Skip to content

Packaging

Plamen Dimitrov edited this page Feb 2, 2022 · 7 revisions

Installation

There are four main ways to deploy guibot on your platform: two platform independent ways through PyPI or Git as well as two platform-specific ways through an RPM or Debian package. The most standard and universal way is to use PyPI but the other three ways include additional CV backend data that could be used for more advanced forms of matching like e.g. OpenCV and Tesseract data. Of course this data can also be downloaded or provided separately.

All dependencies listed below are optional in general and you only have to install the ones you actually need. If for instance you need just guibot with OCR you should be fine with a default OpenCV pick and a choice of Tesseract dependency for your CV backends (image matching) and a default PyAutoGUI pick for your DC backends (display control). Even the current defaults are optional if you don't need them and prefer a different set of backends instead.

PyPI

To install guibot simply tun

pip install guibot

This will install only the most important dependencies leaving the optional ones out. To install all pip provided dependencies run the following within the guibot folder

pip install -r packaging/pip_requirements.txt

Depending on your choise of OS these requirements might have dependencies that are outside of the scope of pip. We provide an example script for RPM-based systems that you can also consult with for any potential missing dependencies for your system of choice that you might still need depending on your usage:

bash packaging/packager_pip.sh

For some additional details on Windows dependencies you can also run or read the following batch script

guibot\packaging\packager_win.bat

Only Python 3 is supported and Python 2 support was deprecated in 2018.

NOTE: The PyPI OpenCV version with contribution modules does not support text matching via OpenCV so you will need an official RPM/Debian package for it. For regular OCR you should be fine with pytesseract or tesserocr though.

Git

If you want to use the source and all provided CV data directly

git clone --depth 1 https://github.com/intra2net/guibot

or download the most recent release tarball. All packaging commands you might run in case you want to build deployable packages assume you are running from within the guibot folder and the same applies to installing additional PyPI dependencies.

RPM package

You can download the most recent RPM package we have provided for you from the project web page or build a new RPM package after you clone the repository with

bash packaging/packager_rpm.sh

or otherwise simply view the contents for RPM requirements for running as well as building split into the respective stages.

Debian package

You can download the most recent Debian package we have provided for you from the project web page or build a new Debian package after you clone the repository with

bash packaging/packager_deb.sh

or otherwise use it in the same way as the RPM packager script.

NOTE: The OpenCV version on some Ubuntu releases is currently old and you need at least 3.0.0 to be able to use feature matching.