Add Central Bank of India and YES Bank statement parsers

2026-08-03 16:32:40 +05:30
parent b1b088d4f2
commit d447e6be8b
3 changed files with 222 additions and 1 deletions
@@ -0,0 +1,116 @@
+from __future__ import annotations
+
+import re
+from pathlib import Path
+
+import pandas as pd
+import pdfplumber
+
+from .base import BaseParser, StatementMeta, amount, extract_text, finalize, norm
+from .common import date_iso, find, infer_mode
+
+
+class CentralBankOfIndiaParser(BaseParser):
+    bank_name = "Central Bank of India"
+    parser_name = "CentralBankOfIndiaParser"
+
+    @classmethod
+    def detect(cls, text: str) -> float:
+        upper = (text or "").upper()
+        score = 0.0
+        if "CENTRAL BANK OF INDIA" in upper:
+            score += 0.60
+        if "CBIN0" in upper or "IFSC CODE: CBIN" in upper:
+            score += 0.20
+        if all(token in upper for token in ("POST DATE", "VALUE DATE", "ACCOUNT DESCRIPTION", "DEBIT", "CREDIT", "BALANCE")):
+            score += 0.18
+        if "END OF STATEMENT - FROM INTERNET BANKING" in upper:
+            score += 0.02
+        return min(score, 0.99)
+
+    @staticmethod
+    def _signed_balance(value) -> float | None:
+        if value is None:
+            return None
+        text = norm(value).upper()
+        parsed = amount(text)
+        if parsed is None:
+            return None
+        return -abs(parsed) if text.endswith("DR") else abs(parsed)
+
+    @staticmethod
+    def _reference_number(narration: str, cheque_no: str) -> str:
+        if norm(cheque_no):
+            return norm(cheque_no)
+        for pattern in (
+            r"\bRRN\s*[:/]?\s*(\d{10,18})\b",
+            r"\b(CBINN\d{10,})\b",
+            r"\b([A-Z]{4,8}R\d{10,})\b",
+            r"\b(\d{12,18})\b",
+        ):
+            match = re.search(pattern, narration, re.I)
+            if match:
+                return match.group(1)
+        return ""
+
+    def parse(self, path, text=None):
+        pdf_path = Path(path)
+        text = text or extract_text(pdf_path)
+        meta = StatementMeta(
+            bank_name=self.bank_name,
+            source_file=pdf_path.name,
+            parser_name=self.parser_name,
+            confidence="High",
+        )
+        meta.customer_name = norm(find(r"Account Number:\s*\d+.*?\n(?:Product Type:.*?\n)?([A-Z][A-Z .]+)", text, flags=re.I | re.S))
+        meta.account_number = find(r"Account Number:\s*([0-9]+)", text)
+        meta.ifsc = find(r"IFSC Code:\s*([A-Z0-9]+)", text)
+        period = re.search(r"STATEMENT OF ACCOUNT\s+from\s+(\d{2}/\d{2}/\d{4})\s+to\s+(\d{2}/\d{2}/\d{4})", text, re.I)
+        if period:
+            meta.period_from = date_iso(period.group(1))
+            meta.period_to = date_iso(period.group(2))
+
+        rows: list[dict] = []
+        with pdfplumber.open(str(pdf_path)) as pdf:
+            for page_no, page in enumerate(pdf.pages, start=1):
+                for table in page.extract_tables() or []:
+                    if not table:
+                        continue
+                    header = [norm(cell).upper() for cell in table[0]]
+                    if not ("POST DATE" in header and "ACCOUNT DESCRIPTION" in header and "BALANCE" in header):
+                        continue
+                    for raw in table[1:]:
+                        cells = list(raw) + [None] * (8 - len(raw))
