Richard Zowalla created OPENNLP-1949:
----------------------------------------
Summary: ADSentenceStream.read() loops forever on an empty
sentence at end of file
Key: OPENNLP-1949
URL: https://issues.apache.org/jira/browse/OPENNLP-1949
Project: OpenNLP
Issue Type: Bug
Reporter: Richard Zowalla
{{ADSentenceStream.read\(\)}} \(lines 481\-531\): at EOF with {{sentenceStarted
== true}} and an empty buffer, neither branch returns, so the {{while
\(true\)}} loop calls {{samples.read\(\)}} forever.
_Repro:_
{code:java}
new ADSentenceStream\(ObjectStreamUtils.createObjectStream\(""\)\).read\(\); //
never returns
{code}
Verified: still running after 3 s.
A truncated or damaged Arvores Deitadas file hangs training instead of failing.
_Fix:_ return {{null}} \(or throw {{InvalidFormatException}} naming the
unclosed sentence\) when EOF is reached with an empty started sentence, and add
a test.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
