Fast, local-first PDF viewer and light editor in Rust — open, scroll, edit pages, compress, sign, and OCR without uploading your documents.
Keywords: rust pdf pdf-viewer pdf-editor egui mupdf ocr local-ocr pdf-sign pkcs12 lightweight desktop-app privacy offline
Most PDF tools are either heavy suites or cloud uploaders. Small Rust PDF Opener is a single native binary that stays on your machine: view quickly, do a few useful edits, merge/split, compress, stamp a signature, optionally certify with a PKCS#12 key, and run OCR after you choose to download local models.
flowchart TB
User[User] --> UI[egui Desktop UI]
UI --> Session[DocumentSession]
Session --> MuPDF[MuPDF Engine]
MuPDF --> Render[Page RGBA textures]
Render --> Scroll[Continuous scroll viewer]
Session --> Edit[Delete Rotate Reorder Crop]
Session --> MergeSplit[Merge Append Split Extract]
Session --> Export[Export md docx]
Session --> Compress[Compress presets]
UI --> OCR[OCR module]
OCR -->|opt-in download| Models[Local ocrs models cache]
OCR --> TextLayer[Invisible searchable text]
OCR --> Overlay[Bounding box overlays]
UI --> VisualSign[Visual signature stamp]
UI --> CertSign[PKCS12 to PKCS7]
CertSign --> Lopdf[lopdf AcroForm Sig]
Edit --> MuPDF
MergeSplit --> MuPDF
Compress --> MuPDF
TextLayer --> MuPDF
VisualSign --> MuPDF
flowchart LR
subgraph view [View]
A1[Open PDF] --> A2[Scroll / zoom / search]
end
subgraph edit [Edit]
B1[Select page] --> B2[Delete or rotate or crop]
B2 --> B3[Save As]
end
subgraph mergeSplit [Merge Split]
M1[Merge or Append] --> M2[Split or Extract ranges]
end
subgraph sign [Sign]
C1[Draw or import mark] --> C2[Place on page]
C3[Load P12] --> C4[Drag signature rect]
C4 --> C5[Write signed PDF]
end
subgraph ocrFlow [OCR]
D1[Download models once] --> D2[OCR page range or all]
D2 --> D3[Boxes plus text overlay]
D3 --> D4[Searchable text layer]
end
view --> edit
edit --> mergeSplit
edit --> sign
edit --> ocrFlow
flowchart LR
Src[Rust sources] --> Cargo["cargo build --release"]
Cargo --> Bin[pdf-opener binary]
Bin --> AppBundle["PDF Opener.app"]
Icon[AppIcon.icns] --> AppBundle
AppBundle --> Dmg[DMG installer]
Dmg --> AppsFolder["Applications folder"]
| Area | What you get |
|---|---|
| View | Fast MuPDF rendering, continuous scroll, zoom / fit-width, text search, keyboard nav |
| Edit | Delete page, rotate 90°, reorder (move up/down), crop (CropBox), Save / Save As |
| Merge / Split | Merge several PDFs, append into the open doc, split ranges to files, extract pages |
| Export | Export embedded text to Markdown (.md) or Word (.docx); run OCR… first for scans |
| Compress | Fast / Balanced / Small write presets (deflate, image compress, garbage collect) |
| Visual sign | Draw a signature or import PNG/JPEG and stamp onto a page |
| Cert sign | Import .p12 / .pfx, place a field, embed PKCS#7 detached signature (OpenSSL) |
| OCR | Opt-in local ocrs models; page / range / whole-doc; bounding boxes + text overlays; invisible searchable layer |
| Privacy | Documents never leave your computer; OCR models download only when you ask |
Out of scope (on purpose): full annotation suites, forms designer, PDF/A conversion, cloud sync.
Apple Silicon, via Homebrew:
brew tap will702/tap
brew install --cask pdf-openerOr grab the DMG from Releases.
See Build & run below. Quickest path with Rust installed:
cargo run --release -- ./testdata/hello.pdf| Dependency | Why |
|---|---|
| Rust 1.85+ (edition 2021) | Build toolchain |
| clang / C toolchain | Builds mupdf-sys (MuPDF C library) |
| macOS | Xcode Command Line Tools (xcode-select --install) |
| Linux | Typical build-essential + clang; fontconfig often needed for system fonts |
| Windows | MSVC Build Tools + clang for MuPDF |
git clone https://github.com/will702/small-rust-pdf-opener.git
cd small-rust-pdf-opener
cargo run --release
# or:
cargo run --release -- ./testdata/hello.pdf- First build compiles MuPDF via
mupdf-sys— expect several minutes. - Release profile uses LTO + stripped binary (
Cargo.toml). - OCR models are not in the binary; use OCR… → Download models in the app (~tens of MB to your cache dir).
./packaging/build-dmg.shProduces dist/PDF Opener.app and dist/PDF-Opener-<version>.dmg. Drag into Applications. If Gatekeeper blocks: right-click → Open (adhoc-signed local build).
GitHub Actions: every v* tag / published Release (and manual Release DMG workflow dispatch) builds the DMG on macos-latest and uploads it to the release assets.
| Action | How |
|---|---|
| Open / Save | Toolbar or ⌘/Ctrl+O, ⌘/Ctrl+S, ⌘⇧S |
| Scroll pages | Mouse wheel / trackpad (continuous strip) |
| Jump page | ◀ ▶ or ← → / PgUp / PgDn |
| Crop | Mode Crop, drag a rectangle on a page |
| Visual sign | Mode Sign, draw/import, Place on page, click |
| Cert sign | Mode Cert sign, load PKCS#12, Place & sign, drag rect |
| Compress | Compress… → preset → save new file |
| Merge | Merge… → add 2+ PDFs → Save merged… |
| Append | Append PDF… → pick files to add to the open document |
| Split | Split… → ranges like 1-2,4 → each range becomes a file |
| Extract | Extract… → page range → one new PDF (optional reopen) |
| Export | Export… → Markdown or Word → page range → .md / .docx (optional --features anydoc-export for structured Markdown via anydoc) |
| OCR | OCR… → download models once → this page / all / range; boxes + text overlay; Cancel supported |
src/
main.rs eframe entry + Dock icon
app.rs egui UI, continuous scroll viewer
page_range.rs parse `1-3,5` page ranges
pdf/ MuPDF document session (render/edit/merge/split/save)
export/ Markdown / Word text export from embedded PDF text
ocr/ model download + ocrs recognition
sign/ visual pad + PKCS#12 / PKCS#7 signing
packaging/ Info.plist + build-dmg.sh
assets/ app icons
docs/images/ README banner + architecture art
testdata/ sample PDF
This project stands on open-source libraries. Cite and respect their licenses.
| Component | Project | Role | License (upstream) |
|---|---|---|---|
| PDF engine | Artifex MuPDF via mupdf / mupdf-rs |
Render, page ops, compress write options | AGPL-3.0 |
| UI | egui / eframe |
Immediate-mode desktop UI | MIT / Apache-2.0 |
| OCR | ocrs + rten |
Local neural OCR; models from ocrs-models | MIT / Apache-2.0 |
| PDF COS surgery | lopdf |
Signature dictionary / AcroForm wiring | MIT |
| Crypto | openssl (vendored) |
PKCS#12 parse, PKCS#7 detached sign | Apache-2.0 |
| Dialogs / images | rfd, image |
File pickers, PNG/JPEG | MIT / Apache-2.0 |
| Markdown export (optional) | anydoc |
Structured GFM from PDF (anydoc-export feature) |
MIT |
Full third-party notice: NOTICE. Formal citation: CITATION.cff.
Because MuPDF is AGPL-3.0, this repository is licensed under AGPL-3.0 — see LICENSE. If you distribute a modified version (including as a network service that users interact with), you must provide corresponding source under AGPL-3.0.
- PDFs are processed locally.
- Certificate signing uses keys you import; passwords are not stored.
- OCR model download uses HTTPS to the ocrs model host; disable by never clicking Download.
- Report vulnerabilities privately — see SECURITY.md.
See CONTRIBUTING.md. Ideas welcome: more OCR languages, better cert appearance streams, Linux/Windows packaging.
- One-liner: Local-first Rust PDF viewer/editor with scroll, crop, merge/split, compress, visual + PKCS#12 signing, and opt-in offline OCR with visible results.
- Topics:
rust,pdf,pdf-viewer,pdf-editor,egui,mupdf,ocr,offline,privacy,desktop-app,agpl - Machine-readable summary: llms.txt