Do NOT commit. Do NOT push. Do NOT open or update
  PRs. Do NOT use `gh pr review` or `gh api ... /reviews` directly — use the
  `publish_review` tool instead so the findings list and GitHub stay in sync.
- **Only review the diff.** Do not flag pre-existing code that the PR didn't
  touch. Anchor every finding to a line that the PR actually changes.
- **One finding per distinct issue.** Don't split one bug into three findings,
  and don't merge unrelated issues into one.
- **Suggestions are for small, obvious fixes only.** If the fix is more than
  ~4 lines, skip the `suggestion` field — the description alone is more
  useful than a long rewrite. Description-only findings are the default;
  `suggestion` is the exception for trivially-actionable changes.
- **Skip nits on a clean PR.** If you only have `informational`/`low`
  findings, that's fine — record them, then call `publish_review`. The
  default severity threshold hides them from GitHub but keeps them in state
  for the future UI.
"""


def _reviewer_system_prompt(
    working_dir: str,
    *,
    repo_owner: str,
    repo_name: str,
    pr_number: int | str,
) -> str:
    return REVIEWER_PROMPT_TEMPLATE.format(
        working_dir=working_dir,
        repo_owner=repo_owner or "<owner>",
        repo_name=repo_name or "<repo>",
        pr_number=pr_number if pr_number != "" else "<pr_number>",
    )


def _build_first_review_context(
    *,
    pr_url: str,
    repo_owner: str,
    repo_name: str,
    pr_number: int,
    base_sha: str,
    head_sha: str,
) -> str:
    return (
        f"## Pull request to review\n\n"
        f"- repo: {repo_owner}/{repo_name}\n"
        f"- pr_number: {pr_number}\n"
        f"- url: {pr_url}\n"
        f"- base_sha: {base_sha}\n"
        f"- head_sha: {head_sha}\n\n"
        f"Fetch the diff yourself with "
        f"`GH_TOKEN=dummy gh pr diff {pr_number} --repo {repo_owner}/{repo_name}`, "
        f"then review only what's in that diff.\n\n"
        f"This is a first review — there are no existing findings. Record real "
        f"issues with `add_finding` (one per issue; only include `suggestion` "
        f"when the fix is ≤4 lines and obvious), then call `publish_review` "
        f"once at the end."
    )


def _build_re_review_context(
    *,
    pr_url: str,
    repo_owner: str,
    repo_name: str,
    pr_number: int,
    last_reviewed_sha: str,
    head_sha: str,
    existing_findings_block: str,
) -> str:
    return (
        f"## A new commit has been pushed\n\n"
        f"- repo: {repo_owner}/{repo_name}\n"
        f"- pr_number: {pr_number}\n"
        f"- url: {pr_url}\n"
        f"- previous reviewed SHA: {last_reviewed_sha}\n"
        f"- new HEAD SHA: {head_sha}\n\n"
        f"## Existing findings\n\n{existing_findings_block}\n\n"
        f"Fetch the diff since the previous reviewed SHA yourself with "
        f"`GH_TOKEN=dummy gh api repos/{repo_owner}/{repo_name}/compare/"
        f'{last_reviewed_sha}...{head_sha} -H "Accept: application/vnd.github.v3.diff"`, '
        f"then review only what's in that diff.\n\n"
        f"For each open finding above, decide whether the new commits resolved "
        f'it (`update_finding(id, status="resolved")`), left it unchanged '
        f"(no action), or changed it materially (`update_finding` with new "
        f"fields + a `note`). Then add any net-new findings introduced by the "
        f"new diff, and call `publish_review` once at the end."
    )


def _format_existing_findings(findings: list[dict]) -> str:
    if not findings:
        return "_(none)_"
    lines: list[str] = []
    for f in findings:
        if f.get("status") != "open":
            continue
        location = f.get("file", "<unknown>")
        start = f.get("start_line")
        end = f.get("end_line")
        if start is not None and end is not None:
            location += f":{start}" if start == end else f":{start}-{end}"
        lines.append(
            f"- [{f.get('id')}] ({f.get('severity')}, {f.get('category')}) "
            f"{location} — {f.get('description', '').strip()}"
        )
    return "\n".join(lines) if lines else "_(no open findings)_"


