Refactor with separate "profiles" and "transaction" reports.

Also adds automatic querying of time periods <= 31 days to meet PayPal API limitations.

paypal_report.py

@@ -4,18 +4,56 @@ Run it like this:
 
 $ export PAYPAL_CLIENT_ID=XXX
 $ export PAYPAL_CLIENT_SECRET=YYY
-$ python3 paypal_report.py 2021-11-01T00:00:00-0700 2021-11-30T23:59:59-0700 > out.csv
+$ python3 paypal_report.py profiles 2021-11-01T00:00:00-0700 2021-11-30T23:59:59-0700 > out.txt
+$ python3 paypal_report.py transactions 2021-11-01T00:00:00-0700 2021-11-30T23:59:59-0700 > out.txt
 
 """
 import argparse
-import csv
+import collections
+import datetime
 import sys
 import os
 
 import requests
 
 
-def get_access_token(client_id, client_secret):
+def main():
+    args = parse_args()
+    client_id, client_secret = load_paypal_keys()
+    access_token = get_paypal_access_token(client_id, client_secret)
+    transactions = get_all_transactions(access_token, args.start_date, args.end_date)
+    if args.report == 'profiles':
+        report_on_unique_profiles(transactions)
+    else:
+        report_on_transactions(transactions)
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(description='Download PayPal subscriber info.')
+    parser.add_argument('report', help='report to run"', choices=['profiles', 'transactions'])
+    parser.add_argument('start_date', help='start date inclusive (eg. 2021-11-01T00:00:00-07:00)', type=parse_iso_time)
+    parser.add_argument('end_date', help='end date inclusive (eg. 2021-11-30T23:59:59-07:00)', type=parse_iso_time)
+    return parser.parse_args()
+
+
+def parse_iso_time(time_str):
+    try:
+        time = datetime.datetime.fromisoformat(time_str)
+    except ValueError:
+        raise argparse.ArgumentTypeError('Date did not match ISO format eg. 2021-11-01T00:00:00-07:00')
+    return time
+
+
+def load_paypal_keys():
+    try:
+        client_id = os.environ['PAYPAL_CLIENT_ID']
+        client_secret = os.environ['PAYPAL_CLIENT_SECRET']
+    except KeyError:
+        sys.exit('Please set PAYPAL_CLIENT_ID and PAYPAL_CLIENT_SECRET environment variables.')
+    return client_id, client_secret
+
+
+def get_paypal_access_token(client_id, client_secret):
     url = 'https://api-m.paypal.com/v1/oauth2/token?grant_type=client_credentials'
     response = requests.post(
         url,
@@ -29,9 +67,31 @@ def get_access_token(client_id, client_secret):
     return data['access_token']
 
 
-def get_transactions(access_token, start_date, end_date, page=1):
+def get_all_transactions(access_token, start_date, end_date):
+    periods = get_time_periods(start_date, end_date)
+    transactions = []
+    for start, end in periods:
+        print(f'Period {start} - {end}.', file=sys.stderr)
+        transactions += get_transactions_for_period(access_token, start, end)
+    return transactions
+
+
+def get_time_periods(start_date, end_date):
+    """Break a time period into smaller periods of up to 31 days.
+
+    PayPal only supports querying for up to 31 days of transactions at a time.
+    """
+    periods = []
+    while start_date < end_date:
+        periods.append((start_date, start_date + datetime.timedelta(days=31)))
+        start_date += datetime.timedelta(days=31)
+    return periods
+
+
+def get_transactions_for_period(access_token, start_date, end_date, page=1):
+    """Downloads all pages of transactions for a time period of 31 days or less."""
     print(f'Fetching transactions page {page}.', file=sys.stderr)
-    url = f'https://api-m.paypal.com/v1/reporting/transactions?start_date={start_date}&end_date={end_date}&fields=all&page_size=500&page={page}'
+    url = f'https://api-m.paypal.com/v1/reporting/transactions?start_date={start_date.isoformat()}&end_date={end_date.isoformat()}&fields=all&page_size=500&page={page}'
     response = requests.get(
         url,
         headers={
@@ -41,32 +101,41 @@ def get_transactions(access_token, start_date, end_date, page=1):
     )
     data = response.json()
     if page < data['total_pages']:
-        return data['transaction_details'] + get_transactions(access_token, start_date, end_date, page + 1)
+        return data['transaction_details'] + get_transactions_for_period(access_token, start_date, end_date, page + 1)
     else:
         return data['transaction_details']
 
 
-if __name__ == '__main__':
-    parser = argparse.ArgumentParser(description='Download PayPal subscriber info.')
-    parser.add_argument('start_date', help='start date inclusive (eg. 2021-11-01T00:00:00-0700)')
-    parser.add_argument('end_date', help='end date inclusive (eg. 2021-11-30T23:59:59-0700)')
-    args = parser.parse_args()
-    try:
-        client_id = os.environ['PAYPAL_CLIENT_ID']
-        client_secret = os.environ['PAYPAL_CLIENT_SECRET']
-    except KeyError:
-        sys.exit('Please set PAYPAL_CLIENT_ID and PAYPAL_CLIENT_SECRET environment variables.')
-    access_token = get_access_token(client_id, client_secret)
-    transactions = get_transactions(access_token, args.start_date, args.end_date)
-    writer = csv.writer(sys.stdout)
+def report_on_unique_profiles(transactions):
+    records = set()
     for t in transactions:
         transaction_info = t['transaction_info']
         if 'paypal_reference_id' in transaction_info:
-            writer.writerow(
+            records.add(
                 (
                     transaction_info['paypal_reference_id'],
-                    transaction_info['transaction_subject'],
+                    transaction_info.get('transaction_subject', 'NO TRANSACTION SUBJECT'),
                 ),
             )
         else:
             print(f'Skipping transaction {transaction_info["transaction_id"]} with no PayPal Reference ID.', file=sys.stderr)
+
+    for id, desc in sorted(records):
+        print(f'{id}   {desc}')
+    count = collections.Counter([id for id, _ in records])
+    dupes = [id for id, total in count.items() if total > 1]
+    print(f'Reference IDs with changing subject: {dupes}', file=sys.stderr)
+
+
+def report_on_transactions(transactions):
+    for t in transactions:
+        transaction_info = t['transaction_info']
+        if 'paypal_reference_id' in transaction_info:
+            print("%-19s %s" % (transaction_info['transaction_id'], transaction_info['paypal_reference_id']))
+            print("%-19s %s" % (transaction_info['paypal_reference_id'], transaction_info['paypal_reference_id']))
+        else:
+            print(f'Skipping transaction {transaction_info["transaction_id"]} with no PayPal Reference ID.', file=sys.stderr)
+
+    
+if __name__ == '__main__':
+    main()