wrongmove/api/app.py

"""FastAPI application for the Real Estate Crawler API."""
from datetime import datetime, timedelta
import json
import logging
import logging.config
from typing import Annotated, AsyncGenerator, Optional
from api.auth import get_current_user
from api.config import DEV_TIER_ORIGINS, PROD_TIER_ORIGINS, APP_ENV
from api.decision_routes import decision_router
from api.passkey_routes import passkey_router
from api.poi_routes import poi_router
from api.ws_routes import ws_router
from api.rate_limit_config import RateLimitConfig
from api.rate_limiter import RateLimitMiddleware
from api.audit_middleware import AuditLogMiddleware
from api.metrics_guard import MetricsGuardMiddleware
from api.security_headers import SecurityHeadersMiddleware
from api.origin_validator import OriginValidatorMiddleware
from dotenv import load_dotenv
from fastapi import Depends, FastAPI, HTTPException, Query
from fastapi.responses import JSONResponse, StreamingResponse
from pydantic import BaseModel
from starlette.requests import Request
from api.auth import User
from models.listing import QueryParameters, ListingType, FurnishType
from notifications import send_notification
from repositories.listing_repository import ListingRepository
from database import engine
from fastapi.middleware.cors import CORSMiddleware
from ui_exporter import convert_to_geojson_feature, convert_row_to_geojson

from services import listing_service, export_service, district_service, task_service
from services import decision_service
from services.listing_cache import (
    get_cached_count,
    get_cached_features,
    begin_cache_population,
    cache_features_batch_staged,
    finalize_cache_population,
    delete_staging_key,
)
from repositories.poi_repository import POIRepository
from repositories.decision_repository import DecisionRepository
from repositories.user_repository import UserRepository
from opentelemetry.instrumentation.fastapi import FastAPIInstrumentor
from api.metrics import init_metrics, get_metrics_asgi_app
import api.metrics as app_metrics
from logging_config import configure_logging


load_dotenv()
configure_logging("api")
logger = logging.getLogger(__name__)

DEFAULT_BATCH_SIZE = 50
_rate_limit_config = RateLimitConfig.from_env()


def get_query_parameters(
    listing_type: ListingType,
    min_bedrooms: int = 1,
    max_bedrooms: int = 999,
    min_price: int = 0,
    max_price: int = 10_000_000,
    min_sqm: Optional[int] = None,
    max_sqm: Optional[int] = None,
    min_price_per_sqm: Optional[float] = None,
    max_price_per_sqm: Optional[float] = None,
    last_seen_days: Optional[int] = None,
    let_date_available_from: Optional[datetime] = None,
    furnish_types: Optional[str] = None,  # comma-separated list
    district_names: Optional[str] = None,  # comma-separated list
) -> QueryParameters:
    """Parse query parameters into QueryParameters model."""
    parsed_furnish_types = None
    if furnish_types:
        parsed_furnish_types = [FurnishType(f.strip()) for f in furnish_types.split(",")]

    parsed_district_names: set[str] = set()
    if district_names:
        parsed_district_names = {d.strip() for d in district_names.split(",") if d.strip()}

    return QueryParameters(
        listing_type=listing_type,
        min_bedrooms=min_bedrooms,
        max_bedrooms=max_bedrooms,
        min_price=min_price,
        max_price=max_price,
        min_sqm=min_sqm,
        max_sqm=max_sqm,
        min_price_per_sqm=min_price_per_sqm,
        max_price_per_sqm=max_price_per_sqm,
        last_seen_days=last_seen_days,
        let_date_available_from=let_date_available_from,
        furnish_types=parsed_furnish_types,
        district_names=parsed_district_names,
    )


app = FastAPI(
    docs_url=None if APP_ENV == "production" else "/docs",
    redoc_url=None if APP_ENV == "production" else "/redoc",
    openapi_url=None if APP_ENV == "production" else "/openapi.json",
)
app.include_router(passkey_router)
app.include_router(poi_router)
app.include_router(decision_router)
app.include_router(ws_router)
init_metrics("realestate-crawler-api")
app.mount("/metrics", get_metrics_asgi_app())


