Kristian Rickert created OPENNLP-1989:
-----------------------------------------
Summary: Regex removal (5b/10): Make event parsing deterministic
and preserve feature identities
Key: OPENNLP-1989
URL: https://issues.apache.org/jira/browse/OPENNLP-1989
Project: OpenNLP
Issue Type: Improvement
Reporter: Kristian Rickert
Correct parsing in {{RealValueFileEventStream}}, {{RealBasicEventStream}}, and
{{SimpleEventStreamBuilder}} using explicit, documented delimiters independent
of global tokenizer settings. Preserve Unicode characters within feature names,
eliminate accidental empty features, and prevent outcome-only events from
prematurely ending a stream. Report malformed records clearly and document the
shared parsing API. Test exact outcomes, feature arrays, values, and continued
reading after outcome-only events.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
