"""Support for RSS/Atom feeds."""
from datetime import datetime, timedelta
from logging import getLogger
from os.path import exists
import pickle
from threading import Lock

import feedparser
import voluptuous as vol

from homeassistant.const import CONF_SCAN_INTERVAL, EVENT_HOMEASSISTANT_START
import homeassistant.helpers.config_validation as cv
from homeassistant.helpers.event import track_time_interval

_LOGGER = getLogger(__name__)

CONF_URLS = "urls"
CONF_MAX_ENTRIES = "max_entries"

DEFAULT_MAX_ENTRIES = 20
DEFAULT_SCAN_INTERVAL = timedelta(hours=1)

DOMAIN = "feedreader"

EVENT_FEEDREADER = "feedreader"

CONFIG_SCHEMA = vol.Schema(
    {
        DOMAIN: {
            vol.Required(CONF_URLS): vol.All(cv.ensure_list, [cv.url]),
            vol.Optional(
                CONF_SCAN_INTERVAL, default=DEFAULT_SCAN_INTERVAL
            ): cv.time_period,
            vol.Optional(
                CONF_MAX_ENTRIES, default=DEFAULT_MAX_ENTRIES
            ): cv.positive_int,
        }
    },
    extra=vol.ALLOW_EXTRA,
)


def setup(hass, config):
    """Set up the Feedreader component."""
    urls = config.get(DOMAIN)[CONF_URLS]
    scan_interval = config.get(DOMAIN).get(CONF_SCAN_INTERVAL)
    max_entries = config.get(DOMAIN).get(CONF_MAX_ENTRIES)
    data_file = hass.config.path(f"{DOMAIN}.pickle")
    storage = StoredData(data_file)
    feeds = [
        FeedManager(url, scan_interval, max_entries, hass, storage) for url in urls
    ]
    return len(feeds) > 0


class FeedManager:
    """Abstraction over Feedparser module."""

    def __init__(self, url, scan_interval, max_entries, hass, storage):
        """Initialize the FeedManager object, poll as per scan interval."""
        self._url = url
        self._scan_interval = scan_interval
        self._max_entries = max_entries
        self._feed = None
        self._hass = hass
        self._firstrun = True
        self._storage = storage
        self._last_entry_timestamp = None
        self._last_update_successful = False
        self._has_published_parsed = False
        self._event_type = EVENT_FEEDREADER
        self._feed_id = url
        hass.bus.listen_once(EVENT_HOMEASSISTANT_START, lambda _: self._update())
        self._init_regular_updates(hass)

    def _log_no_entries(self):
        """Send no entries log at debug level."""
        _LOGGER.debug("No new entries to be published in feed %s", self._url)

    def _init_regular_updates(self, hass):
        """Schedule regular updates at the top of the clock."""
        track_time_interval(hass, lambda now: self._update(), self._scan_interval)

    @property
    def last_update_successful(self):
        """Return True if the last feed update was successful."""
        return self._last_update_successful

    def _update(self):
        """Update the feed and publish new entries to the event bus."""
        _LOGGER.info("Fetching new data from feed %s", self._url)
        self._feed = feedparser.parse(
            self._url,
            etag=None if not self._feed else self._feed.get("etag"),
            modified=None if not self._feed else self._feed.get("modified"),
        )
        if not self._feed:
            _LOGGER.error("Error fetching feed data from %s", self._url)
            self._last_update_successful = False
        else:
            # The 'bozo' flag really only indicates that there was an issue
            # during the initial parsing of the XML, but it doesn't indicate
            # whether this is an unrecoverable error. In this case the
            # feedparser lib is trying a less strict parsing approach.
            # If an error is detected here, log error message but continue
            # processing the feed entries if present.
            if self._feed.bozo != 0:
                _LOGGER.error(
                    "Error parsing feed %s: %s", self._url, self._feed.bozo_exception
                )
            # Using etag and modified, if there's no new data available,
            # the entries list will be empty
            if self._feed.entries:
                _LOGGER.debug(
                    "%s entri(es) available in feed %s",
                    len(self._feed.entries),
                    self._url,
                )
                self._filter_entries()
                self._publish_new_entries()
                if self._has_published_parsed:
                    self._storage.put_timestamp(
                        self._feed_id, self._last_entry_timestamp
                    )
            else:
                self._log_no_entries()
            self._last_update_successful = True
        _LOGGER.info("Fetch from feed %s completed", self._url)

    def _filter_entries(self):
        """Filter the entries provided and return the ones to keep."""
        if len(self._feed.entries) > self._max_entries:
            _LOGGER.debug(
                "Processing only the first %s entries in feed %s",
                self._max_entries,
                self._url,
            )
            self._feed.entries = self._feed.entries[0 : self._max_entries]

    def _update_and_fire_entry(self, entry):
        """Update last_entry_timestamp and fire entry."""
        # Check if the entry has a published date.
        if "published_parsed" in entry.keys() and entry.published_parsed:
            # We are lucky, `published_parsed` data available, let's make use of
            # it to publish only new available entries since the last run
            self._has_published_parsed = True
            self._last_entry_timestamp = max(
                entry.published_parsed, self._last_entry_timestamp
            )
        else:
            self._has_published_parsed = False
            _LOGGER.debug("No published_parsed info available for entry %s", entry)
        entry.update({"feed_url": self._url})
        self._hass.bus.fire(self._event_type, entry)

    def _publish_new_entries(self):
        """Publish new entries to the event bus."""
        new_entries = False
        self._last_entry_timestamp = self._storage.get_timestamp(self._feed_id)
        if self._last_entry_timestamp:
            self._firstrun = False
        else:
            # Set last entry timestamp as epoch time if not available
            self._last_entry_timestamp = datetime.utcfromtimestamp(0).timetuple()
        for entry in self._feed.entries:
            if self._firstrun or (
                "published_parsed" in entry.keys()
                and entry.published_parsed > self._last_entry_timestamp
            ):
                self._update_and_fire_entry(entry)
                new_entries = True
            else:
                _LOGGER.debug("Entry %s already processed", entry)
        if not new_entries:
            self._log_no_entries()
        self._firstrun = False


class StoredData:
    """Abstraction over pickle data storage."""

    def __init__(self, data_file):
        """Initialize pickle data storage."""
        self._data_file = data_file
        self._lock = Lock()
        self._cache_outdated = True
        self._data = {}
        self._fetch_data()

    def _fetch_data(self):
        """Fetch data stored into pickle file."""
        if self._cache_outdated and exists(self._data_file):
            try:
                _LOGGER.debug("Fetching data from file %s", self._data_file)
                with self._lock, open(self._data_file, "rb") as myfile:
                    self._data = pickle.load(myfile) or {}
                    self._cache_outdated = False
            except:  # noqa: E722 pylint: disable=bare-except
                _LOGGER.error(
                    "Error loading data from pickled file %s", self._data_file
                )

    def get_timestamp(self, feed_id):
        """Return stored timestamp for given feed id (usually the url)."""
        self._fetch_data()
        return self._data.get(feed_id)

    def put_timestamp(self, feed_id, timestamp):
        """Update timestamp for given feed id (usually the url)."""
        self._fetch_data()
        with self._lock, open(self._data_file, "wb") as myfile:
            self._data.update({feed_id: timestamp})
            _LOGGER.debug(
                "Overwriting feed %s timestamp in storage file %s",
                feed_id,
                self._data_file,
            )
            try:
                pickle.dump(self._data, myfile)
            except:  # noqa: E722 pylint: disable=bare-except
                _LOGGER.error("Error saving pickled data to %s", self._data_file)
        self._cache_outdated = True