Bug #3211: ArrayIndexOutOfBoundsException in ARegexIterator.computeNestingTable with complex regex - Saxon - Saxonica Developer Community

Actions

Send by e-mail Copy link

Bug #3211

closed

ArrayIndexOutOfBoundsException in ARegexIterator.computeNestingTable with complex regex

Added by Henry S Thompson about 7 years ago. Updated over 4 years ago.

Status:

Closed

Priority:

Normal

Assignee:

Michael Kay

Category:

XSLT conformance

Sprint/Milestone:

Start date:

2017-04-26

Due date:

% Done:

100%

Estimated time:

Legacy ID:

Applies to branch:

9.7

Fix Committed on Branch:

9.7, trunk

Fixed in Maintenance Release:

9.7.0.19

Platforms:

Description

Some combination of nesting and non-capturing groups is the cause. Test case attached -- just run it with empty input.

The 7th branch is the problem, simplifying it in a variety of ways makes the problem go away.

For testing purposes, you can pass in simplified patterns and/or formulae from the command line, so e.g. this, with one fewer non-capturing group, works:

saxon97 test.xsl pat='("[^"]*")|(\{[^}]+})|(,)|([^=\-+*/();:,.$&lt;>^!]+(?:\.[^=\-+*/();:,.$&lt;>^!]+)*\()|([)])|(^=|\()|((?:(?:'\''[^'\'']+'\'')))|(\$?[A-Z]+\$?[0-9]+)|([a-zA-Z_\\][a-zA-Z0-9._]*)|(.)'

Files

test.xsl (730 Bytes) test.xsl

Test case

Henry S Thompson, 2017-04-26 11:43

Actions

Copy link

Updated by Michael Kay about 7 years ago

Category set to XSLT conformance
Status changed from New to In Progress
Assignee set to Michael Kay
Applies to branch 9.8 added

Thanks for reporting it.

It looks as if this code isn't handling non-capturing groups correctly: the start of such a group (that is (?) is properly detected, but the corresponding end of the group isn't.

For background, the nesting table is computed only in the case where there is a zero-length captured group. A comment in the code explains:

The problem here is that the information available from Java isn't sufficient to determine the nesting of groups: match("a", "(a(b?))") and match("a", "(a)(b?)") will both give the same result for group 2 (start=1, end=1). So we need to go back to the original regex to determine the group nesting.

Actions

Copy link

Updated by Michael Kay almost 7 years ago

Status changed from In Progress to Resolved
Fix Committed on Branch 9.7, 9.8 added

Patch committed. Test case analyze-string-097 added to W3C test suite.

Actions

Copy link

Updated by Henry S Thompson almost 7 years ago

Thanks for prompt action!

Actions

Copy link

Updated by Michael Kay almost 7 years ago

Slight revision the patch: (a) there was a regression test analyze-string-017 which failed; (b) the same incorrect code was present for .NET in DotNetRegexIterator and the two classes have been changed to use a common method.

Actions

Copy link