[
https://issues.apache.org/jira/browse/CALCITE-5157?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17539359#comment-17539359
]
Viliam Durina commented on CALCITE-5157:
----------------------------------------
When the parser sees {{a.b.c}}, it can't tell whether {{a}} is a column name
and {{.b.c}} is field access, or whether {{a.b}} is table and column name, and
{{c}} is field access. However, when it sees {{(a).b.c}}, it knows that {{a}}
is column name. If {{a}} in {{(a).b.c}} is the table name, such as in {{select
(a).b.c from a}}, then calcite throws {{Column 'a' not found in any table}}.
However, I checked Postgre, and it allows queries like {{select a from a}}.
This query returns a single column, with a record-typed value. Then it makes
sense in PG that {{select (a).b.c from a}} also works. Oracle, MySql or SQL
server don't allow this, I'm not sure about the SQL standard...
> ClassCastException in checkRollUp with DOT operator
> ---------------------------------------------------
>
> Key: CALCITE-5157
> URL: https://issues.apache.org/jira/browse/CALCITE-5157
> Project: Calcite
> Issue Type: Bug
> Components: core
> Affects Versions: 1.30.0
> Reporter: Viliam Durina
> Priority: Major
> Labels: pull-request-available
> Time Spent: 10m
> Remaining Estimate: 0h
>
> When a query contains nested field access and is using parentheses to
> disambiguate the identifier, the {{SqlValidatorImpl.checkRollup()}} method
> throws a {{{}ClassCastException{}}}, assuming that the input of the DOT
> operator is a {{{}SqlCall{}}}. I think this assumption is wrong, the DOT
> operator typically has {{SqlIdentifier}} as an input, probably also other
> classes.
> Here's the stack trace:
> {{java.lang.ClassCastException: class org.apache.calcite.sql.SqlIdentifier
> cannot be cast to class org.apache.calcite.sql.SqlCall
> (org.apache.calcite.sql.SqlIdentifier and org.apache.calcite.sql.SqlCall are
> in unnamed module of loader 'app')}}
> {{ at
> org.apache.calcite.sql.validate.SqlValidatorImpl.checkRollUp(SqlValidatorImpl.java:3730)}}
> {{ at
> org.apache.calcite.sql.validate.SqlValidatorImpl.checkRollUp(SqlValidatorImpl.java:3749)}}
> {{ at
> org.apache.calcite.sql.validate.SqlValidatorImpl.checkRollUpInSelectList(SqlValidatorImpl.java:3673)}}
> {{ at
> org.apache.calcite.sql.validate.SqlValidatorImpl.validateSelect(SqlValidatorImpl.java:3661)}}
> {{ at
> org.apache.calcite.sql.validate.SelectNamespace.validateImpl(SelectNamespace.java:64)}}
> {{ at
> org.apache.calcite.sql.validate.AbstractNamespace.validate(AbstractNamespace.java:89)}}
> {{ at
> org.apache.calcite.sql.validate.SqlValidatorImpl.validateNamespace(SqlValidatorImpl.java:1100)}}
> {{ at
> org.apache.calcite.sql.validate.SqlValidatorImpl.validateQuery(SqlValidatorImpl.java:1071)}}
> {{ at org.apache.calcite.sql.SqlSelect.validate(SqlSelect.java:247)}}
> {{ at
> org.apache.calcite.sql.validate.SqlValidatorImpl.validateScopedExpression(SqlValidatorImpl.java:1046)}}
> {{ at
> org.apache.calcite.sql.validate.SqlValidatorImpl.validate(SqlValidatorImpl.java:752)}}
> {{ at
> org.apache.calcite.sql2rel.SqlToRelConverter.convertQuery(SqlToRelConverter.java:587)}}
> {{ at org.apache.calcite.prepare.Prepare.prepareSql(Prepare.java:257)}}
> {{ at org.apache.calcite.prepare.Prepare.prepareSql(Prepare.java:220)}}
> {{ at
> org.apache.calcite.prepare.CalcitePrepareImpl.prepare2_(CalcitePrepareImpl.java:648)}}
> {{ at
> org.apache.calcite.prepare.CalcitePrepareImpl.prepare_(CalcitePrepareImpl.java:514)}}
> {{ at
> org.apache.calcite.prepare.CalcitePrepareImpl.prepareSql(CalcitePrepareImpl.java:484)}}
> {{ at
> org.apache.calcite.jdbc.CalciteConnectionImpl.parseQuery(CalciteConnectionImpl.java:234)}}
> {{ at
> org.apache.calcite.jdbc.CalciteMetaImpl.prepareAndExecute(CalciteMetaImpl.java:623)}}
> {{ at
> org.apache.calcite.avatica.AvaticaConnection.prepareAndExecuteInternal(AvaticaConnection.java:677)}}
> {{ at
> org.apache.calcite.avatica.AvaticaStatement.executeInternal(AvaticaStatement.java:156)}}
> {{ ... 67 more}}
> The problem can be reproduced by modifying the
> {{ReflectiveSchemaTest.testSelectWithFieldAccessOnFirstLevelRecordType()}}
> test and putting {{au."birthPlace"}} into parentheses:
> {{select (au."birthPlace")."city" as city from ... }}
> Putting identifiers into parentheses is common to disambiguate field access
> from identifier qualification. For exmaple, if the {{au}} prefix is removed
> from the query in the {{testSelectWithFieldAccessOnFirstLevelRecordType}}
> test, it will fail with {{{}Table 'birthPlace' not found{}}}. But if we put
> {{\"birthPlace\"}} into parentheses, then it is correctly recognized as a
> column of the {{authors}} table.
> I'm not sure about the correct fix to this issue which would not break the
> roll-up functionality, perhaps its author [~zhumayun] can help reviewing the
> PR or suggesting another fix. I'll soon create a PR with a proposed fix.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