+                        post_date, value_date, branch_code, cheque_no, description, debit, credit, balance = cells[:8]
+                        post_date = norm(post_date)
+                        if not re.fullmatch(r"\d{2}/\d{2}/\d{4}", post_date):
+                            continue
+                        narration = norm(description)
+                        debit_value = amount(debit)
+                        credit_value = amount(credit)
+                        balance_value = self._signed_balance(balance)
+                        if debit_value is None and credit_value is None:
+                            continue
+                        rows.append({
+                            "transaction_date": post_date,
+                            "value_date": norm(value_date) or post_date,
+                            "narration": narration,
+                            "reference_no": self._reference_number(narration, norm(cheque_no)),
+                            "debit": debit_value,
+                            "credit": credit_value,
+                            "balance": balance_value,
+                            "source_page": page_no,
+                            "mode": infer_mode(narration),
+                        })
+
+        data = pd.DataFrame(rows)
+        if not data.empty:
+            first = data.iloc[0]
+            first_balance = float(first["balance"])
+            first_debit = float(first["debit"]) if pd.notna(first["debit"]) else 0.0
+            first_credit = float(first["credit"]) if pd.notna(first["credit"]) else 0.0
+            meta.opening_balance = round(first_balance + first_debit - first_credit, 2)
+            meta.total_debit = round(float(pd.to_numeric(data["debit"], errors="coerce").fillna(0).sum()), 2)
+            meta.total_credit = round(float(pd.to_numeric(data["credit"], errors="coerce").fillna(0).sum()), 2)
+            meta.closing_balance = round(float(data.iloc[-1]["balance"]), 2)
+        return meta, finalize(data, meta)
@@ -1,4 +1,4 @@
-from __future__ import annotations
+from __future__ import annotations

 from .idfc import IDFCFirstParser
 from .axis import AxisParser
@@ -9,11 +9,15 @@ from .indian_bank import IndianBankModernParser, IndianBankLegacyParser
 from .indusind import IndusIndParser
 from .kotak import KotakParser
 from .sbi import SBIModernParser, SBIOtherParser
+from .central_bank_of_india import CentralBankOfIndiaParser
+from .yes_bank import YesBankParser
 from .base import extract_text

 PARSERS = [
    IDFCFirstParser,
    AxisParser,
+    CentralBankOfIndiaParser,
+    YesBankParser,
    HDFCParser,
    ICICIParser,
    HSBCParser,
@@ -28,6 +32,8 @@ PARSERS = [
 BANK_OPTIONS = [
    ("auto", "Auto Detect"),
    ("axis", "Axis Bank"),
+    ("central_bank_of_india", "Central Bank of India"),
+    ("yes_bank", "YES Bank"),
    ("hdfc", "HDFC Bank"),
    ("icici", "ICICI Bank"),
    ("hsbc", "HSBC Bank"),
@@ -40,6 +46,8 @@ BANK_OPTIONS = [

 BANK_PARSERS = {
    "axis": [AxisParser],
+    "central_bank_of_india": [CentralBankOfIndiaParser],
+    "yes_bank": [YesBankParser],
    "hdfc": [HDFCParser],
    "icici": [ICICIParser],
    "hsbc": [HSBCParser],
@@ -95,3 +103,4 @@ def parse_pdf(path, bank_hint: str | None = None):
            "Unsupported statement format. Select the bank manually or add a bank-specific parser for this statement layout."
