Clarify the situation when generated extension property causes ClassCast or NPE exceptions

Author: koperagen

core/src/main/kotlin/org/jetbrains/kotlinx/dataframe/exceptions/ColumnValuesMismatchColumnTypeException.kt

@@ -0,0 +1,16 @@
+package org.jetbrains.kotlinx.dataframe.exceptions
+
+import org.jetbrains.kotlinx.dataframe.AnyCol
+import org.jetbrains.kotlinx.dataframe.DataColumn
+import org.jetbrains.kotlinx.dataframe.columns.values
+import org.jetbrains.kotlinx.dataframe.type
+
+/**
+ * Extension properties are generated according to [DataColumn.type] property
+ * [DataColumn.type] must match types of [DataColumn.values], but it can fail to do so.
+ * This causes [ClassCastException] or [NullPointerException] when you use extension property and actual value is of different type or is null.
+ * If generated extension property causes this exception, this is a bug in the library
+ * You can work around this problem by referring to [column] using String API
+ */
+public class ColumnValuesMismatchColumnTypeException(public val column: AnyCol, cause: Throwable) :
+    RuntimeException("Failed to convert column '${column.name()}'", cause)

core/src/main/kotlin/org/jetbrains/kotlinx/dataframe/impl/api/convert.kt

@@ -31,6 +31,7 @@ import org.jetbrains.kotlinx.dataframe.columns.values
 import org.jetbrains.kotlinx.dataframe.dataTypes.IFRAME
 import org.jetbrains.kotlinx.dataframe.dataTypes.IMG
 import org.jetbrains.kotlinx.dataframe.exceptions.CellConversionException
+import org.jetbrains.kotlinx.dataframe.exceptions.ColumnValuesMismatchColumnTypeException
 import org.jetbrains.kotlinx.dataframe.exceptions.TypeConversionException
 import org.jetbrains.kotlinx.dataframe.exceptions.TypeConverterNotFoundException
 import org.jetbrains.kotlinx.dataframe.impl.columns.DataColumnInternal
@@ -60,7 +61,16 @@ internal fun <T, C, R> Convert<T, C>.withRowCellImpl(
     type: KType,
     infer: Infer,
     rowConverter: RowValueExpression<T, C, R>,
-): DataFrame<T> = to { col -> df.newColumn(type, col.name, infer) { rowConverter(it, it[col]) } }
+): DataFrame<T> =
+    to { col ->
+        try {
+            df.newColumn(type, col.name, infer) { rowConverter(it, it[col]) }
+        } catch (e: ClassCastException) {
+            throw ColumnValuesMismatchColumnTypeException(col, e)
+        } catch (e: NullPointerException) {
+            throw ColumnValuesMismatchColumnTypeException(col, e)
+        }
+    }
 
 @PublishedApi
 internal fun <T, C, R> Convert<T, C>.convertRowColumnImpl(

core/src/test/kotlin/org/jetbrains/kotlinx/dataframe/api/convert.kt

@@ -6,10 +6,12 @@ import io.kotest.matchers.shouldBe
 import kotlinx.datetime.Clock
 import kotlinx.datetime.Instant
 import kotlinx.datetime.LocalTime
+import org.jetbrains.kotlinx.dataframe.ColumnsContainer
 import org.jetbrains.kotlinx.dataframe.DataColumn
 import org.jetbrains.kotlinx.dataframe.DataFrame
 import org.jetbrains.kotlinx.dataframe.annotations.DataSchema
 import org.jetbrains.kotlinx.dataframe.exceptions.CellConversionException
+import org.jetbrains.kotlinx.dataframe.exceptions.ColumnValuesMismatchColumnTypeException
 import org.jetbrains.kotlinx.dataframe.exceptions.TypeConversionException
 import org.jetbrains.kotlinx.dataframe.exceptions.TypeConverterNotFoundException
 import org.jetbrains.kotlinx.dataframe.hasNulls
@@ -172,4 +174,15 @@ class ConvertTests {
         val res = col.convertTo<String>()
         res.print()
     }
+
+    private interface Marker
+
+    private val ColumnsContainer<Marker>.a get() = this["a"] as DataColumn<String>
+
+    @Test
+    fun `convert with buggy extension property`() {
+        shouldThrow<ColumnValuesMismatchColumnTypeException> {
+            dataFrameOf("a")(1, 2, 3).cast<Marker>().convert { a }.with { it }
+        }
+    }
 }