"""Taddy Podcast API client for finding podcast appearances"""
import asyncio
import re
from html import unescape
from datetime import datetime, timedelta
from typing import Dict, List, Optional
from web.backend.core.http_client import http_client
from modules.universal_logger import get_logger

logger = get_logger('Taddy')


def strip_html(text: str) -> str:
    """Strip HTML tags and decode entities from text"""
    if not text:
        return text
    # Remove HTML tags
    clean = re.sub(r'<[^>]+>', ' ', text)
    # Decode HTML entities
    clean = unescape(clean)
    # Normalize whitespace
    clean = re.sub(r'\s+', ' ', clean).strip()
    return clean


class TaddyClient:
    """Client for interacting with the Taddy Podcast API (GraphQL)

    Supports primary and fallback accounts for quota management.
    When the primary account fails (500 error / quota exceeded),
    automatically switches to the fallback account.
    """

    BASE_URL = "https://api.taddy.org"

    def __init__(self, user_id: str, api_key: str,
                 user_id_2: str = None, api_key_2: str = None):
        # Primary account
        self.user_id = user_id
        self.api_key = api_key

        # Fallback account (optional)
        self.user_id_2 = user_id_2
        self.api_key_2 = api_key_2
        self.has_fallback = bool(user_id_2 and api_key_2)

        # Track which account is active
        self.using_fallback = False

        self._update_headers()

    def _update_headers(self):
        """Update headers based on current active account"""
        if self.using_fallback and self.has_fallback:
            self.headers = {
                "Content-Type": "application/json",
                "X-USER-ID": self.user_id_2,
                "X-API-KEY": self.api_key_2
            }
        else:
            self.headers = {
                "Content-Type": "application/json",
                "X-USER-ID": self.user_id,
                "X-API-KEY": self.api_key
            }

    def _switch_to_fallback(self) -> bool:
        """Switch to fallback account if available. Returns True if switched."""
        if self.has_fallback and not self.using_fallback:
            self.using_fallback = True
            self._update_headers()
            logger.info("Switched to fallback Taddy account")
            return True
        return False

    async def _graphql_query(self, query: str, variables: Dict = None, retry_on_fallback: bool = True) -> Optional[Dict]:
        """Execute a GraphQL query against the Taddy API

        If the primary account fails with a 500 error (quota exceeded),
        automatically retries with the fallback account if configured.
        """
        try:
            payload = {"query": query}
            if variables:
                payload["variables"] = variables

            response = await http_client.post(
                self.BASE_URL,
                json=payload,
                headers=self.headers
            )

            data = response.json()

            if "errors" in data:
                logger.error(f"Taddy API error: {data['errors']}")
                return None

            return data.get("data")

        except Exception as e:
            error_str = str(e).lower()
            # Check for 500 error (quota exceeded) - http_client raises ServiceError
            if "500" in error_str or "server error" in error_str:
                account_type = "fallback" if self.using_fallback else "primary"
                logger.warning(f"Taddy API returned 500 on {account_type} account (likely quota exceeded)")

                # Try fallback if available and we haven't already
                if retry_on_fallback and self._switch_to_fallback():
                    logger.info("Retrying with fallback Taddy account...")
                    return await self._graphql_query(query, variables, retry_on_fallback=False)

            logger.error(f"Taddy API request failed: {e}")
            return None

    async def search_podcast_appearances(
        self,
        celebrity_name: str,
        lookback_days: int = 730,  # 2 years
        lookahead_days: int = 30,
        limit: int = 25,
        max_pages: int = 10
    ) -> List[Dict]:
        """
        Search for podcast episodes featuring a celebrity.

        Args:
            celebrity_name: Name of the celebrity to search for
            lookback_days: How many days back to search
            lookahead_days: How many days forward to search (for scheduled releases)
            limit: Maximum results per page

        Returns:
            List of podcast appearance dicts
        """
        appearances = []

