"""
Populate the CodeReview Bench leaderboard from HuggingFace datasets.
"""

import json
import os
import pandas as pd
import tempfile
from typing import Dict, List, Optional
from datetime import datetime
import numpy as np

from huggingface_hub import hf_hub_download, HfApi
from datasets import load_dataset

from src.display.utils import CODEREVIEW_COLUMN, DISPLAY_COLS, CATEGORIES
from src.envs import RESULTS_DATASET_ID, TOKEN, CACHE_PATH
from src.leaderboard.processor import leaderboard_to_dataframe


def get_latest_leaderboard(version="v0") -> Optional[Dict]:
    """
    Get the latest leaderboard data from HuggingFace dataset.
    Fallback to local JSON file if HF download fails or is unavailable.
    """
    # First try to fetch from HuggingFace Hub
    try:
        leaderboard_path = hf_hub_download(
            repo_id=RESULTS_DATASET_ID,
            filename=f"leaderboards/leaderboard_{version}.json",
            repo_type="dataset",
            token=TOKEN
        )
        with open(leaderboard_path, 'r') as f:
            return json.load(f)
    except Exception as hf_err:
        print(f"HF download failed or unavailable: {hf_err}. Trying local fallback...")

    # Fallback: attempt to load a local leaderboard_data.json located at the project root
    project_root = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
    local_path_candidates = [
        os.path.join(project_root, "leaderboard_data.json"),  # legacy path in root
        os.path.join(project_root, "data", "leaderboard.json"),  # path defined in envs.py
    ]

    for local_path in local_path_candidates:
        if os.path.exists(local_path):
            try:
                with open(local_path, 'r') as f:
                    return json.load(f)
            except Exception as local_err:
                print(f"Error loading local leaderboard file {local_path}: {local_err}")

    # If nothing found, return None
    return None


def get_model_entry(model_name: str, mode: str, version="v0") -> Optional[Dict]:
    """
    Get a specific model's entry from the entries folder, uniquely identified by model_name, mode, and version.
    """
    try:
        model_name_safe = model_name.replace("/", "_").replace(" ", "_")
        mode_safe = str(mode).replace("/", "_").replace(" ", "_").lower()
        entry_path = hf_hub_download(
            repo_id=RESULTS_DATASET_ID,
            filename=f"entries/entry_{model_name_safe}_{mode_safe}_{version}.json",
            repo_type="dataset",
            token=TOKEN
        )
        with open(entry_path, 'r') as f:
            return json.load(f)
    except Exception as e:
        print(f"Error downloading model entry: {e}")
        return None


def get_all_entries(version="v0") -> List[Dict]:
    """
    Get all entries from the HuggingFace dataset.
    """
    try:
        api = HfApi(token=TOKEN)
        files = api.list_repo_files(repo_id=RESULTS_DATASET_ID, repo_type="dataset")
        entry_files = [f for f in files if f.startswith("entries/") and f.endswith(f"_{version}.json")]

        all_entries = []
        for entry_file in entry_files:
            try:
                entry_path = hf_hub_download(
                    repo_id=RESULTS_DATASET_ID,
                    filename=entry_file,
                    repo_type="dataset",
                    token=TOKEN
                )
                with open(entry_path, 'r') as f:
                    entry_data = json.load(f)
                    all_entries.append(entry_data)
            except Exception as e:
                print(f"Error loading entry {entry_file}: {e}")

        return all_entries
    except Exception as e:
        print(f"Error getting all entries: {e}")
        return []


def get_leaderboard_df(version="v0") -> pd.DataFrame:
    """
    Get the leaderboard data as a DataFrame.
    """
    # Get latest leaderboard data
    leaderboard_data = get_latest_leaderboard(version)

    if not leaderboard_data:
        # If no leaderboard exists, try to build it from entries
        entries = get_all_entries(version)
        if entries:
            leaderboard_data = {
                "entries": entries,
                "last_updated": datetime.now().isoformat(),
                "version": version
            }
        else:
            # Return empty DataFrame if no data available
            return pd.DataFrame(columns=DISPLAY_COLS)

    # Convert to DataFrame
    return leaderboard_to_dataframe(leaderboard_data)


def get_category_leaderboard_df(category: str, version="v0") -> pd.DataFrame:
    """
    Get the leaderboard data filtered by a specific programming language category.
    """
    # Get latest leaderboard data
    leaderboard_data = get_latest_leaderboard(version)

    if not leaderboard_data:
        # If no leaderboard exists, try to build it from entries
        entries = get_all_entries(version)
        if entries:
            leaderboard_data = {
                "entries": entries,
                "last_updated": datetime.now().isoformat(),
                "version": version
            }
        else:
            # Return empty DataFrame if no data available
            return pd.DataFrame(columns=DISPLAY_COLS)

    # Filter entries to only include those with data for the specified programming language
    filtered_entries = []
    for entry in leaderboard_data.get("entries", []):
        # Check if entry has data for this programming language
        programming_language = entry.get("programming_language", "").lower()
        if programming_language == category.lower() or category.lower() == "other":
            # For "other" category, include entries that don't match any specific language
            if category.lower() == "other":
                if programming_language not in [cat.lower() for cat in CATEGORIES[:-1]]:  # Exclude "Other" from check
                    filtered_entries.append(entry)
            else:
                filtered_entries.append(entry)

    # Create a new leaderboard data structure with the filtered entries
    filtered_leaderboard = {
        "entries": filtered_entries,
        "last_updated": leaderboard_data.get("last_updated", datetime.now().isoformat()),
        "version": version
    }

    # Convert to DataFrame
    return leaderboard_to_dataframe(filtered_leaderboard)


def get_detailed_model_data(model_name: str, mode: str, version="v0") -> Dict:
    """
    Get detailed data for a specific model and mode.
    """
    entry = get_model_entry(model_name, mode, version)
    if entry:
        return entry
    leaderboard_data = get_latest_leaderboard(version)
    if leaderboard_data:
        for entry in leaderboard_data.get("entries", []):
            if entry.get("model_name") == model_name and str(entry.get("mode")).lower() == str(mode).lower():
                return entry
    return {}