# Allow CORS (for React frontend)
app.add_middleware(
    CORSMiddleware,
    allow_origins=[*DEV_TIER_ORIGINS, *PROD_TIER_ORIGINS],
    allow_methods=["GET", "POST", "PUT", "DELETE"],
    allow_headers=["Authorization", "Content-Type"],
)

app.add_middleware(
    OriginValidatorMiddleware,
    allowed_origins=[*DEV_TIER_ORIGINS, *PROD_TIER_ORIGINS],
)

# Security middleware (added bottom-to-top; last added = outermost)
# 3. Rate limiting — enforces per-user limits
app.add_middleware(RateLimitMiddleware, config=_rate_limit_config)
# 2. Metrics guard — blocks unauthorized /metrics access
app.add_middleware(MetricsGuardMiddleware, config=_rate_limit_config)
# 1. Audit logging — logs everything including 429s and 403s
app.add_middleware(AuditLogMiddleware)
# 0. Security headers — adds standard security headers to all responses
app.add_middleware(SecurityHeadersMiddleware)


@app.exception_handler(Exception)
async def unhandled_exception_handler(request: Request, exc: Exception) -> JSONResponse:
    logger.exception("Unhandled exception")
    return JSONResponse(
        status_code=500,
        content={"detail": "Internal server error"},
    )


@app.get("/api/status")
async def get_status() -> dict[str, str | None]:
    repository = ListingRepository(engine)
    last_updated = repository.get_last_updated()
    return {
        "status": "OK",
        "last_updated": last_updated.isoformat() if last_updated else None,
    }


@app.get("/api/listing")
async def get_listing(
    user: Annotated[User, Depends(get_current_user)],
    limit: int = 5,
) -> dict[str, list]:
    """Get listings from the database."""
    limit = min(limit, _rate_limit_config.listing_limit_cap)
    repository = ListingRepository(engine)
    result = await listing_service.get_listings(repository, limit=limit)
    logger.info(f"Fetched {result.total_count} listings for {user.email}")
    return {"listings": result.listings}


@app.get("/api/listing_geojson")
async def get_listing_geojson(
    user: Annotated[User, Depends(get_current_user)],
    query_parameters: Annotated[QueryParameters, Depends(get_query_parameters)],
    limit: int | None = None,
) -> dict:
    """Get listings as GeoJSON for map display."""
    if limit is not None:
        limit = min(limit, _rate_limit_config.geojson_limit_cap)
    else:
        limit = _rate_limit_config.geojson_limit_cap
    repository = ListingRepository(engine)
    result = await export_service.export_to_geojson(
        repository,
        query_parameters=query_parameters,
        limit=limit,
    )
    return result.data



def _build_poi_distances_lookup(
    user_email: str,
    listing_type: ListingType,
) -> dict[int, list[dict[str, str | int]]] | None:
    """Build POI distance lookup for a user, or None if no POIs configured."""
    user_repo = UserRepository(engine)
    db_user = user_repo.get_user_by_email(user_email)
    if not db_user or db_user.id is None:
        return None

    poi_repo = POIRepository(engine)
    pois = {p.id: p for p in poi_repo.get_pois_for_user(db_user.id)}
    if not pois:
        return None

    listing_repo = ListingRepository(engine)
    all_ids = list(listing_repo.get_listing_ids(listing_type))
    if not all_ids:
        return None

    distances = poi_repo.get_distances_for_listings(all_ids, listing_type, db_user.id)
    lookup: dict[int, list[dict[str, str | int]]] = {}
    for d in distances:
        poi_name = pois[d.poi_id].name if d.poi_id in pois else "Unknown"
        lookup.setdefault(d.listing_id, []).append({
            "poi_id": d.poi_id,
            "poi_name": poi_name,
            "travel_mode": d.travel_mode,
            "duration_seconds": d.duration_seconds,
            "distance_meters": d.distance_meters,
        })
    return lookup