        # Calculate date range
        now = datetime.now()
        start_date = now - timedelta(days=lookback_days)
        # Convert to Unix timestamp (seconds)
        start_timestamp = int(start_date.timestamp())

        query = """
        query SearchPodcastEpisodes($term: String!, $limitPerPage: Int, $page: Int, $filterForPublishedAfter: Int) {
            search(
                term: $term,
                filterForTypes: PODCASTEPISODE,
                matchBy: EXACT_PHRASE,
                limitPerPage: $limitPerPage,
                page: $page,
                filterForPublishedAfter: $filterForPublishedAfter
            ) {
                searchId
                podcastEpisodes {
                    uuid
                    name
                    description
                    datePublished
                    audioUrl
                    persons {
                        uuid
                        name
                        role
                    }
                    podcastSeries {
                        uuid
                        name
                        imageUrl
                    }
                    websiteUrl
                }
            }
        }
        """

        # Paginate through results (max 20 pages API limit, 25 per page = 500 max)
        # max_pages passed as parameter from config
        all_episodes = []

        for page in range(1, max_pages + 1):
            variables = {
                "term": celebrity_name,
                "limitPerPage": limit,
                "page": page,
                "filterForPublishedAfter": start_timestamp
            }

            data = await self._graphql_query(query, variables)

            if not data or not data.get("search"):
                break

            episodes = data["search"].get("podcastEpisodes", [])
            if not episodes:
                break  # No more results

            all_episodes.extend(episodes)

            # If we got fewer than limit, we've reached the end
            if len(episodes) < limit:
                break

            # Small delay between pages
            await asyncio.sleep(0.2)

        episodes = all_episodes

        for ep in episodes:
            try:
                # Parse the episode data
                podcast_series = ep.get("podcastSeries", {})
                ep_name = (ep.get("name") or "")
                podcast_name = (podcast_series.get("name") or "")
                name_lower = celebrity_name.lower()
                name_parts = name_lower.split()

                # ===== USE PERSONS METADATA FOR ACCURATE FILTERING =====
                # Check if celebrity is listed in the persons array with a role
                persons = ep.get("persons", []) or []
                person_match = None
                credit_type = None

                for person in persons:
                    person_name = (person.get("name") or "").lower()
                    # Match full name or last name
                    if name_lower in person_name or person_name in name_lower:
                        person_match = person
                        role = (person.get("role") or "").lower()
                        # Map Taddy roles to our credit types
                        if "host" in role:
                            credit_type = "host"
                        elif "guest" in role:
                            credit_type = "guest"
                        elif role:
                            credit_type = role  # Use whatever role they have
                        else:
                            credit_type = "guest"  # Default to guest if role not specified
                        break
                    # Also check by last name for partial matches
                    elif len(name_parts) >= 2:
                        last_name = name_parts[-1]
                        first_name = name_parts[0]
                        if len(last_name) >= 4 and (last_name in person_name or first_name in person_name):
                            person_match = person
                            role = (person.get("role") or "").lower()
                            if "host" in role:
                                credit_type = "host"
                            elif "guest" in role:
                                credit_type = "guest"
                            elif role:
                                credit_type = role
                            else:
                                credit_type = "guest"
                            break

                # If person is in the persons list, include the episode
                if person_match:
                    logger.debug(f"Accepting '{ep_name}' - {celebrity_name} listed as {credit_type} in persons metadata")
                    is_host = (credit_type == "host")
                else:
                    # Fallback: check if they're the host via podcast series name
                    podcast_name_lower = podcast_name.lower()
                    is_host = name_lower in podcast_name_lower
                    if not is_host and len(name_parts) >= 2:
                        last_name = name_parts[-1]
                        first_name = name_parts[0]
                        if len(last_name) >= 4:
                            is_host = (f"with {last_name}" in podcast_name_lower or
                                       f"with {first_name}" in podcast_name_lower or
                                       f"{first_name} {last_name}" in podcast_name_lower)

                    if is_host:
                        credit_type = "host"
                        logger.debug(f"Accepting '{ep_name}' - host podcast (name in series title)")
                    else:
                        # No persons metadata - use WHITELIST approach
                        # Only accept if title clearly indicates an interview/guest appearance
                        ep_name_lower = ep_name.lower()
                        if name_lower not in ep_name_lower:
                            logger.debug(f"Skipping '{ep_name}' - name not in title")
                            continue

                        # Check podcast name for news/gossip shows first
                        garbage_podcast_names = ['news', 'gossip', 'rumor', 'daily', 'trending', 'tmz', 'variety', 'march madness', 'cruz show', 'aesthetic arrest', 'devious maids']
                        if any(word in podcast_name_lower for word in garbage_podcast_names):
                            logger.debug(f"Skipping '{ep_name}' - podcast name suggests news/gossip")
                            continue

