Skip to main content

Project logo

OCR application for reading manga in Japanese, made for AJATTers🇯🇵

Lancet

PyPi Chat Support

Lancet is an OCR application for reading manga in Japanese. It uses PyTorch models to recognize Japanese text in images. Lancet works best with Goldendict-ng or Rikaitan.

https://github.com/user-attachments/assets/8859adac-1bd8-435a-9c81-945f875dc205

User Guide

Prerequisites:

Launch Goldendict, then open a manga folder in nsxiv. Start Lancet and wait for the OCR model to load. Press the OCR shortcut (default Alt+O) to show the snipping window, then drag and hold the mouse to perform OCR.

Lancet adds an icon to the system tray. Right‑click the icon to access actions, including Preferences, where you can change the OCR shortcut.

Anki screenshots

Lancet can attach a selected screen region to the Image field of the newest Anki note added within the last day. This replaces a separate screenshot utility for image-card creation.

  1. Install and enable AnkiConnect, then launch Anki.

  2. Create or add the target note.

  3. Press Alt+I (the default shortcut), choose the screen region, and Lancet appends the image to that note's Image field.

    Finish active Browser edits before attaching an image. AnkiConnect cannot synchronously confirm that the Browser has saved pending editor changes.

Preferences has an Anki tab with the AnkiConnect URL, optional API key, target field, maximum image width/height, quality, and output format. The shortcut can be changed on the Main tab with Lancet's other shortcuts. Width and height are maximum bounds. 0 leaves an axis unconstrained, and images are never enlarged. The API key is masked in Preferences but stored in Lancet's plaintext JSON configuration file. On POSIX systems, Lancet writes that file with owner-only permissions. Literal localhost and numeric loopback AnkiConnect endpoints bypass HTTP_PROXY and HTTPS_PROXY. Custom remote endpoints retain normal proxy behavior. The field separator controls the HTML placed before each appended image. It defaults to <br> and may be empty.

Executable overrides

"Preferences" > "Advanced" lets you set the GoldenDict executable. Leave it empty to use automatic discovery through your system PATH.

Installation

All operating systems

Install via pipx from pypi.

pipx install ajt-lancet

The lancet executable will be available in your $PATH.

Windows-like operating systems

Download lancet-windows.exe.

GNU/Linux

Download lancet-gnulinux.

System Requirements

Lancet depends on large Python libraries. pipx installs everything in an isolated location (~/.local/share/pipx) so you can cleanly uninstall with pipx uninstall later. The PyTorch models are downloaded from HuggingFace and saved to ~/.cache/huggingface.

  • Disk space: ~8 GiB (PyTorch and dependencies)
  • RAM: at least 2 GiB

Wayland support

Wayland support has been tested on Sway 1.11. Wayland support is incomplete. It works best on Hyprland and Sway, but on other compositors you may face issues if you have multiple monitors. 🙏 Please suggest improvements via pull requests on GitHub.

Upgrading

Running pipx upgrade upgrades lancet to the latest version.

pipx upgrade ajt-lancet

Running pipx upgrade-all upgrades all packages installed with pipx.

pipx upgrade-all

Development Setup

Since Ajatt-Tools is a distributed effort, we highly welcome new contributors! Install the project in development mode to easily test and commit your changes using hatch.

  • Clone the repo.

  • Install hatch:

    pipx install hatch
    pipx upgrade hatch
    
  • CD into the repo and run:

    hatch shell
    
  • Inside the hatch shell run:

    pip install -e .
    
  • To run the app, run:

    hatch run lancet
    

Try these libre code editors with python support:

Autostart

Add the lancet command to your autostart so Lancet launches at login.

Here's an example for i3wm:

exec --no-startup-id lancet

System keyboard shortcuts

On some setups (like Wayland), applications can't register global keyboard shortcuts, so Lancet's shortcuts won't work. You can define shortcuts in your Wayland compositor instead.

Below is an example for Sway. Add the following to your ~/.config/sway/config: The example also works for i3wm (~/.config/i3/config).

# Run OCR
bindsym --release $mod+o          exec lancet ocr
# Run OCR with the speech-bubble detector
bindsym --release $mod+Shift+o    exec lancet ocr --detect
# Take a screenshot
bindsym --release $mod+Print      exec lancet screenshot
# Attach a selected screen region to the most recently added Anki note.
bindsym --release $mod+i          exec lancet screenshot --add-to-anki

Announcements

  1. Sunsetting Transformers OCR
  2. Meet Lancet

Metadata

Release files for ajt-lancet 26.9.15.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for ajt-lancet 26.9.15.0
File Size Uploaded Provenance Actions
ajt_lancet-26.9.15.0.tar.gz 7.3 MB With provenance Details for ajt_lancet-26.9.15.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for ajt-lancet 26.9.15.0
File Interpreter ABI Platform Actions
ajt_lancet-26.9.15.0-py3-none-any.whl Python 3 none any Details for ajt_lancet-26.9.15.0-py3-none-any.whl

Total release size: 14.5 MB

Release files / ajt_lancet-26.9.15.0.tar.gz

Download URL ajt_lancet-26.9.15.0.tar.gz
Size 7.3 MB
Tags Source
SHA-256 checksum
How to use checksums
18ab0d29670fa7c3455a147f4009aed8ab480a0d29b31cc2fb1000df966dc90d
BLAKE2b-256 checksum
How to use checksums
7b9bd6cb728b387e3d13c4e22d325f2f2788c9c1a9b819d9990720f3ec89de02
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 15, 2026.

Transparency log

Release files / ajt_lancet-26.9.15.0-py3-none-any.whl

Download URL ajt_lancet-26.9.15.0-py3-none-any.whl
Size 7.2 MB
Tags Python 3
SHA-256 checksum
How to use checksums
5cb08142f057d89c5ae33b3f47e5395f32d8dcd6be790a18d1d1802ad2c8ac6d
BLAKE2b-256 checksum
How to use checksums
3e4f0e2542444ad6c6d97ecb90a9bc854b74371aaeeab8f54981c27fea20c2c7
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 15, 2026.

Transparency log

Security for ajt-lancet 26.9.15.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

All 2 files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for ajt_lancet-26.9.15.0.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 15, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page