cleveragents-core/tools/controller/master/forgejo_writes.py

"""Forgejo write protocol for the master.

The master writes three categories of operator-visible state into
Forgejo:
1. Status comments (verbose what-happened summaries on the PR
   timeline)
2. Labels (machine-readable + glanceable state markers)
3. Merge calls (Phase 1d-3c-3, separate module)

Per plan v9 simplified write ordering:
- Forgejo write FIRST with fingerprint-based idempotency
- DB commit SECOND
- Reconciliation tick syncs DB↔Forgejo if a crash happens between

This module ships the controller-side write logic. Forgejo HTTP is
injected via callbacks so tests don't need a live Forgejo instance.
Production wires the callbacks to the existing
``tools/_claim_runtime.get/post`` + ``_status_comments`` helpers.

Fingerprint pattern (carried forward from the existing
``_status_comments`` module): each operator-visible state has a
deterministic fingerprint hash of (workflow_id, event_kind, content).
The comment body includes a hidden HTML marker
``<!-- controller:fingerprint:abc123 -->``. Before posting, we list
existing comments + check for the marker; if found, skip the post.
This gives idempotency without needing the DB to be the source of
truth.
"""
from __future__ import annotations

import hashlib
import logging
from collections.abc import Callable
from dataclasses import dataclass
from typing import Any

logger = logging.getLogger(__name__)


# Fingerprint marker prefix. Searchable + visible-but-quiet in
# Forgejo's UI (HTML comments don't render).
FINGERPRINT_MARKER_PREFIX = "<!-- controller:fingerprint:"
FINGERPRINT_MARKER_SUFFIX = " -->"


# Callback signatures. Each callback returns "what Forgejo said back"
# in a normalized shape.
#
# list_comments(owner, repo, pr_number) -> list of {"body": str, "id": int, ...}
# post_comment(owner, repo, pr_number, body) -> {"id": int, "body": str}
# get_labels(owner, repo, pr_number) -> list of {"name": str, "id": int, ...}
# add_label(owner, repo, pr_number, label_name) -> True on success
# remove_label(owner, repo, pr_number, label_name) -> True on success
ListCommentsCallback = Callable[[str, str, int], list[dict]]
PostCommentCallback = Callable[[str, str, int, str], dict]
GetLabelsCallback = Callable[[str, str, int], list[dict]]
AddLabelCallback = Callable[[str, str, int, str], bool]
RemoveLabelCallback = Callable[[str, str, int, str], bool]


@dataclass
class StatusCommentResult:
    """What ``post_status_comment`` returns.

    Three outcomes:
    - posted: a new comment was successfully posted.
    - duplicate: a comment with the same fingerprint already exists;
      this call was a no-op (idempotency working as intended).
    - failed: the post-comment callback raised; caller decides
      whether to retry on next tick.
    """

    status: str  # 'posted' | 'duplicate' | 'failed'
    comment_id: int | None = None
    fingerprint: str | None = None
    error: str | None = None


@dataclass
class LabelAdjustResult:
    """Summary of one label adjust (set OR unset)."""

    label: str
    action: str  # 'added' | 'removed' | 'no-op' | 'failed'
    error: str | None = None


# ─── fingerprint helpers ──────────────────────────────────────────────


def compute_fingerprint(
    *, workflow_id: int, event_kind: str, content_key: str,
) -> str:
    """Build the deterministic fingerprint for a write.

    The fingerprint is a short SHA256 prefix of
    (workflow_id, event_kind, content_key) — enough to disambiguate
    same-workflow / same-event-kind writes that differ only in content
    (e.g., two STUCK transitions with different reasons).
    """
    payload = f"{workflow_id}:{event_kind}:{content_key}".encode("utf-8")
    return hashlib.sha256(payload).hexdigest()[:16]


def build_marker(fingerprint: str) -> str:
    """Build the HTML marker the comment body must include."""
    return f"{FINGERPRINT_MARKER_PREFIX}{fingerprint}{FINGERPRINT_MARKER_SUFFIX}"


def comment_has_fingerprint(comment_body: str, fingerprint: str) -> bool:
    """Check if a comment's body contains the marker for this fingerprint."""
    return build_marker(fingerprint) in (comment_body or "")


# ─── status comment ───────────────────────────────────────────────────


def post_status_comment(
    *,
    owner: str,
    repo: str,
    pr_number: int,
    workflow_id: int,
    event_kind: str,
    content_key: str,
    body_text: str,
    list_comments: ListCommentsCallback,
    post_comment: PostCommentCallback,
) -> StatusCommentResult:
    """Post a status comment with fingerprint-based idempotency.

