Save Spark DataFrame to shape file (.shp format)

Bartek — Fri, 27 Jan 2023 13:07:14 GMT

Hello,

I know how to create .shp file from Geopandas dataframe using code similar to this, also mentioned on SO:

gpd_df = geopandas.GeoDataFrame(pandas_df, geometry='geom')
gpd_df .to_file("username/nh.shp")

However I have .parquet files that I can load directly to Spark DataFrame and I want to create and save shape file this way. Unfortunately I'm not sure if that's possible. I can't see .shp format in supported formats. I checked also Sedona but found only Shapefilereader not allowing to save/write. What is the state-of-the-art to operate on shape files?

Re: Save Spark DataFrame to shape file (.shp format)

Anonymous — Thu, 09 Mar 2023 04:14:39 GMT

@Bartosz Maciejewski :

Spark does not have native support for writing Shapefiles directly. However, you can use a third-party library such as GeoPandas or PyShp to write your Spark DataFrame to a Shapefile.

Here's an example of how to use GeoPandas to convert a Spark DataFrame to a GeoDataFrame and save it to a Shapefile.

import geopandas as gpd
from pyspark.sql import SparkSession
from shapely.geometry import Point
 
# create SparkSession
spark = SparkSession.builder.appName("SparkGeoPandas").getOrCreate()
 
# create sample Spark DataFrame
df = spark.createDataFrame([(1, Point(0, 0)), (2, Point(1, 1))], ["id", "geometry"])
 
# convert Spark DataFrame to GeoDataFrame using GeoPandas
*** = gpd.GeoDataFrame(df.toPandas(), geometry="geometry")
 
# save GeoDataFrame to Shapefile
***.to_file("path/to/shapefile.shp", driver="ESRI Shapefile")

You can also use another library 'PyShp' instead of GeoPandas.

topic Save Spark DataFrame to shape file (.shp format) in Data Engineering

Save Spark DataFrame to shape file (.shp format)

Re: Save Spark DataFrame to shape file (.shp format)