"""HTTP API: one process, one serialized SQLite writer, a small reader pool.""" from __future__ import annotations import asyncio import fcntl import logging import mimetypes import os import random import shutil import sqlite3 import threading import time import tempfile from concurrent.futures import ThreadPoolExecutor from contextlib import asynccontextmanager, suppress from pathlib import Path from urllib.parse import unquote import anyio from fastapi import FastAPI, Query, Request from fastapi.responses import FileResponse, JSONResponse, PlainTextResponse, Response from starlette.background import BackgroundTask from . import db as dbmod from . import guide from . import store from .blobs import LocalBlobStore, new_hasher from .config import DB_NAME, LIMITS from .errors import ApiError, bad_request, forbidden from .presence import PRESENCE_FLUSH_SECONDS, Presence from . import transfer from .reservations import Reservations class NotInitialized(RuntimeError): pass class AlreadyRunning(RuntimeError): pass class Database: """Serialized writes on one connection/thread; reads on a bounded thread pool. Every write runs in BEGIN IMMEDIATE ... COMMIT and returns only after COMMIT, so an HTTP success always follows a durable commit (synchronous=FULL). """ def __init__(self, path: Path): self.path = path self._writer_conn = dbmod.connect(path) self._writer = ThreadPoolExecutor(1, thread_name_prefix="msgboard-writer") self._readers = ThreadPoolExecutor(LIMITS.reader_threads, thread_name_prefix="msgboard-reader") self._local = threading.local() self._reader_conns: list[sqlite3.Connection] = [] self._lock = threading.Lock() self.pending_writes = 0 self.queued_reads = 0 self.shed_reads = 0 self.slow_reads = 0 # token -> (Principal, checked at): see LIMITS.auth_cache_seconds self.auth_cache: dict[str, tuple] = {} # Bumped after every committed write; the replicator pushes when it changes. self.generation = 0 self.closing = False def _reader_conn(self) -> sqlite3.Connection: conn = getattr(self._local, "conn", None) if conn is None: conn = dbmod.connect(self.path) local = self._local def over_time(): deadline = getattr(local, "deadline", None) return 1 if deadline is not None and time.monotonic() > deadline else 0 conn.set_progress_handler(over_time, 20_000) self._local.conn = conn with self._lock: self._reader_conns.append(conn) return conn async def read(self, fn, *args, shed: bool = True): """Run a read on the reader pool. Under overload, refuse early (503) rather than queue work whose client will have timed out before it runs: that is what turned a burst into an outage.""" if shed and self.queued_reads >= LIMITS.max_queued_reads: self.shed_reads += 1 raise ApiError(503, "overloaded", "the board is busy; retry shortly", {"Retry-After": "2"}) queued_at = time.monotonic() def run(): if time.monotonic() - queued_at > LIMITS.read_queue_timeout: self.shed_reads += 1 raise ApiError(503, "overloaded", "the board is busy; retry shortly", {"Retry-After": "2"}) conn = self._reader_conn() self._local.deadline = time.monotonic() + LIMITS.max_read_seconds try: return fn(conn, *args) except sqlite3.OperationalError as e: if "interrupt" in str(e): self.slow_reads += 1 raise ApiError(503, "too_slow", "this request took too long; try a narrower one", {"Retry-After": "5"}) raise finally: self._local.deadline = None self.queued_reads += 1 # queued or running; only touched on the event loop try: return await asyncio.get_running_loop().run_in_executor(self._readers, run) finally: self.queued_reads -= 1 def _run_write(self, fn, args): conn = self._writer_conn conn.execute("BEGIN IMMEDIATE") try: result = fn(conn, *args) except BaseException: conn.execute("ROLLBACK") raise conn.execute("COMMIT") return result async def write(self, fn, *args): if self.closing: raise ApiError(503, "shutting_down", "server is shutting down; retry shortly", {"Retry-After": "2"}) if self.pending_writes >= LIMITS.max_pending_writes: raise ApiError(503, "overloaded", "write queue is full", {"Retry-After": "1"}) self.pending_writes += 1 try: loop = asyncio.get_running_loop() result = await loop.run_in_executor(self._writer, self._run_write, fn, args) self.generation += 1 if fn not in AUTH_SAFE_WRITES: self.auth_cache.clear() return result finally: self.pending_writes -= 1 def close(self) -> None: self._writer.shutdown(wait=True) self._readers.shutdown(wait=True) for conn in self._reader_conns: conn.close() try: self._writer_conn.execute("PRAGMA wal_checkpoint(TRUNCATE)") finally: self._writer_conn.close() ANY_CHANNEL = -1 # notifier key woken by a post in any channel (activity long-polls wait on it) # Writes that cannot change who a token belongs to or what it may do; every other write clears the auth cache. AUTH_SAFE_WRITES = frozenset({store.post_message, store.react, store.mark_read, store.mark_thread_read, store.mark_all_read, store.save_presence}) class Notifier: """In-memory per-channel wakeups. Hints only: waiters always re-query the database.""" def __init__(self): self._waiters: dict[int, set[asyncio.Future]] = {} self.loop: asyncio.AbstractEventLoop | None = None def register(self, channel_id: int) -> asyncio.Future: fut = asyncio.get_running_loop().create_future() self._waiters.setdefault(channel_id, set()).add(fut) return fut def unregister(self, channel_id: int, fut: asyncio.Future) -> None: waiters = self._waiters.get(channel_id) if waiters is not None: waiters.discard(fut) if not waiters: del self._waiters[channel_id] def notify(self, channel_id: int) -> None: for fut in self._waiters.pop(channel_id, ()): if not fut.done(): fut.set_result(None) def wake_all(self) -> None: for channel_id in list(self._waiters): self.notify(channel_id) class Board: """Everything one running server owns for a data directory.""" def __init__(self, data_dir: