通过阅读Dataset[String]中的数据并按指示符过滤数据，并分配正确的模式，可以从一个文本文件中提取三个不同的模式：

val mixedData: Dataset[String] = spark.read.textFile("sampleData.csv")

def readWithSchema(indicator: String, schema: StructType): DataFrame = {
  val segmentData = mixedData.filter(_.startsWith(indicator))
  spark.read.schema(schema).csv(segmentData)
}

val borrowerSchema = StructType(
  Seq(
    StructField(name = "Indicator", dataType = StringType),
    StructField(name = "First name", dataType = StringType),
    StructField(name = "Last name", dataType = StringType),
    StructField(name = " Company name", dataType = StringType),
    StructField(name = " joining date", dataType = StringType)
  )
)
val borrowers = readWithSchema("BS", borrowerSchema)
// Declare schema for `Relationship` and `Credity facility`, and read with `readWithSchema`

输出量：

+---------+----------+---------+-------------+-------------+
|Indicator|First name|Last name| Company name| joining date|
+---------+----------+---------+-------------+-------------+
|BS       |Rohan     |Mundle   |Infy         |20230101     |
|BS       |Priyatee  |Sinha    |L&T          |20220101     |
+---------+----------+---------+-------------+-------------+