vynl/backend/app/api/endpoints/timeline.py

import json
import logging

import anthropic
from fastapi import APIRouter, Depends, HTTPException
from pydantic import BaseModel
from sqlalchemy import select
from sqlalchemy.ext.asyncio import AsyncSession

from app.core.config import settings
from app.core.database import get_db
from app.core.security import get_current_user
from app.models.user import User
from app.models.playlist import Playlist
from app.models.track import Track
from app.models.recommendation import Recommendation

logger = logging.getLogger(__name__)

router = APIRouter(prefix="/profile", tags=["profile"])


class DecadeData(BaseModel):
    decade: str
    artists: list[str]
    count: int
    percentage: float


class TimelineResponse(BaseModel):
    decades: list[DecadeData]
    total_artists: int
    dominant_era: str
    insight: str


@router.get("/timeline", response_model=TimelineResponse)
async def get_timeline(
    user: User = Depends(get_current_user),
    db: AsyncSession = Depends(get_db),
):
    """Build a music timeline showing which eras/decades define the user's taste."""

    # Get all tracks from user's playlists
    result = await db.execute(
        select(Playlist).where(Playlist.user_id == user.id)
    )
    playlists = list(result.scalars().all())

    all_artists: set[str] = set()
    for p in playlists:
        result = await db.execute(select(Track).where(Track.playlist_id == p.id))
        tracks = result.scalars().all()
        for t in tracks:
            if t.artist:
                all_artists.add(t.artist)

    # Get artists from saved recommendations
    result = await db.execute(
        select(Recommendation).where(
            Recommendation.user_id == user.id,
            Recommendation.saved == True,  # noqa: E712
        )
    )
    saved_recs = result.scalars().all()
    for r in saved_recs:
        if r.artist:
            all_artists.add(r.artist)

    if not all_artists:
        raise HTTPException(
            status_code=404,
            detail="No artists found. Import some playlists first.",
        )

    # Cap at 50 artists for the Claude call
    artist_list = sorted(all_artists)[:50]

    # Call Claude once to categorize all artists by era
    client = anthropic.AsyncAnthropic(api_key=settings.ANTHROPIC_API_KEY)

    prompt = f"""Categorize these artists by their primary era/decade. For each artist, pick the decade they are MOST associated with (when they were most active/influential).

Artists: {', '.join(artist_list)}

Respond with a JSON object with two keys:
1. "decades" - keys are decade strings, values are lists of artists from the input:
{{
  "1960s": ["artist1"],
  "1970s": ["artist2"],
  "1980s": [],
  "1990s": ["artist3"],
  "2000s": ["artist4", "artist5"],
  "2010s": ["artist6"],
  "2020s": ["artist7"]
}}

2. "insight" - A single engaging sentence about their taste pattern across time, like "Your taste peaks in the 2000s indie explosion, with strong roots in 90s alternative." Make it specific to the actual artists and eras present.

Return ONLY a valid JSON object with "decades" and "insight" keys. No other text."""

    try:
        message = await client.messages.create(
            model="claude-sonnet-4-20250514",
            max_tokens=1024,
            messages=[{"role": "user", "content": prompt}],
        )

        response_text = message.content[0].text.strip()

        # Try to extract JSON if wrapped in markdown code blocks
        if response_text.startswith("```"):
            lines = response_text.split("\n")
            json_lines = []
            in_block = False
            for line in lines:
                if line.startswith("```") and not in_block:
                    in_block = True
                    continue
                elif line.startswith("```") and in_block:
                    break
                elif in_block:
                    json_lines.append(line)
            response_text = "\n".join(json_lines)

        parsed = json.loads(response_text)
        decades_data = parsed.get("decades", parsed)
        insight = parsed.get("insight", "")

    except (json.JSONDecodeError, KeyError, IndexError) as e:
        logger.error(f"Failed to parse Claude timeline response: {e}")
        raise HTTPException(
            status_code=500,
            detail="Failed to analyze your music timeline. Please try again.",
        )
    except anthropic.APIError as e:
        logger.error(f"Claude API error in timeline: {e}")
        raise HTTPException(
            status_code=502,
            detail="AI service unavailable. Please try again later.",
        )

    # Build the response
    total_categorized = 0
    decade_results: list[DecadeData] = []

    all_decades = ["1960s", "1970s", "1980s", "1990s", "2000s", "2010s", "2020s"]

    for decade in all_decades:
        artists = decades_data.get(decade, [])
        if isinstance(artists, list):
            total_categorized += len(artists)

    dominant_decade = ""
    max_count = 0

    for decade in all_decades:
        artists = decades_data.get(decade, [])
        if not isinstance(artists, list):
            artists = []
        count = len(artists)
        percentage = round((count / total_categorized * 100), 1) if total_categorized > 0 else 0.0

        if count > max_count:
            max_count = count
            dominant_decade = decade

        decade_results.append(
            DecadeData(
                decade=decade,
                artists=artists,
                count=count,
                percentage=percentage,
            )
        )

    if not insight:
        insight = f"Your music taste is centered around the {dominant_decade}."

    return TimelineResponse(
        decades=decade_results,
        total_artists=len(all_artists),
        dominant_era=dominant_decade,
        insight=insight,
    )