大数据+GIS：别光想着看地图，人家早就开始“算”地图了！

pip install apache-sedona

from sedona.register.geo_registrator import SedonaRegistrator
from sedona.utils.adapter import Adapter
from sedona.core.SpatialRDD import PointRDD, PolygonRDD
from pyspark.sql import SparkSession
from sedona.core.enums import FileDataSplitter
from sedona.core.geom.envelope import Envelope

# 创建 Spark 会话
spark = SparkSession.builder \
    .appName("MetroZoneAnalysis") \
    .config("spark.jars.packages", "org.apache.sedona:sedona-python-adapter-3.0_2.12:1.4.0-incubating") \
    .getOrCreate()

SedonaRegistrator.registerAll(spark)

# 读取地铁站点数据（假设是 WKT 格式）
metro_df = spark.read.option("delimiter", ",").csv("metro_stations.csv").toDF("id", "wkt")
metro_df.createOrReplaceTempView("metro")

# 转为空间点数据
spark.sql("SELECT ST_PointFromText(wkt, ',') AS geometry, id FROM metro").createOrReplaceTempView("metro_points")

# 构造以地铁站为中心的缓冲区（10分钟步行，大约800米）
spark.sql("""
    SELECT ST_Buffer(geometry, 800) AS buffer_zone, id FROM metro_points
""").createOrReplaceTempView("metro_buffers")

# 导入居民小区点位
residential_df = spark.read.option("delimiter", ",").csv("residential.csv").toDF("id", "name", "wkt")
residential_df.createOrReplaceTempView("res")

spark.sql("SELECT ST_PointFromText(wkt, ',') AS geometry, name FROM res").createOrReplaceTempView("res_points")

# 空间连接：找出每个地铁站10分钟圈内的住宅
result = spark.sql("""
    SELECT m.id AS metro_id, r.name AS residential_name
    FROM metro_buffers m, res_points r
    WHERE ST_Contains(m.buffer_zone, r.geometry)
""")

result.show(10, False)