src/cache.py

"""
SQLite-based caching layer for nutrition and literature API responses.
Works both locally and on HF Spaces (file-based, no server needed).
"""

import sqlite3
import json
import hashlib
import time
import os
from pathlib import Path
from typing import Optional


# Default cache location: project root or HF Space persistent storage
CACHE_DIR = os.environ.get("CACHE_DIR", str(Path(__file__).parent.parent))
CACHE_DB = os.path.join(CACHE_DIR, "cache.db")


def _get_connection() -> sqlite3.Connection:
    """Get a SQLite connection with WAL mode for concurrent reads."""
    conn = sqlite3.connect(CACHE_DB, timeout=10)
    conn.execute("PRAGMA journal_mode=WAL")
    conn.execute("PRAGMA synchronous=NORMAL")
    return conn


def init_cache():
    """Create cache tables if they don't exist."""
    conn = _get_connection()
    try:
        conn.executescript("""
            CREATE TABLE IF NOT EXISTS nutrition_cache (
                query_key TEXT PRIMARY KEY,
                response_json TEXT NOT NULL,
                created_at REAL NOT NULL,
                ttl_days INTEGER DEFAULT 30
            );
            CREATE TABLE IF NOT EXISTS literature_cache (
                query_key TEXT PRIMARY KEY,
                response_json TEXT NOT NULL,
                created_at REAL NOT NULL,
                ttl_days INTEGER DEFAULT 7
            );
        """)
        conn.commit()
    finally:
        conn.close()


def _make_key(text: str) -> str:
    """Create a normalized cache key from query text."""
    normalized = text.strip().lower()
    return hashlib.sha256(normalized.encode()).hexdigest()[:32]


def get_cached(table: str, query: str) -> Optional[dict]:
    """
    Retrieve a cached response if it exists and hasn't expired.

    Args:
        table: 'nutrition_cache' or 'literature_cache'
        query: the search query string

    Returns:
        Parsed JSON dict if cache hit and not expired, None otherwise.
    """
    init_cache()
    key = _make_key(query)
    conn = _get_connection()
    try:
        row = conn.execute(
            f"SELECT response_json, created_at, ttl_days FROM {table} WHERE query_key = ?",
            (key,),
        ).fetchone()
        if row is None:
            return None

        response_json, created_at, ttl_days = row
        age_days = (time.time() - created_at) / 86400
        if age_days > ttl_days:
            # Expired, clean up
            conn.execute(f"DELETE FROM {table} WHERE query_key = ?", (key,))
            conn.commit()
            return None

        return json.loads(response_json)
    finally:
        conn.close()


def set_cached(table: str, query: str, data: dict, ttl_days: int = None):
    """
    Store a response in the cache.

    Args:
        table: 'nutrition_cache' or 'literature_cache'
        query: the search query string
        data: response dict to cache
        ttl_days: override default TTL
    """
    init_cache()
    key = _make_key(query)
    ttl = ttl_days or (30 if table == "nutrition_cache" else 7)
    conn = _get_connection()
    try:
        conn.execute(
            f"""INSERT OR REPLACE INTO {table}
                (query_key, response_json, created_at, ttl_days)
                VALUES (?, ?, ?, ?)""",
            (key, json.dumps(data), time.time(), ttl),
        )
        conn.commit()
    finally:
        conn.close()


def clear_expired():
    """Remove all expired entries from all cache tables."""
    init_cache()
    conn = _get_connection()
    now = time.time()
    try:
        for table in ["nutrition_cache", "literature_cache"]:
            conn.execute(
                f"DELETE FROM {table} WHERE (? - created_at) / 86400 > ttl_days",
                (now,),
            )
        conn.commit()
    finally:
        conn.close()


def cache_stats() -> dict:
    """Return count and size info for monitoring."""
    init_cache()
    conn = _get_connection()
    try:
        stats = {}
        for table in ["nutrition_cache", "literature_cache"]:
            count = conn.execute(f"SELECT COUNT(*) FROM {table}").fetchone()[0]
            stats[table] = count
        db_size_mb = os.path.getsize(CACHE_DB) / (1024 * 1024)
        stats["db_size_mb"] = round(db_size_mb, 2)
        return stats
    finally:
        conn.close()