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

1. Go to the Environments tab.

2. Click the Channels button.

3. Click the Add button.

4. Enter the channel URL https://conda.anaconda.org/conda-forge/

5. Hit your keyboard’s Enter key.

6. Click the Update channels button.

7. 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

In rare cases you may have to install CFFI separately for plotting to work.

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.