如何修改pyspark数据框嵌套结构列

How to modify pyspark dataframe nested struct column

我正在尝试 anonymize/hash 嵌套列,但没有成功。架构看起来像这样:

-- abc: struct (nullable = true)
|    |-- xyz: struct (nullable = true)
|    |    |-- abc123: string (nullable = true)
|    |    |-- services: struct (nullable = true)
|    |    |    |-- service: array (nullable = true)
|    |    |    |    |-- element: struct (containsNull = true)
|    |    |    |    |    |-- type: string (nullable = true)
|    |    |    |    |    |-- subtype: string (nullable = true)

我需要更改 type 列的 (anonymize/hash) 值。

对于 Spark 3.1+,有一个列方法 withField 可用于更新结构字段。

假设这是您的输入数据框(对应于您提供的架构):

from pyspark.sql import Row

df = spark.createDataFrame([
    Row(abc=Row(xyz=Row(abc123="value123", services=[Row(type="type1", subtype="subtype1")])))
])

df.show(truncate=False)
#+---------------------------------+
#|abc                              |
#+---------------------------------+
#|{{value123, [{type1, subtype1}]}}|
#+---------------------------------+

你可以使用 transform on the array services to hash the field type for each struct element (here I used xxhash64 函数来实现,例如:

import pyspark.sql.functions as F

df2 = df.withColumn(
    "abc",
    F.col("abc").withField(
        "xyz",
        F.col("abc.xyz").withField(
            "services",
            F.expr("transform(abc.xyz.services, x -> struct(xxhash64(x.type) as type, x.subtype))")
        )
    )
)

df2.show(truncate=False)
#+-----------------------------------------------+
#|abc                                            |
#+-----------------------------------------------+
#|{{value123, [{2134479862461603894, subtype1}]}}|
#+-----------------------------------------------+

对于较旧的 Spark 版本,您需要重新创建整个结构以更新字段,这在有许多嵌套字段时会变得乏味。在你的情况下它会是这样的:

df2 = df.withColumn(
    "abc",
    F.struct(
        F.struct(
            F.col("abc.xyz.abc123"),
            F.expr(
                "transform(abc.xyz.services, x -> struct(xxhash64(x.type) as type, x.subtype))"
            ).alias("services")
        ).alias("xyz")
    )
)