Word wants text. The PDF wanted a printer.
People google “PDF to Word” expecting a clone of a designed page. What they get from upload farms is often a worse clone plus a copy of the file on someone else’s disk. This page extracts the text layer with pdf.js and packs paragraphs into a .docx with the docx library. That is the job: a draft you can edit, not a facsimile.
Worked example: 8-page employment contract
Convert. Open in Word. Clause numbers should be there; the signature block image will not. That is correct. You wanted the words.
Worked example: two-column judgment PDF
pdf.js reads in painting order. Columns will interleave. If the download reads like a blender, it was a layout PDF. Keep the original for reading; use this only to search-and-replace a phrase.
Worked example: scanned rent agreement
Status will say almost no text. Stop. Photograph-PDFs need OCR, which is a different promise. I will not invent clauses.
Questions
Is the PDF uploaded?
No. Extraction and .docx packing stay in this tab.
Will my brochure look like InDesign?
No. You get paragraphs of extracted text. Tables become lines. Headers repeat.
Scanned PDF?
If almost no characters come out, it is a picture of pages. This tool does not OCR. Use a later OCR path or type.
Password?
Enter it here. Unlock-first is optional and would flatten text away — do not unlock if you want words.
Images?
Not copied into the Word file. Text only.
Track changes / comments?
Gone. New document.