A small GNOME app for marking up PDFs with a stylus, mouse or keyboard: signing documents, scribbling notes, updating hit points on a character sheet. The toolbar is a single header bar so the page gets the rest of the window (F11 hides even that).
Needs only Fedora packages (all present on a default Workstation install):
sudo dnf install python3-gobject gtk4 libadwaita poppler-glib python3-cairo python3-pikepdf python3-cryptography
Then:
./install.sh # installs to ~/.local, adds a launcher
./install.sh --uninstall
Or run straight from the checkout: ./bin/pdffer some.pdf.
| Tool | Key | What it does |
|---|---|---|
| Select | S | Drag any mark to move it; drag an image's corner handles to resize. Delete removes the selected mark. |
| Pen | P | Draw. Pressure-sensitive with a stylus. |
| Text | T | Click to type; Enter to finish (Shift+Enter for a new line), Esc to cancel. Click existing text to edit it, drag it to move it. |
| Eraser | E | Wipe over marks to remove them. The stylus eraser end works in any tool. |
The color dot next to the tools sets color, pen width and text size.
Images: the image button (or Ctrl+I) inserts a PNG/JPEG/etc. You can also paste one with Ctrl+V or drag an image file onto the page. A transparent PNG of your signature works well for signing.
Pages drawer: the sidebar button (or F9) shows thumbnails of every page, including your marks. Click one to jump there.
Fingers scroll and pinch-zoom; the stylus and mouse draw. Middle-drag pans with a mouse.
| Shortcut | |
|---|---|
| Ctrl+O / Ctrl+S / Ctrl+Shift+S | Open / Save / Save As |
| Ctrl+Z / Ctrl+Shift+Z (Ctrl+Y) | Undo / Redo |
| Ctrl+I / Ctrl+V | Insert image / paste image |
| F9 | Show/hide page thumbnails |
| Ctrl+scroll, Ctrl+plus/minus, Ctrl+0 | Zoom, fit width |
| F11 | Fullscreen |
Marks are saved as standard PDF annotations (ink, free text and image stamps) with proper appearance streams, so they show up in Evince/Papers, Okular, Firefox, Chrome, Acrobat and friends. Each one also stores its editable data, so when you reopen a file here your old marks can still be erased, moved and edited. That's handy for hit points that change every session.
Save Flattened Copy… (main menu) writes a copy with the marks burned into the page content instead. Use that for signed documents you send to someone, so the signature can't be stripped or edited as an annotation.
Saves are incremental: the file you opened is kept byte-for-byte and your changes are appended after it. So if someone sent you a PDF that is already digitally signed (DocuSign, Adobe Sign, …), their signature stays valid after you add yours; viewers will note that annotations were added after signing. Unchanged marks are reused, so saving repeatedly doesn't grow the file.
Encrypted PDFs (e.g. ones that open without a password but restrict editing or copying) stay encrypted with the same restrictions, including in flattened copies. Damaged PDFs that can't take an incremental update are rewritten in full instead.
Saves are atomic (written to a temp file, then renamed), so a crash mid-save won't corrupt your PDF.
Error messages explain the problem in plain language; the dialog's
Technical Details section has the underlying cause. Running
pdffer from a terminal also logs warnings there, and
PDFFER_DEBUG=1 pdffer file.pdf logs more detail (for example
about stylus input).
python3 -m unittest discover tests
On a desktop this includes GUI tests that drive the real window (windows flash briefly; they use a separate app instance and a throwaway settings file, so a copy of the app you have open is unaffected). Without a display the GUI tests are skipped; to run them headless:
dbus-run-session -- xvfb-run -a env GDK_BACKEND=x11 GSK_RENDERER=cairo \
python3 -m unittest discover tests
Lint with ruff (configured in
pyproject.toml; CI pins the version):
pipx run ruff==0.17.3 check .
GitHub Actions runs the linter and all tests on each push in a Fedora container
(.github/workflows/tests.yml). python3 tests/benchmark.py prints
redraw, hover and memory numbers to compare before and after a change.
document.py: loading, marks, undo/redo, PDF writing (incremental, encrypted)render.py: cairo drawing shared by screen and PDF outputcanvas.py: the scrolling page view; routes input to the active tooltools.py: pen, text, eraser and select toolslayout.py: page layout geometry (pure Python, unit-tested)page_renderer.py: background Poppler rendering threadtext_editor.py: the in-place text boxstylus.py: stylus pressure, eraser tip and coalesced samplesthumbnails.py: page thumbnail drawerstyle_button.py: color, pen width and text size popovererrors.py: plain-language error messageswindow.py,app.py,settings.py: the window, application and preferences
MIT; see LICENSE.