from __future__ import annotations import logging from datetime import datetime from threading import Event, Thread from typing import Any import diskcache import pytz from requests.exceptions import ConnectionError, ConnectTimeout, ReadTimeout from core.constants import HTTP_HEADERS from core.url_data_cache import URLDataCache from providers.callsigndata.callsign_data_provider import CallsignDataProvider logger = logging.getLogger(__name__) class FileDownloadCallsignDataProvider(CallsignDataProvider): """Generic callsign data provider class for providers that fetch their data from the web by downloading a file.""" def __init__( self, name: str, provider_config: dict[str, Any], url: str, cache_file_path: str, poll_interval: int, storage: diskcache.Cache, ) -> None: """Set up the provider, note poll_interval is in *days*.""" super().__init__(name, provider_config, storage) self._url = url self._cache_file_path = cache_file_path self._poll_interval = poll_interval self._thread: Thread | None = None self._stop_event = Event() self._url_data_cache = URLDataCache(f"callsigndata_{name}") if self.enabled: self.status = "Ready" def start(self) -> None: # Fire off the polling thread. It will poll immediately on startup, then sleep for poll_interval between # subsequent polls, so start() returns immediately and the application can continue starting. logger.info(f"Set up query of {self.name} callsign reference data every {self._poll_interval!s} days.") self._thread = Thread(target=self._run, name=f"FileDownloadCallsignDataProvider-{self.name}", daemon=True) self._thread.start() def stop(self) -> None: self._stop_event.set() if self._thread: self._thread.join(timeout=12) if self._thread.is_alive(): logger.warning(f"{self.name} callsign data worker thread did not exit on time and will be killed.") def _run(self) -> None: while True: self._poll() if self._stop_event.wait(timeout=self._poll_interval * 60 * 60 * 24): break def _poll(self) -> None: try: # Request the file. Use the data cache (with a TTL of 1 day) here, not as the main mechanism for # caching, but just so continual restarts of the software during testing don't hammer the servers. logger.debug(f"Downloading {self.name} callsign reference data...") http_response = self._url_data_cache.get(self._url, headers=HTTP_HEADERS) # Check response code was good if http_response.ok: # Save the data to a local file with open(self._cache_file_path, "wb") as f: f.write(http_response.content) f.flush() # Pass off to the subclass for processing ok = self._handle_file(self._cache_file_path) if ok: self.status = "OK" self.last_update_time = datetime.now(pytz.UTC) logger.info(f"Updated callsign reference data from {self.name}") else: self.status = "Error" logger.warning(f"Error updating callsign reference data from {self.name}.") else: self.status = "Error" logger.warning( f"HTTP {http_response.status_code} when downloading callsign reference data from {self.name}." ) except ConnectionError: self.status = "Error" logger.warning(f"Connection error when downloading callsign reference data from {self.name}.") except (ConnectTimeout, ReadTimeout): self.status = "Error" logger.warning(f"Timeout when downloading callsign reference data from {self.name}.") except Exception: self.status = "Error" logger.exception(f"Exception in callsign reference data provider ({self.name})") self._stop_event.wait(timeout=1) def _handle_file(self, path: str) -> bool: """Handle an updated file downloaded from the server. Return true if successful, false otherwise.""" raise NotImplementedError("Subclasses must implement this method")