Add common OCR fallback for image-only bank statements

app/modules/bank_statement_analyzer/parsers/sbi.py

@@ -8,13 +8,26 @@ class SBIModernParser(BaseParser):
     bank_name='State Bank of India'; parser_name='SBIModernParser'
     @classmethod
     def detect(cls,text):
-        u=text.upper(); return 0.98 if 'STATE BANK OF INDIA' in u and 'REF NO./CHEQUE' in u and 'DETAILS' in u else 0
+        u=norm(text).upper()
+        bank_marker='STATE BANK OF INDIA' in u
+        digital_layout='REF NO./CHEQUE' in u and 'DETAILS' in u
+        yono_print_layout=(
+            'STATEMENT OF ACCOUNT' in u
+            and 'VALUE DATE' in u
+            and 'POST DATE' in u
+            and 'DEBIT' in u
+            and 'CREDIT' in u
+            and 'BALANCE' in u
+        )
+        return 0.98 if bank_marker and (digital_layout or yono_print_layout) 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='High')
         meta.customer_name=find(r'Account Name\s*:?\s*([^\n]+)',text)
+        if not meta.customer_name:
+            meta.customer_name=find(r'\bMs\.?\s+([A-Z][A-Z .]{3,})',text)
         meta.account_number=find(r'Account Number\s*:?\s*([0-9X*]+)',text)
-        meta.ifsc=find(r'IFS Code\s*:?\s*([A-Z0-9]+)',text)
-        m=re.search(rf'Account Statement from\s*({DATE_TOKEN_PATTERN})\s+to\s+({DATE_TOKEN_PATTERN})', text, re.I)
+        meta.ifsc=find(r'(?:IFS|IFSC) Code\s*:?\s*([A-Z0-9]+)',text)
+        m=re.search(rf'(?:Account Statement from|Statement From\s*:?)\s*({DATE_TOKEN_PATTERN})\s+to\s+({DATE_TOKEN_PATTERN})', text, re.I)
         if m:
             meta.period_from=date_iso(m.group(1)); meta.period_to=date_iso(m.group(2))
         meta.opening_balance=amount(find(r'Balance as on[^\n]*\n\s*([\d,]+\.\d{2})',text))
@@ -43,8 +56,15 @@ class SBIModernParser(BaseParser):
                     if abs((prev-txn)-bal)<0.05: debit=txn
                     elif abs((prev+txn)-bal)<0.05: credit=txn
                 if debit is None and credit is None:
-                    if p >= 78: credit=txn
-                    else: debit=txn
+                    upper=alltxt.upper()
+                    if '/CR/' in upper or upper.startswith('DEP TFR') or 'INTEREST CREDIT' in upper:
+                        credit=txn
+                    elif '/DR/' in upper or upper.startswith('WDL TFR') or upper.startswith('DEBIT '):
+                        debit=txn
+                    elif p >= 78:
+                        credit=txn
+                    else:
+                        debit=txn
             narr=first[:nums[-2].start() if len(nums)>=2 else nums[-1].start()].strip()+' '+' '.join(x.strip() for x in r['lines'][1:])
             ref=''; z=re.search(r'(?:UPI|NEFT|IMPS|RTGS)[/A-Z0-9-]{6,}',alltxt,re.I); ref=z.group(0) if z else ''
             if debit is None and credit is None: continue