tgit/app/repo/publisher.py

"""Turns a member's draft into one commit on the uploads repo's main branch (SPEC C-20…C-25, ARCHITECTURE §2.3).

Pushes are serialized with a lock: two members pushing at once never fight over the shared working tree.
"""

import asyncio
import logging
from collections.abc import Awaitable, Callable
from dataclasses import dataclass, field
from datetime import datetime, timedelta
from enum import StrEnum
from pathlib import Path

from sqlalchemy import func, select

from app.db.models import ChangeOp, PipelineStatus, Push, User, utcnow
from app.db.session import SessionMaker
from app.repo.drafts import DraftView
from app.repo.git import GitError, UploadsRepo

log = logging.getLogger(__name__)

BOT_NAME = "tgit-bot"
BOT_EMAIL = "tgit-bot@users.noreply.tgit"
PUSHES_PER_HOUR = 20
PUSH_ATTEMPTS = 3


class PushStatus(StrEnum):
    OK = "ok"
    CONFLICT = "conflict"  # someone changed the same files in main since the draft began
    EMPTY = "empty"  # the draft matches main already
    LIMIT = "limit"
    FAILED = "failed"


@dataclass(frozen=True)
class PushResult:
    status: PushStatus
    sha: str | None = None
    conflicts: list[str] = field(default_factory=list)


class Publisher:
    def __init__(
        self,
        repo: UploadsRepo,
        sessions: SessionMaker,
        clock: Callable[[], datetime] = utcnow,
        before_push: Callable[[], Awaitable[None]] | None = None,  # test hook: simulate a concurrent push
    ) -> None:
        self._repo = repo
        self._sessions = sessions
        self._clock = clock
        self._before_push = before_push
        self._lock = asyncio.Lock()

    async def push(self, member: User, draft: DraftView, message: str, *, overwrite: bool = False) -> PushResult:
        if await self._pushes_last_hour(member.tg_id) >= PUSHES_PER_HOUR:
            return PushResult(PushStatus.LIMIT)
        async with self._lock:
            try:
                return await self._push(member, draft, message, overwrite)
            except GitError:
                log.exception("push failed")
                await self._reset()
                return PushResult(PushStatus.FAILED)

    async def record(
        self,
        member: User,
        result: PushResult,
        message: str,
        draft: DraftView,
        chat_id: int,
        message_id: int,
        status: tuple[PipelineStatus, PipelineStatus],
    ) -> Push:
        """status — (ci, cd) to start from; the PipelineWatcher takes it from there (#11)."""
        assert result.sha is not None
        push = Push(
            user_id=member.tg_id,
            sha=result.sha,
            message=message,
            paths=[c.path for c in draft.changes],
            chat_id=chat_id,
            message_id=message_id,
            ci_status=status[0],
            cd_status=status[1],
            created_at=self._clock(),
        )
        async with self._sessions.begin() as s:
            s.add(push)
        return push

    async def _push(self, member: User, draft: DraftView, message: str, overwrite: bool) -> PushResult:
        repo = self._repo
        await repo.refresh(force=True)
        if not overwrite:
            conflicts = await self._conflicts(draft)
            if conflicts:
                return PushResult(PushStatus.CONFLICT, conflicts=conflicts)

        await self._reset()
        self._apply(draft)
        await repo.git("add", "-A")
        try:
            await repo.git("diff", "--cached", "--quiet")
            return PushResult(PushStatus.EMPTY)  # exit 0: nothing staged
        except GitError:
            pass  # exit 1: there are changes

        handle = f"@{member.username}" if member.username else member.full_name
        full_message = f"{message.strip()}\n\nTelegram-User: {handle} ({member.tg_id})"
        await repo.git(
            "-c", f"user.name={BOT_NAME}", "-c", f"user.email={BOT_EMAIL}", "commit", "-q", "-m", full_message
        )

        branch = repo.branch
        for attempt in range(1, PUSH_ATTEMPTS + 1):
            if self._before_push is not None:
                await self._before_push()
            try:
                await repo.git("push", "-q", "origin", f"HEAD:{branch}")
                break
            except GitError:
                if attempt == PUSH_ATTEMPTS:
                    raise
                log.info("push rejected, rebasing (attempt %d)", attempt)
                await repo.git("fetch", "-q", "origin")
                try:
                    await repo.git(
                        "-c", f"user.name={BOT_NAME}", "-c", f"user.email={BOT_EMAIL}", "rebase", "-q", repo.ref
                    )
                except GitError:
                    await repo.git("rebase", "--abort")
                    return PushResult(PushStatus.CONFLICT, conflicts=await self._conflicts(draft))
        sha = (await repo.git("rev-parse", "HEAD")).decode().strip()
        return PushResult(PushStatus.OK, sha=sha)

    async def _conflicts(self, draft: DraftView) -> list[str]:
        try:
            out = await self._repo.git("diff", "--name-only", draft.base_sha, self._repo.ref)
        except GitError:
            return [c.path for c in draft.changes]  # base unknown (e.g. history rewritten): be safe
        changed = set(out.decode().splitlines())
        return [c.path for c in draft.changes if c.path in changed]

    async def _reset(self) -> None:
        repo = self._repo
        await repo.git("checkout", "-q", "-B", repo.branch, repo.ref)
        await repo.git("reset", "-q", "--hard", repo.ref)
        await repo.git("clean", "-q", "-fdx")

    def _apply(self, draft: DraftView) -> None:
        root = self._repo.path.resolve()
        for change in draft.changes:
            target = (root / change.path).resolve()
            if root not in target.parents:  # validated earlier; this is the last line of defence
                raise GitError(f"path escapes the repo: {change.path}")
            if change.op is ChangeOp.DELETE:
                target.unlink(missing_ok=True)
                _prune_empty_dirs(target.parent, root)
            else:
                target.parent.mkdir(parents=True, exist_ok=True)
                target.write_bytes(change.content or b"")

    async def _pushes_last_hour(self, user_id: int) -> int:
        async with self._sessions() as s:
            count = await s.scalar(
                select(func.count())
                .select_from(Push)
                .where(Push.user_id == user_id, Push.created_at > self._clock() - timedelta(hours=1))
            )
            return count or 0


def _prune_empty_dirs(directory: Path, root: Path) -> None:
    while directory != root and directory.is_dir() and not any(directory.iterdir()):
        directory.rmdir()
        directory = directory.parent