Add recursive descent XQuery parser (opt-in via feature flag)#6220
Draft
joewiz wants to merge 4 commits intoeXist-db:developfrom
Draft
Add recursive descent XQuery parser (opt-in via feature flag)#6220joewiz wants to merge 4 commits intoeXist-db:developfrom
joewiz wants to merge 4 commits intoeXist-db:developfrom
Conversation
4 tasks
joewiz
changed the title
joewiz
force-pushed
the
v2/new-parser
branch
7 times, most recently
from
e352bc1 to
b53f2ee
Compare
Add a recursive descent parser as an alternative to the ANTLR 2 generated parser. Enable with -Dexist.parser=rd (default remains antlr2). The parser supports XQuery 3.1, XQuery 4.0, XQUF 3.0, and XQFT 3.0 in 5,500 lines (6x smaller than the 914KB ANTLR 2 generated code) at 3-5x faster parse times with zero keyword overhead for XQUF/XQFT. Parser (exist-core/src/main/java/org/exist/xquery/parser/next/): - XQueryParser.java: recursive descent, builds Expression tree directly - XQueryLexer.java: tokenizer with character-level XML scanning - Token.java, Keywords.java, ParseError.java: supporting infrastructure - FTExpressions.java, XQUFExpressions.java: stub expression classes (to be replaced with real org.exist.xquery.ft/xquf imports) Feature flag (XQuery.java): - exist.parser system property: "antlr2" (default) or "rd" - compileWithNativeParser() bypasses ANTLR 2 pipeline entirely Tests (247 parser tests + 23 integration tests): - XQueryParserTest: 170 tests covering expressions, FLWOR, constructors, type expressions, XQ4 syntax, XQUF, XQFT - NativeParserIntegrationTest: 23 tests via eXist's XQuery.execute() path - XQueryLexerTest: 54 lexer unit tests - LexerBenchmark, ParserBenchmark: performance validation Validation: 93.1% pass rate on exist-core's full test suite (3,942 tests) with -Dexist.parser=rd. Remaining 7% is long-tail edge cases and infrastructure test failures unrelated to parsing. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
XQ4 syntax is now only available when the query declares
xquery version "4.0". Default (no declaration) is 3.1 behavior.
Gated features:
- Pipeline operator (->)
- Mapping arrow (=!>)
- Otherwise expression
- Braced if (no else clause)
- Keyword arguments (name := value)
- Focus functions (fn { })
- String templates (``[...]``)
- QName literals (#name)
- Default parameter values
- for member clause
- while clause
- try/finally
Not gated (available in 3.1):
- Arrow operator (=>)
- Simple map (!)
- Arrays, maps, lookups
- Inline functions, function references
- try/catch (without finally)
- XQUF, XQFT, eXist legacy update
Error messages are helpful:
Pipeline operator '->' requires xquery version "4.0".
Add 'xquery version "4.0";' to enable XQuery 4.0 features.
Responds to community call feedback (line-o, 2026-03-23).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Consolidates several test improvements and fixes accumulated during rd parser validation against ANTLR 2 and the XQTS: - Add regression tests confirming array:get#3 and parse-xml behave identically in both parsers (the 32 apparent XQTS regressions were build-version differences, not parser bugs) - Update XQueryParserTest assertions to use real XQUF/FT class names (XQUFTransformExpr, FTContainsExpr) instead of stub names; fixes 14 tests - Add 18 FunctX-pattern integration tests exercising real-world XQuery patterns through both parsers; all 18 pass, confirming rd handles higher-order functions, FLWOR, typeswitch, namespaces, etc. - Add grammar-dispatch-audit.py: cross-references 367 EBNF productions against 113 rd parse methods, flags Expr/ExprSingle mismatches; result is ALL CLEAR after the FLWOR return fix - Fix XQUF keyword conflicts in test modules: add missing closing brace in flwor.xql and rename $copy → $expanded-set in test.xq to avoid conflict with the XQUF `copy` keyword Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The rebase onto develop picked up the three orderby-empty-ordering-spec tests added by PR eXist-db#6073. The v2/new-parser test commit had independently added the same three tests, causing XQST0034 duplicate function errors. Remove the duplicates introduced by this branch. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
duncdrum
added
enhancement
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds a recursive descent (rd) parser as an alternative to the ANTLR 2 generated parser. Opt-in via
-Dexist.parser=rd(default remains ANTLR 2). Zero impact on existing behavior.Note: This branch depends on v2/w3c-xquery-update-3.0, v2/xqft-phase2, and v2/xquery-4.0-parser. Must merge after those PRs.
What Changed
org.exist.xquery.parser.next(~5,700 lines)System.getProperty("exist.parser", "antlr2")— set to"rd"to activateBenchmark Results
Tests
Test plan
-Dexist.parser=rdactivates the rd parser🤖 Generated with Claude Code