lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Robert Muir <rcm...@gmail.com>
Subject Re: Problem with TermVector offsets and positions not being preserved
Date Fri, 20 Jul 2012 13:10:37 GMT
Hi Mike:

I wrote up some tests last night against 3.6 trying to find some way
to reproduce what you are seeing, e.g. adding additional segments with
the field specified without term vectors, without tv offsets, omitting
TF, and merging them and checking everything out. I couldnt find any
problems.

Can you provide more information?

On Thu, Jul 19, 2012 at 7:16 PM, Mike O'Leary <tmoleary@uw.edu> wrote:
> I created an index using Lucene 3.6.0 in which I specified that a certain text field
in each document should be indexed, stored, analyzed with no norms, with term vectors, offsets
and positions. Later I looked at that index in Luke, and it said that term vectors were created
for this field, but offsets and positions were not. The code I used for indexing couldn't
be simpler. It looks like this for the relevant field:
>
> doc.add(new Field("ReportText", reportTextContents, Field.Store.YES, Field.Index.ANALYZED_NO_NORMS,
Field.TermVector.WITH_POSITIONS_OFFSETS);
>
> The indexer adds these documents to the index and commits them. I ran the indexer in
a debugger and watched the Lucene code set the Field instance variables called storeTermVector,
storeOffsetWithTermVector and storePositionWithTermVector to true for this field.
>
> When the indexing was done, I ran a simple program in a debugger that opens an index,
reads each document and writes out its information as XML. The values of storeOffsetWithTermVector
and storePositionWithTermVector in the ReportText Field objects were false. Is there something
other than specifying Field.TermVector.WITH_POSITIONS_OFFSETS when constructing a Field that
needs to be done in order for offsets and positions to be saved in the index? Or are there
circumstances under which the Field.TermVector setting for a Field object is ignored? This
doesn't make sense to me, and I could swear that offsets and positions were being saved in
some older indexes I created that I unfortunately no longer have around for comparison. I'm
sure that I am just overlooking something or have made some kind of mistake, but I can't see
what it is at the moment. Thanks for any help or advice you can give me.
> Mike



-- 
lucidimagination.com

---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org


Mime
View raw message