andygrove opened a new issue, #289:
URL: https://github.com/apache/datafusion-spark/issues/289
Part of #271 and apache/datafusion-comet#7300.
### What the function does
`xpath_string(xml, path)` evaluates an XPath 1.0 expression and converts the
result with the XPath `string()`
rules (`XPathConstants.STRING`), giving a `STRING`. For a node-set this is
the string value of the first node in
document order, and for an element that is the concatenation of all its
descendant text. Both arguments are
strings (collation-aware from 4.0, and the result has the session default
string type). `path` must be
foldable. The result is always nullable. Available since Spark 2.0.0.
Parsing, DTD and entity handling, namespaces, NULL handling, invalid-XML and
invalid-path errors, and lazy path
compilation are shared with the family. They are specified in the `xpath`
issue. This issue covers only the
string conversion.
### Examples
```sql
SELECT xpath_string('<a><b>b</b><c>cc</c></a>','a/c');
-- cc
SELECT xpath_string('<a><b>bb</b><c>cc</c></a>', 'a');
-- bbcc
SELECT xpath_string('<a><b>bb</b><c>cc</c></a>', 'a/b');
-- bb
SELECT xpath_string('<a><b>bb</b><c>cc</c></a>', 'a/d');
-- (empty string)
SELECT xpath_string('<a><b>b1</b><b>b2</b></a>', '//b');
-- b1
SELECT xpath_string('<a><b>b1</b><b>b2</b></a>', 'a/b[1]');
-- b1
SELECT xpath_string('<a><b>b1</b><b>b2</b></a>', 'a/b[2]');
-- b2
SELECT xpath_string('<a><b>b1</b><b id="b_2">b2</b></a>', 'a/b[@id="b_2"]');
-- b2
```
These come from the `ExpressionDescription`, the Hive `udf_xpath_string`
golden files run by
`HiveCompatibilitySuite`, and `XPathExpressionSuite`. `UDFXPathUtilSuite`
also has these cases:
- `a/b[3]/text()` gives `b3`;
- `a/b[4]/text()` (no match) gives `''`;
- `a/b[2]/@k` gives `foo`;
- an external entity (`<!ENTITY embed SYSTEM "...">`) expands to `''`.
### Edge cases to test
- No match gives `''`, not NULL (`a/d` above). NULL comes only from a NULL
or `''` argument.
- Conversion rules from XPath 1.0 `string()`:
- an element gives its concatenated descendant text, including
whitespace-only text nodes and CDATA;
- comment and processing-instruction text inside the element is not
included;
- an attribute gives its value;
- a boolean gives `true` or `false`.
- Numbers are formatted the XPath 1.0 way, not with Java's
`Double.toString`. Integers have no decimal point
(`sum(a/b)` over `1` and `2` gives `3`), and the special values are `NaN`,
`Infinity` and `-Infinity`.
Derive the outputs from the spec and confirm them against Spark, because
the formatting of large and small
magnitudes differs between engines.
- XPath string functions in the path: `concat`, `substring` (1-based, with
rounding rules), `normalize-space`,
`translate`, `substring-before`/`after`, and `string-length` (which XPath
counts in characters; check
supplementary code points).
- Entity and character references (`&`, `😀`) are decoded in the
output. So is an internal DTD
entity.
- 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. From 4.0 the result
is the session default string
type rather than plain `StringType`.
### Implementation notes
- Spark: `XPathString` in
`sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/xml/xpath.scala`.
It calls `XPathStringEvaluator`, which calls `UDFXPathUtil.evalString`.
- DataFusion core has no equivalent.
- Build it on the shared evaluator from the `xpath` issue in
`crates/spark/src/function/xml/`. The existing
`crates/sqllogictest/test_files/xml/xpath_string.slt` is a commented-out
stub that should be enabled and
extended. 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._
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]