Manage PDFs Offline: Merge, Split, Compress Without Uploading

A visa agent once emailed me a 34-file application — scans, forms, bank statements — as 34 separate attachments with names like “scan_final_FINAL2.pdf”. The consulate asked for one ordered PDF. Merging took four minutes; renaming discipline would have saved forty. That afternoon taught me PDF workflow: merge, split, compress and convert as one local, private pipeline. This guide covers all eight operations with the exact limits (20 files, 200 pages, 10MB each) so you never discover them mid-deadline.

Here is the promise: you will learn to merge PDFs in order, split by page ranges, compress to 1MB for email, convert scans to PDF and pages to JPG, rotate sideways scans and stamp watermarks — entirely offline, files never uploaded. I ran every flow below in October 2026 with real documents: a 5-report bundle, a 120-page manual, an 18MB photo deck. Work alongside me starting at our free PDF merge tool.

Part of the blog guides. Merge in PDF merge; split in PDF split; compress in PDF compress; scans to PDF in image to PDF.

The 8-tool offline toolbox (what each does)

Everything above runs on pdf-lib, jsPDF and pdf.js inside your browser — I verified with Wi-Fi off after page load. Contrast with cloud tools that cap free use at 2 files a day and keep your uploads: for visa bundles, bank statements and contracts, local-first is not a feature, it is the requirement. Each operation below links its deep tutorial.

Merge in order: cover + chapters → 20 pages

The classic job: cover.pdf + chapter1 (12 pages) + chapter2 (8 pages) = one 20-page book. Steps: select up to 20 files (10MB each), drag rows into reading order — the badge confirms 12 + 8 = 20 — click merge, download book-combined.pdf. What survives: text layers, hyperlinks, outline hierarchy. What flattens: form fields, signatures, annotations become static ink. Encrypted bank statements refuse until unlocked with the owner password — remove restrictions first; brute force violates policy and our terms. Walkthrough: merge in order.

Merge vs combined limits

Small booklet (2 files, 20 pages) flies on phones; archive jobs (20 files, 200 pages) need a laptop and patience. If your set exceeds 200 pages, split first, merge in parts, then merge the parts.

JPG scans → single PDF (A4, 150 DPI)

Phone scans arrive as loose JPGs. Batch them: select up to 50 JPG/PNG (10MB each), pick A4 portrait for documents (Letter for US forms, landscape for slides), drag thumbnails into order, generate. A 12-shot set at 150 DPI assembles a 4.2MB PDF in seconds. HEIC photos and Word files must convert to JPG/PNG first — direct ingestion supports only those two. Keep scan-01.jpg naming so order is obvious. Guide: scans to PDF. Pair with image compression when scans are oversized.

Extract pages by range without fear

Syntax 1-3,5 means pages 1, 2, 3 and 5 — the live badge counts 4 before you commit, catching reversed typos like 5-3 instantly. Researchers pull chapters (10–20), admins quarantine invoices, litigators segregate exhibits. Periodic mode (every Nth page) samples surveys. From a 120-page manual, 1-3,5,8-10 yields 7 pages saved as manual-pages-1-3-5-8-10.pdf. Tutorial: range extraction. Remember the workflow pair: split singles out, merge puts back — merging is splitting in reverse.

Compress to 1MB for Gmail and portals

Gmail blocks 25MB+ attachments; portals cap at 10MB. Pick a level: Light (~20% off, invisible), Medium (~50% off at 150 DPI, email-ready), Strong (~70% off, check small text at 100% zoom). My 18MB photo deck hit 6MB at Medium — 67% off, text still vector-sharp because only images downsample. Scanned pages (flat images, no text layer) save less, 20–30%. Encrypted files need unlocking first. Levels demo: 1MB compression.

Pages → JPG at 2× for print

Need slides as images? Render at 1× (~96 DPI) for web previews, 2× (~192 DPI) for print, 85% JPEG quality for photos (PNG for diagrams to avoid artifacts). A 6-page brochure exports page-01.jpg through page-06.jpg into one ZIP. Memory rule: under 50 pages per batch on phones, 1× scale on old devices. Guide: PDF to JPG.

Text extraction vs OCR: know which you have

Born-digital PDFs embed selectable glyphs — extraction is instant and exact. Scanned pages are flat pictures with no text layer: extraction returns empty per-page flags, correctly. That is not a bug; OCR (optical recognition) is a different technology and out of scope here — use dedicated OCR software for scans, then bring the result back. For digital files, select ranges like 1-3,5, extract, copy or download chapter.txt. Method: text without OCR.

Fix sideways scans (split-rotate-merge for mixed bundles)

Flatbeds produce sideways pages; phone scans inherit orientation flags. If every page is wrong, rotate all 90/180/270 in one click. If only pages 5–8 are sideways (mixed bundle), the pro move: split out 5–8 at PDF split, rotate that segment, merge back at PDF merge. Preview the first thumbnail before committing — double-rotating an already-correct file wastes a round trip. Fixes: sideways scan repair.

Watermark DRAFT at 25% without ruining readability

Twenty-five percent gray, center-diagonal, 48pt — visible enough to deter screenshots, light enough to read through. Eighty-percent black screams and obscures; corners suit near-final samples. Underlying text stays selectable because the overlay draws independently. Use DRAFT, CONFIDENTIAL, SAMPLE or DO NOT COPY per audience, and archive unwatermarked masters. Stamping: watermark guide.

Limits and honest notes

Ceilings exist to protect your tab: 20 files, 200 pages, 10MB each, 100MB total per batch. Password-locked files need owner credentials first. Interactive forms flatten on merge. Image-only scans need external OCR. Split oversized jobs into passes on phones, keep originals archived, and verify pagination by sampling first, middle and final pages.

General guidance only. Verify sensitive documents page by page; encrypted or signed files may need desktop software.