Skip to main content

Language Bindings

GeoBrix exposes one canonical function set across SQL, Python, and Scala. The per-function reference — signatures, parameters, return types, and runnable examples — lives on the package pages: RasterX, GridX, VectorX, and PMTiles. This page orients you to the registration and import patterns each language uses, and to the lightweight-vs-heavyweight tier choice for raster.

One function, three surfaces​

Every function has a single canonical name; the surface determines only how you call it:

SurfaceFormExample
SQLgbx_<name> (prefixed)gbx_rst_boundingbox(tile)
Python<name> via a package modulerx.rst_boundingbox(tile)
Scala<name> via a package importrst_boundingbox(col("tile"))

The gbx_ SQL prefix distinguishes GeoBrix functions from Databricks built-in st_* functions. Python and Scala use the bare rst_* / bng_* / quadbin_* / st_* / pmtiles_* names.

Registration and imports​

SQL​

SQL functions are not available until they are registered into the Spark catalog from Python or Scala. Once registered, they work in any SQL context (notebooks, Databricks SQL, jobs):

from databricks.labs.gbx.rasterx import functions as rx
rx.register(spark)            # installs gbx_rst_* into the catalog
SELECT gbx_rst_width(tile) FROM rasters;

Use SHOW USER FUNCTIONS LIKE 'gbx_*' to list what is registered, and DESCRIBE FUNCTION gbx_rst_clip for a signature.

Python​

Import a package's functions module and call the bare names. register(spark) is only needed if you also want the SQL surface; the Python column expressions work without it:

from databricks.labs.gbx.rasterx import functions as rx     # heavyweight raster
from databricks.labs.gbx.gridx.bng import functions as bx    # BNG
from databricks.labs.gbx.vectorx import functions as vx      # vector

df.select(rx.rst_boundingbox("tile"))

Wrappers accept Column or scalar values; non-string scalars are auto-wrapped, while a bare string follows PySpark's column-reference convention (wrap a string literal in f.lit("...")).

Scala​

GeoBrix is natively written in Scala. Import a package's functions object to bring the column functions into scope, and call register(spark) for the SQL surface:

import com.databricks.labs.gbx.rasterx.functions._

functions.register(spark)                 // optional: installs gbx_rst_*
df.select(rst_boundingbox(col("tile")))

Typed overloads accept Boolean / Int / Double / String value parameters directly (no lit(...) wrapping); Column arguments such as geometry and tile still take Column.

Lightweight vs heavyweight: a one-line swap (raster)​

RasterX ships in two execution tiers that share the same function names. The only thing that changes is the import:

from databricks.labs.gbx.rasterx import functions as rx     # heavyweight (GDAL/JVM)
from databricks.labs.gbx.pyrx     import functions as rx     # lightweight (pure-Python)

Downstream code is identical. See Choosing an Execution Tier for the comparison and the lightweight install, and Performance & Benchmarking for tier-by-tier timings.

Next Steps​