        )
    return parser.parse(path, text)
+
@@ -0,0 +1,96 @@
+from __future__ import annotations
+
+import re
+from pathlib import Path
+
+import pandas as pd
+import pdfplumber
+
+from .base import BaseParser, StatementMeta, amount, extract_text, finalize, norm
+from .common import date_iso, find, infer_mode
+
+
+class YesBankParser(BaseParser):
+    bank_name = "YES Bank"
+    parser_name = "YesBankParser"
+
+    @classmethod
+    def detect(cls, text: str) -> float:
+        upper = (text or "").upper()
+        score = 0.0
+        if "YES BANK" in upper:
+            score += 0.55
+        if "YESB000" in upper or "IFSC CODE: YESB" in upper:
+            score += 0.20
+        if "TRANSACTION DETAILS FOR YOUR ACCOUNT NUMBER" in upper:
+            score += 0.15
+        if all(token in upper for token in ("TRANSACTION DATE", "VALUE DATE", "WITHDRAWALS", "DEPOSITS", "RUNNING")):
+            score += 0.10
+        return min(score, 0.99)
+
+    def parse(self, path, text=None):
+        pdf_path = Path(path)
+        text = text or extract_text(pdf_path)
+        meta = StatementMeta(
+            bank_name=self.bank_name,
+            source_file=pdf_path.name,
+            parser_name=self.parser_name,
+            confidence="High",
+        )
+        meta.customer_name = norm(find(r"Primary Holder:\s*([^\n]+?)(?:\s+A/C Opening Date:|\n)", text, flags=re.I))
+        if not meta.customer_name:
+            meta.customer_name = norm(find(r"Primary Account Holder Name:\s*([^\n]+)", text, flags=re.I))
+        meta.customer_id = find(r"(?:Cust Id|Customer Id):\s*([0-9]+)", text, flags=re.I)
+        meta.account_number = find(r"Statement of account:\s*([0-9]+)", text, flags=re.I)
+        if not meta.account_number:
+            meta.account_number = find(r"account number\s+([0-9]+)", text, flags=re.I)
+        meta.ifsc = find(r"IFSC Code:\s*([A-Z0-9]+)", text, flags=re.I)
+        period = re.search(r"Period:\s*From\s+(.+?)\s+To\s+([^\n]+)", text, re.I)
+        if period:
+            meta.period_from = date_iso(period.group(1).strip())
+            meta.period_to = date_iso(period.group(2).strip())
+
+        rows: list[dict] = []
+        with pdfplumber.open(str(pdf_path)) as pdf:
+            for page_no, page in enumerate(pdf.pages, start=1):
+                for table in page.extract_tables() or []:
+                    if not table:
+                        continue
+                    header = [norm(cell).upper() for cell in table[0]]
+                    if not ("TRANSACTION DATE" in header and "VALUE DATE" in header and "DESCRIPTION" in header):
+                        continue
+                    for raw in table[1:]:
+                        cells = list(raw) + [None] * (7 - len(raw))
+                        txn_date, value_date, reference_no, description, withdrawal, deposit, balance = cells[:7]
+                        txn_date = norm(txn_date)
+                        if not re.fullmatch(r"\d{2}-[A-Za-z]{3}-\d{4}", txn_date):
+                            continue
+                        debit_value = amount(withdrawal)
+                        credit_value = amount(deposit)
+                        balance_value = amount(balance)
+                        if debit_value is None and credit_value is None:
+                            continue
+                        narration = norm(description)
+                        rows.append({
+                            "transaction_date": txn_date,
+                            "value_date": norm(value_date) or txn_date,
+                            "narration": narration,
+                            "reference_no": norm(reference_no),
+                            "debit": debit_value,
+                            "credit": credit_value,
+                            "balance": balance_value,
+                            "source_page": page_no,
+                            "mode": infer_mode(narration),
+                        })
+
+        data = pd.DataFrame(rows)
+        if not data.empty:
+            first = data.iloc[0]
+            first_balance = float(first["balance"])
+            first_debit = float(first["debit"]) if pd.notna(first["debit"]) else 0.0
+            first_credit = float(first["credit"]) if pd.notna(first["credit"]) else 0.0
+            meta.opening_balance = round(first_balance + first_debit - first_credit, 2)
+            meta.total_debit = round(float(pd.to_numeric(data["debit"], errors="coerce").fillna(0).sum()), 2)
+            meta.total_credit = round(float(pd.to_numeric(data["credit"], errors="coerce").fillna(0).sum()), 2)
+            meta.closing_balance = round(float(data.iloc[-1]["balance"]), 2)
+        return meta, finalize(data, meta)