fix(go/adbc/driver/snowflake): Made GetObjects case insensitive #1328
Conversation
…and added Skip capability for ValueTests TestConfiguration is now a part of SnowflakeTestingUtils, and can be used across the tests (no redundant reading the configuration) ValueTests were marked as Theory, changed them to SkippableTheory to skip the tests, if a XUnit.SkipException is thrown
| conditions = append(conditions, ` CATALOG_NAME LIKE '`+*catalog+`'`) | ||
| conditions = append(conditions, ` CATALOG_NAME ILIKE '`+*catalog+`'`) | ||
| } | ||
| if dbSchema != nil && *dbSchema != "" { | ||
| conditions = append(conditions, ` SCHEMA_NAME LIKE '`+*dbSchema+`'`) | ||
| conditions = append(conditions, ` SCHEMA_NAME ILIKE '`+*dbSchema+`'`) |
There was a problem hiding this comment.
would it be worthwhile adding tests in the Go side for this?
There was a problem hiding this comment.
I was looking at the existing tests in driver_test.go and they seemed like integration tests similar to what we have in DriverTest.cs. I am not sure if adding the same kind of test would add as much value. Is there a specific test that you had in mind, that I can draw inspiration from?
There was a problem hiding this comment.
At a minimum perhaps checking that the query is the right text. I don't think that there's a test in driver_test.go which verifies the case sensitivity though, which would be good to have.
|
|
||
| public static IEnumerable<object[]> CatalogNamePatternData() | ||
| { | ||
| Skip.IfNot(Utils.CanExecuteTestConfig(SnowflakeTestingUtils.SNOWFLAKE_TEST_CONFIG_VARIABLE)); |
There was a problem hiding this comment.
why would we skip here instead of just skipping the test that calls this?
There was a problem hiding this comment.
same with the other areas, including the SnowflakeTestingUtils class
Refactoring includes being able to match the exact catalog, schema, and table names when wild cards are involved in the name indentifiers of the GetObjects patterns For example: A pattern name like FOO_BAR will match against FOO1BAR and FOOABAR. Similarly FOO%BAR will match with foohellobar, FOObyeBAR etc.. Therefore it is essential to filter out the returned matches with the expected name identifiers. Similar checks would have to be added by authors using the Driver when calling the GetObjects API.
|
Refactoring includes being able to match the exact catalog, schema, and table names when wild cards are involved in the name indentifiers of the For example: A pattern name like Note: Similar checks would have to be added by authors using the Driver when calling the GetObjects API. |
Description:
GetObjectsAPI was inconsistent case sensitivity for patterns.getObjectsDbSchemasdriver implementation usedLIKEwhereasgetObjectsTablesusedILIKESolution:
Based on discussion here: #1314 changed
GetObjectsAPI to useILIKEthroughout instead ofLIKETesting:
Added tests for lowercase, uppercase and
_wildcardFixes #1314.