def _get_user_id_safe(user_email: str) -> int | None:
    """Get database user ID by email, or None if user doesn't exist."""
    try:
        user_repo = UserRepository(engine)
        db_user = user_repo.get_user_by_email(user_email)
        if db_user is None or db_user.id is None:
            return None
        return db_user.id
    except Exception:
        logger.debug("Could not look up user ID for %s", user_email)
        return None


def _should_include(
    feature_id: int,
    decision_filter: str,
    disliked_ids: set[int] | None,
    liked_ids: set[int] | None,
) -> bool:
    """Determine if a listing should be included based on decision filter."""
    if decision_filter == "everything":
        return True
    if decision_filter == "liked":
        return liked_ids is not None and feature_id in liked_ids
    # default "all": hide disliked
    return disliked_ids is None or feature_id not in disliked_ids


async def _stream_from_cache(
    query_parameters: QueryParameters,
    batch_size: int,
    limit: int | None,
    disliked_ids: set[int] | None = None,
    liked_ids: set[int] | None = None,
    decision_filter: str = "all",
) -> AsyncGenerator[str, None]:
    """Stream GeoJSON features from the Redis cache (cache-hit path)."""
    cached_count = get_cached_count(query_parameters)
    effective_total = min(limit, cached_count) if limit and cached_count else cached_count

    yield json.dumps({
        "type": "metadata",
        "batch_size": batch_size,
        "total_expected": effective_total,
        "cached": True,
    }) + "\n"

    count = 0
    for feature_batch in get_cached_features(query_parameters, batch_size=batch_size):
        # Apply decision filtering
        if decision_filter != "everything":
            feature_batch = [
                f for f in feature_batch
                if _should_include(
                    f.get("properties", {}).get("id", 0),
                    decision_filter,
                    disliked_ids,
                    liked_ids,
                )
            ]
        if limit and count + len(feature_batch) > limit:
            feature_batch = feature_batch[:limit - count]
        count += len(feature_batch)
        if feature_batch:
            yield json.dumps({"type": "batch", "features": feature_batch}) + "\n"
        if limit and count >= limit:
            break

    yield json.dumps({"type": "complete", "total": count}) + "\n"


async def _stream_from_db(
    query_parameters: QueryParameters,
    batch_size: int,
    limit: int | None,
    poi_distances_lookup: dict[int, list[dict[str, str | int]]] | None = None,
    skip_cache: bool = False,
    disliked_ids: set[int] | None = None,
    liked_ids: set[int] | None = None,
    decision_filter: str = "all",
) -> AsyncGenerator[str, None]:
    """Stream GeoJSON features from the database, populating the cache as we go."""
    repository = ListingRepository(engine)

    total = repository.count_listings(query_parameters)
    effective_total = min(limit, total) if limit else total

    yield json.dumps({
        "type": "metadata",
        "batch_size": batch_size,
        "total_expected": effective_total,
        "cached": False,
    }) + "\n"

    staging_key: str | None = None
    if not skip_cache:
        staging_key = begin_cache_population(query_parameters)

    try:
        count = 0
        batch: list[dict] = []
        for row in repository.stream_listings_optimized(
            query_parameters, limit=limit, page_size=batch_size
        ):
            feature = convert_row_to_geojson(row, query_parameters.listing_type.value)
            # Inject POI distances if available
            if poi_distances_lookup and row['id'] in poi_distances_lookup:
                feature['properties']['poi_distances'] = poi_distances_lookup[row['id']]

            # Apply decision filtering
            if not _should_include(row['id'], decision_filter, disliked_ids, liked_ids):
                # Still cache the feature (it's valid data), just don't stream it
                if staging_key:
                    cache_features_batch_staged(staging_key, [feature])
                continue

            batch.append(feature)
            count += 1

            if len(batch) >= batch_size:
                if staging_key:
                    cache_features_batch_staged(staging_key, batch)
                yield json.dumps({"type": "batch", "features": batch}) + "\n"
                batch = []

        if batch:
            if staging_key:
                cache_features_batch_staged(staging_key, batch)
            yield json.dumps({"type": "batch", "features": batch}) + "\n"

