Converting an Indian bank statement PDF to Excel or CSV breaks most generic converters in three predictable places. If you are writing your own parser (or choosing a tool), test these first.
1. Lakh/crore digit grouping
Indian statements print 1,23,456.78, not 123,456.78. A regex that expects groups of three digits after each comma silently drops these rows. Accept groups of two or three:
const AMOUNT = /^\(?(\d{1,3}(?:,\d{2,3})*|\d+)\.\d{2}\)?\s*(Cr|Dr)?$/i;
2. Cr / Dr instead of minus signs
Many statements have a single Amount column with a Cr or Dr suffix, and the Balance column can also be Dr when the account is overdrawn. Treat Dr as negative for the amount, and as a negative balance in the balance column.
3. DD-MM-YYYY dates
03-04-2026 is 3 April, not March 4. Detect the order across the whole statement: if any first component is above 12 it is day-first. If every date is ambiguous, default to day-first for Indian banks and let the user flip it.
4. Verify with the running balance
The most useful check is free, because every statement prints it:
previous balance + amount == this row's balance
opening balance + total == closing balance
If every row passes, nothing was dropped or misread. If a sign is ambiguous, the balance change tells you whether the row was a credit or a debit.
A ready-made option
Disclosure: I built Backfill, which handles all of the above. It runs in the browser (the PDF is never uploaded, useful for client statements), opens password-protected PDFs, verifies every row against the balance column, and exports CSV/Excel, OFX or QIF. Preview and verification are free.
It reads digital PDFs downloaded from net banking; scanned copies are not supported yet.
If your bank's layout trips it up, share the column headers (no personal data) in the comments.
Top comments (0)