"""This script will check that the list of tracks provided (UPC_VOL_TRK.wav)
from a text file matches the list of files on an asset path."""

import os

from integration_scripts import file_utils as fu
from integration_scripts import logger

# Define vars
file_name = '/Users/lvona/Documents/python_projects/ripper-feeder/' \
            'Suburban_2020_ingestion_list.dat'

root_path = '/Volumes/ripped_assets/ripper_feeder_staging/Suburban/' \
            'ready_to_copy/2020_01_14'


def main():
    # Get list of files from root path.
    with open(file_name) as f:
        metadata_processed = f.readlines()

    metadata_processed = [line.rstrip('\n') for line in metadata_processed]
    metadata_processed.sort()

    # Get list of files from text file.
    data_source_assets = fu.get_all_files_in_subdirs(root_path, '*.wav')
    data_source_assets.sort()

    # Report list lengths
    logger.info('Metadata reports {} assets.', len(metadata_processed))
    logger.info('Data Source reports {} assets.', len(data_source_assets))

    # Compare Lists
    if metadata_processed != data_source_assets:

        # -- DUMB CHECKS ------------------------------------------------

        # Check for dupes in each list
        metadata_set = set(metadata_processed)
        data_source_set = set(data_source_assets)

        # Report list lengths
        logger.info('Metadata reports {} unique assets.',
                    len(metadata_set))
        logger.info('Data Source reports {} unique assets.',
                    len(data_source_set))

        if metadata_set == data_source_set:
            logger.info('File lists match.')
            return

        if len(metadata_set) < len(metadata_processed):
            logger.warning('There are duplicates in the METADATA list.')

        if len(data_source_set) < len(data_source_assets):
            logger.warning('There are duplicates in the DATA SOURCE list.')

        # Get list lengths
        metadata_set_len = len(metadata_set)
        prepped_set_len = len(data_source_set)

        if metadata_set_len > prepped_set_len:
            bigger = 'METADATA has more files.'
        elif metadata_set_len < prepped_set_len:
            bigger = 'DATA SOURCE has more files.'
        else:
            bigger = 'Lists are the same size, but differ.'

        # Report to User
        logger.warning('Files Do No Match. {}', bigger)

        # -- FILTERED CHECKS --------------------------------------------

        diff_set_1 = list(set([x for x in metadata_set if x not in
                          data_source_set]))
        diff_set_1.sort()

        diff_set_2 = list(set([x for x in data_source_set if x not in
                          metadata_set]))
        diff_set_2.sort()

        diff_set = [*diff_set_1, *diff_set_2]
        diff_set.sort()

        intersection = [x for x in metadata_set if x in data_source_set]
        intersection_upcs = list(set([
            li.split('_')[0] + '\n' for li in intersection]))
        intersection_upcs.sort()

        # Check for different values
        # In Metadata, Not in Data Source
        not_in_data_source = [
            x for x in metadata_set if x not in data_source_set]
        not_in_data_source.sort()

        # In Data Source, Not in Metadata
        not_in_metadata = [
            x for x in data_source_set if x not in metadata_set]
        not_in_metadata.sort()

        # UPC's where in Metadata, Not in Data Source
        not_in_data_source_upc_list = list(set([
            li.split('_')[0] + '\n' for li in not_in_data_source]))
        not_in_data_source_upc_list.sort()

        # UPC's where in Data Source, Not in Metadata
        not_in_metadata_upc_list = list(set([
            li.split('_')[0] + '\n' for li in not_in_metadata]))
        not_in_metadata_upc_list.sort()

        # Len of upc where not in Metadata
        not_in_metadata_upc_list_len = len(not_in_metadata_upc_list)

        # Len of upc where not in Data Source
        not_in_data_source_upc_list_len = len(not_in_data_source_upc_list)

        logger.warning(
            'UPC\'s with files not in METADATA set - {} UPC\'s:',
            not_in_metadata_upc_list_len)

        for line in not_in_metadata_upc_list:
            logger.warning(line.rstrip('\n'))

        logger.warning(
            'UPC\'s with files not in DATA SOURCE set - {} UPC\'s:',
            not_in_data_source_upc_list_len)

        for line in not_in_data_source_upc_list:
            logger.warning(line.rstrip('\n'))

        # Name output file
        output_file = os.path.splitext(file_name)[0] + '.txt'
        logger.warning('Writing diff file {}', output_file)

        # Full Diff
        output_list = [li + '\n' for li in list(diff_set)]
        output_list.sort()

        not_in_data_source = [li + '\n' for li in list(not_in_data_source)]
        not_in_data_source.sort()

        not_in_metadata = [li + '\n' for li in list(not_in_metadata)]
        not_in_metadata.sort()

        not_in_data_source_upc_list = [
            li for li in list(not_in_data_source_upc_list)]
        not_in_data_source_upc_list.sort()

        not_in_metadata_upc_list = [
            li for li in list(not_in_metadata_upc_list)]
        not_in_metadata_upc_list.sort()

        intersection = [li + '\n' for li in list(intersection)]
        intersection.sort()

        # Print track list
        # for line in output_list:
        #     logger.warning(line.rstrip('\n'))

        # Get upc list
        upc_list = set([li.split('_')[0] + '\n' for li in output_list])

        # logger.warning('UPC List')

        # Print upc list
        # for line in upc_list:
        #     logger.warning(line.rstrip('\n'))

        with open(output_file, 'w') as of:

            of.write(
                'UPC\'s with all files found in both sources - '
                '{} UPC\'s\n'.format(len(intersection_upcs)))
            of.writelines(intersection_upcs)

            of.write(
                '\n\nFiles found in both sources - {} Tracks\n'.format(len(
                    intersection)))
            of.writelines(intersection)

            of.write('\n\nNot In Data Source - UPC\'s - {} UPC\'s\n'.format(
                not_in_data_source_upc_list_len))
            of.writelines(not_in_data_source_upc_list)

            of.write(
                '\n\nNot In Metadata - UPC\'s - {} UPC\'s\n'.format(
                    not_in_metadata_upc_list_len))
            of.writelines(not_in_metadata_upc_list)

            of.write(
                '\n\nNot In Data Source - Track List - {} Tracks\n'.format(
                    len(not_in_data_source)))
            of.writelines(not_in_data_source)

            of.write(
                '\n\nNot In Metadata - Track List - {} Tracks\n'.format(
                    len(not_in_metadata)))
            of.writelines(not_in_metadata)

            of.write('\n\nTOTAL DIFF - UPC List - {} UPC\'s\n'.format(
                len(upc_list)))
            of.writelines(upc_list)


if __name__ == '__main__':
    main()