republisher/repub/entrypoint.py

212 lines
6.3 KiB
Python
Raw Normal View History

2026-03-29 13:52:23 +02:00
from __future__ import annotations
import argparse
2026-03-31 12:47:36 +02:00
import asyncio
import logging
2026-03-30 11:42:13 +02:00
import os
2026-03-31 12:47:36 +02:00
import signal
2026-03-29 13:52:23 +02:00
import sys
2026-03-31 12:47:36 +02:00
from contextlib import suppress
2026-05-27 13:04:47 +02:00
from pathlib import Path
2026-03-31 12:47:36 +02:00
from hypercorn.asyncio import serve as hypercorn_serve
from hypercorn.config import Config as HypercornConfig
2026-03-29 13:52:23 +02:00
2026-03-30 11:42:13 +02:00
import repub.crawl as crawl_module
2026-05-27 13:04:47 +02:00
from repub.cleanup import DEFAULT_MEDIA_DIRS, cleanup_media
from repub.config import (
AUDIO_DIR,
FILE_DIR,
IMAGE_DIR,
VIDEO_DIR,
build_base_settings,
load_config,
)
2026-03-31 12:47:36 +02:00
from repub.web import SHUTDOWN_EVENT_KEY, create_app
2026-03-30 11:42:13 +02:00
FeedNameFilter = crawl_module.FeedNameFilter
check_runtime = crawl_module.check_runtime
__all__ = ["FeedNameFilter", "check_runtime", "entrypoint", "parse_args"]
2024-04-18 11:57:24 +02:00
logger = logging.getLogger(__name__)
logger.setLevel(logging.DEBUG)
2026-03-29 13:52:23 +02:00
logger.propagate = False
if not logger.handlers:
handler = logging.StreamHandler()
handler.setLevel(logging.DEBUG)
handler.setFormatter(
logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
)
logger.addHandler(handler)
2024-04-18 11:57:24 +02:00
2026-03-30 11:42:13 +02:00
def parse_args(argv: list[str] | None = None) -> tuple[str, argparse.Namespace]:
raw_args = list(argv) if argv is not None else sys.argv[1:]
2024-04-18 11:57:24 +02:00
2026-03-30 11:42:13 +02:00
parser = argparse.ArgumentParser(description="Mirror RSS and Atom feeds")
subparsers = parser.add_subparsers(dest="command")
2024-04-18 11:57:24 +02:00
2026-03-30 11:42:13 +02:00
serve_parser = subparsers.add_parser("serve", help="Start the republisher web UI")
serve_parser.add_argument(
"--host",
2026-03-30 15:23:34 +02:00
default=os.environ.get("REPUBLISHER_HOST", "127.0.0.1"),
2026-03-30 11:42:13 +02:00
help="Host interface for the web UI",
)
serve_parser.add_argument(
"--port",
2026-03-30 15:23:34 +02:00
default=os.environ.get("REPUBLISHER_PORT", "8080"),
2026-03-30 11:42:13 +02:00
help="Port for the web UI",
)
2026-03-30 15:36:12 +02:00
serve_parser.add_argument(
"--dev-mode",
action="store_true",
help="Serve published feeds from /feeds for local development",
)
2024-04-18 11:57:24 +02:00
2026-03-30 11:42:13 +02:00
crawl_parser = subparsers.add_parser("crawl", help="Run the feed crawler once")
crawl_parser.add_argument(
2026-03-29 13:52:23 +02:00
"-c",
"--config",
default="repub.toml",
help="Path to runtime config TOML file",
)
2026-05-27 13:04:47 +02:00
cleanup_parser = subparsers.add_parser(
"cleanup-media",
help="Delete old unreferenced published media",
)
cleanup_parser.add_argument(
"-c",
"--config",
default=None,
help="Read output and media directory settings from runtime config TOML",
)
cleanup_parser.add_argument(
"--feeds-dir",
default=None,
help="Published feeds directory to clean (default: config out_dir/feeds or out/feeds)",
)
cleanup_parser.add_argument(
"--days",
type=int,
default=25,
help="Delete unreferenced media older than this many days",
)
cleanup_parser.add_argument(
"--dry-run",
action="store_true",
help="Report cleanup matches without deleting files",
)
