"""Utils for manipulating data."""
from collections import OrderedDict

from bulk_metadata_ingester_common.constants import spreadsheet as tab_consts
from bulk_metadata_ingester_common.constants.data_headers import FIELD_JOIN_CONST  # noqa: E501
from bulk_metadata_ingester_common.exceptions import ChunkException
import pandas as pd


def pivot_bulk_file(bulk_upload_rows, logger):
    """Collate file rows into groups of project/product keys.

    Args:
        bulk_upload_rows (pandas.DataFrame):    The CSV data as a DataFrame.
        logger (Logger)

    Returns:
        dict: A collated dict of lists of tracks

    """
    # Init vars
    project_code = None
    product_code = None
    data_by_release = OrderedDict()
    product_data_list = list()  # the list of tracks for a project
    row_number = 2  # Excel spreadsheets start at row 2
    total_rows = 0
    release_count = 0

    # Syntactic Sugar - Shorten constants
    PROJ_CONST = tab_consts.folder_name_project_code
    PROD_CONST = tab_consts.product_code

    # Get a list of unique keys
    package_list = list(
        dict.fromkeys(
            [f'{r[PROJ_CONST]}{FIELD_JOIN_CONST}{r[PROD_CONST]}'
                for _, r in bulk_upload_rows.iterrows()]
        )
    )

    # Get the count
    count_of_packages = len(package_list)

    # Iterate over DB result rows
    for i, row in bulk_upload_rows.iterrows():  # each row (multiple projects)

        total_rows += 1

        # Checks for initial app state
        if not project_code:
            project_code = row.get(PROJ_CONST)
        if not product_code:
            product_code = row.get(PROD_CONST)

        # Check for change in product_code values
        if (product_code != row.get(PROD_CONST)) or \
                (project_code != row.get(PROJ_CONST)):

            # Store finished product
            data_by_release[f'{project_code}{FIELD_JOIN_CONST}{product_code}']\
                = pd.DataFrame(product_data_list).to_dict(orient='records')

            # Bump release count
            release_count += 1

            # Debug
            logger.info(
                f'Release {project_code}{FIELD_JOIN_CONST}{product_code} '
                f'prepared.')

            # Reset product dict
            product_data_list = list()

            # Update project and product codes
            product_code = row[PROD_CONST]
            project_code = row[PROJ_CONST]

        # Add row number field
        row[tab_consts.catalog_ingest_order] = row_number
        row_number += 1

        # Append product to project
        product_data_list.append(row.copy())

    # Shorten the key
    release_key = f'{project_code}{FIELD_JOIN_CONST}{product_code}'

    # Append final product to list
    if product_data_list:
        data_by_release[release_key] = \
            pd.DataFrame(product_data_list).to_dict(orient='records')

        logger.info(f'Release {release_count} prepared.')

    # Check the processed count matches the expected amount
    if len(data_by_release) != count_of_packages:
        # TODO: Throw Exception instead of die
        raise ChunkException('Something did not get chunked properly.')

    return data_by_release