import glob
import json
import math
import os
from dataclasses import dataclass

import dateutil
import numpy as np

from src.display.formatting import make_clickable_model
from src.display.utils import AutoEvalColumn, ModelType, Tasks, Precision, WeightType
from src.submission.check_validity import is_model_on_hub


@dataclass
class EvalResult:
    """Represents one full evaluation. Built from a combination of the result and request file for a given run.
    """
    model_name: str
    student_id: str
    results: dict

    @classmethod
    def init_from_json_file(self, json_filepath):
        """Inits the result from the specific model result file"""
        with open(json_filepath) as fp:
            data = json.load(fp)

        config = data.get("config")

        # Extract results available in this file (some results are split in several files)
        results = {}
        for task in Tasks:
            task = task.value

            # We average all scores of a given metric (not all metrics are present in all files)
            accs = np.array([v.get(task.metric, None) for k, v in data["results"].items() if task.benchmark == k])
            if accs.size == 0 or any([acc is None for acc in accs]):
                continue
            results[task.col_name] = accs.mean()

        return self(
            model_name=config.get("model_name", None),
            student_id=config.get("student_id", None),
            results=results,
        )

    def update_with_request_file(self, requests_path, model_name, student_id):
        """Finds the relevant request file for the current model and updates info with it"""
        request_file = get_request_file_for_model(requests_path, model_name, student_id)
        try:
            with open(request_file, "r") as f:
                request = json.load(f)
            self.date = request.get("submitted_time", "")
        except Exception:
            print(f"Could not find request file for {student_id}_{model_name}")

    def to_dict(self):
        """Converts the Eval Result to a dict compatible with our dataframe display"""
        data_dict = {
            "eval_name": self.eval_name,  # not a column, just a save name
            "Model Name": self.model_name,
        }

        # Add task-specific metrics
        for task in Tasks:
            data_dict[task.value.col_name] = self.results.get(task.value.col_name, None)

        # Add student ID and submission date
        data_dict["Student ID"] = self.student_id
        data_dict["Submission Date"] = self.date

        return data_dict


def get_request_file_for_model(requests_path, model_name, student_id):
    """Selects the correct request file for a given model."""
    request_files = os.path.join(
        requests_path, student_id,
        f"request_{student_id}_{model_name}*.json",
    )
    request_files = glob.glob(request_files)

    # Select the latest request file based on the modification date
    request_file = ""
    request_files = sorted(request_files, key=lambda x: os.path.getmtime(x), reverse=True)
    if len(request_files) > 0:
        request_file = request_files[0]

    return request_file


def get_raw_eval_results(results_path: str, requests_path: str) -> list[EvalResult]:
    """From the path of the results folder root, extract all needed info for results"""
    model_result_filepaths = []

    for root, _, files in os.walk(results_path):
        # Filter out non-JSON files
        files = [f for f in files if f.endswith(".json") and f.startswith("result")]

        # Sort the files by date
        try:
            files.sort(key=lambda x: x.removesuffix(".json").removeprefix("result")[:-7])
        except dateutil.parser._parser.ParserError:
            files = [files[-1]]

        for file in files:
            model_result_filepaths.append(os.path.join(root, file))

    eval_results = {}
    for model_result_filepath in model_result_filepaths:
        # Creation of result
        eval_result = EvalResult.init_from_json_file(model_result_filepath)
        eval_result.update_with_request_file(requests_path, eval_result.model_name, eval_result.student_id)

        # Store results of same eval together
        eval_name = f"{eval_result.student_id}_{eval_result.model_name}"
        eval_result.eval_name = eval_name
        if eval_name in eval_results.keys():
            eval_results[eval_name].results.update({k: v for k, v in eval_result.results.items() if v is not None})
        else:
            eval_results[eval_name] = eval_result

    results = []
    for v in eval_results.values():
        try:
            v.to_dict() # we test if the dict version is complete
            results.append(v)
        except KeyError:  # not all eval values present
            continue

    return results