Fix Indian Bank extraction and add shared multi-format dates for all banks

This commit is contained in:
A R R R Associates
2026-07-16 21:56:08 +05:30
parent 43ab9c9aae
commit d9ae1519a4
10 changed files with 340 additions and 311 deletions
@@ -2,7 +2,7 @@ from __future__ import annotations
import re, pandas as pd
from pathlib import Path
from .base import *
from .common import find
from .common import DATE_TOKEN_PATTERN, date_iso, find
class IndusIndParser(BaseParser):
bank_name='IndusInd Bank'; parser_name='IndusIndParser'
@@ -13,13 +13,13 @@ class IndusIndParser(BaseParser):
meta.account_number=find(r'Account Number\s*:\s*([0-9X*]+)',text)
meta.customer_id=find(r'Cust\.Reln\.No\s*:\s*([0-9X*]+)',text)
meta.ifsc=find(r'IFSC Code\s*:\s*([A-Z0-9]+)',text)
m=re.search(r'Period\s*:\s*(\d{2}[/-][A-Za-z0-9]{2,3}[/-]\d{2,4})\s*(?:to|TO|-)\s*(\d{2}[/-][A-Za-z0-9]{2,3}[/-]\d{2,4})',text,re.I)
m=re.search(rf'Period\s*:\s*({DATE_TOKEN_PATTERN})\s*(?:to|-)\s*({DATE_TOKEN_PATTERN})', text, re.I)
if m:
meta.period_from=pd.to_datetime(m.group(1),dayfirst=True).strftime('%Y-%m-%d'); meta.period_to=pd.to_datetime(m.group(2),dayfirst=True).strftime('%Y-%m-%d')
meta.period_from=date_iso(m.group(1)); meta.period_to=date_iso(m.group(2))
meta.total_debit=amount(find(r'Total Withdrawal Amount\s*:\s*([\d,]+\.\d{2})',text))
meta.total_credit=amount(find(r'Total Deposit Amount\s*:\s*([\d,]+\.\d{2})',text))
# rows start with date and often end with amount Dr/Cr and balance Cr
date_re=re.compile(r'^\s*(\d{2}[-/]?[A-Za-z]{3}[-/]?\d{2,4}|\d{2}-\d{2}-\d{4})\s+(.*)$')
date_re=re.compile(rf'^\s*({DATE_TOKEN_PATTERN})\s+(.*)$', re.I)
rows=[]; cur=None; page=1
for line in text.splitlines():
if '\f' in line: page+=line.count('\f')