Mercurial > public > finance-parser
changeset 9:bf19235a9636
minor bugs and add sample reports
| author | Dennis C. M. <dennis@denniscm.com> |
|---|---|
| date | Wed, 07 Jun 2023 11:07:14 +0100 |
| parents | f2442b8f43de |
| children | 2350662483a3 |
| files | analyze_document/app.py process_document/app.py reports/2022/acx_balance_1.pdf reports/2022/acx_balance_2.pdf reports/2022/acx_pnl_1.pdf reports/2022/itx_balance_1.pdf reports/2022/itx_pnl_1.pdf reports/2022/san_balance_1.pdf reports/2022/san_balance_2.pdf reports/2022/san_balance_3.pdf reports/2022/san_balance_4.pdf reports/2022/san_pnl_1.pdf reports/2022/san_pnl_2.pdf reports/itx_balance.pdf reports/san_balance.pdf |
| diffstat | 15 files changed, 2 insertions(+), 2 deletions(-) [+] |
line wrap: on
line diff
--- a/analyze_document/app.py Mon Jun 05 20:07:22 2023 +0100 +++ b/analyze_document/app.py Wed Jun 07 11:07:14 2023 +0100 @@ -14,7 +14,7 @@ object_key = event_detail['object']['key'] company_ticker = re.search('unprocessed/(.*)_', object_key).group(1) - doc_type = re.search(f'unprocessed/{company_ticker}_(.*).pdf', object_key).group(1) + doc_type = re.search(f'unprocessed/{company_ticker}_(.*)_', object_key).group(1) file_id = str(uuid.uuid4()) data_dict = textract_client.analyze_document(
--- a/process_document/app.py Mon Jun 05 20:07:22 2023 +0100 +++ b/process_document/app.py Wed Jun 07 11:07:14 2023 +0100 @@ -152,7 +152,7 @@ for amount_format in amount_formats: plural_amount_format = f'{amount_format}s' - if amount_format in phrase or plural_amount_format in phrase: + if amount_format in phrase.lower() or plural_amount_format in phrase.lower(): return amount_format