dg-unmatched.py

import dlx_dl
import boto3
import re
from pymongo import MongoClient

ssm_client = boto3.client('ssm')
connect_string = ssm_client.get_parameter(Name='connect-string')['Parameter']['Value']
api_key = ssm_client.get_parameter(Name='undl-dhl-metadata-api-key')['Parameter']['Value']
nonce_key = ssm_client.get_parameter(Name='undl-callback-nonce')['Parameter']['Value']
callback_url = ssm_client.get_parameter(Name='undl-callback-url')['Parameter']['Value']

client = MongoClient(connect_string)
result = client['undlFiles']['files'].aggregate([
    {
        '$match': {
            'source': 'adhoc::digitization'
        }
    },
    {
        '$project': {
            '_id': 0, 
            'symbol': {
                '$arrayElemAt': [
                    '$identifiers.value', 0
                ]
            }
        }
    }, {
        '$group': {
            '_id': {
                'symbol': '$symbol'
            }, 
            'count': {
                '$sum': 1
            }
        }
    }
])

symbols = []
for res in result:
    symbol = res["_id"]["symbol"]
    symbol_count = client['undlFiles']['bibs'].aggregate([
        {
            '$match': {
                '191.subfields.value': symbol
            }
        }, {
            '$count': 'count'
        }
    ])
    for sc in symbol_count:
        #print(symbol, sc['count'])
        if int(sc['count']) == 0:
            print(symbol, sc['count'])