2026-03-30 11:42:13 +02:00
if not raw_args:
2026-03-30 15:36:12 +02:00
raw_args = ["serve", "--dev-mode"]
2026-03-30 11:42:13 +02:00
elif raw_args[0] in {"-c", "--config"}:
raw_args = ["crawl", *raw_args]
2026-05-27 13:04:47 +02:00
elif raw_args[0] not in {"serve", "crawl", "cleanup-media"}:
2026-03-30 15:36:12 +02:00
raw_args = ["serve", "--dev-mode", *raw_args]
2026-03-29 13:52:23 +02:00
2026-03-30 11:42:13 +02:00
args = parser.parse_args(raw_args)
command = args.command or "serve"
return command, args
2026-03-29 13:52:23 +02:00
2026-05-27 13:04:47 +02:00
def _cleanup_config(args: argparse.Namespace) -> tuple[Path, tuple[str, ...]]:
feeds_dir = Path(args.feeds_dir) if args.feeds_dir else Path("out/feeds")
media_dirs = DEFAULT_MEDIA_DIRS
if args.config is None:
return feeds_dir, media_dirs
config = load_config(args.config)
settings = build_base_settings(config)
media_dirs = (
str(settings.get("REPUBLISHER_IMAGE_DIR", IMAGE_DIR)),
str(settings.get("REPUBLISHER_AUDIO_DIR", AUDIO_DIR)),
str(settings.get("REPUBLISHER_VIDEO_DIR", VIDEO_DIR)),
str(settings.get("REPUBLISHER_FILE_DIR", FILE_DIR)),
)
if args.feeds_dir is None:
feeds_dir = config.out_dir / "feeds"
return feeds_dir, media_dirs
2026-03-31 12:47:36 +02:00
def _install_signal_handlers(stop_event: asyncio.Event) -> None:
loop = asyncio.get_running_loop()
def request_stop(*_: object) -> None:
if not stop_event.is_set():
stop_event.set()
for signum in (signal.SIGINT, signal.SIGTERM):
try:
loop.add_signal_handler(signum, request_stop)
except NotImplementedError:
signal.signal(signum, request_stop)
async def _serve_app(*, host: str, port: int, dev_mode: bool) -> None:
stop_event = asyncio.Event()
_install_signal_handlers(stop_event)
app = create_app(dev_mode=dev_mode)
app.extensions[SHUTDOWN_EVENT_KEY] = stop_event
config = HypercornConfig()
config.bind = [f"{host}:{port}"]
config.use_reloader = False
config.accesslog = "-"
config.errorlog = "-"
async def shutdown_trigger() -> None:
await stop_event.wait()
try:
await hypercorn_serve(app, config, shutdown_trigger=shutdown_trigger)
finally:
stop_event.set()
2026-03-30 11:42:13 +02:00
def entrypoint(argv: list[str] | None = None) -> int:
command, args = parse_args(argv)
2026-03-29 13:52:23 +02:00
2026-03-30 11:42:13 +02:00
if command == "crawl":
crawl_module.check_runtime = check_runtime
return crawl_module.crawl_from_config(args.config)
2026-03-29 13:52:23 +02:00
2026-05-27 13:04:47 +02:00
if command == "cleanup-media":
try:
feeds_dir, media_dirs = _cleanup_config(args)
except FileNotFoundError as error:
missing_path = (
Path(error.filename).expanduser()
if error.filename
else Path(args.config).expanduser()
)
logger.error("Config file not found: %s", missing_path)
return 2
except ValueError as error:
logger.error("Invalid config: %s", error)
return 2
result = cleanup_media(
feeds_dir=feeds_dir,
retention_days=args.days,
dry_run=bool(args.dry_run),
media_dirs=media_dirs,
)
return 1 if result.failures else 0
2026-03-29 13:52:23 +02:00
try:
2026-03-30 11:42:13 +02:00
port = int(args.port)
except ValueError:
2026-03-30 15:23:34 +02:00
logger.error("Invalid REPUBLISHER_PORT/--port value: %s", args.port)
return 2
2026-03-29 13:52:23 +02:00
2026-03-31 12:47:36 +02:00
with suppress(KeyboardInterrupt):
asyncio.run(_serve_app(host=args.host, port=port, dev_mode=bool(args.dev_mode)))
2026-03-30 11:42:13 +02:00
return 0
2026-03-29 13:52:23 +02:00
if __name__ == "__main__":
sys.exit(entrypoint())