The submission system wants the manuscript, supplementary tables and cover letter in one PDF, the scan from your supervisor cannot be searched, and a table needs to go into Excel. Many people reach for Acrobat or upload the file to some website. Most of these jobs have better options, and an unpublished manuscript does not need to leave your computer.
Bottom line: merging, splitting, rotating, OCR and extracting tables from text-based PDFs usually do not need Acrobat. Editing body text, complex forms, electronic signatures and compliance archiving need separate evaluation, and this article does not claim these tools replace all of Acrobat.
What was tested (2026-11-09, Linux): the command-line tools qpdf 11.9.0, OCRmyPDF 17.14.0, Tesseract 5.3.4 and tabula-py 2.10.0 were run on self-made sample PDFs; the script is pdf_tools_demo.py. Graphical tools such as PDF Arranger, PDFsam Basic, Tabula desktop and Sioyek have not yet been operated, and their descriptions rely on official sources.
1. Tools by task
| Task | Tool | Cost and licence | Platforms | Does the file leave your computer? |
|---|---|---|---|---|
| Merge, split, rotate (visual) | PDF Arranger | Free, GPL-3.0 | Windows, Linux | Local |
| Merge, split, extract pages | PDFsam Basic | Free, AGPL-3.0; paid products also exist | Windows, macOS, Linux | Local |
| Batch page operations | qpdf | Free, Apache-2.0 | Windows, macOS, Linux | Local |
| Scan to searchable PDF | OCRmyPDF | Free, MPL-2.0 | Windows, macOS, Linux | Local |
| Tables from text-based PDFs | Tabula | Free, MIT-style licence | See official downloads | Local |
| Reading papers | Sioyek | Free, GPL-3.0 | See official downloads | Local |
"Free" and "open source" are different things. The main tools here are both, but PDFsam also has paid products, so do not count their features as part of Basic.
2. Merge, split, rearrange
For a few files, a graphical tool is convenient: PDF Arranger for dragging thumbnails, rotating and cropping; PDFsam Basic for splitting or merging by page number.
For dozens of files, the command line is more dependable. These commands were actually run (a.pdf has 3 pages, b.pdf has 2):
# Merge, giving 5 pages
qpdf --empty --pages a.pdf b.pdf -- merged.pdf
# Extract pages 2-3
qpdf merged.pdf --pages . 2-3 -- part.pdf
# Rearrange: page 5 first, then page 1
qpdf merged.pdf --pages . 5,1 -- reordered.pdf
# Rotate page 1 clockwise by 90 degrees
qpdf merged.pdf --rotate=+90:1 rotated.pdf
Checking the output: the merge had 5 pages, the extract had 2, the rearranged text order was "Paper B page 2, Paper A page 1", and page 1 had a rotation of 90.
3. Tables: first check that the text is real
If you can select the text in a PDF, it is text-based; if not, it is probably an image. pdftotext can tell you too:
pdftotext file.pdf -
Tabula works on text-based PDFs. The same table was tried both ways:
| Input | Result |
|---|---|
Text-based PDF with ruled lines (lattice mode) |
4 rows, 3 columns, identical to the original |
| Scan (image only) | No table extracted |
Always compare the extraction with the original. Merged cells, tables across pages and tables without ruled lines easily shift columns.
4. Scan to searchable PDF
OCRmyPDF adds a text layer under the image, leaving the page looking essentially the same.
# English; for Chinese use -l chi_sim and install the language pack
ocrmypdf -l eng --output-type pdf scan.pdf scan_ocr.pdf
# Process the current folder in a batch (bash)
for f in *.pdf; do ocrmypdf -l eng --output-type pdf "$f" "ocr_$f"; done
No text could be extracted from the sample scan before OCR. Afterwards, all 3 sentences were recognised correctly, but in the table "S1" was read as "Sl" (the digit 1 became a lower-case L).
It is a useful reminder: OCR text is good for searching and locating, and should not be treated as data. Check sample IDs, numbers and units against the original image.
OCRmyPDF outputs PDF/A by default, and the example here requests a normal PDF. If you need long-term archiving, validate the output against your institution's requirements.
5. Reading papers
The main irritation in reading papers is jumping between equations and references. Sioyek is a PDF reader aimed at papers and technical documents, and its official repository uses GPL-3.0. If literature management is your main need, Zotero's built-in reader may already be enough.
6. Sensitive files: first ask where the file goes
Unpublished data, manuscripts under review, medical records and contracts should not be casually uploaded to online tools.
- The tools in the table above work locally, which is their main advantage for sensitive files.
- PDF24 Creator in the ZIRULAB library has Windows and web versions, so know which you are using. Web tools generally involve uploading the file, so read the privacy statement first.
- Stirling-PDF can be self-hosted. If you use someone else's deployment, the file still leaves your computer.
- If your organisation has information-security rules, they take priority.
Also, covering text with a black box does not remove it, and the text may still be there underneath. True redaction needs a specific feature and the result needs checking.
7. What may still need Acrobat
- Directly editing body text and layout.
- Complex form design, electronic signatures and approval workflows.
- True redaction, preflight and other compliance needs.
- Workflows where your organisation specifies Acrobat.
Check Adobe's official information for the scope and price of these features. Many universities and institutions already hold licences, so ask before buying.
8. Who this is for
Suitable for: researchers who often merge submission files, process scanned literature or extract tables from papers.
Not suitable for: people whose main job is editing PDF body text or maintaining complex forms.
To chain these steps together, see ZIRULAB's workflow for OCR of scanned PDFs and literature management.