    Procedure:
    1. Build the fingerprint from (workflow_id, event_kind, content_key).
    2. Build the marker + check existing comments for it (via
       ``list_comments`` callback).
    3. If a matching comment exists → return duplicate (no-op).
    4. Else post a new comment with body = body_text + marker.

    Failure of ``post_comment`` is returned as ``StatusCommentResult(
    status='failed', error=...)`` — the caller decides whether to
    retry on the next tick. ``list_comments`` failure falls through:
    we conservatively post (risk: occasional duplicate; mitigated by
    fingerprint match on next sweep).
    """
    fingerprint = compute_fingerprint(
        workflow_id=workflow_id,
        event_kind=event_kind, content_key=content_key,
    )
    marker = build_marker(fingerprint)

    try:
        existing = list_comments(owner, repo, pr_number) or []
    except Exception as exc:  # noqa: BLE001 — Forgejo flake; fall through to post
        logger.warning(
            "list_comments failed for #%d (%s); will post (risks duplicate): %s",
            pr_number, repo, exc,
        )
        existing = []

    for c in existing:
        if comment_has_fingerprint(c.get("body") or "", fingerprint):
            return StatusCommentResult(
                status="duplicate",
                comment_id=c.get("id"),
                fingerprint=fingerprint,
            )

    body = f"{body_text}\n\n{marker}"
    try:
        posted = post_comment(owner, repo, pr_number, body)
    except Exception as exc:  # noqa: BLE001
        return StatusCommentResult(
            status="failed",
            fingerprint=fingerprint,
            error=str(exc),
        )

    return StatusCommentResult(
        status="posted",
        comment_id=(posted or {}).get("id"),
        fingerprint=fingerprint,
    )


# ─── labels ───────────────────────────────────────────────────────────


def adjust_labels(
    *,
    owner: str,
    repo: str,
    pr_number: int,
    add: list[str] | None = None,
    remove: list[str] | None = None,
    get_labels: GetLabelsCallback,
    add_label: AddLabelCallback,
    remove_label: RemoveLabelCallback,
) -> list[LabelAdjustResult]:
    """Add + remove labels with no-op-on-already-set/missing.

    Procedure:
    1. List the PR's current labels (via ``get_labels``).
    2. For each label in ``add``: if already present → no-op; else
       add via ``add_label``.
    3. For each label in ``remove``: if not present → no-op; else
       remove via ``remove_label``.

    Returns one LabelAdjustResult per label across both lists.
    Failures are per-label (a failed add doesn't prevent a remove
    of a different label).
    """
    add = add or []
    remove = remove or []
    if not add and not remove:
        return []

    try:
        current_labels = get_labels(owner, repo, pr_number) or []
    except Exception as exc:  # noqa: BLE001
        # Defensive: if we can't list, mark every requested action
        # as failed so the caller can retry on next tick.
        err_msg = f"get_labels failed: {exc}"
        return [
            LabelAdjustResult(label=lbl, action="failed", error=err_msg)
            for lbl in (add + remove)
        ]

    current_names = {(c.get("name") or "") for c in current_labels}
    results: list[LabelAdjustResult] = []

    for lbl in add:
        if lbl in current_names:
            results.append(LabelAdjustResult(label=lbl, action="no-op"))
            continue
        try:
            ok = add_label(owner, repo, pr_number, lbl)
            results.append(LabelAdjustResult(
                label=lbl, action="added" if ok else "failed",
                error=None if ok else "add_label returned False",
            ))
        except Exception as exc:  # noqa: BLE001
            results.append(LabelAdjustResult(
                label=lbl, action="failed", error=str(exc),
            ))

    for lbl in remove:
        if lbl not in current_names:
            results.append(LabelAdjustResult(label=lbl, action="no-op"))
            continue
        try:
            ok = remove_label(owner, repo, pr_number, lbl)
            results.append(LabelAdjustResult(
                label=lbl, action="removed" if ok else "failed",
                error=None if ok else "remove_label returned False",
            ))
        except Exception as exc:  # noqa: BLE001
            results.append(LabelAdjustResult(
                label=lbl, action="failed", error=str(exc),
            ))

    return results


__all__ = [
    "AddLabelCallback",
    "FINGERPRINT_MARKER_PREFIX",
    "FINGERPRINT_MARKER_SUFFIX",
    "GetLabelsCallback",
    "LabelAdjustResult",
    "ListCommentsCallback",
    "PostCommentCallback",
    "RemoveLabelCallback",
    "StatusCommentResult",
    "adjust_labels",
    "build_marker",
    "comment_has_fingerprint",
    "compute_fingerprint",
    "post_status_comment",
]