Rakseeibra/arrr-erp
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add SBI Account Summary layout template parser

Browse Source
This commit is contained in:
A R R R Associates
2026-08-04 10:29:49 +05:30
parent 7db38c7b04
commit 1dbde5e0b9
2 changed files with 107 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
app/modules/bank_statement_analyzer/parsers/registry.py
+4 -3
View File
@@ -1,4 +1,4 @@
from __future__ import annotations
from __future__ import annotations
from .idfc import IDFCFirstParser
from .axis import AxisParser
@@ -8,7 +8,7 @@ from .hsbc import HSBCParser
from .indian_bank import IndianBankModernParser, IndianBankLegacyParser
from .indusind import IndusIndParser
from .kotak import KotakParser
from .sbi import SBIModernParser, SBIOtherParser
from .sbi import SBIAccountSummaryParser, SBIModernParser, SBIOtherParser
from .central_bank_of_india import CentralBankOfIndiaParser
from .yes_bank import YesBankParser
from .city_union_bank import CityUnionBankParser
@@ -31,6 +31,7 @@ PARSERS = [
IndianBankLegacyParser,
IndusIndParser,
KotakParser,
SBIAccountSummaryParser,
SBIOtherParser,
SBIModernParser,
]
@@ -67,7 +68,7 @@ BANK_PARSERS = {
"indian_bank": [IndianBankModernParser, IndianBankLegacyParser],
"indusind": [IndusIndParser],
"kotak": [KotakParser],
"sbi": [SBIOtherParser, SBIModernParser],
"sbi": [SBIAccountSummaryParser, SBIOtherParser, SBIModernParser],
}
app/modules/bank_statement_analyzer/parsers/sbi.py
+103 -4
View File
@@ -26,7 +26,39 @@ _HEADER_RE = re.compile(
def _summary_metrics(text: str) -> dict[str, float | int] | None:
"""Extract SBI's printed statement summary without depending on line wrapping."""
match = re.search(
r"Brought\s+Forward.*?Dr\s+Count\s+Cr\s+Count.*?"
r"Total\s+Debits.*?Total\s+Credits.*?Closing\s+Balance.*?"
r"([\d,]+\.\d{2})\s*(CR|DR)?\s+"
r"(\d{1,6})\s+(\d{1,6})\s+"
r"([\d,]+\.\d{2})\s+([\d,]+\.\d{2})\s+"
r"([\d,]+\.\d{2})\s*(CR|DR)?",
text,
re.I | re.S,
)
if not match:
return None
def signed(value: str, suffix: str | None) -> float:
parsed = float(value.replace(",", ""))
return -parsed if (suffix or "").upper() == "DR" else parsed
return {
"opening_balance": signed(match.group(1), match.group(2)),
"debit_count": int(match.group(3)),
"credit_count": int(match.group(4)),
"total_debit": float(match.group(5).replace(",", "")),
"total_credit": float(match.group(6).replace(",", "")),
"closing_balance": signed(match.group(7), match.group(8)),
}
def _expected_summary_transactions(text: str) -> int | None:
summary = _summary_metrics(text)
if summary:
return int(summary["debit_count"]) + int(summary["credit_count"])
counts = re.findall(
r"(?:Brought\s+Forward.*?)(\d{1,3})\s+(\d{1,3})\s+[\d,]+\.\d{2}\s+[\d,]+\.\d{2}",
text,
@@ -37,6 +69,16 @@ def _expected_summary_transactions(text: str) -> int | None:
return sum(int(debit_count) + int(credit_count) for debit_count, credit_count in counts)
def _apply_printed_summary(meta: StatementMeta, text: str) -> None:
summary = _summary_metrics(text)
if not summary:
return
meta.opening_balance = float(summary["opening_balance"])
meta.total_debit = float(summary["total_debit"])
meta.total_credit = float(summary["total_credit"])
meta.closing_balance = float(summary["closing_balance"])
def _ocr_yono_page(pdf_path: Path, page_number: int, work_dir: Path, dpi: int) -> tuple[int, str]:
prefix = work_dir / f"sbi_yono_{page_number:05d}"
image_path = prefix.with_suffix(".png")
