[
https://issues.apache.org/jira/browse/CALCITE-6494?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Dmitry Sysolyatin updated CALCITE-6494:
---------------------------------------
Description:
MongoAdapter throws an exception if any function unsupported by MongoProject is
used
A simple example [1]:
{code}
@Test void testCalciteFunc() {
assertModel(MODEL)
.query("select CHAR_LENGTH(state) from zips")
.runs();
}
{code}
throws
{code}
Unable to implement MongoToEnumerableConverter: rowcount = 100.0, cumulative
cost = {30.0 rows, 30.1 cpu, 0.0 io}, id = 39
MongoProject(EXPR$0=[CHAR_LENGTH(CAST(ITEM($0, 'state')):VARCHAR(2))]):
rowcount = 100.0, cumulative cost = {20.0 rows, 20.1 cpu, 0.0 io}, id = 37
MongoTableScan(table=[[mongo_raw, zips]]): rowcount = 100.0, cumulative
cost = {10.0 rows, 10.100000000000001 cpu, 0.0 io}, id = 13
......
Suppressed: java.lang.IllegalArgumentException: Translation of
CHAR_LENGTH(CAST(ITEM($0, 'state')):VARCHAR(2)) is not supported by MongoProject
{code}
The reason is that MongoProjectRule does not properly filter projects from
LogicalProject.getNamedProjects that can be used. I think the proper
implementation of this rule should be similar to DruidProjectRule [2]. It
should extract input references from the project if it cannot be fully
supported and add an additional Project on top of MongoProject.
[1] -
https://github.com/dssysolyatin/calcite/commit/96cddae9c6516701e1aa44f7c75df5afdc497f8d
[2] -
https://github.com/apache/calcite/blob/ea1a255fd071a518fe8d30e361efa0696164a037/druid/src/main/java/org/apache/calcite/adapter/druid/DruidRules.java#L353
was:
MongoAdapter throws an exception if any function unsupported by MongoProject is
used
A simple example [1]:
{code}
@Test void testCalciteFunc() {
assertModel(MODEL)
.query("select CHAR_LENGTH(state) from zips")
.runs();
}
{code}
throws
{code}
Unable to implement MongoToEnumerableConverter: rowcount = 100.0, cumulative
cost = {30.0 rows, 30.1 cpu, 0.0 io}, id = 39
MongoProject(EXPR$0=[CHAR_LENGTH(CAST(ITEM($0, 'state')):VARCHAR(2))]):
rowcount = 100.0, cumulative cost = {20.0 rows, 20.1 cpu, 0.0 io}, id = 37
MongoTableScan(table=[[mongo_raw, zips]]): rowcount = 100.0, cumulative
cost = {10.0 rows, 10.100000000000001 cpu, 0.0 io}, id = 13
......
Suppressed: java.lang.IllegalArgumentException: Translation of
CHAR_LENGTH(CAST(ITEM($0, 'state')):VARCHAR(2)) is not supported by MongoProject
{code}
The reason is that MongoProjectRule does not properly filter projects from
LogicalProjects.getNamedProjects that can be used. I think the proper
implementation of this rule should be similar to DruidProjectRule [2]. It
should extract input references from the project if it cannot be fully
supported and add an additional Project on top of MongoProject.
[1] -
https://github.com/dssysolyatin/calcite/commit/96cddae9c6516701e1aa44f7c75df5afdc497f8d
[2] -
https://github.com/apache/calcite/blob/ea1a255fd071a518fe8d30e361efa0696164a037/druid/src/main/java/org/apache/calcite/adapter/druid/DruidRules.java#L353
> MongoAdapter throws an exception if any function unsupported by MongoProject
> is used
> ------------------------------------------------------------------------------------
>
> Key: CALCITE-6494
> URL: https://issues.apache.org/jira/browse/CALCITE-6494
> Project: Calcite
> Issue Type: Bug
> Components: mongodb-adapter
> Affects Versions: 1.37.0
> Reporter: Dmitry Sysolyatin
> Priority: Major
>
> MongoAdapter throws an exception if any function unsupported by MongoProject
> is used
> A simple example [1]:
> {code}
> @Test void testCalciteFunc() {
> assertModel(MODEL)
> .query("select CHAR_LENGTH(state) from zips")
> .runs();
> }
> {code}
> throws
> {code}
> Unable to implement MongoToEnumerableConverter: rowcount = 100.0, cumulative
> cost = {30.0 rows, 30.1 cpu, 0.0 io}, id = 39
> MongoProject(EXPR$0=[CHAR_LENGTH(CAST(ITEM($0, 'state')):VARCHAR(2))]):
> rowcount = 100.0, cumulative cost = {20.0 rows, 20.1 cpu, 0.0 io}, id = 37
> MongoTableScan(table=[[mongo_raw, zips]]): rowcount = 100.0, cumulative
> cost = {10.0 rows, 10.100000000000001 cpu, 0.0 io}, id = 13
> ......
> Suppressed: java.lang.IllegalArgumentException: Translation of
> CHAR_LENGTH(CAST(ITEM($0, 'state')):VARCHAR(2)) is not supported by
> MongoProject
> {code}
> The reason is that MongoProjectRule does not properly filter projects from
> LogicalProject.getNamedProjects that can be used. I think the proper
> implementation of this rule should be similar to DruidProjectRule [2]. It
> should extract input references from the project if it cannot be fully
> supported and add an additional Project on top of MongoProject.
> [1] -
> https://github.com/dssysolyatin/calcite/commit/96cddae9c6516701e1aa44f7c75df5afdc497f8d
> [2] -
> https://github.com/apache/calcite/blob/ea1a255fd071a518fe8d30e361efa0696164a037/druid/src/main/java/org/apache/calcite/adapter/druid/DruidRules.java#L353
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
