import re
import io
from datetime import datetime, date
from typing import List, Optional

import config
from services.s3_client import S3Client

from workers.base_worker import BaseWorker
from utils import db_utils, encoding
from workers.prs.models import PRSDataFile
from models.projects_import_info import ProjectsImportInfo

FILE_REGEX = r"^prs\/Decibel_(?P<type>MPS_budget|PRS_po|Sub_Project|mps_category|mps_group)_(?P<timestamp>.*)\.txt"

FILE_ORDER = {
    "sub_project": 1,
    "mps_category": 2,
    "mps_group": 3,
    "mps_budget": 4,
    "prs_po": 5,
}

TABLE_NAME_MAPS = {
    "sub_project": "RawPRSProjects",
    "mps_category": "RawPRSBudgetCategory",
    "mps_group": "RawPRSBudgetGroup",
    "mps_budget": "RawPRSBudgets",
    "prs_po": "RawPRSPurchaseOrders",
}


class PRSDataImportError(Exception):
    details: str

    def __init__(self, details: str):
        self.details = details


class PRSDataImporter(BaseWorker):
    client: S3Client
    worker_name = "PRSDataImporter"

    def should_log_exceptions(self):
        return True

    def execute(self):
        self.logger.info("PRS Data import", "Started")

        self.client = S3Client(config.SFTP_BUCKET)

        last_import = ProjectsImportInfo.get_imported_prs_date() or date.min
        aws_files = self.get_file_list()
        most_recent_files = sorted(aws_files, key=lambda x: x.timestamp, reverse=True)[:5]
        files_to_import = list(filter(lambda x: x.timestamp.date() > last_import, most_recent_files))
        if not files_to_import:
            files_names = ", ".join([file.file_name for file in aws_files])
            error_details = f"No data files found since {last_import}. Got: {files_names}"
            self.logger.error("PRS Data import error", error_details)
            raise PRSDataImportError(error_details)
        for item in files_to_import:
            file_object = self.client.get_object(item.file_name, encoding.UTF8).replace("\n\r\n", "\n")
            with io.StringIO(file_object) as file:
                if item.is_budget_taxonomy_file():
                    db_utils.update_data_from(file=file, table_name=TABLE_NAME_MAPS[item.type.lower()])
                else:
                    db_utils.import_data_from(file=file, table_name=TABLE_NAME_MAPS[item.type.lower()])

        ProjectsImportInfo.update_imported_prs_date()
        self.logger.success("PRS Data import", "Finished!")

    def get_file_list(self) -> List[PRSDataFile]:
        files = self.client.get_objects_list("prs/")
        data = [self.__map_from_file(file["Key"]) for file in files]
        data = [item for item in data if item is not None]
        data.sort(key=lambda x: FILE_ORDER[x.type.lower()])
        return data

    def __map_from_file(self, file_name: str) -> Optional[PRSDataFile]:
        match = re.compile(FILE_REGEX, re.IGNORECASE).match(file_name)
        if not match:
            if not file_name.endswith(".trg"):
                self.logger.error("PRS Data file error", f"Invalid file name: {file_name}")
            return None
        type = match.group("type")
        timestamp = match.group("timestamp")
        try:
            timestamp = datetime.strptime(timestamp, "%Y%m%d")
        except ValueError:
            self.logger.error("PRS Data file error", f"Invalid date format: {timestamp}")
            return None
        return PRSDataFile(type, timestamp, file_name)