Path, replicator=None): self.data_dir = Path(data_dir) db_path = self.data_dir / DB_NAME if not db_path.exists(): raise NotInitialized(f"no board in {self.data_dir}; run `python -m msgboard init` first") self._lock_fh = open(self.data_dir / "server.lock", "a") try: fcntl.flock(self._lock_fh, fcntl.LOCK_EX | fcntl.LOCK_NB) except BlockingIOError: self._lock_fh.close() raise AlreadyRunning(f"another server is already using {self.data_dir}") check = dbmod.connect(db_path) try: version = dbmod.schema_version(check) if version > dbmod.SCHEMA_VERSION: raise dbmod.SchemaError(f"database schema {version} is newer than this software") if version < dbmod.SCHEMA_VERSION: raise NotInitialized("database schema is outdated; run `python -m msgboard init` to migrate") if not store.admin_exists(check): raise NotInitialized("board has no operator; run `python -m msgboard init`") except BaseException: check.close() self._release_lock() raise check.close() self.db = Database(db_path) # one-time maintenance on the writer connection before serving with self.db._lock: conn = self.db._writer_conn conn.execute("BEGIN IMMEDIATE") try: store.backfill_mentions(conn) store.backfill_mentioned_as(conn) conn.execute("COMMIT") except BaseException: conn.execute("ROLLBACK") raise self.presence = Presence(store.load_presence(conn)) self.blobs = LocalBlobStore(self.data_dir / "blobs") # storage folders live in the replication store (the Space's bucket) or, locally, under the data dir; # downloads are served from a local cache of those objects from .replicate import DirStore self.storage = replicator.store if replicator is not None else DirStore(self.data_dir) self.storage_cache = self.data_dir / "storage-cache" self.storage_cache.mkdir(exist_ok=True) self.notifier = Notifier() self.shutting_down = False self.open_polls = 0 self.open_uploads = 0 self.commit_lock: asyncio.Lock | None = None self.storage_reservations = Reservations(paths=True) self.import_reservations = Reservations() self.started = time.time() self.counters = {"posts": 0, "posts_replayed": 0, "edits": 0, "polls": 0, "errors_5xx": 0} self.replicator = replicator if replicator is not None: replicator.start(lambda: self.db.generation) def begin_shutdown(self) -> None: """Release held long-polls so a Ctrl-C does not wait for them to time out.""" if self.replicator is not None and not self.shutting_down: self.replicator.record("shutdown-signal") self.shutting_down = True if self.replicator is not None: # With replication, stop taking writes now: the final snapshot must # include every acknowledged write, and the next instance is waiting. self.db.closing = True loop = self.notifier.loop if loop is not None and not loop.is_closed(): loop.call_soon_threadsafe(self.notifier.wake_all) def _release_lock(self) -> None: try: fcntl.flock(self._lock_fh, fcntl.LOCK_UN) finally: self._lock_fh.close() async def flush_presence(self) -> None: versions, records = self.presence.pending() history = self.presence.pending_history() if records or history: await self.db.write(store.save_presence, records, history) self.presence.saved(versions) self.presence.history_saved(history) async def presence_loop(self) -> None: while True: await asyncio.sleep(PRESENCE_FLUSH_SECONDS) if self.shutting_down: return self.presence.sample(time.time()) try: await self.flush_presence() except Exception: # Observations remain dirty and retry on the next interval. Presence # must not cause regular requests to fail when the writer is busy. logging.getLogger(__name__).exception("could not persist agent presence") def close(self) -> None: if getattr(self, "_closed", False): return self._closed = True # The event loop has stopped. Queue the last observations after in-flight # writes, before the replicator takes its final snapshot. This also works # when begin_shutdown has closed the normal HTTP writer entry point. versions, records = self.presence.pending() history = self.presence.pending_history() if records or history: try: self.db._writer.submit(self.db._run_write, store.save_presence, (records, history)).result() self.db.generation += 1 self.presence.saved(versions) self.presence.history_saved(history) except Exception: # Ephemeral presence must never prevent the final snapshot of # acknowledged messages or leave the database/lease open. logging.getLogger(__name__).exception("could not persist final agent presence") if self.replicator is not None: self.replicator.stop(final_push=True) # after in-flight writes, before closing self.db.close() self._release_lock() def _bearer(request: Request) -> str | None: # X-Msgboard-Token lets clients reach the board through a proxy that uses # Authorization for its own login (for example a private HF Space). direct = request.headers.get("x-msgboard-token") if direct: return direct.strip() header = request.headers.get("authorization", "") scheme, _, value = header.partition(" ") return value.strip() if scheme.lower() == "bearer" else None async def _json_body(request: Request, max_bytes: int = 256 * 1024): import json data = bytearray() async for chunk in request.stream(): data += chunk if len(data) > max_bytes: raise ApiError(413, "too_large", "request body too large") if not data: return {} try: return json.loads(data) except ValueError: raise bad_request("request body is not valid JSON") def create_app(board: Board, dashboard: bool = True) -> FastAPI: @asynccontextmanager async def lifespan(app): board.notifier.loop = asyncio.get_running_loop() board.commit_lock = asyncio.Lock() board.presence.sample(time.time()) task = asyncio.create_task(board.presence_loop()) try: yield finally: board.begin_shutdown() task.cancel() with suppress(asyncio.CancelledError): await task app = FastAPI(title="msgboard", version="0.1.0", lifespan=lifespan, docs_url=None, redoc_url=None, openapi_url="/v1/openapi.json") app.state.board = board db = board.db @app.exception_handler(ApiError) async def api_error(request: Request, exc: ApiError): if exc.status >= 500: board.counters["errors_5xx"] += 1 body = {"code": exc.code, "message": exc.message} if exc.details: body["details"] = exc.details return JSONResponse({"error": body}, status_code=exc.status, headers=exc.headers) async def authn(request: Request, *, observe: bool = True): token = _bearer(request) now = time.monotonic() hit = db.auth_cache.get(token) if token else None if hit is not None and now - hit[1] < LIMITS.auth_cache_seconds: if observe: board.presence.seen(hit[0], time.time()) return hit[0] p = await db.read(store.authenticate, token, time.time()) if len(db.auth_cache) > 20_000: db.auth_cache.clear() db.auth_cache[token] = (p, now) if observe: board.presence.seen(p, time.time()) return p async def with_presence(p, items): return await db.read(store.attach_presence, p, items, board.presence.snapshot()) # ------------------------------------------------------------- health @app.get("/v1/health") async def health(): return {"status": "ok"} @app.get("/v1/ready") async def ready(): await db.read(lambda c: c.execute("SELECT 1").fetchone()) return {"status": "ready", "shutting_down": board.shutting_down} @app.get("/v1/metrics") async def metrics(request: Request): p = await authn(request) if not p.is_admin: raise forbidden("admin only") return health_summary() def health_summary(): db_path = board.data_dir / DB_NAME wal = Path(str(db_path) + "-wal") return { **board.counters, "open_polls": board.open_polls, "open_uploads": board.open_uploads, "pending_writes": db.pending_writes, "queued_reads": db.queued_reads, "shed_reads": db.shed_reads, "slow_reads": db.slow_reads, "db_bytes": db_path.stat().st_size, "wal_bytes": wal.stat().st_size if wal.exists() else 0, "disk_free_bytes": shutil.disk_usage(board.data_dir).free, "uptime_seconds": time.time() - board.started, "replication": board.replicator.status if board.replicator else None, } # ------------------------------------------------------- identities def body_dict(body): if not isinstance(body, dict): raise bad_request("request body must be a JSON object") return body @app.get("/v1/me") async def get_me(request: Request): p = await authn(request) return (await with_presence(p, [await db.read(store.me, p)]))[0] @app.post("/v1/me/read-all") async def mark_all_read(request: Request): """Mark everything you can read as read (channels, DMs and threads).""" p = await authn(request) return await db.write(store.mark_all_read, p, time.time()) @app.put("/v1/me/prefs") async def put_prefs(request: Request): p = await authn(request) return await db.write(store.update_prefs, p, await _json_body(request, 64 * 1024), time.time()) @app.get("/v1/principals") async def list_principals(request: Request): p = await authn(request) return {"principals": await with_presence(p, await db.read(store.list_principals, p))} @app.get("/v1/presence") async def list_presence(request: Request): p = await authn(request) return {"principals": await db.read(store.list_presence, p, board.presence.snapshot())} @app.patch("/v1/principals/{principal_id}") async def update_principal(principal_id: int, request: Request): p = await authn(request) body = body_dict(await _json_body(request)) result = await db.write(store.update_principal, p, principal_id, body, time.time()) return (await with_presence(p, [result]))[0] @app.post("/v1/principals/{principal_id}/suspend") async def suspend(principal_id: int, request: Request): p = await authn(request) raw = await request.body() body = body_dict(await _json_body(request)) if raw.strip() else {} return await db.write(store.suspend, p, principal_id, body.get("reason"), time.time()) @app.post("/v1/principals/{principal_id}/unsuspend") async def unsuspend(principal_id: int, request: Request): p = await authn(request) return await db.write(store.unsuspend, p, principal_id, time.time()) @app.get("/v1/suspended") async def list_suspended(request: Request): p = await authn(request) return {"suspended": await db.read(store.list_suspended, p)} @app.post("/v1/principals/{principal_id}/token") async def rotate_token(principal_id: int, request: Request): p = await authn(request) result = await db.write(store.rotate_agent_token, p, principal_id, time.time()) return JSONResponse(result, status_code=201, headers={"Cache-Control": "no-store"}) @app.post("/v1/principals/{principal_id}/signin-link") async def signin_link(principal_id: int, request: Request): p = await authn(request) result = await db.write(store.create_signin_link, p, principal_id, time.time()) return JSONResponse(result, status_code=201, headers={"Cache-Control": "no-store"}) @app.delete("/v1/tokens/{token_id}") async def revoke_token(token_id: str, request: Request): p = await authn(request) return await db.write(store.revoke_token, p, token_id, time.time()) # ------------------------------------------------- invites and joining @app.post("/v1/invites") async def create_invite(request: Request): p = await authn(request) body = body_dict(await _json_body(request)) return JSONResponse(await db.write(store.create_invite, p, body, time.time()), status_code=201, headers={"Cache-Control": "no-store"}) @app.get("/v1/invites") async def list_invites(request: Request): p = await authn(request) return {"invites": await db.read(store.list_invites, p, time.time())} @app.delete("/v1/invites/{invite_id}") async def revoke_invite(invite_id: int, request: Request): p = await authn(request) return await db.write(store.revoke_invite, p, invite_id, time.time()) @app.get("/v1/join/{code}") async def lookup_invite(code: str): return await db.read(store.lookup_invite, code, time.time()) @app.post("/v1/join") async def join(request: Request): body = body_dict(await _json_body(request, 4096)) result = await db.write(store.join, body.get("code"), body.get("name"), body.get("display_name"), time.time()) return JSONResponse(result, status_code=201 if result["created"] else 200, headers={"Cache-Control": "no-store"}) # ------------------------------------------------------------ settings @app.get("/v1/settings") async def get_settings(request: Request): await authn(request) return await db.read(store.get_settings) @app.patch("/v1/settings") async def update_settings(request: Request): p = await authn(request) body = body_dict(await _json_body(request)) return await db.write(store.update_settings, p, body, time.time()) # --------------------------------------------------------- channels @app.get("/v1/channels") async def list_channels(request: Request, name: str | None = None): p = await authn(request) return {"channels": await db.read(store.list_channels, p, name)} @app.post("/v1/channels") async def create_channel(request: Request): p = await authn(request) body = body_dict(await _json_body(request)) channel, created = await db.write(store.create_channel, p, body.get("name"), body.get("description"), body.get("private", False), time.time()) return JSONResponse(channel, status_code=201 if created else 200) @app.get("/v1/messages/{message_id}") async def get_message(message_id: int, request: Request): p = await authn(request) return await db.read(store.get_message, p, message_id) @app.get("/v1/channels/{channel_id}/export") async def export_channel(channel_id: int, request: Request): p = await authn(request) transfer.require_admin(p) manifest, keys = await db.read(transfer.export_manifest, p, channel_id, _origin(request)) scratch = tempfile.TemporaryDirectory(prefix="msgboard-export-") path = Path(scratch.name) / "channel.zip" try: await anyio.to_thread.run_sync(transfer.build_bundle, manifest, keys, board.blobs, path, board.storage) except BaseException: scratch.cleanup() raise return FileResponse(path, media_type="application/zip", filename=f"channel-{channel_id}.zip", background=BackgroundTask(scratch.cleanup)) @app.post("/v1/channels/{channel_id}/import") async def import_channel(channel_id: int, request: Request, dry_run: bool = False, keep_target_doc: bool = False): p = await authn(request, observe=not dry_run) await db.read(transfer.public_channel, p, channel_id) declared = request.headers.get("content-length", "") if declared.isdigit() and int(declared) > transfer.MAX_BUNDLE_BYTES: raise ApiError(413, "too_large", "channel bundle exceeds 512 MiB") with tempfile.TemporaryDirectory(prefix="msgboard-import-") as scratch: path = Path(scratch) / "bundle.zip" with path.open("xb") as fh: size = 0 async for chunk in request.stream(): size += len(chunk) if size > transfer.MAX_BUNDLE_BYTES: raise ApiError(413, "too_large", "channel bundle exceeds 512 MiB") await anyio.to_thread.run_sync(fh.write, chunk) manifest, digest, files = await anyio.to_thread.run_sync(transfer.read_bundle, path, Path(scratch)) if dry_run: report = await db.read(transfer.plan_import, p, channel_id, manifest, digest, keep_target_doc) report = await anyio.to_thread.run_sync(transfer.verify_reused_storage, report, board.storage) else: # A disconnected caller must not remove staged local files while # its publication task is still using them. Retries are idempotent. task = asyncio.create_task(apply_history(p, channel_id, manifest, digest, files, keep_target_doc)) try: report = await asyncio.shield(task) except asyncio.CancelledError: with suppress(Exception): await task raise return {**report, "dry_run": dry_run} async def apply_history(p, channel_id, manifest, digest, files, keep_target_doc): keys = [("channel", str(channel_id))] + [("author", a["external_id"].lower()) for a in manifest["principals"]] async with board.import_reservations.hold(keys): prepared = await db.write(transfer.prepare_import, p, channel_id, manifest, digest, time.time(), keep_target_doc) paths = [(prepared["people"][f["owner_id"]], f["path"]) for f in manifest["storage_files"]] published = False uploaded_storage, uploaded_attachments = [], [] try: async with board.storage_reservations.hold(paths): try: prepared["report"] = await db.read(transfer.checked_import_plan, p, channel_id, manifest, digest, prepared) await anyio.to_thread.run_sync(transfer.stage_import, channel_id, manifest, files, prepared, board.blobs, board.storage, uploaded_storage, uploaded_attachments) report = await db.write(transfer.import_channel, p, channel_id, manifest, digest, prepared, time.time()) published = True return report finally: if not published: if uploaded_storage: try: await anyio.to_thread.run_sync(board.storage.delete, uploaded_storage) except Exception: logging.getLogger(__name__).exception("could not remove unindexed import files") for key in uploaded_attachments: board.blobs.discard(board.blobs.path_for(key)) finally: if not published: await db.write(transfer.cleanup_ghosts, prepared) @app.get("/v1/channels/{channel_id}/doc") async def get_doc(channel_id: int, request: Request): p = await authn(request) return await db.read(store.get_doc, p, channel_id) @app.put("/v1/channels/{channel_id}/doc") async def put_doc(channel_id: int, request: Request): p = await authn(request) doc, changed = await db.write(store.save_doc, p, channel_id, await _json_body(request, store.MAX_DOC_BYTES * 2 + 4096), time.time()) return doc @app.get("/v1/channels/{channel_id}/doc/versions") async def doc_versions(channel_id: int, request: Request): p = await authn(request) return await db.read(store.doc_versions, p, channel_id) @app.get("/v1/channels/{channel_id}/doc/versions/{version}") async def doc_version(channel_id: int, version: int, request: Request): p = await authn(request) return await db.read(store.doc_version, p, channel_id, version) @app.delete("/v1/channels/{channel_id}") async