view mupdf-source/thirdparty/tesseract.txt @ 38:8934ac156ef5

Allow to build with the PyPI package "clang" instead of "libclang". 1. It seems to be maintained. 2. In the FreeBSD base system there is no pre-built libclang.so. If you need this library you have to install llvm from ports additionally. 2. On FreeBSD there is no pre-built wheel "libclang" with a packaged libclang.so.
author Franz Glasner <fzglas.hg@dom66.de>
date Tue, 23 Sep 2025 10:27:15 +0200
parents b50eed0cc0ef
children
line wrap: on
line source

If you want to build with Tesseract functionality, you need to run make
with a "tesseract=yes" argument.

You will also need a suitable set of traineddata for the languages you
wish to run. Only the LSTM engine (the latest and most accurate engine)
is built into Tesseract, so the traineddata contained within the
repository itself is no good.

Suitable data can be retrieved from either:

  https://github.com/tesseract-ocr/tessdata_best

or

  https://github.com/tesseract-ocr/tessdata_fast

e.g.

  wget https://github.com/tesseract-ocr/tessdata_fast/raw/master/eng.traineddata