[xwiki-users] Lucene Search does not correctly filter by space
Hello all, we've recently launched a new wiki and have been forced to switch to Database Search because of an odd problem with Lucene. We have multiple spaces, and when using Lucene, selecting a space from the drop-down list does NOT return the expected (filtered) results. We still get results from multiple spaces. Lucene seems to be prioritizing 'weight' over our selection of a specific space. We do not have this same problem with Database Search, which filters correctly every time. Of course, we'd prefer to use Lucene because of the strong query syntax and weighting. I have searched forums for days and cannot seem to find someone who has had the same issue. I feel like it's something simple ... will someone PLEASE help? Thank you much! -- View this message in context: http://xwiki.475771.n2.nabble.com/Lucene-Search-does-not-correctly-filter-by... Sent from the XWiki- Users mailing list archive at Nabble.com.
On 09/06/2012 05:48 PM, iceicelady wrote:
Hello all, we've recently launched a new wiki and have been forced to switch to Database Search because of an odd problem with Lucene. We have multiple spaces, and when using Lucene, selecting a space from the drop-down list does NOT return the expected (filtered) results. We still get results from multiple spaces. Lucene seems to be prioritizing 'weight' over our selection of a specific space.
We do not have this same problem with Database Search, which filters correctly every time. Of course, we'd prefer to use Lucene because of the strong query syntax and weighting.
I have searched forums for days and cannot seem to find someone who has had the same issue. I feel like it's something simple ... will someone PLEASE help?
Thank you much!
Well, there are two "bugs" involved here. One is that when there are spaces (as in white space) in the space name, the part of the query that is supposed to filter only results from that space is broken: instead of searching for documents in the space "a space with whitespace", it searches for documents in the space "a" and also containing "space with whitespace" as words in the text. I just fixed this issue for the upcoming 4.3 release: http://jira.xwiki.org/browse/XWIKI-8191 The second one is that the space name is not stored as it is, but it's tokenized and analyzed so that it supports non-exact matches. But this means that if we have two spaces, "Main" and "The Main Space", trying to get results from the "Main" space will also include results from the "The Main Space" space. Fixing this would be a bit harder, since it means indexing not just the "space" field, but also an "exactspace" field that would allow exact matches on the space name. I've created https://jira.xwiki.org/browse/XWIKI-8192 to track this issue. -- Sergiu Dumitriu http://purl.org/net/sergiu/
Sergiu Dumitriu-2 wrote
Well, there are two "bugs" involved here.
One is that when there are spaces (as in white space) in the space name, the part of the query that is supposed to filter only results from that space is broken: instead of searching for documents in the space "a space with whitespace", it searches for documents in the space "a" and also containing "space with whitespace" as words in the text. I just fixed this issue for the upcoming 4.3 release: http://jira.xwiki.org/browse/XWIKI-8191
The second one is that the space name is not stored as it is, but it's tokenized and analyzed so that it supports non-exact matches. But this means that if we have two spaces, "Main" and "The Main Space", trying to get results from the "Main" space will also include results from the "The Main Space" space. Fixing this would be a bit harder, since it means indexing not just the "space" field, but also an "exactspace" field that would allow exact matches on the space name. I've created https://jira.xwiki.org/browse/XWIKI-8192 to track this issue.
Thank you very much for your response, Sergiu. That really helps to explain things. Our two main spaces have "white spaces" in their names AND have some similar letters in their names. Less than a month ago, wupgraded from XWIKI 2.4 to 4.2 milestone 1. I read the JIRAs you mentioned, and it's good to know that if we upgrade to 4.3 (or higher) sometime in the future, those issues have been captured and are being worked. -- View this message in context: http://xwiki.475771.n2.nabble.com/Lucene-Search-does-not-correctly-filter-by... Sent from the XWiki- Users mailing list archive at Nabble.com.
Sergiu Dumitriu-2 wrote
Well, there are two "bugs" involved here.