                        # Reject listicles (multiple comma-separated topics)
                        comma_count = ep_name_lower.count(',')
                        if comma_count >= 3:
                            logger.debug(f"Skipping '{ep_name}' - listicle format ({comma_count} commas)")
                            continue

                        # WHITELIST: Only accept if title matches clear interview patterns
                        interview_patterns = [
                            # Direct interview indicators
                            rf'(interview|interviews|interviewing)\s+(with\s+)?{re.escape(name_lower)}',
                            rf'{re.escape(name_lower)}\s+(interview|interviewed)',
                            # Guest indicators
                            rf'(guest|featuring|feat\.?|ft\.?|with guest|special guest)[:\s]+{re.escape(name_lower)}',
                            rf'{re.escape(name_lower)}\s+(joins|joined|stops by|sits down|talks|speaks|discusses|shares|reveals|opens up|gets real|gets honest)',
                            # "Name on Topic" format (common interview title)
                            rf'^{re.escape(name_lower)}\s+on\s+',
                            # Episode number + name format ("Ep 123: Name...")
                            rf'^(ep\.?|episode|#)\s*\d+[:\s]+{re.escape(name_lower)}',
                            # Name at start followed by colon or dash (interview format)
                            rf'^{re.escape(name_lower)}\s*[:\-–—]\s*',
                            # "Conversation with Name"
                            rf'(conversation|chat|talk|talking|speaking)\s+with\s+{re.escape(name_lower)}',
                            # "Name Returns" / "Name is Back"
                            rf'{re.escape(name_lower)}\s+(returns|is back|comes back)',
                            # Q&A format
                            rf'(q&a|q\s*&\s*a|ama)\s+(with\s+)?{re.escape(name_lower)}',
                            # Podcast-specific patterns
                            rf'{re.escape(name_lower)}\s+(live|in studio|in the studio|on the show|on the pod)',
                        ]

                        is_interview = False
                        for pattern in interview_patterns:
                            if re.search(pattern, ep_name_lower):
                                is_interview = True
                                logger.debug(f"Accepting '{ep_name}' - matches interview pattern")
                                break

                        if not is_interview:
                            logger.debug(f"Skipping '{ep_name}' - no interview pattern match (name just mentioned)")
                            continue

                        credit_type = "guest"

                # Get the artwork URL from podcast series
                artwork_url = podcast_series.get("imageUrl")

                # Parse date
                date_published = ep.get("datePublished")
                if date_published:
                    # Taddy returns Unix timestamp in seconds
                    try:
                        pub_date = datetime.fromtimestamp(date_published)
                        appearance_date = pub_date.strftime("%Y-%m-%d")
                        status = "upcoming" if pub_date.date() > now.date() else "aired"
                    except (ValueError, TypeError):
                        appearance_date = None
                        status = "aired"
                else:
                    appearance_date = None
                    status = "aired"

                # Get episode URL
                episode_url = ep.get("websiteUrl")

                appearance = {
                    "appearance_type": "Podcast",
                    "show_name": podcast_series.get("name", "Unknown Podcast"),
                    "episode_title": ep.get("name"),
                    "appearance_date": appearance_date,
                    "status": status,
                    "description": strip_html(ep.get("description")),
                    "poster_url": artwork_url,
                    "audio_url": ep.get("audioUrl"),
                    "url": episode_url,
                    "credit_type": credit_type or ("host" if is_host else "guest"),
                    "character_name": "Self",
                    "taddy_episode_uuid": ep.get("uuid"),
                    "taddy_podcast_uuid": podcast_series.get("uuid"),
                    "duration_seconds": None,  # Duration removed from query to reduce complexity
                }

                appearances.append(appearance)
                logger.info(f"Found podcast appearance: {celebrity_name} on '{podcast_series.get('name')}' - {ep.get('name')}")

            except Exception as e:
                logger.error(f"Error parsing Taddy episode: {e}")
                continue

        return appearances

    async def test_connection(self) -> bool:
        """Test if the API credentials are valid"""
        query = """
        query TestConnection {
            search(term: "test", filterForTypes: PODCASTSERIES, limitPerPage: 1) {
                searchId
            }
        }
        """

        data = await self._graphql_query(query)
        return data is not None