NPE in FrameColumnImpl.schema property

[koperagen]

NPE happens in this line.

values.mapNotNull { it.takeIf { it.nrow > 0 }?.schema() }.intersectSchemas()

Code that leads to it:

I briefly looked into csv.kt and found that only tryParseImpl method could potentially create FrameColumn and provide null there. Need to confirm that it's possible.
Another thing that could cause problem is read method itself that actually tries to parse the file as JSON, CSV, TSV, Excel and others until it succeeds. So, if that file cannot be parsed as CSV, it continues and can produce strange result too

Full stack trace below

The problem is found in one of the loaded libraries: check library converters (fields callbacks)
java.lang.NullPointerException: Parameter specified as non-null is null: method org.jetbrains.kotlinx.dataframe.DataFrameKt.getNrow, parameter <this>
org.jetbrains.kotlinx.jupyter.exceptions.ReplLibraryException: The problem is found in one of the loaded libraries: check library converters (fields callbacks)
	at org.jetbrains.kotlinx.jupyter.exceptions.CompositeReplExceptionKt.throwLibraryException(CompositeReplException.kt:50)
	at org.jetbrains.kotlinx.jupyter.codegen.FieldsProcessorImpl.process(FieldsProcessorImpl.kt:68)
	at org.jetbrains.kotlinx.jupyter.repl.impl.CellExecutorImpl$execute$1$1.invoke(CellExecutorImpl.kt:94)
	at org.jetbrains.kotlinx.jupyter.repl.impl.CellExecutorImpl$execute$1$1.invoke(CellExecutorImpl.kt:93)
	at org.jetbrains.kotlinx.jupyter.config.LoggingKt.catchAll(logging.kt:42)
	at org.jetbrains.kotlinx.jupyter.config.LoggingKt.catchAll$default(logging.kt:41)
	at org.jetbrains.kotlinx.jupyter.repl.impl.CellExecutorImpl.execute(CellExecutorImpl.kt:93)
	at org.jetbrains.kotlinx.jupyter.repl.CellExecutor$DefaultImpls.execute$default(CellExecutor.kt:14)
	at org.jetbrains.kotlinx.jupyter.ReplForJupyterImpl$evalEx$1.invoke(repl.kt:500)
	at org.jetbrains.kotlinx.jupyter.ReplForJupyterImpl$evalEx$1.invoke(repl.kt:478)
	at org.jetbrains.kotlinx.jupyter.ReplForJupyterImpl.withEvalContext(repl.kt:441)
	at org.jetbrains.kotlinx.jupyter.ReplForJupyterImpl.evalEx(repl.kt:478)
	at org.jetbrains.kotlinx.jupyter.messaging.ProtocolKt$shellMessagesHandler$2$res$1.invoke(protocol.kt:320)
	at org.jetbrains.kotlinx.jupyter.messaging.ProtocolKt$shellMessagesHandler$2$res$1.invoke(protocol.kt:314)
	at org.jetbrains.kotlinx.jupyter.JupyterExecutorImpl$runExecution$execThread$1.invoke(execution.kt:38)
	at org.jetbrains.kotlinx.jupyter.JupyterExecutorImpl$runExecution$execThread$1.invoke(execution.kt:33)
	at kotlin.concurrent.ThreadsKt$thread$thread$1.run(Thread.kt:30)
Caused by: java.lang.NullPointerException: Parameter specified as non-null is null: method org.jetbrains.kotlinx.dataframe.DataFrameKt.getNrow, parameter <this>
	at org.jetbrains.kotlinx.dataframe.DataFrameKt.getNrow(DataFrame.kt)
	at org.jetbrains.kotlinx.dataframe.impl.columns.FrameColumnImpl$schema$1.invoke(FrameColumnImpl.kt:43)
	at org.jetbrains.kotlinx.dataframe.impl.columns.FrameColumnImpl$schema$1.invoke(FrameColumnImpl.kt:42)
	at kotlin.SynchronizedLazyImpl.getValue(LazyJVM.kt:74)
	at org.jetbrains.kotlinx.dataframe.impl.schema.UtilsKt.extractSchema(Utils.kt:92)
	at org.jetbrains.kotlinx.dataframe.impl.schema.UtilsKt.extractSchema(Utils.kt:26)
	at org.jetbrains.kotlinx.dataframe.api.SchemaKt.schema(schema.kt:17)
	at org.jetbrains.kotlinx.dataframe.impl.codeGen.ReplCodeGeneratorImpl.process(ReplCodeGeneratorImpl.kt:50)
	at org.jetbrains.kotlinx.dataframe.jupyter.Integration.updateAnyFrameVariable(Integration.kt:132)
	at org.jetbrains.kotlinx.dataframe.jupyter.Integration.access$updateAnyFrameVariable(Integration.kt:73)
	at org.jetbrains.kotlinx.dataframe.jupyter.Integration$onLoaded$4.invoke(Integration.kt:295)
	at org.jetbrains.kotlinx.dataframe.jupyter.Integration$onLoaded$4.invoke(Integration.kt:290)
	at org.jetbrains.kotlinx.jupyter.api.libraries.FieldHandlerFactory.createUpdateExecution$lambda$0(FieldHandlerFactory.kt:38)
	at org.jetbrains.kotlinx.jupyter.codegen.FieldsProcessorImplKt.executeEx(FieldsProcessorImpl.kt:88)
	at org.jetbrains.kotlinx.jupyter.codegen.FieldsProcessorImplKt.access$executeEx(FieldsProcessorImpl.kt:1)
	at org.jetbrains.kotlinx.jupyter.codegen.FieldsProcessorImpl.process(FieldsProcessorImpl.kt:47)
	... 15 more

[koperagen]

So indeed some JSON value in the cell + null value in other causes an issue in CSV reading

[koperagen]

val df2 = DataFrame.readDelimStr("""name
"[""str""]"
null
""")

[Jolanrensen]

What is the expected result? A FrameColumn cannot contain nulls, right?

Should we:

• throw exception, because FrameColumn cannot contain null
• convert null to empty dataframe
• Don't parse the value as JSON, but keep it a String