pipedrag

Drag files found on the command line into any app that accepts drops:

find ~/Downloads -name '*.pdf' | pipedrag

pipedrag reads file paths from stdin (newline- or NUL-separated), opens a small window with a multi-file icon, and lets you drag it into Dolphin, a browser, a chat window … as if you had selected the files in a file manager.

Drops can be chunked (pipedrag -n 1000), most browser uploads have some sort of limit. Chunks already dropped are marked as "green".

$> find . -name '*.jpeg' -type f | pipedrag -n 1000

How to install a release version

Install as python image

pip install pipedrag

Debian / Ubuntu

Download the latest release from https://git.jtheuer.de/pipedrop/releases and install it via

sudo dpkg -i <filename.deb>

Development and contribution

make setup          # creates .venv and installs the project + dev tools

Requires Python ≥ 3.10. If python3 -m venv fails on Debian/Ubuntu, install python3-venv first.

How the drop works

• The drag speaks XDND — the same protocol Dolphin uses — via text/uri-list.
• The dragged files are registered with the FileTransfer portal, so sandboxed targets (snap/Flatpak apps) can read them even from directories their sandbox blocks directly — e.g. snap Firefox silently denies dot-directories like ~/.local/share/. This is the same mechanism KDE/Dolphin uses.

Drop inspector

upload-test.html is a browser page that prints everything the browser exposes about dropped files (metadata, mime types, filesystem handle) and their SHA-256/SHA-384/SHA-512 checksums — useful for verifying a drop end to end and cross-checking with sha256sum.

Notes

• The drag starts from inside the mouse-press handler, as Qt requires on Wayland.

AI transparency

This project was largely coded via claude using the deepseek api. I couldn't have built a QT app so quickly and the drag'n'drop part with the portals was a lot more messy than expected.