Skip to content

Implement Spark-compatible xpath_float #284

Open

Description

@andygrove

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

What the function does

xpath_float(xml, path) evaluates an XPath 1.0 expression, converts the result with XPath number() into a
Java Double, and then narrows it with Double.floatValue(), giving a FLOAT. Both arguments are strings
(collation-aware from 4.0), and path must be foldable. The result is always nullable. Available since Spark
2.0.0.

The number() conversion is in the xpath_double issue, and xpath_float is that value cast to f32.
Parsing, DTD and entity handling, namespaces, NULL handling and errors are in the xpath issue. This issue
covers only the double-to-float step.

Examples

SELECT xpath_float('<a><b>1</b><b>2</b></a>', 'sum(a/b)');
-- 3.0
SELECT xpath_float('<a>this is not a number</a>', 'a');
-- NaN
SELECT xpath_float('<a>this 2 is not a number</a>', 'a');
-- NaN
SELECT xpath_float('<a><b>2000000000</b><c>40000000000</c></a>', 'a/b * a/c');
-- 8.0E19
SELECT xpath_float('<a>try a boolean</a>', 'a = 10');
-- 0.0
SELECT xpath_float('<a><b class="odd">1</b><b class="even">2</b><b class="odd">4</b><c>8</c></a>', 'a/b');
-- 1.0
SELECT xpath_float('<a><b class="odd">1</b><b class="even">2</b><b class="odd">4</b><c>8</c></a>', 'sum(a/*)');
-- 15.0
SELECT xpath_float('<a><b class="odd">1</b><b class="even">2</b><b class="odd">4</b><c>8</c></a>', 'sum(a/b[@class="odd"])');
-- 5.0
SELECT xpath_float('<a><b>1.0</b><b>2.1</b></a>', 'sum(a/b)');
-- 3.1

These come from the ExpressionDescription, the Hive udf_xpath_float golden files run by
HiveCompatibilitySuite, XPathExpressionSuite and XPathFunctionsSuite. The last result is
3.1.toFloat, which is the rounded f32 of the double sum 3.1.

Edge cases to test

  • The narrowing is Java's (float) d, which uses IEEE round-to-nearest-even, so in Rust it is d as f32.
    A finite double above f32::MAX becomes Infinity. XPath has no exponent literal, so build the large value
    from element text, for example <a>1000000000000000000000000000000000000000</a> with path 'a'. A double
    below the smallest f32 subnormal becomes 0.0 or -0.0. 1 div 0 stays Infinity.
  • Check that rounding happens once, from the double result. sum() adds in double and narrows at the end,
    so it is not equivalent to adding f32 values.
  • NaN stays NaN (non-numeric text). A false boolean gives 0.0.
  • The empty node-set is NaN by XPath 1.0 rules, although the usage string says zero. See the note in the
    xpath_double issue and verify against Spark.
  • NULL or '' arguments give NULL. A non-literal path gives DATATYPE_MISMATCH.NON_FOLDABLE_INPUT.
    Malformed XML and an invalid path raise runtime errors (see the xpath issue).

Differences across Spark versions

None found beyond the shared ones in the xpath issue.

Implementation notes

  • Spark: XPathFloat in sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/xml/xpath.scala.
    It calls XPathFloatEvaluator (ret.floatValue()) in .../xml/XmlExpressionEvalUtils.scala.
  • DataFusion core has no equivalent.
  • Reuse the shared number() evaluator (see the xpath and xpath_double issues) in
    crates/spark/src/function/xml/, and cast the result to f32.
  • Tests: add crates/sqllogictest/test_files/xml/xpath_float.slt and document the function with #[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