@@ -211,7 +253,7 @@ def _parse_rows(
# physical line. Split before every detected date pair so each transaction
# reaches the normal row collector independently.
text = re.sub(
r"(?<!\n)(?=(?:\d{1,2}[/-]\d{1,2}[/-]\d{2,4})\s+"
r"(?<=[^\n\d])(?=(?:\d{1,2}[/-]\d{1,2}[/-]\d{2,4})\s+"
r"(?:\d{1,2}[/-]\d{1,2}[/-]\d{2,4})\s+)",
"\n",
text,
@@ -226,9 +268,10 @@ def _parse_rows(
rows: list[dict] = []
current: dict | None = None
pending_prefix: list[str] = []
page = 1
for raw_line in text.splitlines():
for raw_line in text.split("\n"):
page += _page_increment(raw_line)
line = raw_line.replace("\f", "").rstrip()
match = start_re.match(line)
@@ -242,11 +285,24 @@ def _parse_rows(
"transaction_date": transaction_date,
"value_date": value_date,
"lines": [remainder],
"prefix_lines": list(pending_prefix),
"source_page": page,
}
pending_prefix = []
continue
stripped = line.strip()
is_mode_prefix = bool(
stripped
and re.fullmatch(
r"(?:DEP|WDL)\s+TFR(?:\s+INB.*)?|INTEREST\s+CREDIT|CEMTEX\s+DEP",
stripped,
re.I,
)
)
if is_mode_prefix:
pending_prefix = [stripped]
continue
if current and stripped and not _HEADER_RE.match(stripped):
current["lines"].append(stripped)
@@ -257,7 +313,7 @@ def _parse_rows(
previous_balance = meta.opening_balance
for row in rows:
first_line = row["lines"][0]
full_text = norm(" ".join(row["lines"]))
full_text = norm(" ".join([*row.get("prefix_lines", []), *row["lines"]]))
debit, credit, balance, first_narration = _row_values(
first_line,
full_text,
@@ -266,7 +322,7 @@ def _parse_rows(
if balance is None:
continue
narration = norm(" ".join([first_narration, *row["lines"][1:]]))
narration = norm(" ".join([*row.get("prefix_lines", []), first_narration, *row["lines"][1:]]))
if debit is None and credit is None and previous_balance is not None:
delta = round(balance - previous_balance, 2)
if delta < -0.01:
@@ -354,6 +410,49 @@ class SBIYONORelationshipParser(BaseParser):
return meta, frame
class SBIAccountSummaryParser(BaseParser):
"""SBI Account Summary layout with dual dates and a printed summary page."""
bank_name = "State Bank of India"
parser_name = "SBIAccountSummaryParser"
@classmethod
def detect(cls, text: str) -> float:
upper = norm(text).upper()
required = (
"STATE BANK OF INDIA",
"ACCOUNT SUMMARY",
"STATEMENT OF ACCOUNT",
"STATEMENT SUMMARY",
"BROUGHT FORWARD",
"DR COUNT",
"CR COUNT",
)
if all(marker in upper for marker in required):
return 0.998
return 0.0
def parse(self, path, text=None):
extracted_text = text or extract_text(path)
meta = _extract_common_meta(extracted_text, path, self.parser_name)
_apply_printed_summary(meta, extracted_text)
if meta.opening_balance is None:
meta.opening_balance = _opening_from_summary(extracted_text)
start_re = re.compile(
rf"^\s*({DATE_TOKEN_PATTERN})\s+({DATE_TOKEN_PATTERN})\s+(.*)$",
re.I,
)
frame = _parse_rows(extracted_text, meta, start_re, 2)
summary = _summary_metrics(extracted_text)
if summary and len(frame) != int(summary["debit_count"]) + int(summary["credit_count"]):
raise ValueError(
"SBI Account Summary rows could not be fully reconciled with the printed "
"debit and credit counts. Please retain the PDF and contact support."
)
return meta, frame
class SBIStandardParser(BaseParser):
bank_name = "State Bank of India"
parser_name = "SBIStandardParser"