def delete_channel(channel_id: int, request: Request, confirm: str = ""): p = await authn(request) result = await db.write(store.delete_channel, p, channel_id, confirm, time.time()) for key in result.pop("storage_keys"): # the rows are gone; now the files try: board.blobs.path_for(key).unlink() except FileNotFoundError: pass board.notifier.notify(channel_id) return result @app.get("/v1/channels/{channel_id}/info") async def channel_info(channel_id: int, request: Request): p = await authn(request) result = await db.read(store.channel_info, p, channel_id) result["members"] = await with_presence(p, result["members"]) result["posters"] = await with_presence(p, result["posters"]) return result @app.patch("/v1/channels/{channel_id}") async def update_channel(channel_id: int, request: Request): p = await authn(request) return await db.write(store.update_channel, p, channel_id, await _json_body(request, 4096), time.time()) @app.post("/v1/channels/{channel_id}/archive") async def archive_channel(channel_id: int, request: Request): p = await authn(request) return await db.write(store.archive_channel, p, channel_id, time.time()) @app.post("/v1/channels/{channel_id}/threads/{thread_id}/read") async def mark_thread_read(channel_id: int, thread_id: int, request: Request): p = await authn(request) body = body_dict(await _json_body(request, 1024)) return await db.write(store.mark_thread_read, p, channel_id, thread_id, body.get("seq"), time.time()) @app.post("/v1/channels/{channel_id}/read") async def mark_read(channel_id: int, request: Request): p = await authn(request) body = body_dict(await _json_body(request, 1024)) return await db.write(store.mark_read, p, channel_id, body.get("seq"), time.time()) @app.get("/v1/activity") async def get_activity(request: Request, limit: int = Query(50, ge=1, le=200), after: int | None = Query(None, ge=0), wait: float = Query(0, ge=0, le=LIMITS.max_wait_seconds), channels: str | None = None, reactions: bool = False): """Everything for you (mentions, replies, DMs, plus `channels` you follow). With `after`, a cursor read, oldest first; add `wait` to long-poll it.""" p = await authn(request) parts = [x.strip() for x in channels.split(",") if x.strip()] if channels else [] joined = "joined" in parts # every channel you are a member of try: follow = [int(x) for x in parts if x != "joined"] except ValueError: raise bad_request("channels must be comma-separated channel ids, or `joined`") if joined: follow += await db.read(store.joined_channel_ids, p) if len(follow) > 100: raise bad_request("follow at most 100 channels") if wait > 0 and after is None: raise bad_request("wait needs a cursor: pass `after` (start from the next_cursor of a plain read)") token = board.presence.next_poll() polled_at = time.time() mode = "joined" if joined else ("channels" if follow else "addressed") def observe(result): board.presence.polled(p, token, polled_at, after, mode, follow, bool(result["messages"])) if wait <= 0: result = await db.read(store.activity, p, limit, after, follow, reactions) observe(result) return result if board.open_polls >= LIMITS.max_open_polls: raise ApiError(429, "too_many_polls", "too many open long-polls", {"Retry-After": "1"}) board.open_polls += 1 board.counters["polls"] += 1 fut = board.notifier.register(ANY_CHANNEL) try: result = await db.read(store.activity, p, limit, after, follow) observe(result) if result["messages"] or board.shutting_down: return result deadline = time.monotonic() + wait while True: # any post wakes us; keep waiting until one is actually for this caller left = deadline - time.monotonic() if left <= 0: break done, _ = await asyncio.wait({fut}, timeout=left) if done and not board.shutting_down: # let a burst of posts land, then answer them with one query instead of one per post await asyncio.sleep(min(LIMITS.poll_coalesce_seconds * (0.75 + random.random() / 2), max(0.0, deadline - time.monotonic()))) if await request.is_disconnected(): return result board.notifier.unregister(ANY_CHANNEL, fut) fut = board.notifier.register(ANY_CHANNEL) p = await authn(request, observe=False) after = result["next_cursor"] # previous query exhausted the older history result = await db.read(store.activity, p, limit, after, follow, shed=False) board.presence.observed(p, token, bool(result["messages"]), time.time()) if result["messages"] or board.shutting_down: return result return result finally: board.notifier.unregister(ANY_CHANNEL, fut) board.open_polls -= 1 @app.get("/v1/search") async def search(request: Request, q: str = "", channel: int | None = None, sender: int | None = None, after: str | None = None, before: str | None = None, sort: str = "newest", limit: int = Query(20, ge=1, le=100), offset: int = Query(0, ge=0, le=10_000)): """Full-text search over messages you can read, plus matching channel and people names. `q` takes words, "phrases" and filters: from:@name in:#channel after:/before:/on:DATE has:file ...""" p = await authn(request) if len(q) > 500: raise bad_request("q is at most 500 characters") return await db.read(lambda conn: store.search(conn, p, q, channel_id=channel, sender_id=sender, after=after, before=before, sort=sort, limit=limit, offset=offset, now=time.time())) @app.get("/v1/analytics") async def get_analytics(request: Request, days: int = Query(7, ge=1, le=365), start: float | None = Query(None, alias="from"), end: float | None = Query(None, alias="to"), top_k: int = Query(10, ge=1, le=100)): p = await authn(request) result = await db.read(store.analytics, p, time.time(), days, start, end, top_k, board.presence.pending_history()) result["listening"]["now"] = board.presence.listening_now(time.time()) result["health"] = health_summary() return result def _origin(request: Request) -> str: proto = request.headers.get("x-forwarded-proto") or request.url.scheme host = request.headers.get("x-forwarded-host") or request.headers.get("host") or request.url.netloc return f"{proto}://{host}{request.scope.get('root_path', '')}" @app.get("/agents", include_in_schema=False) async def agents_guide(request: Request): return PlainTextResponse(guide.render(_origin(request), public=True), media_type="text/markdown; charset=utf-8") @app.get("/v1/guide") async def get_guide(request: Request): await authn(request) settings = await db.read(store.get_settings) return PlainTextResponse(guide.render(_origin(request), settings["guidelines"], docs=settings["channel_docs"]), media_type="text/markdown; charset=utf-8") @app.get("/v1/channels/{channel_id}/members") async def list_members(channel_id: int, request: Request): p = await authn(request) return {"members": await with_presence(p, await db.read(store.list_members, p, channel_id))} @app.post("/v1/channels/{channel_id}/members") async def add_member(channel_id: int, request: Request): p = await authn(request) body = body_dict(await _json_body(request)) members = await db.write(store.add_member, p, channel_id, body.get("principal_id"), time.time()) return {"members": await with_presence(p, members)} @app.delete("/v1/channels/{channel_id}/members/{principal_id}") async def remove_member(channel_id: int, principal_id: int, request: Request): p = await authn(request) result = await db.write(store.remove_member, p, channel_id, principal_id, time.time()) return result if isinstance(result, dict) else {"members": await with_presence(p, result)} @app.post("/v1/dms") async def open_dm(request: Request): p = await authn(request) body = body_dict(await _json_body(request)) channel, created = await db.write(store.open_dm, p, body.get("with"), time.time()) return JSONResponse(channel, status_code=201 if created else 200) # --------------------------------------------------------- messages @app.post("/v1/channels/{channel_id}/messages") async def post_message(channel_id: int, request: Request): p = await authn(request) request_id = request.headers.get("idempotency-key") if not request_id or not store.CLIENT_ID.match(request_id): raise bad_request("Idempotency-Key header is required and must match " + store.CLIENT_ID.pattern) payload = await _json_body(request, LIMITS.max_body_bytes * 2 + 64 * 1024) normalized, digest = store.normalize_payload(channel_id, payload) message, created = await db.write( store.post_message, p, channel_id, request_id, normalized, digest, time.time()) if created: board.counters["posts"] += 1 board.notifier.notify(channel_id) board.notifier.notify(ANY_CHANNEL) return JSONResponse(message, status_code=201) board.counters["posts_replayed"] += 1 return JSONResponse(message, status_code=200, headers={"Idempotent-Replayed": "true"}) @app.patch("/v1/channels/{channel_id}/messages/{message_id}") async def edit_message(channel_id: int, message_id: int, request: Request): p = await authn(request) payload = await _json_body(request, LIMITS.max_body_bytes * 2 + 1024) message, changed = await db.write(store.edit_message, p, channel_id, message_id, payload, time.time()) if changed: board.counters["edits"] += 1 board.notifier.notify(channel_id) return message @app.put("/v1/channels/{channel_id}/messages/{message_id}/reactions/{emoji}") async def add_reaction(channel_id: int, message_id: int, emoji: str, request: Request): p = await authn(request) message, changed = await db.write(store.react, p, channel_id, message_id, unquote(emoji), True, time.time()) if changed: board.notifier.notify(channel_id) return message @app.delete("/v1/channels/{channel_id}/messages/{message_id}/reactions/{emoji}") async def remove_reaction(channel_id: int, message_id: int, emoji: str, request: Request): p = await authn(request) message, changed = await db.write(store.react, p, channel_id, message_id, unquote(emoji), False, time.time()) if changed: board.notifier.notify(channel_id) return message @app.delete("/v1/channels/{channel_id}/messages/{message_id}") async def delete_message(channel_id: int, message_id: int, request: Request): p = await authn(request) result = await db.write(store.delete_message, p, channel_id, message_id, time.time()) for key in result.pop("storage_keys"): try: board.blobs.path_for(key).unlink() except FileNotFoundError: pass board.notifier.notify(channel_id) return result @app.get("/v1/channels/{channel_id}/messages/{message_id}/revisions") async def message_revisions(channel_id: int, message_id: int, request: Request): p = await authn(request) return await db.read(store.message_revisions, p, channel_id, message_id) @app.get("/v1/channels/{channel_id}/messages") async def read_messages( channel_id: int, request: Request, after: int = Query(0, ge=0), limit: int = Query(LIMITS.default_page_size, ge=1, le=LIMITS.max_page_size), wait: float = Query(0, ge=0, le=LIMITS.max_wait_seconds), before: int | None = Query(None, ge=1), top_level: bool = False, thread: int | None = Query(None, ge=1), edits_after: int | None = Query(None, ge=0), ): p = await authn(request) if before is not None or top_level or thread is not None: if wait > 0 or edits_after is not None: raise bad_request("wait and edits_after only work with plain `after` reads") return await db.read(store.read_messages, p, channel_id, after, limit, before, top_level, thread) async def page(p): result = await db.read(store.read_messages, p, channel_id, after, limit) if edits_after is not None: # also hand back messages edited since this edit cursor result.update(await db.read(store.read_edits, p, channel_id, edits_after, limit)) return result if wait <= 0: return await page(p) if board.open_polls >= LIMITS.max_open_polls: raise ApiError(429, "too_many_polls", "too many open long-polls", {"Retry-After": "1"}) board.open_polls += 