eric-wang-1990 commented on code in PR #2884:
URL: https://github.com/apache/arrow-adbc/pull/2884#discussion_r2110506372
##########
csharp/src/Drivers/Databricks/DatabricksStatement.cs:
##########
@@ -151,5 +158,233 @@ internal void SetMaxBytesPerFile(long maxBytesPerFile)
{
this.maxBytesPerFile = maxBytesPerFile;
}
+
+ /// <summary>
+ /// Helper method to handle the special case for the "SPARK" catalog
in metadata queries.
+ ///
+ /// Why:
+ /// - In Databricks, the legacy "SPARK" catalog is used as a
placeholder to represent the default catalog.
+ /// - When a client requests metadata for the "SPARK" catalog, the
underlying API expects a null catalog name
+ /// to trigger default catalog behavior. Passing "SPARK" directly
would not return the expected results.
+ ///
+ /// What it does:
+ /// - If the CatalogName property is set to "SPARK"
(case-insensitive), this method sets it to null.
+ /// - This ensures that downstream API calls behave as if no catalog
was specified, returning default catalog metadata.
+ ///
+ /// This logic is required to maintain compatibility with legacy tools
and standards that expect "SPARK" to act as a default catalog alias.
+ /// </summary>
+ private void HandleSparkCatalog()
+ {
+ if (CatalogName != null && CatalogName.Equals("SPARK",
StringComparison.OrdinalIgnoreCase))
+ {
+ CatalogName = null;
+ }
+ }
+
+ /// <summary>
+ /// Overrides the GetCatalogsAsync method to handle
EnableMultipleCatalogSupport flag.
+ /// When EnableMultipleCatalogSupport is false, returns a single
catalog "SPARK" without making an RPC call.
+ /// When EnableMultipleCatalogSupport is true, delegates to the base
class implementation to retrieve actual catalogs.
+ /// </summary>
+ /// <param name="cancellationToken">Cancellation token</param>
+ /// <returns>Query result containing catalog information</returns>
+ protected override async Task<QueryResult>
GetCatalogsAsync(CancellationToken cancellationToken = default)
+ {
+ // If EnableMultipleCatalogSupport is false, return a single
catalog "SPARK" without making an RPC call
+ if (enableMultipleCatalogSupport)
Review Comment:
Thanks for catching, Fixed
##########
csharp/src/Drivers/Apache/Hive2/HiveServer2Statement.cs:
##########
@@ -352,7 +352,7 @@ protected void ValidateOptions(IReadOnlyDictionary<string,
string> properties)
}
}
- private async Task<QueryResult>
ExecuteMetadataCommandQuery(CancellationToken cancellationToken)
+ protected virtual async Task<QueryResult>
ExecuteMetadataCommandQuery(CancellationToken cancellationToken)
Review Comment:
Thanks for catching, this is not used anymore will revert
##########
csharp/src/Drivers/Databricks/DatabricksStatement.cs:
##########
@@ -151,5 +158,233 @@ internal void SetMaxBytesPerFile(long maxBytesPerFile)
{
this.maxBytesPerFile = maxBytesPerFile;
}
+
+ /// <summary>
+ /// Helper method to handle the special case for the "SPARK" catalog
in metadata queries.
+ ///
+ /// Why:
+ /// - In Databricks, the legacy "SPARK" catalog is used as a
placeholder to represent the default catalog.
+ /// - When a client requests metadata for the "SPARK" catalog, the
underlying API expects a null catalog name
+ /// to trigger default catalog behavior. Passing "SPARK" directly
would not return the expected results.
+ ///
+ /// What it does:
+ /// - If the CatalogName property is set to "SPARK"
(case-insensitive), this method sets it to null.
+ /// - This ensures that downstream API calls behave as if no catalog
was specified, returning default catalog metadata.
+ ///
+ /// This logic is required to maintain compatibility with legacy tools
and standards that expect "SPARK" to act as a default catalog alias.
+ /// </summary>
+ private void HandleSparkCatalog()
+ {
+ if (CatalogName != null && CatalogName.Equals("SPARK",
StringComparison.OrdinalIgnoreCase))
+ {
+ CatalogName = null;
+ }
+ }
+
+ /// <summary>
+ /// Overrides the GetCatalogsAsync method to handle
EnableMultipleCatalogSupport flag.
+ /// When EnableMultipleCatalogSupport is false, returns a single
catalog "SPARK" without making an RPC call.
+ /// When EnableMultipleCatalogSupport is true, delegates to the base
class implementation to retrieve actual catalogs.
