src/api/routes/news.py

from fastapi import APIRouter, Query, HTTPException, Depends, Response
from typing import Optional
from src.core.ports.vector_store_port import VectorStorePort
from src.core.ports.embedder_port import EmbedderPort
from src.core.use_cases.analytics_use_case import AnalyticsUseCase
from src.api.dependencies import get_vector_store_port, get_embedder_port, get_analytics_use_case
from src.core.domain.schemas import BrowseResponse, SearchResponse, NewsArticle

router = APIRouter()

def _dict_to_article(payload: dict, score: float = None) -> NewsArticle:
    if payload is None:
        payload = {}
    
    # Safely get metadata
    metadata = payload.get("metadata") if payload else {}
    if metadata is None:
        metadata = {}

    # Title can be stored at top-level payload OR nested inside metadata
    title = (
        payload.get("title")
        or (metadata.get("title") if isinstance(metadata, dict) else None)
    )
        
    return NewsArticle(
        doc_id=payload.get("doc_id", "unknown"),
        url=payload.get("url"),
        title=title,
        content=payload.get("text", payload.get("content", "")),
        source=payload.get("source"),
        published_at=payload.get("published_at"),
        score=score,
        metadata=metadata if isinstance(metadata, dict) else {}
    )

@router.get("/latest", response_model=BrowseResponse)
def get_latest_news(
    response: Response,
    limit: int = Query(10, le=50),
    source: Optional[str] = None,
    language: Optional[str] = None,
    days_back: int = Query(7, ge=1, le=30, description="Number of days to look back (default: 7)"),
    vector_store: VectorStorePort = Depends(get_vector_store_port)
):
    """Get latest news articles sorted by publication date (default: last 7 days)"""
    # Prevent caching of news results - always fetch fresh data
    response.headers["Cache-Control"] = "no-cache, no-store, must-revalidate, max-age=0"
    response.headers["Pragma"] = "no-cache"
    response.headers["Expires"] = "0"
    
    try:
        result = vector_store.browse(
            limit=limit, 
            offset=0, 
            source=source, 
            language=language,
            days_back=days_back
        )
        
        # Convert Qdrant points to articles (already sorted by Qdrant)
        articles = []
        for p in result["articles"]:
            article = _dict_to_article(p.payload or {}, getattr(p, "score", None))
            articles.append(article)
        
        # Ensure next_offset is an integer or None
        next_offset = result.get("next_offset")
        if next_offset is not None and not isinstance(next_offset, int):
            next_offset = None  # If it's not an int, set to None
        
        return BrowseResponse(
            total_returned=len(articles),
            articles=articles,
            next_offset=next_offset
        )
    except Exception as e:
        # Return empty response on error instead of 500
        print(f"Error fetching news: {e}")
        return BrowseResponse(
            total_returned=0,
            articles=[],
            next_offset=None
        )

@router.get("/browse", response_model=BrowseResponse)
def browse_news(
    limit: int = Query(20, le=100), 
    offset: int = 0, 
    source: Optional[str] = None, 
    language: Optional[str] = None,
    vector_store: VectorStorePort = Depends(get_vector_store_port)
):
    result = vector_store.browse(limit=limit, offset=offset, source=source, language=language)
    
    # Qdrant scroll returns points with .payload
    articles = []
    for p in result["articles"]:
        articles.append(_dict_to_article(p.payload or {}, getattr(p, "score", None)))
    
    return BrowseResponse(
        total_returned=len(articles),
        articles=articles,
        next_offset=result["next_offset"]
    )

@router.get("/search", response_model=SearchResponse)
def search_news(
    q: str = Query(..., min_length=1),
    top_k: int = Query(10, le=50),
    source: Optional[str] = None,
    language: Optional[str] = None,
    embedder: EmbedderPort = Depends(get_embedder_port),
    vector_store: VectorStorePort = Depends(get_vector_store_port)
):
    try:
        query_vector = embedder.encode_query(q)
        results = vector_store.search(
            query_vectors=query_vector, 
            limit=top_k, 
            source_filter=source, 
            language_filter=language
        )
        # VectorStorePort.search returns SearchResult objects (content, metadata, score, doc_id)
        articles = []
        for r in results:
            articles.append(_dict_to_article(r.metadata, r.score))
            
        return SearchResponse(results=articles)
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

@router.get("/sources")
def get_sources(analytics: AnalyticsUseCase = Depends(get_analytics_use_case)):
    query = "SELECT source, count() as article_count FROM sentiment_results GROUP BY source ORDER BY article_count DESC"
    res = analytics.execute_raw_query(query)
    if res and res.get("rows"):
         sources = [row[0] for row in res["rows"] if row[0]]
         return {"sources": sources}
    return {"sources": []}

@router.get("/{doc_id}", response_model=NewsArticle)
def get_news_article(doc_id: str, vector_store: VectorStorePort = Depends(get_vector_store_port)):
    result = vector_store.get_by_doc_id(doc_id)
    if not result:
        raise HTTPException(status_code=404, detail="Article not found")
    
    return _dict_to_article(result.metadata, result.score)