One is that when there are spaces (as in white space) in the space name, the part of the query that is supposed to filter only results from that space is broken: instead of searching for documents in the space "a space with whitespace", it searches for documents in the space "a" and also containing "space with whitespace" as words in the text. I just fixed this issue for the upcoming 4.3 release: http://jira.xwiki.org/browse/XWIKI-8191
The second one is that the space name is not stored as it is, but it's tokenized and analyzed so that it supports non-exact matches. But this means that if we have two spaces, "Main" and "The Main Space", trying to get results from the "Main" space will also include results from the "The Main Space" space. Fixing this would be a bit harder, since it means indexing not just the "space" field, but also an "exactspace" field that would allow exact matches on the space name. I've created https://jira.xwiki.org/browse/XWIKI-8192 to track this issue. -- Sergiu Dumitriu
Hi Sergiu, Do you have any time (or XWIKI version) estimate for XWIKI-8192 on the roadmap? We started testing XWIKI 4.3 on a development server, but did not deploy it to our users. For our next deployment at our company, we would ideally like to wait for the Lucene Search fix. We understand that most developers are donating their time to XWIKI, but we just wondered if there was any proposed time frame for this particular issue. Thanks, Lisa -- View this message in context: http://xwiki.475771.n2.nabble.com/Lucene-Search-does-not-correctly-filter-by... Sent from the XWiki- Users mailing list archive at Nabble.com.
lisalisa wrote
Hi Sergiu,
Do you have any time (or XWIKI version) estimate for XWIKI-8192 on the roadmap? We started testing XWIKI 4.3 on a development server, but did not deploy it to our users. For our next deployment at our company, we would ideally like to wait for the Lucene Search fix. We understand that most developers are donating their time to XWIKI, but we just wondered if there was any proposed time frame for this particular issue.
Thanks, Lisa
Oh, I was watching XWIKI-8192 in JIRA and see that it will be resolved for XWIKI 4.4.1 ... awesome!! Lisa -- View this message in context: http://xwiki.475771.n2.nabble.com/Lucene-Search-does-not-correctly-filter-by... Sent from the XWiki- Users mailing list archive at Nabble.com.
On 12/08/2012 06:58 AM, lisalisa wrote:
lisalisa wrote
Hi Sergiu,
Do you have any time (or XWIKI version) estimate for XWIKI-8192 on the roadmap? We started testing XWIKI 4.3 on a development server, but did not deploy it to our users. For our next deployment at our company, we would ideally like to wait for the Lucene Search fix. We understand that most developers are donating their time to XWIKI, but we just wondered if there was any proposed time frame for this particular issue.
Thanks, Lisa
Oh, I was watching XWIKI-8192 in JIRA and see that it will be resolved for XWIKI 4.4.1 ... awesome!!
Lisa
Sorry for the delay, I started working on it a while ago but ran into some problems. -- Sergiu Dumitriu http://purl.org/net/sergiu/
Hi, I have checked that the lucene index folder has proper write permission and I have emptied that folder before restart the server, but still the lucene indexing stucked once again. I am using xwiki 4.3 stable and enable the lucene search from admin section search configuration and configured the lucene index directory from xwiki.cfg file as xwiki.plugins.lucene.indexdir=/opt/apache-tomcat-6.0.26/luceneindex/xwiki/lucene Any help would be greatly appreciated. Thanks & Regards, Sanjib Pal -- View this message in context: http://xwiki.475771.n2.nabble.com/Lucene-Search-does-not-correctly-filter-by... Sent from the XWiki- Users mailing list archive at Nabble.com.
Deleting the index folder (rather than emptying it) before restart solved the issue for us. Thanks, Lisa sanjib wrote
Hi,
I have checked that the lucene index folder has proper write permission and I have emptied that folder before restart the server, but still the lucene indexing stucked once again. I am using xwiki 4.3 stable and enable the lucene search from admin section search configuration and configured the lucene index directory from xwiki.cfg file as
xwiki.plugins.lucene.indexdir=/opt/apache-tomcat-6.0.26/luceneindex/xwiki/lucene
Any help would be greatly appreciated.
Thanks, Sanjib Pal
-- View this message in context: http://xwiki.475771.n2.nabble.com/Lucene-Search-does-not-correctly-filter-by... Sent from the XWiki- Users mailing list archive at Nabble.com.
participants (5)
-
iceicelady -
lisalisa -
sanjib -
Sergiu Dumitriu -
Sergiu Dumitriu