Skip to content

Implement Spark-compatible xpath_number (alias of xpath_double) #287

Open

Description

@andygrove

Part of #271 and apache/datafusion-comet#7300.

What the function does

xpath_number(xml, path) is a registered alias of xpath_double: expression[XPathDouble]("xpath_number", true) in FunctionRegistry. It evaluates an XPath 1.0 expression, converts the result with XPath number(),
and returns DOUBLE. Both arguments are strings (collation-aware from 4.0), path must be foldable, and the
result is always nullable. Available since Spark 2.0.0. The only observable difference from xpath_double is
the display name: prettyName reads the FUNC_ALIAS tag, so the column header is xpath_number(...).

The numeric conversion rules (node-set, boolean and string to number, NaN, the empty node-set) are in the
xpath_double issue. Parsing, DTD and entity handling, namespaces, NULL handling and the error behavior are
in the xpath issue. This one only needs the alias.

Examples

SELECT xpath_number('<a><b>1</b><b>2</b></a>','sum(a/b)');
-- 3.0
SELECT xpath_number('<a><b>1.0</b><b>2.1</b></a>', 'sum(a/b)');
-- 3.1

The first comes from the Hive udf_xpath_double golden file (DESCRIBE FUNCTION EXTENDED xpath_number) and
the second from XPathFunctionsSuite.

Edge cases to test

  • Run the same cases as xpath_double and check that the results match exactly, including NaN for
    non-numeric text and 0.0 for a false boolean expression.
  • The output column name is xpath_number(xml, path), not xpath_double(...).
  • NULL or '' arguments give NULL. A non-literal path gives DATATYPE_MISMATCH.NON_FOLDABLE_INPUT.

Differences across Spark versions

None found. The alias has existed since 2.0.0.

Implementation notes

  • Spark: XPathDouble in sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/xml/xpath.scala,
    registered as xpath_number in sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/FunctionRegistry.scala.
  • DataFusion core has no equivalent.
  • Implement it as an alias (ScalarUDFImpl::aliases) of the xpath_double UDF in
    crates/spark/src/function/xml/, rather than as a separate kernel. This can be done in the same PR as
    xpath_double.
  • Tests: add xpath_number cases to crates/sqllogictest/test_files/xml/xpath_double.slt, or put them in a
    separate xpath_number.slt. Mention the alias in the #[user_doc].

This issue was generated by an LLM (Claude) from the Apache Spark source code and tests. Examples come from Spark's ExpressionDescription examples, test suites and .sql.out golden files, but please verify them against Spark before relying on them.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions