[
https://issues.apache.org/jira/browse/OPENNLP-1990?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18120943#comment-18120943
]
ASF GitHub Bot commented on OPENNLP-1990:
-----------------------------------------
krickert commented on PR #1322:
URL: https://github.com/apache/opennlp/pull/1322#issuecomment-5902817292
Anchors now go through the same ASCII digit parser as the identifiers, and
the signed and other-script forms moved to the rejected cases. The helpers live
in `MascParserTestUtil`, the whitespace characters are constants, the node id
and `ref` checks are separate tests, and the null and empty cases use
`NullAndEmptySource`. The MASC section now says what the corpus is and how to
use the reader, with one command. The old masc-structure link was dead, so it
points at the corpus page. Description cut down to what changes for users.
> Regex removal (5c/10): Validate MASC identifiers and anchors consistently
> -------------------------------------------------------------------------
>
> Key: OPENNLP-1990
> URL: https://issues.apache.org/jira/browse/OPENNLP-1990
> Project: OpenNLP
> Issue Type: Improvement
> Reporter: Kristian Rickert
> Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Unify MASC identifier, target-list, and anchor parsing across the relevant
> readers. Validate expected prefixes, numeric ranges, and exactly two ordered,
> nonnegative anchors for supported text spans, preserving useful SAX error
> causes. Apply whitespace rules supported by the MASC/GrAF format and document
> their relationship to XML attribute normalization:
> https://www.w3.org/TR/xml/#AVNormalize. Preserve source text and offsets, add
> malformed-input and Unicode regressions, and validate the changes against a
> real MASC corpus.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
