[LUCENE-2587] Highlighter picks wrong offset for fragment boundaries

• Type: Bug
• Status: Open
• Priority: Trivial
• Resolution: Unresolved
• Affects Version/s: 3.0.2
• Component/s: modules/highlighter
• Labels: newdev

Environment

Java 6 + Lucene 3.0.2

• Lucene Fields: New
• Assignee: Unassigned
• Reporter: Terje Eggestad
• Created: 04/Aug/10 11:46
• Updated: 28/Aug/22 12:30

Description

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) (j) (K) (l) (m) (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

Issue Links

• links to: GitHub Pull Request #1069