1 board.counters["polls"] += 1 # Register before checking history so a commit between the query and # the wait cannot be missed. fut = board.notifier.register(channel_id) try: result = await page(p) if result["messages"] or result.get("edits") or board.shutting_down: return result await asyncio.wait({fut}, timeout=wait) # Re-authenticate and re-authorize: the token may have been revoked # or the grant removed while this request was held. p = await authn(request) return await page(p) finally: board.notifier.unregister(channel_id, fut) board.open_polls -= 1 # ------------------------------------------------- profiles, invites @app.get("/v1/principals/{principal_id}/profile") async def get_profile(principal_id: int, request: Request): p = await authn(request) result = await db.read(store.profile, p, principal_id, time.time()) result = (await with_presence(p, [result]))[0] result["agents"] = await with_presence(p, result["agents"]) return result @app.put("/v1/principals/{principal_id}/avatar") async def put_avatar(principal_id: int, request: Request): p = await authn(request) data = bytearray() async for chunk in request.stream(): data += chunk if len(data) > LIMITS.max_avatar_bytes: raise ApiError(413, "too_large", f"pictures are limited to {LIMITS.max_avatar_bytes // 1024} KB") return await db.write(store.set_avatar, p, principal_id, bytes(data), time.time()) @app.delete("/v1/principals/{principal_id}/avatar") async def delete_avatar(principal_id: int, request: Request): p = await authn(request) return await db.write(store.delete_avatar, p, principal_id, time.time()) @app.put("/v1/principals/{principal_id}/pattern") async def put_pattern(principal_id: int, request: Request): p = await authn(request) body = await _json_body(request, 1024) seed = body.get("seed") if isinstance(body, dict) else None return await db.write(store.set_pattern, p, principal_id, seed, time.time()) @app.get("/v1/principals/{principal_id}/avatar") async def get_avatar(principal_id: int, request: Request): await authn(request) # any member of the board may see pictures ctype, data, digest = await db.read(store.get_avatar, principal_id) return Response(data, media_type=ctype, headers={ "ETag": f'"{digest[:16]}"', "Cache-Control": "private, max-age=86400", "X-Content-Type-Options": "nosniff", "Content-Security-Policy": "default-src 'none'"}) # ------------------------------------------------------ attachments @app.put("/v1/channels/{channel_id}/attachments/{upload_id}") async def upload(channel_id: int, upload_id: str, request: Request): p = await authn(request) declared = request.headers.get("content-length") if declared and declared.isdigit() and int(declared) > LIMITS.max_upload_bytes: raise ApiError(413, "too_large", f"uploads are limited to {LIMITS.max_upload_bytes} bytes") if board.open_uploads >= LIMITS.max_concurrent_uploads: raise ApiError(429, "too_many_uploads", "too many concurrent uploads", {"Retry-After": "1"}) # Browsers can only send ASCII headers, so the filename may be percent-encoded. filename = os.path.basename(unquote(request.headers.get("x-filename", "")))[:255] or None if filename and any(ord(ch) < 32 for ch in filename): raise bad_request("invalid filename") content_type = request.headers.get("content-type", "application/octet-stream")[:128] board.open_uploads += 1 tmp_path = None try: row = await db.write(store.begin_upload, p, channel_id, upload_id, filename, content_type, time.time()) tmp_path, fh = await anyio.to_thread.run_sync(board.blobs.open_temp) hasher, size = new_hasher(), 0 try: async for chunk in request.stream(): size += len(chunk) if size > LIMITS.max_upload_bytes: raise ApiError(413, "too_large", f"uploads are limited to {LIMITS.max_upload_bytes} bytes") hasher.update(chunk) await anyio.to_thread.run_sync(fh.write, chunk) await anyio.to_thread.run_sync(board.blobs.finish_temp, fh) finally: if not fh.closed: fh.close() digest = hasher.hexdigest() expected = request.headers.get("x-checksum-sha256") if expected and expected.lower() != digest: raise bad_request("X-Checksum-Sha256 does not match the uploaded content") async with board.commit_lock: current = await db.read(store.attachment_state, row["id"]) if current["state"] == "ready": result = await db.write(store.finish_upload, row["id"], size, digest) return JSONResponse(result, status_code=200) await anyio.to_thread.run_sync(board.blobs.commit, tmp_path, row["storage_key"]) tmp_path = None result = await db.write(store.finish_upload, row["id"], size, digest) return JSONResponse(result, status_code=201) finally: board.open_uploads -= 1 if tmp_path is not None: board.blobs.discard(tmp_path) @app.get("/v1/attachments/{att_id}/info") async def attachment_info(att_id: str, request: Request): p = await authn(request) info = await db.read(store.get_attachment, p, att_id) info.pop("storage_key") return info @app.get("/v1/attachments/{att_id}") async def download(att_id: str, request: Request): p = await authn(request) info = await db.read(store.get_attachment, p, att_id) return FileResponse( board.blobs.path_for(info["storage_key"]), media_type="application/octet-stream", filename=info["filename"] or att_id, headers={ "X-Checksum-Sha256": info["sha256"], "X-Content-Type-Options": "nosniff", "X-Upload-Content-Type": info["content_type"] or "application/octet-stream", }, ) # ------------------------------------------------------------------ storage folders def _cache_put(src: Path, sha: str) -> Path: dest = board.storage_cache / sha os.replace(src, dest) files = sorted(board.storage_cache.iterdir(), key=lambda f: f.stat().st_atime) total = sum(f.stat().st_size for f in files) for f in files: # keep the cache under its size, oldest first if total <= LIMITS.storage_cache_bytes or f == dest: continue total -= f.stat().st_size f.unlink(missing_ok=True) return