Stop and back up...... It's very unusual to use KeywordTokenizer with WDDF, it's far more common to use something like StandardTokenizer, WhitespaceTokenizer, etc.
Using keyword along with WDDF is kind of working, but probably not doing what you expect. Get familiar with the admin/analysis page to see. Suffix Search is supported by ReversedWildcardFilter. But really spend some time understanding your analysis chain before you go there. KeywordTokenizer is intended to be for input that you want to treat as a single unit rather than a series of tokens. FWIW, Erick On Thu, Jun 12, 2014 at 1:03 AM, Shay Sofer <sha...@checkpoint.com> wrote: > Thanks for your reply. > > How can I support suffix search? > > Name: Hello_world > Search: *world > > And I'll get hello_world as a result. > > Thanks in advance. > > > > -----Original Message----- > From: Shawn Heisey [mailto:s...@elyograg.org] > Sent: Wednesday, June 11, 2014 5:47 PM > To: solr-user@lucene.apache.org > Subject: Re: Solr search > >> Hi, >> >> Any suggestion for tokenizer / filter / other solutions that support >> search in Solr as following - >> >> Use Case >> >> Input >> >> Solr should return >> >> All Results >> >> * >> >> All results >> >> Prefix Search >> >> Text* >> >> All data started by Text* (Prefix search) >> >> Exact Search >> >> "Auto Text" >> >> Exact match. Only Auto Text >> >> Partial (substring) >> >> *Text* >> >> All strings contains the text >> >> >> Now I'm using KeywordTokenizerFactory and WordDelimiterFilterFactory. >> >> My issue is with exact search: >> When I have document named hello_world, and I'm trying to do Exact >> Search of hello, I got "hello_world" as a result (I want to get only >> hello named doccuments). > > The WordDelimeterFilter will split on the underscore, which means that the > term "hello" is in the index for that document. Leave that filter out if you > really do want an exact match. > > Searching for "*" by itself is not how you match all documents. It may work, > but it is a wildcard search, which means under the covers that it's a search > for every term in the index for that field. It's SLOW. The special shortcut > *:* (this must be the entire query with no field name, and I'm assuming the > standard query parser here) is what you want for all documents. In terms of > user input, this is what you want to use when the user leaves the search box > empty. If you're using dismax or edismax, then you would send an empty q > parameter or leave it off entirely, and define a default q.alt parameter in > solrconfig.xml, set to *:* for all docs. > > Thanks, > Shawn > > > > > > Email secured by Check Point
