[
https://issues.apache.org/jira/browse/PARQUET-1441?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16700445#comment-16700445
]
Nandor Kollar commented on PARQUET-1441:
----------------------------------------
After a bit more investigation, I think this is probably a Parquet issue,
however it isn't clean to me how can this be a regression. The failing code
path in Parquet and in Avro was committed long ago, what I can think of as a
possibility is that Spark might have recently moved from 2-level list structure
to 3-level lists.
The unit test attached to this PR doesn't reflect the problem, because I think
it tests the correct behaviour: in the converter one can switch between 2 and 3
level list with {{parquet.avro.add-list-element-records}} property. The test
for the Spark Jira is a lot more informative.
I think that the problem is that
[AvroRecordConverter|https://github.com/apache/parquet-mr/blob/master/parquet-avro/src/main/java/org/apache/parquet/avro/AvroRecordConverter.java#L865]
tries to decide between 3 and 2 level list by first trying to interpret the
schema as 2 level, and check the compatibility with the expected Avro schema.
Normally, the two are incompatible (if it was written as 3 level), and Parquet
will know that it is a 3-level list. This works fine when lists are not nested
into other lists, but if we try to represent the 3 level nested list Parquet
structure as 2 level, the resulting 2 level Avro schema is not even a valid
Avro schema!
> SchemaParseException: Can't redefine: list in AvroIndexedRecordConverter
> ------------------------------------------------------------------------
>
> Key: PARQUET-1441
> URL: https://issues.apache.org/jira/browse/PARQUET-1441
> Project: Parquet
> Issue Type: Bug
> Components: parquet-avro
> Reporter: Michael Heuer
> Priority: Major
> Labels: pull-request-available
>
> The following unit test added to TestAvroSchemaConverter fails
> {code:java}
> @Test
> public void testConvertedSchemaToStringCantRedefineList() throws Exception {
> String parquet = "message spark_schema {\n" +
> " optional group annotation {\n" +
> " optional group transcriptEffects (LIST) {\n" +
> " repeated group list {\n" +
> " optional group element {\n" +
> " optional group effects (LIST) {\n" +
> " repeated group list {\n" +
> " optional binary element (UTF8);\n" +
> " }\n" +
> " }\n" +
> " }\n" +
> " }\n" +
> " }\n" +
> " }\n" +
> "}\n";
> Configuration conf = new Configuration(false);
> AvroSchemaConverter avroSchemaConverter = new AvroSchemaConverter(conf);
> Schema schema =
> avroSchemaConverter.convert(MessageTypeParser.parseMessageType(parquet));
> schema.toString();
> }
> {code}
> while this one succeeds
> {code:java}
> @Test
> public void testConvertedSchemaToStringCantRedefineList() throws Exception {
> String parquet = "message spark_schema {\n" +
> " optional group annotation {\n" +
> " optional group transcriptEffects (LIST) {\n" +
> " repeated group list {\n" +
> " optional group element {\n" +
> " optional group effects (LIST) {\n" +
> " repeated group list {\n" +
> " optional binary element (UTF8);\n" +
> " }\n" +
> " }\n" +
> " }\n" +
> " }\n" +
> " }\n" +
> " }\n" +
> "}\n";
>
> Configuration conf = new Configuration(false);
> conf.setBoolean("parquet.avro.add-list-element-records", false);
> AvroSchemaConverter avroSchemaConverter = new AvroSchemaConverter(conf);
> Schema schema =
> avroSchemaConverter.convert(MessageTypeParser.parseMessageType(parquet));
> schema.toString();
> }
> {code}
> I don't see a way to influence the code path in AvroIndexedRecordConverter to
> respect this configuration, resulting in the following stack trace downstream
> {noformat}
> Cause: org.apache.avro.SchemaParseException: Can't redefine: list
> at org.apache.avro.Schema$Names.put(Schema.java:1128)
> at org.apache.avro.Schema$NamedSchema.writeNameRef(Schema.java:562)
> at org.apache.avro.Schema$RecordSchema.toJson(Schema.java:690)
> at org.apache.avro.Schema$ArraySchema.toJson(Schema.java:805)
> at org.apache.avro.Schema$UnionSchema.toJson(Schema.java:882)
> at org.apache.avro.Schema$RecordSchema.fieldsToJson(Schema.java:716)
> at org.apache.avro.Schema$RecordSchema.toJson(Schema.java:701)
> at org.apache.avro.Schema$UnionSchema.toJson(Schema.java:882)
> at org.apache.avro.Schema$RecordSchema.fieldsToJson(Schema.java:716)
> at org.apache.avro.Schema$RecordSchema.toJson(Schema.java:701)
> at org.apache.avro.Schema.toString(Schema.java:324)
> at
> org.apache.avro.SchemaCompatibility.checkReaderWriterCompatibility(SchemaCompatibility.java:68)
> at
> org.apache.parquet.avro.AvroRecordConverter.isElementType(AvroRecordConverter.java:866)
> at
> org.apache.parquet.avro.AvroIndexedRecordConverter$AvroArrayConverter.<init>(AvroIndexedRecordConverter.java:333)
> at
> org.apache.parquet.avro.AvroIndexedRecordConverter.newConverter(AvroIndexedRecordConverter.java:172)
> at
> org.apache.parquet.avro.AvroIndexedRecordConverter.<init>(AvroIndexedRecordConverter.java:94)
> at
> org.apache.parquet.avro.AvroIndexedRecordConverter.newConverter(AvroIndexedRecordConverter.java:168)
> at
> org.apache.parquet.avro.AvroIndexedRecordConverter.<init>(AvroIndexedRecordConverter.java:94)
> at
> org.apache.parquet.avro.AvroIndexedRecordConverter.<init>(AvroIndexedRecordConverter.java:66)
> at
> org.apache.parquet.avro.AvroCompatRecordMaterializer.<init>(AvroCompatRecordMaterializer.java:34)
> at
> org.apache.parquet.avro.AvroReadSupport.newCompatMaterializer(AvroReadSupport.java:144)
> at
> org.apache.parquet.avro.AvroReadSupport.prepareForRead(AvroReadSupport.java:136)
> at
> org.apache.parquet.hadoop.InternalParquetRecordReader.initialize(InternalParquetRecordReader.java:204)
> at
> org.apache.parquet.hadoop.ParquetRecordReader.initializeInternalReader(ParquetRecordReader.java:182)
> at
> org.apache.parquet.hadoop.ParquetRecordReader.initialize(ParquetRecordReader.java:140)
> ...
> {noformat}
> See also downstream issues
> https://issues.apache.org/jira/browse/SPARK-25588
> [https://github.com/bigdatagenomics/adam/issues/2058]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
