"""Utilities for CloudSearch documents."""
from dateutil.parser import parse
import sentry_sdk

import config # noqa
import const
import sentry  # noqa
from utils import date as date_utils
from utils import string as string_utils

ZERO_DATE = '0000-00-00'
MINIMAL_YEAR = 1000
MAX_LIST_LENGTH = 1000


def prepare_for_upload(schema, document):
    """Clean a CloudSearch document before upload.

    Args:
        schema (dict): JSON schema for the document
        document (dict): a dict representing a CloudSearch document

    Returns:
        dict: cleaned document

    """
    if document.get('type') == 'delete':
        document.pop('fields')
    else:
        properties = schema.get('properties')
        fields = document.get('fields')
        for prop, value in fields.copy().items():
            if prop in properties:

                if properties[prop]['type'] == 'integer' and value is None:
                    document['fields'].pop(prop)

                if properties[prop]['type'] == 'array' and value is None:
                    document['fields'][prop] = []

                if properties[prop]['type'] == 'string' and value is None:
                    document['fields'].pop(prop)
                elif properties[prop]['type'] == 'string' and value:
                    document['fields'][prop] = string_utils.cleanup_unicode(
                        value)

                if (properties[prop]['type'] == 'date_utc'
                        and value is None   # noqa
                        or value == ZERO_DATE): # noqa
                    document['fields'].pop(prop)
                elif properties[prop]['type'] == 'date_utc' and value:
                    try:
                        parsed_date = parse(value)
                        formatted_date = parsed_date.strftime('%Y-%m-%dT%H:%M:%SZ')
                        if (int(value[:4]) < MINIMAL_YEAR
                                or not date_utils.is_expected_datetime_format(formatted_date)):  # noqa
                            raise ValueError(const.INVALID_DATE_ERROR)
                        document['fields'][prop] = formatted_date
                    except ValueError:
                        document['fields'].pop(prop)
                if (properties[prop]['type'] == 'array'
                        and len(document['fields'][prop]) > MAX_LIST_LENGTH):  # noqa
                    sentry_sdk.capture_message(
                        'isrc_list is too long for release: {}'.format(
                            document['fields'].get('release_id')))
                    return False
                elif properties[prop]['type'] == 'array' and value:
                    document['fields'][prop] = [
                        string_utils.cleanup_unicode(item) for item in value]
            else:
                document['fields'].pop(prop)

        if not document['fields']:
            return False

    return document