Getting Started

Installing polars-st

pip install polars-st

Basics

polars-st provides geometry operations under the namespace st on Polars Expr, Series, DataFrame and LazyFrame. Functions used to read files or parse geometries are available at the module root. Here's a basic example:

>>> import polars as pl
>>> import polars_st as st
>>> df = pl.DataFrame({
...     "wkt": [
...         "POINT(0 0)",
...         "POINT(1 2)",
...     ]
... })
>>> gdf = df.select(geometry=st.from_wkt("geometry"))
>>> area = gdf.select(pl.col("geometry").st.area())

If you have type checking enabled, you might face this error: Cannot access member « st » for class « Expr ». In order to support autocompletions and type checking for the st namespace, polars-st provides a utility function st.geom, with the same signature as pl.col, but which returns GeoExpr instead of Expr. GeoExpr is (kinda) just a type alias to polars Expr with type annotations added for the st namespace. It is therefore recommended that you use st.geom instead of pl.col to query geometry columns.

In addition to type checking, st.geom also has a trick up its sleeve: assuming the geometry column matches the default name defined in st.Config (the built-in default is "geometry"), you can even omit typing the column name entirely:

>>> area = gdf.select(st.geom().st.area())

Even better, operations that involves a single geometry can be called in a simpler form:

>>> area = gdf.select(st.area())

The default geometry column name can be configured with st.Config. Like polars.Config, st.Config can be used as a context manager, a function context decorator, or as a global configuration object.

>>> gdf = st.GeoSeries("my_geometry", ["POINT(1 2)"]).to_frame()
>>> with st.Config(geometry_column="my_geometry"):
...     x = gdf.select(st.x())
>>> x
shape: (1, 1)
┌─────────────┐
│ my_geometry │
│ ---         │
│ f64         │
╞═════════════╡
│ 1.0         │
└─────────────┘