highlights=x

[tophe]

hello, I have an index build with { var ret=new Document(); ret.add(doc.id, {'field':'id','type':'int'}); ret.add(doc.txt,{'store':'yes','termvector':'with_positions_offsets'}); return ret } } that index is working, but when I get the results : {"limit":25,"etag":"46e09727a512","fetch_duration":2,"q":"default:jahrhunderts","search_duration":0,"total_rows":380,"skip":0,"rows":[{"id":"93898_c0059","score":2.3190693855285645,"highlights":{"default":[]},"fields":{"default":"GEMÄLDE DES 19. JAHRHUNDERTS"}},{"id":"93898_c0092","score":2.3190693855285645,"highlights":{"default":[]},"fields":{"default":"GEMÄLDE DES 19. JAHRHUNDERTS"}},

the highlights is alway empty ! I use couchdb 1.5.0 and the git verstion of couch-lucene, and tomcat 7.0.40, and java 1.7.0 on debian whezzy.

does I missed something ?

[rnewson]

termvectors, which power highlights, currently broken on master while I did the big Lucene 4.7 upgrade. Use the latest version or go back on master to before that change. I'll have it fixed before the 1.0 release.

[rnewson]

what query parameters are you using?

[tophe]

Hi robert,

the query http://couchdb1:5984/_fti/local/catalogues/_design/cataloguepage/by_txt?q=Abbaye&highlights=10

response extract {"limit":25,"etag":"ccd463570545","fetch_duration":145,"q":"default:abbaye","search_duration":178,"total_rows":220,"skip":0,"rows":[{"id":"92093_c0066","score":1.7958104610443115,"highlights":{"default":[]},"fields":{"default":"ABBAYE DE MONDAYE\r\nPRÉMONTRÉS\r\nJUAYE-MONDAYE\r\nCALVADOS\r\nPOUR LA RESTAURATION DE LA FERME CONVENTUELLE\r\nET LA RESTITUTION DU PARVIS DE L'ABBATIALE\r\n"}}, I found another probleme with

http://couchdb1:5984/_fti/local/catalogues/_design/cataloguepage/by_txt?q=Abbaye and idcatalogue:93205&limit=200&include_docs=true

{"limit":200,"etag":"ccea5d3956ec","fetch_duration":523,"q":"default:abbaye idcatalogue:`\b\u0000\u0005X\u0015","search_duration":4,"total_rows":304,"skip":0,"rows":[{"id":"93205_c0053","score":8.486856460571289,"fields":... {"id":"81695_c0015","doc":{"idauctioneer":5394,"_rev":"1-7738f0956555c0261efe9a0f42e559f5","ref":3831,"title":"TABLEAUX ABSTRAITS ET CONTEMPORAINS - SCULPTURES","_id":"81695_c0015","sales":"[170297]","year":2011,"offset":2,"type":"cataloguepage","idcatalogue":81695,"txt": the result do not contain only idcatalogue=93205, as you can see in the last response. (you get the index definition in my first mail) I am currently on the git/master depot, should I use another branch ?

another problem, we use the war file with tomcat 7.40.0, and we have to remove manualy the line 

in the WEB-INF/web.xml to run it with tomcat.

Le Mercredi 9 avril 2014 23h05, Robert Newson [email protected] a écrit :

what query parameters are you using? — Reply to this email directly or view it on GitHub.

[rnewson]

you need to say "AND" not "and" to do a Boolean and operation (https://lucene.apache.org/core/4_6_0/queryparser/org/apache/lucene/queryparser/classic/package-summary.html)

[tophe]

so powerful, with the fuzzy operator support, (that was my next question). sorry I doesn't see the link to lucene query parser in the doc.

I have another probleme, I have put the couchdb-lucene on tomcat mounted as ROOT, because it don't work on a directory as (localhost:8080/couchdb-lucene). it work fine like that.

Le Jeudi 10 avril 2014 9h16, Robert Newson [email protected] a écrit :

you need to say "AND" not "and" to do a Boolean and operation (https://lucene.apache.org/core/4_6_0/queryparser/org/apache/lucene/queryparser/classic/package-summary.html) — Reply to this email directly or view it on GitHub.

[rnewson]

The deploy-inside-tomcat piece is a contribution and I have to confess that I never run couchdb-lucene that way, I use the bin/run script that I wrote. There's a ticket about making the war relocatable that I've never had time to validate or keep current, so I would consider deploying couchdb-lucene in a war as an expert option, I can't really help with that.

I will dig into the termvector/highlight issue at the weekend though. I misspoke earlier though, the include_termvectors=true option is temporarily removed but neither the index-time termvector settings or the highlighting options were removed. Perhaps they're failing here for a different reason. Could you try putting the highlighting field somewhere other than 'default'?