dest @app.put("/v1/storage/{handle}/{path:path}") async def storage_put(handle: str, path: str, request: Request): p = await authn(request) owner = await db.read(store.storage_owner, p, handle, True) path = store.storage_path(path) declared = request.headers.get("content-length") if declared and declared.isdigit(): await db.read(store.storage_check_quota, owner["id"], path, int(declared)) content_type = (request.headers.get("content-type") or mimetypes.guess_type(path)[0] or "application/octet-stream")[:128] tmp_path, fh = await anyio.to_thread.run_sync(board.blobs.open_temp) try: hasher, size = new_hasher(), 0 try: max_file = (await db.read(store.storage_limits))[0] async for chunk in request.stream(): size += len(chunk) if size > max_file: raise ApiError(413, "too_large", f"files are limited to {store._mb(max_file)} on this board") hasher.update(chunk) await anyio.to_thread.run_sync(fh.write, chunk) await anyio.to_thread.run_sync(board.blobs.finish_temp, fh) finally: if not fh.closed: fh.close() digest = hasher.hexdigest() expected = request.headers.get("x-checksum-sha256") if expected and expected.lower() != digest: raise bad_request("X-Checksum-Sha256 does not match the uploaded content") task = asyncio.create_task(publish_storage(p, owner, path, size, digest, content_type, tmp_path)) try: result = await asyncio.shield(task) except asyncio.CancelledError: with suppress(Exception): await task raise tmp_path = None return JSONResponse(result, status_code=201) finally: if tmp_path is not None: board.blobs.discard(tmp_path) async def publish_storage(p, owner, path, size, digest, content_type, tmp_path): async with board.storage_reservations.hold([(owner["id"], path)]): await db.read(store.storage_check_quota, owner["id"], path, size) key = store.storage_key(owner["id"], path) # Replacing a file needs a rollback copy in case the final quota # check loses to an unrelated upload. This I/O holds only its path. try: await db.read(store.storage_lookup, p, owner["external_id"], path) replacing = True except ApiError as exc: if exc.status != 404: raise replacing = False with tempfile.TemporaryDirectory(prefix="msgboard-replace-") as scratch: backup = Path(scratch) / "old" backed_up = replacing and await anyio.to_thread.run_sync(board.storage.get, key, backup) changed = recorded = False try: changed = True await anyio.to_thread.run_sync(lambda: board.storage.put([(tmp_path, key)])) await anyio.to_thread.run_sync(_cache_put, tmp_path, digest) result = await db.write(store.storage_record_checked, p, owner["id"], path, size, digest, content_type, time.time()) recorded = True return result finally: if changed and not recorded: if backed_up: await anyio.to_thread.run_sync(lambda: board.storage.put([(backup, key)])) else: await anyio.to_thread.run_sync(board.storage.delete, [key]) @app.get("/v1/storage/{handle}") async def storage_root(handle: str, request: Request, prefix: str = ""): p = await authn(request) return await db.read(store.storage_list, p, handle, prefix) @app.get("/v1/storage/{handle}/{path:path}") async def storage_get(handle: str, path: str, request: Request, info: bool = False): p = await authn(request) if path.endswith("/"): # a folder: list it return await db.read(store.storage_list, p, handle, path) meta, key = await db.read(store.storage_lookup, p, handle, path) if info: return meta cached = board.storage_cache / meta["sha256"] if not cached.exists(): owner = await db.read(store.storage_owner, p, handle, False) async with board.storage_reservations.hold([(owner["id"], store.storage_path(path))]): # A replacement may have completed while we waited. Read its # current metadata, never old metadata with newly uploaded bytes. meta, key = await db.read(store.storage_lookup, p, handle, path) cached = board.storage_cache / meta["sha256"] if not cached.exists(): fd, name = tempfile.mkstemp(prefix=".storage-", suffix=".part", dir=board.storage_cache) os.close(fd) tmp = Path(name) try: ok = await anyio.to_thread.run_sync(board.storage.get, key, tmp) if not ok: raise ApiError(503, "storage_unavailable", "the file is listed but could not be fetched from storage; try again") # the cache is named by the recorded hash, so only bytes with that hash may go in it: # a storage object overwritten behind the database (e.g. a mismatched restore) is never served h = new_hasher() with tmp.open("rb") as fh: for chunk in iter(lambda: fh.read(1 << 20), b""): h.update(chunk) if h.hexdigest() != meta["sha256"]: raise ApiError(409, "storage_mismatch", "the stored bytes do not match this file's recorded checksum") os.replace(tmp, cached) finally: tmp.unlink(missing_ok=True) # always a download: never let stored HTML or SVG run on the board's own origin return FileResponse(cached, media_type="application/octet-stream", filename=os.path.basename(meta["path"]), headers={"X-Checksum-Sha256": meta["sha256"], "X-Content-Type-Options": "nosniff", "X-Upload-Content-Type": meta["content_type"] or "application/octet-stream", "Content-Security-Policy": "sandbox", "Cache-Control": "private, no-cache"}) @app.delete("/v1/storage/{handle}/{path:path}") async def storage_delete(handle: str, path: str, request: Request): p = await authn(request) owner = await db.read(store.storage_owner, p, handle, True) path = store.storage_path(path) async with board.storage_reservations.hold([(owner["id"], path)]): out = await db.write(store.storage_remove, p, handle, path, time.time()) keys = out.pop("keys") await anyio.to_thread.run_sync(lambda: board.storage.delete(keys)) return out if dashboard: page = Path(__file__).with_name("dashboard.html") @app.get("/", include_in_schema=False) async def dashboard_page(): return FileResponse(page, media_type="text/html", headers={"Cache-Control": "no-cache", "X-Content-Type-Options": "nosniff"}) return app