Question 1

Why does this scan instead of converting?

Accepted Answer

Because extraction is priced per table-bearing page, and you cannot tell how many of those a document has by looking at it. The scan is free and answers that. Converting two pages free and paywalling the third would tell you less.

Question 2

Does it work on tables with no lines around them?

Accepted Answer

Yes. Bank statements and financial appendices are usually set with whitespace alignment and no rulings at all. Those are found by looking at where the columns line up across consecutive rows rather than by looking for lines.

Question 3

What about a scanned PDF?

Accepted Answer

A scan is a picture, so there are no rulings or glyph positions to read and the scanner reports no tables. The parse API with ocr=force reads it as text first, priced per page because OCR costs per page.

Question 4

Will the numbers arrive as numbers?

Accepted Answer

Yes. Values are coerced on the way into the spreadsheet, including accounting negatives written as (1,234.50). A date fragment like 2024-01 is left as text rather than being read as a subtraction.

PDF to Excel

Related