xapian-fu 0.2 → 1.0.1

data/spec/xapian_doc_spec.rb

@@ -13,4 +13,184 @@ describe XapianDoc do
     XapianDoc.new(:id => 666).should_not == XapianDoc.new(:id => 667)
   end
 
+  describe "to_xapian_document" do
+    it "should tokenize strings" do
+      xdb = XapianDb.new
+      xdoc = xdb.documents.new("once upon a time").to_xapian_document
+      xdoc.terms.should be_a_kind_of Array
+      xdoc.terms.last.should be_a_kind_of Xapian::Term
+      xdoc.terms.last.term.should == "upon"
+    end
+
+    it "should store positions by default when tokenizing" do
+      xdb = XapianDb.new
+      doc = xdb.documents.new("once upon a time")
+      doc.save
+      xdb.ro.positionlist(doc.id, "time").first.should_not == nil
+    end
+
+    it "should not store positions when tokenizing when :index_positions is set to false" do
+      xdb = XapianDb.new(:index_positions => false)
+      doc = xdb.documents.new("once upon a time")
+      doc.save
+      xdb.ro.positionlist(doc.id, "once").first.should == nil
+    end
+
+    it "should tokenize a hash" do
+      xdb = XapianDb.new
+      xdoc = xdb.documents.new(:title => 'once upon a time').to_xapian_document
+      xdoc.terms.should be_a_kind_of Array
+      xdoc.terms.last.should be_a_kind_of Xapian::Term
+      xdoc.terms.last.term.should == "upon"
+    end
+    
+    it "should tokenize the fields of a hash separately" do
+      xdb = XapianDb.new
+      xdoc = xdb.documents.new({ :text => "once upon a time", :title => "A story" }).to_xapian_document
+      terms = xdoc.terms.collect { |t| t.term }
+      terms.should include "XTEXTonce"
+      terms.should include "XTITLEstory"
+      terms.should_not include "XTEXTstory"
+    end
+
+    it "should not tokenize fields declared as not to be indexed" do
+      xdb = XapianDb.new(:fields => { :name => { :index => false } })
+      xdoc = xdb.documents.new({ :name => 'John Leach', :quote => 'Xapian Rocks' }).to_xapian_document
+      terms = xdoc.terms.collect { |t| t.term }
+      terms.should_not include 'XNAMEjohn'
+      terms.should_not include 'XNAMEleach'
+      terms.should_not include 'Zjohn'
+      terms.should_not include 'Zleach'
+      terms.should_not include 'john'
+      terms.should_not include 'leach'
+    end
+
+    it "should convert Time instances to a useful format when tokenizing" do
+      time = Time.now
+      xdb = XapianDb.new
+      xdoc = xdb.documents.new(:created_at => time).to_xapian_document
+      terms = xdoc.terms.collect { |t| t.term }
+      terms.should include time.utc.strftime("%Y%m%d%H%M%S")
+    end
+    
+    it "should convert DateTime instances to a useful format when tokenizing" do
+      datetime = DateTime.now
+      xdb = XapianDb.new
+      xdoc = xdb.documents.new(:created_at => datetime).to_xapian_document
+      terms = xdoc.terms.collect { |t| t.term }
+      terms.should include datetime.strftime("%Y%m%d%H%M%S")
+    end
+
+    it "should convert Time instances to a useful format when tokenizing" do
+      date = Date.today
+      xdb = XapianDb.new
+      xdoc = xdb.documents.new(:created_on => date).to_xapian_document
+      terms = xdoc.terms.collect { |t| t.term }
+      terms.should include date.strftime("%Y%m%d")
+    end
+
+    it "should stem English words by default" do
+      xdb = XapianDb.new
+      xdoc = xdb.documents.new("She fished for fish").to_xapian_document
+      terms = xdoc.terms.collect { |t| t.term }
+      terms.should_not include "Zfished"
+      terms.should include "Zfish"
+    end
+
+    it "should inherit the databases stemmer by default" do
+      xdb = XapianDb.new(:stemmer => :french)
+      xdoc = xdb.documents.new("majestueusement").to_xapian_document
+      terms = xdoc.terms.collect { |t| t.term }
+      terms.should include 'Zmajestu'
+    end
+
+    stems = {
+      :german => { "aufeinander" => "aufeinand" },
+      :french => { "majestueusement" => "majestu" },
+      :swedish => { "kloekornas" => "kloek" },
+      :danish => { "indvendingerne" => "indvending" },
+      :russian => { "падшую", "падш" },
+      :italian => { "propagamento" => "propag" }
+    }
+    stems.keys.each do |lang|
+      stems[lang].each do |word, stem|
+        it "should stem #{lang.to_s.capitalize} words when the :stemmer option is set to :#{lang}" do
+          xdb = XapianDb.new
+          xdoc = xdb.documents.new(word, :stemmer => lang).to_xapian_document
+          terms = xdoc.terms.collect { |t| t.term }
+          terms.should include 'Z'+stem
+        end
+      end
+    end
+
+    it "should not stem words when stemmer is set to false" do
+      xdb = XapianDb.new
+      xdoc = xdb.documents.new("She fished for fish", :stemmer => false).to_xapian_document
+      terms = xdoc.terms.collect { |t| t.term if t.term =~ /^Z/ }.compact
+      terms.should be_empty
+    end
+    
+    it "should not stem english stop words by default" do
+      xdb = XapianDb.new
+      xdoc = xdb.documents.new("And they made a cake", :stemmer => :english).to_xapian_document
+      terms = xdoc.terms.collect { |t| t.term }
+      terms.should_not include 'Zand'
+      terms.should_not include 'Za'
+      terms.should include 'Zcake'
+    end
+    
+    it "should allow setting the stopper on initialisation" do
+      xdb = XapianDb.new(:stopper => :english)
+      xdoc = xdb.documents.new("And they made a cake", :stopper => :french)
+      xdoc.stopper.call("ayantes").should == true
+      xdoc.stopper.call("and").should == false
+    end
+    
+    it "should support stop words encoded in utf8" do
+      xdb = XapianDb.new
+      xdoc = xdb.documents.new("и они made a cake", :stemmer => :russian, :stopper => :russian).to_xapian_document
+      terms = xdoc.terms.collect { |t| t.term }
+      terms.should_not include 'Zи'
+      terms.should_not include 'Zони'
+      terms.should include 'Zcake'      
+    end
+  end
+
+  describe "stemmer" do
+    it "should return the same stemmer as the database by default" do
+      xdb = XapianDb.new(:language => :french)
+      xdoc = xdb.documents.new("stink and bones")
+      xdoc.stemmer.call("contournait").should == "contourn"
+    end
+    it "should return a stemmer for the document language, overriding the db" do
+      xdb = XapianDb.new(:language => :english)
+      xdoc = xdb.documents.new("stink and bones", :language => :french)
+      xdoc.stemmer.call("contournait").should == "contourn"
+    end
+    it "should return a stemmer set by the :stemmer option, overriding the :language option and the db stemmer" do
+      xdb = XapianDb.new(:language => :german)
+      xdoc = xdb.documents.new("stink and bones", :language => :english, :stemmer => :french)
+      xdoc.stemmer.call("contournait").should == "contourn"
+    end
+     
+  end
+
+  describe "stopper" do
+    it "should return the same stopper as the database by default" do
+      xdb = XapianDb.new(:language => :french)
+      xdoc = xdb.documents.new("stink and bones")
+      xdoc.stopper.call("avec").should == true
+    end    
+    it "should return a stopper for the document language, overriding the db" do
+      xdb = XapianDb.new(:language => :english)
+      xdoc = xdb.documents.new("stink and bones", :language => :french)
+      xdoc.stopper.call("avec").should == true
+    end
+    it "should return a stopper set by the :stopper option, overriding the :language option and the db stopper" do
+      xdb = XapianDb.new(:language => :german)
+      xdoc = xdb.documents.new("stink and bones", :language => :english, :stopper => :french)
+      xdoc.stopper.call("avec").should == true
+    end    
+  end
+
 end

