Turn a GST return into a spreadsheet

A filed return is a PDF of tables, and everything you want to do with it afterwards needs those tables as numbers. This recovers the grid and writes figures that a spreadsheet can total.

• A figure printed as 48,20,000 is read as forty-eight lakh, not as 48.2.
• The grid is recovered from the page geometry, not from spaces.
• The return never leaves this browser tab.

Indian grouping is not a display preference

A parser that assumes groups of three reads 48,20,000 as forty-eight point two, which is off by a factor of a hundred thousand and looks perfectly reasonable in a cell. That is the sort of error that survives a review, because nothing about it draws the eye.

Grouping is read here rather than assumed: a final group of three with groups of two above it is Indian, and the figure is both parsed and displayed that way. Two groups on their own are not evidence either way, so they stay western, and the decision is made once for a sheet rather than once per figure, because a document prints one way throughout.

What comes across and what does not

The tables come across as tables, with the header row bold on a fill and frozen so it stays in view. Column widths are set to what has to fit, because Excel's default width shows every date and every grouped amount as a row of hashes.

What does not come across is the form itself: the boxes, the signatures and the layout of the return. This recovers data, not a facsimile. For an archived copy of the return as filed, keep the PDF; the spreadsheet is for the arithmetic you are going to do next.

A downloaded return is read straight off the page, which is more accurate than any recogniser because the words come from whoever wrote the file. A printed and scanned one goes through recognition first, which is the first step in the plan and the reason it is there.

Does it work for other tax forms?

It reads tables rather than a specific form, so yes for anything laid out as a grid.

Will the totals match the return?

Check them. The figures are recovered from the page, and a scanned return can lose a digit without saying so.

Support and billing: support@gosmartpdf.com. Operated by Mrityunjay Kumar, India. Payments by Dodo Payments, our merchant of record.

Loading the editor…