core/org.jetbrains.kotlinx.dataframe.api/castTo

castTo

inline fun <reified T> DataFrame<*>.castTo(schemaFrom: DataFrame<T>, verify: Boolean = true): DataFrame<T>

inline fun <reified T> DataFrame<*>.castTo(schemaFrom: Function<DataFrame<T>>, verify: Boolean = true): DataFrame<T>

With the compiler plugin, schema marker T of DataFrame can be a local type. You cannot refer to it directly from your code, like a type argument for cast. The example below shows a situation where you'd need to cast DataFrame<*> to DataFrame. This function helps by inferring type from schemaFrom

// parse listOf("b:1:abc", "c:2:bca")
private fun convert(data: List<String>)/*: DataFrame<plugin generated local type>*/= data.map { it.split(":") }.toDataFrame {
     "part1" from { it[0] }
     "part2" from { it[1].toInt() }
     "part3" from { it[2] }
}

fun serialize(data: List<String>, destination: File) {
     convert(data).writeJson(destination)
}

fun deserializeAndUse(file: File) {
     val df = DataFrame.readJson(file).castTo(schemaFrom = ::convert)
     // Possible to use properties
     df.part1.print()
}

For more information: See cast on the documentation website.