data/db.py

"""
data/db.py — SQLite persistence for feelin'
"""

from __future__ import annotations
import pandas as pd
import uuid
import os
from pathlib import Path

# ── Paths ────────────────────────────────────────────────────────────────────
BASE_DIR = Path("/data")
POSTS_CSV    = BASE_DIR / "posts.csv"
PODCASTS_CSV = BASE_DIR / "podcasts.csv"
SFX_DIR      = BASE_DIR / "sfx"
POD_DIR      = BASE_DIR / "podcasts"
SEG_DIR      = BASE_DIR / "segments"

for d in [SFX_DIR, POD_DIR, SEG_DIR]:
    d.mkdir(parents=True, exist_ok=True)

# ── Schema defaults ───────────────────────────────────────────────────────────
POSTS_COLS = {
    "id": str, "text": str, "created_at": str,
    "is_brag": int, "brag_roast": str, "emotion": str,
    "score_hilarious": float, "score_tragic": float,
    "score_unhinged": float, "score_awkward": float, "score_chaotic": float,
    "top_category": str, "top_score": float, "commentary": str,
}

PODCASTS_COLS = {
    "id": str, "category": str, "script": str,
    "audio_path": str, "created_at": str, "post_count": int,
}

# ── Helpers ───────────────────────────────────────────────────────────────────
def _now() -> str:
    from datetime import datetime, timezone
    return datetime.now(timezone.utc).strftime("%Y-%m-%d %H:%M:%S")

def _load(path: Path, cols: dict) -> pd.DataFrame:
    """Load CSV or return empty DataFrame with correct schema."""
    if path.exists():
        df = pd.read_csv(path, dtype=str)          # read all as str first
        # cast to proper types, fill missing cols
        for col, typ in cols.items():
            if col not in df.columns:
                df[col] = None
            try:
                df[col] = df[col].astype(typ)
            except (ValueError, TypeError):
                pass
        return df
    return pd.DataFrame({c: pd.Series(dtype=t) for c, t in cols.items()})

def _save(df: pd.DataFrame, path: Path):
    df.to_csv(path, index=False)

# ── Init (no-op for pandas, just ensures dirs exist) ─────────────────────────
def init_db():
    for d in [SFX_DIR, POD_DIR, SEG_DIR]:
        d.mkdir(parents=True, exist_ok=True)

# ── Posts ─────────────────────────────────────────────────────────────────────
def insert_post(text: str) -> str:
    post_id = str(uuid.uuid4())
    df = _load(POSTS_CSV, POSTS_COLS)
    new_row = {c: None for c in POSTS_COLS}
    new_row.update({"id": post_id, "text": text, "created_at": _now(),
                    "is_brag": 0, "score_hilarious": 0.0, "score_tragic": 0.0,
                    "score_unhinged": 0.0, "score_awkward": 0.0, "score_chaotic": 0.0,
                    "top_score": 0.0})
    df = pd.concat([df, pd.DataFrame([new_row])], ignore_index=True)
    _save(df, POSTS_CSV)
    return post_id

def flag_brag(post_id: str, roast: str):
    df = _load(POSTS_CSV, POSTS_COLS)
    mask = df["id"] == post_id
    df.loc[mask, "is_brag"] = 1
    df.loc[mask, "brag_roast"] = roast
    _save(df, POSTS_CSV)

def update_post_scores(post_id: str, scores: dict, top_category: str, top_score: float, commentary: str):
    if not scores:
        return
    df = _load(POSTS_CSV, POSTS_COLS)
    mask = df["id"] == post_id
    for cat in ["hilarious", "tragic", "unhinged", "awkward", "chaotic"]:
        df.loc[mask, f"score_{cat}"] = scores.get(cat, 0)
    df.loc[mask, "top_category"] = top_category
    df.loc[mask, "top_score"]    = top_score
    df.loc[mask, "commentary"]   = commentary
    _save(df, POSTS_CSV)

def get_leaderboard(category: str, limit: int = 10) -> list[dict]:
    df = _load(POSTS_CSV, POSTS_COLS)
    col = f"score_{category}"
    df[col] = pd.to_numeric(df[col], errors="coerce").fillna(0)
    df["is_brag"] = pd.to_numeric(df["is_brag"], errors="coerce").fillna(0)

    result = (
        df[(df["is_brag"] == 0) & (df[col] > 0)]
        .sort_values(col, ascending=False)
        .head(limit)
        .copy()
    )

    # Rename BEFORE selecting to avoid duplicate "score" column
    result = result.rename(columns={col: "score"})

    keep = ["id", "text", "score",
            "score_hilarious", "score_tragic", "score_unhinged",
            "score_awkward", "score_chaotic",
            "top_category", "commentary", "created_at"]
    keep = [c for c in keep if c in result.columns]

    return result[keep].to_dict(orient="records")

    
def get_top_posts_for_podcast(category: str, limit: int = 10) -> list[dict]:
    df = _load(POSTS_CSV, POSTS_COLS)
    col = f"score_{category}"
    df[col] = pd.to_numeric(df[col], errors="coerce").fillna(0)
    df["is_brag"] = pd.to_numeric(df["is_brag"], errors="coerce").fillna(0)
    result = (
        df[(df["is_brag"] == 0) & (df[col] >= 5.0)]
        .sort_values(col, ascending=False)
        .head(limit)
    )
    return result[["id", "text", col, "commentary"]].rename(columns={col: "score"}).to_dict(orient="records")

def get_stats() -> dict:
    df = _load(POSTS_CSV, POSTS_COLS)
    if df.empty:
        return {"total_posts": 0, "brags_caught": 0,
                "categories": {c: 0 for c in ["hilarious","tragic","unhinged","awkward","chaotic"]}}
    df["is_brag"] = pd.to_numeric(df["is_brag"], errors="coerce").fillna(0)
    cats = {}
    for cat in ["hilarious", "tragic", "unhinged", "awkward", "chaotic"]:
        col = f"score_{cat}"
        df[col] = pd.to_numeric(df[col], errors="coerce").fillna(0)
        cats[cat] = int(((df["is_brag"] == 0) & (df[col] > 0)).sum())
    return {
        "total_posts":   len(df),
        "brags_caught":  int((df["is_brag"] == 1).sum()),
        "categories":    cats,
    }

# ── Podcasts ──────────────────────────────────────────────────────────────────
def save_podcast(category: str, script: str, audio_path: str, post_count: int) -> str:
    pod_id = str(uuid.uuid4())
    df = _load(PODCASTS_CSV, PODCASTS_COLS)
    new_row = {"id": pod_id, "category": category, "script": script,
               "audio_path": audio_path, "created_at": _now(), "post_count": post_count}
    df = pd.concat([df, pd.DataFrame([new_row])], ignore_index=True)
    _save(df, PODCASTS_CSV)
    return pod_id

def get_latest_podcast(category: str) -> dict | None:
    df = _load(PODCASTS_CSV, PODCASTS_COLS)
    filtered = df[df["category"] == category]
    if filtered.empty:
        return None
    return filtered.sort_values("created_at", ascending=False).iloc[0].to_dict()