        # Atomically promote staged data to live cache
        if staging_key:
            finalize_cache_population(staging_key, query_parameters)
            staging_key = None  # Mark as finalized

        yield json.dumps({"type": "complete", "total": count}) + "\n"
    finally:
        # Clean up orphaned staging key on failure
        if staging_key:
            delete_staging_key(staging_key)


@app.get("/api/listing_geojson/stream")
async def stream_listing_geojson(
    user: Annotated[User, Depends(get_current_user)],
    query_parameters: Annotated[QueryParameters, Depends(get_query_parameters)],
    batch_size: int = DEFAULT_BATCH_SIZE,
    limit: int | None = None,
    include_poi_distances: bool = False,
    decision_filter: str = "all",
) -> StreamingResponse:
    """Stream listings as NDJSON for progressive map loading.

    Returns newline-delimited JSON with three message types:
    - metadata: Initial message with batch_size and total_expected count
    - batch: Array of GeoJSON features
    - complete: Final message with total count

    Decision filter options:
    - "all" (default): Show all listings except disliked ones
    - "liked": Show only liked listings
    - "everything": Show all listings including disliked
    """
    batch_size = min(batch_size, _rate_limit_config.geojson_stream_batch_size_cap)
    if limit is not None:
        limit = min(limit, _rate_limit_config.geojson_stream_limit_cap)
    else:
        limit = _rate_limit_config.geojson_stream_limit_cap

    # Build POI distances lookup if requested
    poi_distances_lookup = _build_poi_distances_lookup(user.email, query_parameters.listing_type) if include_poi_distances else None

    # Build decision filter sets
    disliked_ids: set[int] | None = None
    liked_ids: set[int] | None = None
    if decision_filter != "everything":
        user_id = _get_user_id_safe(user.email)
        if user_id is not None:
            decision_repo = DecisionRepository(engine)
            listing_type_str = query_parameters.listing_type.value
            if decision_filter == "liked":
                liked_ids = decision_service.get_liked_ids(
                    decision_repo, user_id, listing_type_str
                )
            else:
                # default "all": load disliked to exclude
                disliked_ids = decision_service.get_disliked_ids(
                    decision_repo, user_id, listing_type_str
                )

    cached_count = get_cached_count(query_parameters)
    if cached_count is not None and cached_count > 0 and not include_poi_distances:
        app_metrics.geojson_cache_operations.add(1, {"result": "hit"})
        generator = _stream_from_cache(
            query_parameters, batch_size, limit,
            disliked_ids=disliked_ids,
            liked_ids=liked_ids,
            decision_filter=decision_filter,
        )
    else:
        app_metrics.geojson_cache_operations.add(1, {"result": "miss"})
        generator = _stream_from_db(
            query_parameters, batch_size, limit, poi_distances_lookup,
            skip_cache=include_poi_distances,
            disliked_ids=disliked_ids,
            liked_ids=liked_ids,
            decision_filter=decision_filter,
        )

    return StreamingResponse(
        generator,
        media_type="application/x-ndjson",
        headers={
            "Cache-Control": "no-cache",
            "X-Accel-Buffering": "no",  # Disable nginx buffering
        }
    )


@app.post("/api/refresh_listings")
async def refresh_listings(
    user: Annotated[User, Depends(get_current_user)],
    query_parameters: Annotated[QueryParameters, Depends(get_query_parameters)],
) -> dict[str, str]:
    """Trigger a background task to refresh listings."""
    await send_notification(
        f"{user.email} refreshing listings with query parameters {query_parameters.model_dump_json()}"
    )

    repository = ListingRepository(engine)
    result = await listing_service.refresh_listings(
        repository,
        query_parameters,
        async_mode=True,
        user_email=user.email,
    )

