"""Utilities for Opensearch documents."""

import logging
import re
from typing import Any, Optional

import sentry_sdk
from dateutil.parser import parse

from src.constants.fields import (
    CLEANUP_PATTERN,
    INVALID_DATE_ERROR,
    MAX_LIST_LENGTH,
    MINIMAL_YEAR,
    VALID_DATE_FORMAT,
    ZERO_DATE,
)
from src.logic.types import OpensearchBulkDocument


def cleanup_unicode(value: str) -> str:
    """Cleanup characters that are not supported by Cloudsearch.

    Args:
        value (str): string to clean up

    Returns:
        str: cleaned string

    """
    if value:
        value = re.sub(CLEANUP_PATTERN, "", value)
    return value


def is_expected_datetime_format(value: str) -> bool:
    """Check date format that is supported by Opensearch.

    Args:
        value (str): UTC date

    Returns:
        bool

    """
    try:
        return True if re.match(VALID_DATE_FORMAT, value) else False
    except Exception:
        return False


def parse_date(value: str) -> str:
    """Try to parse the value as a date."""
    parsed_date = parse(value)
    formatted_date = parsed_date.strftime("%Y-%m-%dT%H:%M:%SZ")
    if int(value[:4]) < MINIMAL_YEAR or not is_expected_datetime_format(formatted_date):
        raise ValueError(INVALID_DATE_ERROR)
    return formatted_date


def prepare_for_upload(
    schema: dict[str, Any], document: OpensearchBulkDocument
) -> Optional[OpensearchBulkDocument]:
    """Clean ows response document before upload to Opensearch.

    Args:
        schema (dict): JSON schema for the document.
        document (dict): a dict representing an Opensearch document

    Returns:
        dict: cleaned document

    """
    if document.get("_op_type") == "delete":
        document.pop("doc")
        return document

    properties = schema.get("properties")
    fields = document.get("doc")
    if not fields:
        return None
    if not properties:
        return None

    for prop, value in fields.copy().items():
        if prop in properties:
            if properties[prop]["type"] == "integer" and value is None:
                document["doc"].pop(prop)
            elif properties[prop]["type"] == "array" and value is None:
                document["doc"][prop] = []
            elif properties[prop]["type"] == "string" and value is None:
                document["doc"].pop(prop)
            elif properties[prop]["type"] == "string" and value:
                document["doc"][prop] = cleanup_unicode(value)
            elif properties[prop]["type"] == "date_utc" and value is None or value == ZERO_DATE:
                document["doc"].pop(prop)
            elif properties[prop]["type"] == "date_utc" and value:
                try:
                    document["doc"][prop] = parse_date(value)
                except ValueError:
                    document["doc"].pop(prop)
                    msg = "SKIP: invalid date: {} for release: {}".format(
                        prop, document["doc"].get("release_id")
                    )
                    logging.warning(msg)
            elif (
                properties[prop]["type"] == "array" and len(document["doc"][prop]) > MAX_LIST_LENGTH
            ):
                msg = "SKIP: isrc_list is too long for release: {}".format(
                    document["doc"].get("release_id")
                )
                sentry_sdk.capture_message(msg)
                logging.warning(msg)
                return None
            elif properties[prop]["type"] == "array" and value:
                document["doc"][prop] = [cleanup_unicode(item) for item in value]
        else:
            document["doc"].pop(prop)

    if not document["doc"]:
        return None
    return document