+ /// </summary>
+ /// <param name="cancellationToken">Cancellation token</param>
+ /// <returns>Query result containing catalog information</returns>
+ protected override async Task<QueryResult>
GetCatalogsAsync(CancellationToken cancellationToken = default)
+ {
+ // If EnableMultipleCatalogSupport is false, return a single
catalog "SPARK" without making an RPC call
+ if (enableMultipleCatalogSupport)
+ {
+ // Create a schema with a single column TABLE_CAT
+ var field = new Field("TABLE_CAT", StringType.Default, true);
+ var schema = new Schema(new[] { field }, null);
+
+ // Create a single row with value "SPARK"
+ var builder = new StringArray.Builder();
+ builder.Append("SPARK");
+ var array = builder.Build();
+
+ // Return the result without making an RPC call
+ return new QueryResult(1, new
HiveServer2Connection.HiveInfoArrowStream(schema, new[] { array }));
+ }
+
+ // If EnableMultipleCatalogSupport is true, delegate to base class
implementation
+ return await base.GetCatalogsAsync(cancellationToken);
+ }
+
+ /// <summary>
+ /// Overrides the GetSchemasAsync method to handle the SPARK catalog
case.
+ /// When EnableMultipleCatalogSupport is true:
+ /// - If catalog is "SPARK", sets catalogName to null in the API call
+ /// When EnableMultipleCatalogSupport is false:
+ /// - If catalog is not null or SPARK, returns empty result without
RPC call
+ /// </summary>
+ /// <param name="cancellationToken">Cancellation token</param>
+ /// <returns>Query result containing schema information</returns>
+ protected override async Task<QueryResult>
GetSchemasAsync(CancellationToken cancellationToken = default)
+ {
+ // Handle SPARK catalog case
+ HandleSparkCatalog();
+
+ // If EnableMultipleCatalogSupport is false and catalog is not
null or SPARK, return empty result without RPC call
+ if (enableMultipleCatalogSupport && CatalogName != null)
+ {
+ // Create a schema with TABLE_CATALOG and TABLE_SCHEMA columns
Review Comment:
Fixed
##########
csharp/src/Drivers/Apache/Spark/SparkStatement.cs:
##########
@@ -52,5 +54,11 @@ protected override void
SetStatementProperties(TExecuteStatementReq statement)
IntervalTypesAsArrow = false,
};
}
+
+ // Override ExecuteMetadataCommandQuery to allow derived classes to
override it
+ protected override async Task<QueryResult>
ExecuteMetadataCommandQuery(CancellationToken cancellationToken)
Review Comment:
Thanks for catching, this is not used anymore will revert
##########
csharp/src/Drivers/Databricks/DatabricksStatement.cs:
##########
@@ -151,5 +158,233 @@ internal void SetMaxBytesPerFile(long maxBytesPerFile)
{
this.maxBytesPerFile = maxBytesPerFile;
}
+
+ /// <summary>
+ /// Helper method to handle the special case for the "SPARK" catalog
in metadata queries.
+ ///
+ /// Why:
+ /// - In Databricks, the legacy "SPARK" catalog is used as a
placeholder to represent the default catalog.
+ /// - When a client requests metadata for the "SPARK" catalog, the
underlying API expects a null catalog name
+ /// to trigger default catalog behavior. Passing "SPARK" directly
would not return the expected results.
+ ///
+ /// What it does:
+ /// - If the CatalogName property is set to "SPARK"
(case-insensitive), this method sets it to null.
+ /// - This ensures that downstream API calls behave as if no catalog
was specified, returning default catalog metadata.
+ ///
+ /// This logic is required to maintain compatibility with legacy tools
and standards that expect "SPARK" to act as a default catalog alias.
+ /// </summary>
+ private void HandleSparkCatalog()
+ {
+ if (CatalogName != null && CatalogName.Equals("SPARK",
StringComparison.OrdinalIgnoreCase))
+ {
+ CatalogName = null;
+ }
+ }
+
+ /// <summary>
+ /// Overrides the GetCatalogsAsync method to handle
EnableMultipleCatalogSupport flag.
+ /// When EnableMultipleCatalogSupport is false, returns a single
catalog "SPARK" without making an RPC call.
+ /// When EnableMultipleCatalogSupport is true, delegates to the base
class implementation to retrieve actual catalogs.
+ /// </summary>
+ /// <param name="cancellationToken">Cancellation token</param>
+ /// <returns>Query result containing catalog information</returns>
+ protected override async Task<QueryResult>
GetCatalogsAsync(CancellationToken cancellationToken = default)
+ {
Review Comment:
Nope, the GetCatalogs call does not take in the catalogName as parameter so
not needed
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
