services/csv_parser.py

import pandas as pd
from io import BytesIO
from datetime import datetime
from typing import List, Dict, Set


def parse_letterboxd_csv(file_content: bytes) -> List[Dict]:
    try:
        df = pd.read_csv(BytesIO(file_content))
    except Exception as e:
        raise ValueError(f"Impossible de lire le CSV : {e}")

    required = {"Date", "Name", "Year", "Letterboxd URI", "Rating"}
    missing = required - set(df.columns)
    if missing:
        raise ValueError(f"Colonnes manquantes : {missing}")

    records = []
    for _, row in df.iterrows():
        try:
            rating_val = row["Rating"]
            if pd.isna(rating_val):
                continue
            rating = float(rating_val)
            date_viewed = datetime.strptime(str(row["Date"]).strip(), "%Y-%m-%d")
            name = str(row["Name"]).strip()
            year = int(row["Year"]) if pd.notna(row["Year"]) else None
            key = (name + "_" + str(year)) if year else name
            records.append({"key": key, "name": name, "year": year, "rating": rating, "date_viewed": date_viewed})
        except Exception:
            continue
    return records


def parse_watchlist_csv(file_content: bytes) -> Set[str]:
    try:
        df = pd.read_csv(BytesIO(file_content))
    except Exception:
        return set()

    if "Name" not in df.columns:
        return set()

    keys = set()
    for _, row in df.iterrows():
        try:
            name = str(row["Name"]).strip()
            year = int(row["Year"]) if pd.notna(row.get("Year", float("nan"))) else None
            key = (name + "_" + str(year)) if year else name
            keys.add(key)
        except Exception:
            continue
    return keys