Tables and forms to spreadsheet
Goal
Get data that lives in PDFs into a spreadsheet you can sort and add up.
The scenario. Your club sent out a PDF registration form to 40 members and 31 sent it back filled in. Separately, the treasurer has last year's accounts as a PDF with a table of 60 rows. You want both in Excel.
Honest expectation. There is no single "PDF to spreadsheet" button that always works. What works depends on what the PDF is. Choose your road below.
Which road? (decide before you start)
| Your PDF is | Use | Result |
|---|---|---|
| A fillable form, filled in and returned | Extract Form Data | One CSV per form (reliable: it reads the real fields) |
| A text PDF with a table | Converter PDF to Excel with Extract tables only | An XLSX (good for clean grids, may need touch-up) |
| A scan of a table | Searchable with Editable Excel and Extract tables only | An XLSX from recognised text (approximate) |
| A text file with columns (TXT) | Converter TXT to Excel, Delimiter matched | An XLSX |
What does not produce a spreadsheet:
- Parser outputs a ZIP of text, images, attachments, annotations, bookmarks and form data (as an XFDF file meant for software). It has no tables file.
- Extract Text gives plain text with no columns.
- Add Table goes the other way: xlsx into a PDF.
Chain
Form road: (Splitter) → Extract Form Data → open CSV
Table road: (Unlock) → Converter (PDF to Excel, tables only) [scans: Searchable → Excel]
Step by step
Road A: forms to CSV
- Check they are real forms. Open one in a viewer: can you click into a field? If a form was printed and filled by hand, or was flattened, there are no fields to read (see the fillable-form recipe for why order matters).
- If a returned file contains several forms, Splitter first. Unlock protected ones.
- Same template only. Forms from different templates give columns that do not line up; do each template separately.
- Extract Form Data: upload all (it has no options) and press Extract.
- Download the ZIP. You get one CSV per form, form_data_summary.csv (status, fields, a sample), and form_data_export.csv (one line per item, semicolon between name and value).
- Open the CSV in a spreadsheet, importing as UTF-8 to avoid odd characters. Combine the per-form CSVs by pasting them under the same header.
Road B: a table in a text PDF to XLSX
- Converter, the PDF to Excel page.
- Excel export mode: Extract tables only (default is the entire document, which is less tidy).
- Press CONVERT, download, open in Excel.
- Variation: Converter PDF to XML offers a Format option of Excel (Open XML); use it only if the above leaves your table broken (verify on the live app).
Road C: a scanned table to XLSX
- Rotation so pages are upright.
- Searchable: Document language matches, Deskew on, Output format: Editable Excel (.xlsx), Extract tables only on.
- The Excel output is built from recognised text and the page picture is dropped; complex layouts are only approximately preserved.
Expected result
Road A: a CSV with field names in the header and the answers below. Roads B and C: an XLSX with the detected tables, often needing manual cleanup (merged cells, split numbers).
Check it worked
- Road A: count rows against the number of forms returned. Failed files are marked in the summary.
- Roads B and C: check the totals against the PDF; one misread digit looks like a real number. Recognition is an estimate.
- Open one empty CSV value and ask whether the form was flattened or the field was truly blank.
Variations
- Form data for another PDF tool: Parser gives
formdata.xfdf(not spreadsheet-friendly). - Back into a PDF: later, insert a cleaned sheet as a table with Add Table.
- Mass forms: for 200 forms use the CLI (ExtractFormData plugin) and see Make it repeatable.
What can go wrong
- CSV with headers but no values. Not filled in, or flattened.
- "Nothing found". A scan, or a normal PDF with lines drawn on it.
- Converter result is not a clean grid. PDF reconstruction is approximate; tables-only helps, not always.
- Merged cells and multi-line cells confuse any table detection.
- Signatures and unusual widgets may give no useful values (verify with your form).
Why this order
Real form fields beat guessing, so forms use the exact-reader. For tables, scans must get a text layer before anything can find rows, so recognition comes before conversion to Excel. Unlock and Splitter come first because the extractors cannot open locked files or separate stacked forms.