Extract tables from a PDF
Load a PDF with a table (the bank statement, the price list, the timetable) and get it back in the right columns, ready to copy into Excel or download as CSV. No more copy-paste that spreads everything into one column. The file never leaves your browser.
Why copy-pasting a PDF loses the columns
Inside a PDF there are no «tables»: there are many little pieces of text, each with its position on the page. When you copy and paste, the computer lines them up and the columns disappear. Here instead the coordinates of each piece of text are read: those at the same height make a row, those in a column make a column, and the table is rebuilt as you see it on the sheet.
Copy for Excel or download as CSV
The Copy for Excel button puts the table on the clipboard in the format Excel and Google Sheets understand at once: you paste and the columns separate by themselves. The CSV is a file, with the right separator for your language. The first 60 rows show in the preview; the file has them all.
If the columns are off, you fix them
Automatic recognition works well on tables with well-separated columns, but a document can have two columns so close they look like one, or a long description that overflows. That is why there is the Columns: if the recognised number is wrong, you change it and the table is rebuilt split into that many equal columns. It is not magic that always guesses: it is a tool that lets you correct when needed, instead of giving you a wrong result without saying so.
How it is checked to be right
Extracting a table is a heuristic, and a heuristic must be tested against an independent judge: here the table pulled from a PDF is compared, cell by cell, with pdfplumber, a library that does exactly this. And on a fake bank statement the sum of the amounts column must add up to the cent: it is the thing anyone holding the real statement would check first.
What it does not do
It does not read scans: if the PDF is a photo of a sheet, there is no text inside but only an image, and there is nothing to column there (it tells you, and sends you to Image to text). It does not rebuild tables with complicated merged cells, and on very close columns it can slip: that is what the manual correction is for. If you just need the text without columns there is Extract text from PDF, and to open a CSV in Excel without wrecking the numbers CSV to Excel.