Installation¶
textnets is included in conda-forge, the Python Package Index, and nixpkgs. That means you can install the package using conda, pip, or nix.
Note
Please note that textnets requires Python 3.9 or newer to run.
Using conda¶
This is the preferred method for most users. The Anaconda Python distribution is a convenient way to get up and running with Python, especially if you are on a Mac or Windows system.
Once it is installed you can use its package manager conda
to install
textnets:
$ conda install -c conda-forge textnets
This tells conda to install textnets from the conda-forge channel.
If you don’t know how to enter this command, you can use the Anaconda Navigator instead. It provides a graphical interface that allows you to install new packages.
Installing textnets in Anaconda Navigator
Go to the Environments tab.
Click the Channels button.
Click the Add button.
Enter the channel URL https://conda.anaconda.org/conda-forge/
Hit your keyboard’s Enter key.
Click the Update channels button.
Now you can install textnets in a new environment. (Make sure the package filter on the Environments tab is set to “all.”)
Using pip¶
Alternately, if you already have Python installed, you can use its package manger to install textnets. In a virtual environment, run:
$ python -m pip install textnets
Using nix¶
Users of nix
can use the version from nixpkgs, for instance by using
nix-shell
:
$ nix-shell -p 'python3.withPackages (p: with p; [ ipython textnets spacy_models.en_core_web_sm ])' --run ipython
Plotting¶
textnets depends on the Cairo graphics library for plotting. If you are using a Mac without Anaconda or Nix, you will probably have to install Cairo separately using the Homebrew package manager.
Language Support¶
textnets can try to download the language models you need “on the fly”
if you set the autodownload
parameter to True
. (It is off by default
because language models are frequently many hundreds of megabytes in size and
probably shouldn’t be downloaded on a metered connection.)
>>> import textnets as tn
>>> tn.params["autodownload"] = True
You can also install the models manually by issuing a command like:
$ python -m spacy download en_core_web_sm
After updating textnets you may also need to update the language models. Run the following command to check:
$ python -m spacy validate
If there are no language models available for your corpus language, there may be some basic support. Even in that case, some languages (including Japanese, Russian, Thai, Vietnamese, Ukrainian, and Chinese) require additional installs for tokenization support. Consult the spaCy documentation for details.