    # Track task for user
    if result.task_id:
        task_service.add_task_for_user(user.email, result.task_id)

    return {"task_id": result.task_id or "", "message": result.message}


@app.get("/api/task_status")
async def get_task_status(
    user: Annotated[User, Depends(get_current_user)],
    task_id: str,
) -> dict[str, str | int | float | None]:
    """Get the status of a background task."""
    user_tasks = task_service.get_user_tasks(user.email)
    if task_id not in user_tasks:
        raise HTTPException(status_code=404, detail="Task not found")
    status = task_service.get_task_status(task_id)
    return {
        "task_id": status.task_id,
        "status": status.status,
        "result": json.dumps(status.result) if status.result else None,
        "progress": status.progress,
        "processed": status.processed,
        "total": status.total,
        "message": status.message,
        "error": status.error if APP_ENV != "production" else None,
        "traceback": status.traceback if APP_ENV != "production" else None,
    }


@app.get("/api/tasks_for_user")
async def get_tasks_for_user(
    user: Annotated[User, Depends(get_current_user)],
) -> list[str]:
    """Get all task IDs for the current user."""
    return task_service.get_user_tasks(user.email)


@app.post("/api/cancel_task")
async def cancel_task(
    user: Annotated[User, Depends(get_current_user)],
    task_id: str = Query(..., description="The task ID to cancel"),
) -> dict[str, str | bool]:
    """Cancel a running task and remove it from the user's task list."""
    # Verify user owns this task
    user_tasks = task_service.get_user_tasks(user.email)
    if task_id not in user_tasks:
        raise HTTPException(status_code=404, detail="Task not found or not owned by user")

    try:
        task_service.cancel_task(task_id, user_email=user.email)
        logger.info(f"Task {task_id} cancelled by {user.email}")
        return {"success": True, "message": "Task cancelled"}
    except Exception as e:
        logger.error(f"Failed to cancel task {task_id}: {e}")
        return {"success": False, "message": str(e)}


@app.post("/api/clear_all_tasks")
async def clear_all_tasks(
    user: Annotated[User, Depends(get_current_user)],
) -> dict[str, str | int | bool]:
    """Clear all tasks for the current user."""
    try:
        count = task_service.clear_all_tasks(user.email)
        logger.info(f"Cleared {count} tasks for {user.email}")
        return {"success": True, "count": count, "message": f"Cleared {count} tasks"}
    except Exception as e:
        logger.error(f"Failed to clear tasks for {user.email}: {e}")
        return {"success": False, "count": 0, "message": str(e)}


@app.get("/api/get_districts")
async def get_districts(
    user: Annotated[User, Depends(get_current_user)],
) -> dict[str, str]:
    """Get all available districts."""
    return district_service.get_all_districts()


class ListingDetailResponse(BaseModel):
    id: int
    price: float
    number_of_bedrooms: int
    square_meters: float | None
    agency: str | None
    council_tax_band: str | None
    url: str
    listing_type: str
    description: str | None
    display_address: str | None
    property_sub_type: str | None
    key_features: list[str]
    photos: list[dict]
    floorplans: list[dict]
    price_history: list[dict]
    furnish_type: str | None
    available_from: str | None
    service_charge: float | None
    lease_left: int | None
    decision: str | None
    poi_distances: list[dict]


@app.get("/api/listing/{listing_id}/detail", response_model=ListingDetailResponse)
async def get_listing_detail(
    user: Annotated[User, Depends(get_current_user)],
    listing_id: int,
    listing_type: str = Query(default="RENT"),
) -> ListingDetailResponse:
    """Get detailed information for a single listing."""
    repository = ListingRepository(engine)
    lt = ListingType(listing_type)
    listings = await repository.get_listings(
        only_ids=[listing_id], listing_type=lt
    )
    if not listings:
        raise HTTPException(status_code=404, detail="Listing not found")

    listing = listings[0]
    additional_info = listing.additional_info or {}
    property_info = additional_info.get("property", {})

    # Extract description
    text_info = property_info.get("text", {})
    description = text_info.get("description") if isinstance(text_info, dict) else None

