On 01.02.2007, at 13:57, Ngoc Ngoc wrote:
> I want to learn more about ferret. So I downloaded ferret-0.10.14 and
> write a simple test script
>
> Only query = TermQuery.new(:content, 'program') gives result.
> If I change 'program' with 'Good' or 'Extra' -> no result
> and searching on (:title, 'Ruby') -> no result
>
>
> query = TermQuery.new(:content, 'Good')
>
> searcher.search_each(query) do |id, score|
> doc = searcher[id]
> puts "Document #{id} found with a score of #{score}"
> puts doc[:content]
> end
Try 'good' in lowercase and it'll work.
The reason is that your index converts each word in your document to
lowercase. This is due to the default analyzer used by the index
which happens to be Ferret::Analysis::StandardAnalyzer (see the rdocs
for details).
Because of the way you build your query, this lowercase conversion is
not applied to your query string, hence no match.
The trick is, that your queries need to go through the same analyzer
that is used for indexing in order to get the desired results.
Let me suggest a better way to write this script:
require 'rubygems'
require 'ferret'
include Ferret::Index
index = Index.new
index << {
:location => 'here',
:title => 'Programming Ruby',
:content => 'Good excellent program'
}
index << {
:location => 'local',
:title => 'Programming Rubyist',
:content => 'Extra ordinary program'
}
docs = index.search('Good')
docs.hits.each do |hit|
puts hit.inspect
puts "Document #{hit.doc} found with a score of #{hit.score}"
puts index[hit.doc][:content]
end
As you see, you don't need to employ a Searcher, nor do you have to
build a TermQuery explicitely. Just call #search on your Index and
everything works as expected (plus you get all the nice features of
Ferret's query parser).
Since we have used the Index#search method, the search automatically
uses the same analyzer for parsing your query that was used for
indexing the document.
Cheers,
Andy
_______________________________________________
Ferret-talk mailing list
[email protected]
http://rubyforge.org/mailman/listinfo/ferret-talk