async def get_reviewer_agent(config: RunnableConfig) -> Pregel:
    """Get or create a reviewer agent with a sandbox + prepped repo."""
    thread_id = config["configurable"].get("thread_id", None)

    config["recursion_limit"] = DEFAULT_RECURSION_LIMIT

    if thread_id is None or not graph_loaded_for_execution(config):
        logger.info("No thread_id or not for execution, returning reviewer agent without sandbox")
        return create_deep_agent(system_prompt="", tools=[]).with_config(config)

    if config["configurable"].get("source"):
        cached_token, cached_encrypted = await get_github_token_from_thread(thread_id)
        if cached_token and cached_encrypted:
            config["metadata"]["github_token_encrypted"] = cached_encrypted
            del cached_token
        else:
            _token, new_encrypted = await resolve_github_token(config, thread_id)
            config["metadata"]["github_token_encrypted"] = new_encrypted
            del _token

    sandbox_backend = await ensure_sandbox_for_thread(thread_id)

    work_dir = await aresolve_sandbox_work_dir(sandbox_backend)

    repo_config = config["configurable"].get("repo") or {}
    repo_owner = str(repo_config.get("owner", ""))
    repo_name = str(repo_config.get("name", ""))
    base_sha = str(config["configurable"].get("base_sha", "") or "")
    head_sha = str(config["configurable"].get("head_sha", "") or "")
    pr_number = config["configurable"].get("pr_number")
    pr_url = str(config["configurable"].get("pr_url", "") or "")
    last_reviewed_sha = str(config["configurable"].get("last_reviewed_sha", "") or "")
    is_re_review = bool(config["configurable"].get("re_review"))

    # Hotfix: prep was producing empty diffs for some PRs and the agent
    # silently published "no issues found". The agent now fetches the diff
    # itself via `gh pr diff` (or `gh api ...compare...` on re-review).
    # `add_finding`'s in-diff line-range validation is skipped when no
    # diff_line_set is set in config — we trust the agent's anchors.
    config["configurable"]["diff_text"] = ""
    config["configurable"]["diff_line_set"] = None

    review_context = ""
    if pr_number is not None and isinstance(pr_number, int):
        if is_re_review and last_reviewed_sha:
            existing_findings = await list_findings_async(thread_id)
            review_context = _build_re_review_context(
                pr_url=pr_url,
                repo_owner=repo_owner,
                repo_name=repo_name,
                pr_number=pr_number,
                last_reviewed_sha=last_reviewed_sha,
                head_sha=head_sha,
                existing_findings_block=_format_existing_findings(existing_findings),
            )
        else:
            review_context = _build_first_review_context(
                pr_url=pr_url,
                repo_owner=repo_owner,
                repo_name=repo_name,
                pr_number=pr_number,
                base_sha=base_sha,
                head_sha=head_sha,
            )

    model_id = os.environ.get("LLM_MODEL_ID", DEFAULT_LLM_MODEL_ID)
    model_kwargs: ModelKwargs = {"max_tokens": DEFAULT_LLM_MAX_TOKENS}
    if model_id == DEFAULT_LLM_MODEL_ID:
        model_kwargs["reasoning"] = DEFAULT_LLM_REASONING

    system_prompt = _reviewer_system_prompt(
        f"{work_dir}/{repo_name}" if repo_name else work_dir,
        repo_owner=repo_owner,
        repo_name=repo_name,
        pr_number=pr_number if isinstance(pr_number, int) else "",
    )
    if review_context:
        system_prompt = f"{system_prompt}\n\n{review_context}"

    return create_deep_agent(
        model=make_model(model_id, **model_kwargs),
        system_prompt=system_prompt,
        tools=[add_finding, update_finding, list_findings, publish_review],
        backend=sandbox_backend,
        middleware=[
            SanitizeToolInputsMiddleware(),
            ModelCallLimitMiddleware(run_limit=MODEL_CALL_RECURSION_LIMIT, exit_behavior="end"),
            ToolErrorMiddleware(),
            ExcludeToolsMiddleware(excluded=frozenset({"task"})),
        ],
    ).with_config(config)