    # Extract photos
    photos_raw = property_info.get("images", [])
    photos: list[dict] = []
    if isinstance(photos_raw, list):
        for img in photos_raw:
            if isinstance(img, dict):
                photos.append({
                    "url": img.get("url", ""),
                    "caption": img.get("caption", ""),
                    "type": img.get("type", ""),
                })

    # Extract floorplans
    floorplans_raw = property_info.get("floorplans", [])
    floorplans: list[dict] = []
    if isinstance(floorplans_raw, list):
        for fp in floorplans_raw:
            if isinstance(fp, dict):
                floorplans.append({
                    "url": fp.get("url", ""),
                    "caption": fp.get("caption", ""),
                })

    # Extract other fields
    key_features = property_info.get("keyFeatures", [])
    if not isinstance(key_features, list):
        key_features = []
    display_address_info = property_info.get("address", {})
    display_address = (
        display_address_info.get("displayAddress")
        if isinstance(display_address_info, dict)
        else None
    )
    property_sub_type = property_info.get("propertySubType")
    council_tax_band = property_info.get("councilTaxBand") or listing.council_tax_band
    furnish_type_val = property_info.get("letFurnishType")
    available_from_val = property_info.get("letDateAvailable")

    # Price history
    price_history = [item.to_dict() for item in listing.price_history]

    # Service charge and lease (for BuyListing)
    service_charge: float | None = None
    lease_left: int | None = None
    if hasattr(listing, "service_charge"):
        service_charge = listing.service_charge  # type: ignore[union-attr]
    if hasattr(listing, "lease_left"):
        lease_left = listing.lease_left  # type: ignore[union-attr]

    # Available from (for RentListing)
    if available_from_val is None and hasattr(listing, "available_from"):
        af = listing.available_from  # type: ignore[union-attr]
        if af is not None:
            available_from_val = af.isoformat() if hasattr(af, "isoformat") else str(af)

    # Furnish type (for RentListing)
    if furnish_type_val is None and hasattr(listing, "furnish_type"):
        ft = listing.furnish_type  # type: ignore[union-attr]
        if ft is not None:
            furnish_type_val = str(ft)

    # Load user's decision for this listing
    decision_val: str | None = None
    user_id = _get_user_id_safe(user.email)
    if user_id is not None:
        decision_repo = DecisionRepository(engine)
        decisions = decision_repo.get_decisions_for_user(user_id)
        for d in decisions:
            if d.listing_id == listing_id and d.listing_type == listing_type:
                decision_val = d.decision
                break

    # Load POI distances
    poi_distances_list: list[dict] = []
    if user_id is not None:
        poi_repo = POIRepository(engine)
        pois = {p.id: p for p in poi_repo.get_pois_for_user(user_id)}
        if pois:
            distances = poi_repo.get_distances_for_listings(
                [listing_id], lt, user_id
            )
            for d in distances:
                poi_name = pois[d.poi_id].name if d.poi_id in pois else "Unknown"
                poi_distances_list.append({
                    "poi_id": d.poi_id,
                    "poi_name": poi_name,
                    "travel_mode": d.travel_mode,
                    "duration_seconds": d.duration_seconds,
                    "distance_meters": d.distance_meters,
                })

    return ListingDetailResponse(
        id=listing.id,
        price=listing.price,
        number_of_bedrooms=listing.number_of_bedrooms,
        square_meters=listing.square_meters,
        agency=listing.agency,
        council_tax_band=council_tax_band,
        url=listing.url,
        listing_type=listing_type,
        description=description,
        display_address=display_address,
        property_sub_type=property_sub_type,
        key_features=key_features,
        photos=photos,
        floorplans=floorplans,
        price_history=price_history,
        furnish_type=furnish_type_val,
        available_from=available_from_val,
        service_charge=service_charge,
        lease_left=lease_left,
        decision=decision_val,
        poi_distances=poi_distances_list,
    )


FastAPIInstrumentor.instrument_app(app)