[
https://issues.apache.org/jira/browse/LUCENE-2587?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17579967#comment-17579967
]
Elliot Z. Lin commented on LUCENE-2587:
---------------------------------------
Jumping in here to revive this decade-old issue. Roberto, I carried on your
work, applied your patch, and verified that the issue still persists. I wrote
an additional test to demonstrate the contiguity bug and implemented a fix.
This is currently in PR.
> Highlighter picks wrong offset for fragment boundaries
> ------------------------------------------------------
>
> Key: LUCENE-2587
> URL: https://issues.apache.org/jira/browse/LUCENE-2587
> Project: Lucene - Core
> 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
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> 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 was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
