Skip to content

Installation

The library can be installed with pip or by cloning the project repository with git.

Dependencies

ruTS relies on the following third-party libraries:

  • python - 3.11 or newer
  • nltk
  • pymorphy3
  • razdel
  • scipy
  • spaCy - 3.7.0 or newer
  • numpy
  • pandas
  • matplotlib
  • graphviz

Installing with pip

Run in the command line:

pip install ruts

This installs the release version of the library with all dependencies.

Graphviz

The graphviz package among the dependencies is only a wrapper: rendering the word tree and the collocation network requires the Graphviz executables (brew install graphviz, apt install graphviz, conda install graphviz), otherwise render() and displaying the graph in Jupyter fail with ExecutableNotFound. Nothing else needs them.

Fast morphological analysis

For corpora, install the library with the fast extra:

pip install "ruts[fast]"

It adds the DAWG2 C extension for pymorphy3 (CPython only, wheels are available for Linux, macOS and Windows): word form analysis gets about 5x faster (175 vs 34 thousand word forms per second), MorphStats on a 70-thousand-word text about a third faster, because analyses are cached per word form and the rest of the time goes into counting. Without the extra the library works the same, just slower.

spaCy model

The spaCy components, the syntactic statistics and the examples that start with spacy.load("ru_core_news_sm") need the Russian-language model; without it spacy.load fails with OSError: [E050]. The model is installed separately:

python -m spacy download ru_core_news_sm

The larger models ru_core_news_md and ru_core_news_lg make no sense for ruTS statistics: in our measurements on the gold Universal Dependencies annotation (Taiga and GSD) they are less than one UAS point more accurate than sm, the text-level statistics of the three models coincide (correlation above 0.95 for 40 of 53), and they weigh 93 and 670 MB against 47.

Datasets

The datasets are downloaded by default into the ruts_data directory next to the package (DEFAULT_DATA_DIR in ruts.constants): for an installation from a wheel that is site-packages/ruts_data, where writing may be forbidden for a system Python. In that case pass your own directory in the data_dir argument when creating a dataset.

Note

A detailed guide to the pip package manager is available on its website.

Installing with git

This way you get the latest version of the library straight from the project repository. Run the following commands:

  1. Clone the repository into a local directory:

    git clone https://github.com/SergeyShk/ruTS.git

  2. Change into it:

    cd ruTS

  3. Install the library into the current environment:

    pip install .

    For development it is more convenient to use uv, which creates an isolated environment and installs all dependencies, including the development tools:

    uv sync --all-groups --no-group examples

    The examples group with Jupyter is needed only for the example notebooks.

Note

A detailed guide to the git version control system is available on its website.