import datetime
import queue
import threading
import time
from typing import Optional, Collection

import dateutil.parser
import requests

from platypush.context import get_bus, get_plugin
from platypush.message.event.rss import NewFeedEntryEvent
from platypush.plugins import RunnablePlugin, action
from platypush.schemas.rss import RssFeedEntrySchema


class RssPlugin(RunnablePlugin):
    """
    A plugin for parsing and subscribing to RSS feeds.

    Triggers:

        - :class:`platypush.message.event.rss.NewFeedEntryEvent` when a new entry is received on a subscribed feed.

    Requires:

        * **feedparser** (``pip install feedparser``)

    """

    user_agent = 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) ' + \
                 'Chrome/62.0.3202.94 Safari/537.36'

    def __init__(
            self, subscriptions: Optional[Collection[str]] = None, poll_seconds: int = 300,
            user_agent: str = user_agent, **kwargs
    ):
        """
        :param subscriptions: List of feeds to monitor for updates, as URLs.
        :param poll_seconds: How often we should check for updates (default: 300 seconds).
        :param user_agent: Custom user agent to use for the requests.
        """
        super().__init__(**kwargs)
        self.subscriptions = subscriptions or []
        self.poll_seconds = poll_seconds
        self.user_agent = user_agent
        self._latest_timestamps = self._get_latest_timestamps()
        self._feed_worker_queues = [queue.Queue()] * 5
        self._feed_response_queue = queue.Queue()
        self._feed_workers = []
        self._latest_entries = []

    @staticmethod
    def _get_feed_latest_timestamp_varname(url: str) -> str:
        return f'LATEST_FEED_TIMESTAMP[{url}]'

    @classmethod
    def _get_feed_latest_timestamp(cls, url: str) -> Optional[datetime.datetime]:
        t = get_plugin('variable').get(
            cls._get_feed_latest_timestamp_varname(url)
        ).output.get(cls._get_feed_latest_timestamp_varname(url))

        if t:
            return dateutil.parser.isoparse(t)

    def _get_latest_timestamps(self) -> dict:
        return {
            url: self._get_feed_latest_timestamp(url)
            for url in self.subscriptions
        }

    def _update_latest_timestamps(self) -> None:
        variable = get_plugin('variable')
        variable.set(**{
            self._get_feed_latest_timestamp_varname(url): latest_timestamp
            for url, latest_timestamp in self._latest_timestamps.items()
        })

    @staticmethod
    def _parse_content(entry) -> Optional[str]:
        content = getattr(entry, 'content', None)
        if not content:
            return

        if isinstance(content, list):
            return content[0]['value']

        return content

    @action
    def parse_feed(self, url: str):
        """
        Parse a feed URL.

        :param url: Feed URL.
        :return: .. schema:: rss.RssFeedEntrySchema(many=True)
        """
        import feedparser

        feed = feedparser.parse(requests.get(url, headers={'User-Agent': self.user_agent}).text)
        return RssFeedEntrySchema().dump(
            sorted([
                {
                    'feed_url': url,
                    'feed_title': getattr(feed.feed, 'title', None),
                    'id': getattr(entry, 'id', None),
                    'url': entry.link,
                    'published': datetime.datetime.fromtimestamp(time.mktime(entry.published_parsed)),
                    'title': entry.title,
                    'summary': getattr(entry, 'summary', None),
                    'content': self._parse_content(entry),
                }
                for entry in feed.entries
                if getattr(entry, 'published_parsed', None)
            ], key=lambda e: e['published']),
            many=True
        )

    @action
    def get_latest_entries(self, limit: int = 20):
        """
        Get the latest entries from the subscribed feeds, sorted by descending published date.

        :param limit: Maximum number of entries to return (default: 20).
        :return: .. schema:: rss.RssFeedEntrySchema(many=True)
        """
        return sorted(self._latest_entries, key=lambda e: e['published'], reverse=True)[:limit]

    def _feed_worker(self, q: queue.Queue):
        while not self.should_stop():
            try:
                url = q.get(block=True, timeout=1)
            except queue.Empty:
                continue

            try:
                self._feed_response_queue.put({
                    'url': url,
                    'content': self.parse_feed(url).output,
                })
            except Exception as e:
                self._feed_response_queue.put({
                    'url': url,
                    'error': e,
                })

        self._feed_response_queue.put(None)

    def main(self):
        self._feed_workers = [
            threading.Thread(target=self._feed_worker, args=(q,))
            for q in self._feed_worker_queues
        ]

        for worker in self._feed_workers:
            worker.start()

        self.logger.info(f'Initialized RSS plugin with {len(self.subscriptions)} subscriptions')

        while not self.should_stop():
            responses = {}
            for i, url in enumerate(self.subscriptions):
                worker_queue = self._feed_worker_queues[i % len(self._feed_worker_queues)]
                worker_queue.put(url)

            time_start = time.time()
            timeout = 60
            max_time = time_start + timeout
            new_entries = []

            while (
                    not self.should_stop() and
                    len(responses) < len(self.subscriptions) and
                    time.time() - time_start <= timeout
            ):
                try:
                    response = self._feed_response_queue.get(block=True, timeout=max_time-time_start)
                except queue.Empty:
                    self.logger.warning('RSS parse timeout')
                    break

                if not response:
                    continue

                url = response['url']
                error = response.get('error')
                if error:
                    self.logger.error(f'Could not parse feed {url}: {error}')
                    responses[url] = error
                else:
                    responses[url] = response['content']

            responses = {k: v for k, v in responses.items() if not isinstance(v, Exception)}

            for url, response in responses.items():
                latest_timestamp = self._latest_timestamps.get(url)
                new_entries += response

                for entry in response:
                    published = datetime.datetime.fromisoformat(entry['published'])
                    if not latest_timestamp or published > latest_timestamp:
                        latest_timestamp = published
                        get_bus().post(NewFeedEntryEvent(**entry))

                self._latest_timestamps[url] = latest_timestamp

            self._update_latest_timestamps()
            self._latest_entries = new_entries
            time.sleep(self.poll_seconds)

    def stop(self):
        super().stop()
        for worker in self._feed_workers:
            worker.join(timeout=60)

        self.logger.info('RSS integration stopped')


# vim:sw=4:ts=4:et: