Audit GB Shipment

audit_gb_shipment.py

import json


def get_gb_counts(tsv):
    counts = dict()
    for line in open(tsv):
        barcode = line.split('\t')[0].lower()
        # Skip header row.
        if barcode == 'barcode':
            continue
        # Add 1 to GB count because we make a copy of the best version.
        flacs = int(line.split('\t')[1]) + 1
        tifs = int(line.split('\t')[2])

        # i.e. {'GBIA00000X': {'flacs': 9, 'tifs': 1}}
        counts[barcode] = dict(flacs=flacs, tifs=tifs)

    return counts


def audit_shipment(metadata_file, gb_counts):
    """Audit shipment with counts from GB.

    BEFORE running this script, Generate a JSONL document of all metadata for
    all items for the given shipment:

    ia search 'collection:georgeblood AND shiptracking:15446' -p scope:all -i \
        | parallel 'ia md {}' \
        | pv -acbrl > 15446.jsonl
    """

    for line in open(metadata_file):
        j = json.loads(line.strip())
        flacs = len([f for f in j['files'] if f['name'].endswith('.flac')])
        tifs = len([f for f in j['files'] if f['format'] == 'TIFF'])
        barcode = j['metadata']['collection-catalog-number'].lower()

        assert flacs == gb_counts[barcode]['flacs']
        assert tifs == gb_counts[barcode]['tifs']

    print('success, audit complete!')


if __name__ == '__main__':
    counts = get_gb_counts('/Users/archive/78rpm-utils/spreadsheets/iArchiveExport_FileCount_20170315.tsv')
    audit_shipment('15446.jsonl', counts)