Install from PyPI
The client-side install. Use this on any machine that runs the Whistlerlib client: your laptop, a Jupyter server, a CI runner that submits jobs to a Dask cluster, etc.
For the cluster itself (scheduler + workers), see Docker installation.
Requirements
- Python ≥ 3.11.
- A reachable Dask scheduler (a local one via
LocalCluster, or a remote one, see Architecture). The pip install does not start a cluster for you.
Install
pip install whistlerlib
If you use uv:
uv pip install whistlerlib(oruv add whistlerlibif you're managing apyproject.toml).
That's it for the client. Verify:
python -c "import whistlerlib; print(whistlerlib.Context)"
# <class 'whistlerlib.context.Context'>
What you get
A pip install on a bare host gives you the alt-python algorithm surface: everything in whistlerlib.dask.alt_python_algs:
hashtag_histogram_alt_pythonmention_histogram_alt_pythonngram_histogram_alt_pythonsentiment_range_spanish_alt_pythonhashtag_weighted_coonet,mention_weighted_coonet
Plus the core types: Context, TweetDataset.
What you don't get (and why that's fine)
The R-bridge methods, hashtag_histogram_r, mention_histogram_r, ngram_histogram_r, sentiment_histogram_and_sum_r, are not available from a pip install alone. They shell out to Rscript on the worker, which needs R + a curated set of R libraries (tm, syuzhet, RWeka, radvertools, …). Installing all of that on a researcher's laptop is exactly what we don't want you to do.
Instead, the R libraries live inside the published albertogarob/whistlerlib Docker image (Docker installation). The pattern is:
- Client:
pip install whistlerlibon your laptop. Use the alt-python methods locally, or submit work to a remote cluster. - Cluster workers: run the
albertogarob/whistlerlibimage. R + libraries baked in. R-bridge methods just work when called against this cluster.
So:
- ✅
pip install whistlerlib→ client-side, alt-python methods, no R required. - ✅
pip install whistlerlib+ connect to a Dask cluster runningalbertogarob/whistlerlib→ full surface including R-bridge. - ❌
pip install whistlerlib+ R installed manually on the host → not a supported configuration. We don't test it; don't go there.
Next
- Docker installation, bring up a local cluster (Compose) or production cluster (Swarm).
- Tutorial 01, first end-to-end run.