Mailing-List: contact java-user-help@lucene.apache.org; run by ezmlm
Precedence: bulk
Reply-To: java-user@lucene.apache.org
Received-SPF: pass (athena.apache.org: domain of uwe@thetaphi.de designates
 188.138.97.18 as permitted sender)
From: "Uwe Schindler" <uwe@thetaphi.de>
To: <java-user@lucene.apache.org>
References: <DUB121-W48577A0B472F3912BAE09E97DF0@phx.gbl>
In-Reply-To: <DUB121-W48577A0B472F3912BAE09E97DF0@phx.gbl>
Subject: RE: Segment readers in Lucene 4.2
Date: Tue, 2 Apr 2013 15:58:47 +0200
Message-ID: <013b01ce2faa$32ae69c0$980b3d40$@thetaphi.de>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Thread-Index: AQImq9OEv36/XvQxC10Wi+ji35TmeJgR+pEA
Content-Language: de

Hi,

this is all not public tot he code because it is also subject to change!

With Lucene 4.x, you can assume:
directoryReader.leaves().get(i) corresponds to segmentsinfos.info(i)

WARNING: But this is only true if:
- the reader is instanceof DirectoryReader
- the segmentinfos were opened on the exact same commit point like the =
directory reader. If you have IndexWriter concurrently committing in =
between, this is no longer correct. To make this 100% correct, do =
DirectoryReader.listCommits(), chose one single "defined" commit and use =
it for DirectoryReader.open(CommitPoint) and the segmentInfos.read(Dir, =
commitPoint.getSegmentFileName()). Ideally open the DirectoryReader =
first, so it holds a lock, so the segments file may not be deleted =
before/while reading the segmentInfos.

Uwe

-----
Uwe Schindler
H.-H.-Meier-Allee 63, D-28213 Bremen
http://www.thetaphi.de
eMail: uwe@thetaphi.de


> -----Original Message-----
> From: andi rexha [mailto:a_rexha@hotmail.com]
> Sent: Tuesday, April 02, 2013 3:45 PM
> To: java-user@lucene.apache.org
> Subject: Segment readers in Lucene 4.2
>=20
> Hi,
> I have a question about the Index Readers in Lucene.
> As far as I understand from the documentation, with the Lucene 4, we =
can
> create an Index Reader from DirectoryReader.open(directory);
>=20
> From the code of the DirectoryReader, I have seen that it uses the
> SegmentReader to create the reader. To get all the SegmentReaders we
> have to loop through the AtomicReaderContexts that we can get from the
> "dirReader.leaves()";
>=20
> Am I correct until here? Did I miss something from the whole picture?
>=20
> My question is: I want to associate each segment to its
> AtomicReaderContext. I have an application that caches information =
from
> each segment, and I want to keep the reference to the current opened
> segments. I have found a way to do that but that is not elegant :
>=20
> I could associate the reader.toString() method to each of the segment
> information.
>=20
>  SegmentInfos sis =3D new SegmentInfos();
>  sis.read(open);
>  for (int i =3D 0; i < sis.size(); i ++) {
>             segmentInfo[i] =3D sis.info(i);  }
>=20
> List<AtomicReaderContext> leaves =3D dirReader.leaves();
>         System.out.println(leaves.size());
>         for (AtomicReaderContext context : leaves) {
>             checkIfExist(segmentInfo(i), context.reader().toString()); =
//here the
> solution is that I use the toString() method to extract the =
information of the
> segment. I cannot extract the information of the segment because it is
> protected }
>=20
>=20
> Do you have any clue whether it is possible to have this mapping in a
> different way?
>=20
> Thanks in advance,
> Best regards,
> Andi
>=20
>=20
>=20


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