[ https://issues.apache.org/jira/browse/LUCENE-2587?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13143938#comment-13143938 ]
Steven Rowe edited comment on LUCENE-2587 at 11/6/11 2:10 PM: -------------------------------------------------------------- Hi Terje, Thanks for posting your code and a test - I'll take a look. My initial interest is seeing how your code triggers the highlighting bug. I think the ideal way to proceed is create a test fragmenter based on but simpler than your fragmenter, to trigger the bug, and then fix the bug. Roberto may have already essentially done all of this - I haven't had a chance to look at his patch yet. Once that's all done, I think it's appropriate to create a new issue for your sentence fragmenter, since that would be a great addition to Lucene! Thanks again, Steve was (Author: steve_rowe): Hi Terje, Thanks for posting your code and a test - I'll take a look. My initial interest is seeing how your code triggers the highlighting bug. I think the idea way to proceed is create a test fragmenter based on but simpler than your fragmenter, to triggers the bug, and then fix the bug. Roberto may have already essentially done all of this - I haven't had a chance to look at his patch yet. Once that's all done, I think it's appropriate to create a new issue for your sentence fragmenter, since that would be a great addition to Lucene! Thanks again, Steve > Highlighter picks wrong offset for fragment boundaries > ------------------------------------------------------ > > Key: LUCENE-2587 > URL: https://issues.apache.org/jira/browse/LUCENE-2587 > Project: Lucene - Java > Issue Type: Bug > Components: modules/highlighter > Affects Versions: 3.0.2 > Environment: Java 6 + Lucene 3.0.2 > Reporter: Terje Eggestad > Priority: Trivial > Labels: newdev > Attachments: IMSentenceFragmenter.java, LUCENE-2587.patch, > TestIMSentenceFragmenter.java > > > I have written a new Fragmenter since we need fragments for hitlines to be > on sentence boundaries and not cross paragraphs. > When using it with org.apache.lucene.search.highlight.Highlighter, I get > hitlines that starts with ". ", "? ", "! "... > Consider the text "A b c d e. F g h i j! K l m n o. " > which become the tokenstream : (A) (b) (c) (d) (e) (F) (g) (h) (i) (j) (K) > (l) (m) (n) (o) > If the fragmenter return isNewFragment() = true on F and K and Highlighter > pick the middle fragment, lets say we search on "g" the hitline becomes: > ". F <B>g</B> h i j" > The reason, it seems, is that the offset to the fragment boundaries found by > taking the endOffset of the last token in a fragment , > not the startOffset of the first. > TJ -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org