Add bank statement analyzer with automatic work storage

2026-07-13 10:12:19 +05:30
parent d391e9b443
commit bbc5afe1c0
23 changed files with 1047 additions and 0 deletions
@@ -0,0 +1,53 @@
+from __future__ import annotations
+import re, pandas as pd
+from pathlib import Path
+from .base import *
+from .common import find
+
+class IndusIndParser(BaseParser):
+    bank_name='IndusInd Bank'; parser_name='IndusIndParser'
+    @classmethod
+    def detect(cls,text): return 0.98 if 'INDUSIND' in text.upper() and 'STATEMENT OF ACCOUNT' in text.upper() else 0
+    def parse(self,path,text=None):
+        text=text or extract_text(path); meta=StatementMeta(bank_name=self.bank_name,source_file=Path(path).name,parser_name=self.parser_name,confidence='Medium')
+        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)
+        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.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+(.*)$')
+        rows=[]; cur=None; page=1
+        for line in text.splitlines():
+            if '\f' in line: page+=line.count('\f')
+            m=date_re.match(line)
+            if m:
+                if cur: rows.append(cur)
+                cur={'transaction_date':m.group(1),'value_date':m.group(1),'lines':[m.group(2)],'source_page':page}
+            elif cur and line.strip() and not re.match(r'^(Statement Summary|Opening Balance|Total Withdrawal|Branch Address|The limits)',line.strip(),re.I): cur['lines'].append(line)
+        if cur: rows.append(cur)
+        out=[]; prev=None
+        for r in rows:
+            first=r['lines'][0]; alltxt=norm(' '.join(r['lines']))
+            # Typical: narration ref debit Dr credit Cr balance Cr
+            tokens=list(re.finditer(r'([\d,]+\.\d{2})\s*(Dr|Cr)',first,re.I))
+            if not tokens: continue
+            debit=credit=bal=None
+            if len(tokens)>=2:
+                bal=amount(tokens[-1].group(1))
+                txn=amount(tokens[-2].group(1)); typ=tokens[-2].group(2).upper()
+                if typ=='DR': debit=txn
+                else: credit=txn
+            elif len(tokens)==1:
+                # carried forward / balance-only line; skip
+                continue
+            narr=first[:tokens[-2].start()].strip()+' '+' '.join(x.strip() for x in r['lines'][1:])
+            ref=''; z=re.search(r'\b([A-Z0-9]{8,})\b',narr); ref=z.group(1) if z else ''
+            out.append({**r,'narration':norm(narr),'reference_no':ref,'debit':debit,'credit':credit,'balance':bal})
+        if out:
+            f=out[0]; meta.opening_balance=round((f['balance'] or 0)+(f.get('debit') or 0)-(f.get('credit') or 0),2)
+            meta.closing_balance=out[-1]['balance']
+        return meta,finalize(pd.DataFrame(out),meta)