apache / apache/beam

WITH query AS + SELECT query JOIN other throws invalid type

Open
#19,839 0 comments 0 reactions 0 assignees View on GitHub
bug dsl P3 sql
Dominant language
Java
Stars
8.7k
Forks
4.7k
Avg merge
1d 20h
Merged PRs (30d)
196

Description

The first one of the three following queries fails, despite queries being equivalent:
```

Pipeline p = Pipeline.create();

Schema schemaA =
Schema.of(
Schema.Field.of("id",
Schema.FieldType.BYTES),
Schema.Field.of("fA1", Schema.FieldType.STRING));

Schema schemaB
=
Schema.of(
Schema.Field.of("id", Schema.FieldType.STRING),
Schema.Field.of("fB1",
Schema.FieldType.STRING));

PCollection inputA =
p.apply(Create.of(ImmutableList.of()).withCoder(SchemaCoder.of(schemaA)));

PCollection
inputB =
p.apply(Create.of(ImmutableList.of()).withCoder(SchemaCoder.of(schemaB)));

//
Fails
String query1 =
"WITH query AS "
+ "( "
+ " SELECT id, fA1, fA1 AS fA1_2
"
+ " FROM tblA"
+ ") "
+ "SELECT fA1, fB1, fA1_2 "
+ "FROM query
"
+ "JOIN tblB ON (TO_HEX(query.id) = tblB.id)";

// Ok
String query2 =
"WITH query
AS "
+ "( "
+ " SELECT fA1, fB1, fA1 AS fA1_2 "
+ " FROM tblA "

+ " JOIN tblB "
+ " ON (TO_HEX(tblA.id) = tblB.id) "
+ ")"
+ "SELECT fA1,
fB1, fA1_2 "
+ "FROM query ";

// Ok
String query3 =
"WITH query AS "
+ "( "

+ " SELECT TO_HEX(id) AS id, fA1, fA1 AS fA1_2 "
+ " FROM tblA"
+ ") "
+ "SELECT fA1,
fB1, fA1_2 "
+ "FROM query "
+ "JOIN tblB ON (query.id = tblB.id)";

Schema transform3 =

PCollectionTuple.of("tblA", inputA)
.and("tblB", inputB)
.apply(SqlTransform.query(query3))

.getSchema();
System.out.println(transform3);

Schema transform2 =
PCollectionTuple.of("tblA",
inputA)
.and("tblB", inputB)
.apply(SqlTransform.query(query2))
.getSchema();
System.out.println(transform2);

Schema
transform1 =
PCollectionTuple.of("tblA", inputA)
.and("tblB", inputB)
.apply(SqlTransform.query(query1))

.getSchema();
System.out.println(transform1);

```

 

The error is:
```

Exception in thread "main" java.lang.AssertionError: Field ordinal 2 is invalid for  type 'RecordType(VARBINARY
id, VARCHAR fA1)'Exception in thread "main" java.lang.AssertionError: Field ordinal 2 is invalid for 
type 'RecordType(VARBINARY id, VARCHAR fA1)' at org.apache.beam.repackaged.sql.org.apache.calcite.rex.RexBuilder.makeFieldAccess(RexBuilder.java:197)
```

 

If I change `schemaB.id` to `BYTES` (while also avoid using `TO_HEX`), all queries work fine. 

Imported from Jira [BEAM-8896](https://issues.apache.org/jira/browse/BEAM-8896). Original Jira may contain additional context.
Reported by: fdiazgon.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.