Check an invoice table whose values are right aligned

Where the extra work starts

Include that layout in your trial and check the row associations before deciding the output saves work.

Upload a representative supplier invoice, inspect its extracted fields and line items beside the PDF, then download Excel. Put the awkward invoice in your trial too: the useful result is less total work after corrections.

Visual alignment is evidence to inspect, not proof of the parser's mapping. ParseMill reads native invoice text and uses English OCR on textless PDF pages. Either route still needs the quantity, unit price and amount checked against their original columns.

A practical way to get it right

Choose one row with short numbers and one with wider values. Inspect the quantity, price and amount under their headings at a readable zoom. Note whether decimal alignment or right edges differ from the header positions. Copy the text to check that characters survive, but retain the visual page for column relationships.

After a bounded extraction, compare each candidate field with the source location. A correct row total does not prove that quantity and price were assigned correctly, and a value near a column boundary deserves direct inspection. Record any shift or concatenation as an extraction issue.

If the layout does not fit reliably, stop the affected input class and use reviewed transcription or another suitable method. Do not apply a blind global column shift unless the source confirms it for every affected row and page.

Adobe: Copy content from PDFs describes text selection and copying and how to check document restrictions. Compare copied characters with the visual column relationship instead of using a flat text paste as the only mapping check.

What this looks like on a real task

Synthetic example. The people, references and figures below are invented to demonstrate the method.

The concatenated amount is not repaired by assuming which digits belong to which field. Return to the source, correct with evidence if the workflow permits and inspect neighboring rows for the same pattern. Copy the field, visible value and candidate columns into a small alignment review fixture.

Catch the cases that create rework

A thousands separator can make a number wider without changing its meaning, while a missing separator can change the raw string. Preserve the source representation. A heading centered above a numeric column can also have a different anchor from its values; do not assume the visual center is the parser's boundary.

If a screenshot copies as an image, it can help illustrate the layout but does not provide native text for extraction.

Microsoft: Power Query PDF connector exposes detected table structure and transformation options. Inspect the actual candidate cells; detection alone does not establish the intended quantity, price and amount mapping.

Decide whether the change is worth keeping

The accepted result should show source-backed field assignments for both narrow and wide values, with no unexplained concatenation or shift. Include varied numeric widths in a repeatable layout test. Measure correction effort if alignment issues recur, since frequent manual repair can outweigh the apparent convenience of reusing the same extraction setup.