UnresolvedCollation¶

Overview¶

UnresolvedCollation is a temporary leaf expression used during Spark SQL's analysis phase to represent a collation name that has not yet been resolved from a fully qualified collation name. This expression serves as a placeholder until the collation specification can be validated and resolved into a concrete collation implementation.

Syntax¶

-- Used internally during SQL parsing, not directly invokable by users
COLLATE collation_name

Arguments¶

Argument	Type	Description
collationName	Seq[String]	A sequence of strings representing the hierarchical collation name parts (e.g., database, schema, collation)

Return Type¶

This expression throws an UnresolvedException when dataType is accessed, as it is not meant to be evaluated directly. The return type is determined after resolution.

Supported Data Types¶

This expression does not directly support data types as it is an intermediate representation. After resolution, collations typically apply to string-based data types.

Algorithm¶

Stores the unresolved collation name as a sequence of string identifiers
Implements the Unevaluable trait, preventing direct evaluation
Maintains resolved = false to indicate it requires further analysis
Throws UnresolvedException when data type information is requested
Uses the UNRESOLVED_COLLATION tree pattern for identification during analysis

Partitioning Behavior¶

This expression does not affect partitioning behavior as it is resolved away during the analysis phase before physical planning:

Does not preserve or affect partitioning (resolved before physical planning)
No shuffle requirements (not present during execution)

Edge Cases¶

Throws UnresolvedException when dataType is accessed before resolution
Returns false for nullable property by default
Cannot be evaluated directly due to Unevaluable trait implementation
Must be resolved during analysis phase or query compilation will fail

Code Generation¶

This expression does not support code generation as it implements the Unevaluable trait and is designed to be resolved away during the analysis phase before code generation occurs.

Examples¶

-- Internal representation during parsing of:
SELECT col COLLATE utf8_general_ci FROM table
-- Creates UnresolvedCollation(Seq("utf8_general_ci"))

// This is an internal expression, not directly created in DataFrame API
// It would be created during SQL parsing of collation specifications
val unresolvedCollation = UnresolvedCollation(Seq("utf8", "general", "ci"))