data/spec/xapian_doc_value_accessor_spec.rb

@@ -0,0 +1,92 @@
+require 'xapian'
+require 'lib/xapian_fu.rb'
+include XapianFu
+require 'fileutils'
+
+describe XapianDocValueAccessor do
+
+  it "should store and fetch values like a hash" do
+    values = XapianDocValueAccessor.new(XapianDoc.new(nil))
+    values.store(:city, "Leeds").should == "Leeds"
+    values.fetch(:city).should == "Leeds"
+    values[:city] = "London"
+    values[:city].should == "London"
+  end
+
+  it "should add and retrieve values from the Xapian::Document" do
+    doc = XapianDoc.new(nil)
+    values = XapianDocValueAccessor.new(doc)
+    lambda { values[:city] = "London" }.should change(doc.xapian_document, :values_count).by(1)
+  end
+
+  it "should store fields defined as Fixnum as packed Long" do
+    xdb = XapianDb.new(:fields => { :number => { :type => Fixnum, :store => true } })
+    [-83883, 256532, 0, 0x3fffffff].each do |number|
+      doc = xdb.documents.new(:number => number)
+      doc.values.store(:number, number, Fixnum).should == number
+      doc.values.fetch(:number, Fixnum).should == number
+      doc.to_xapian_document.values.first.value.should == [number].pack("l")
+    end
+  end  
+
+  it "should store fields defined as Bignum as packed Double-precision float, big-endian byte order" do
+    xdb = XapianDb.new(:fields => { :number => { :type => Bignum, :store => true } })
+    doc = xdb.documents.new(:number => 0x3fffffffffffff)
+    doc.values.store(:number, 0x3fffffffffffff).should == 0x3fffffffffffff
+    doc.values.fetch(:number).should == 0x3fffffffffffff
+    doc.to_xapian_document.values.first.value.should == [0x3fffffffffffff].pack("G")
+  end
+  
+  it "should store fields defined as Float as packed Double-precision float, big-endian byte order" do
+    xdb = XapianDb.new(:fields => { :number => { :type => Float, :store => true } })
+    [-0.303393984588383833, 8.448488388488384, 1.0].each do |number|
+      doc = xdb.documents.new(:number => number)
+      doc.values.store(:number, number).should == number
+      doc.values.fetch(:number).should == number
+      doc.to_xapian_document.values.first.value.should == [number].pack("G")
+    end      
+  end
+  
+  it "should store fields defined as Time in UTC as packed Double-precision float, big-endian byte order" do
+    xdb = XapianDb.new(:fields => { :created_at => { :type => Time, :store => true }})
+    time = Time.now
+    doc = xdb.documents.new(:created_at => time)
+    doc.values.store(:created_at, time).should == time
+    doc.values.fetch(:created_at).should == time
+    doc.to_xapian_document.values.first.value.should == [time.utc.to_f].pack("G")
+  end
+
+  it "should store fields defined as DateTime as a string" do
+    xdb = XapianDb.new(:fields => { :created_at => { :type => DateTime, :store => true }})
+    datetime = DateTime.now
+    doc = xdb.documents.new(:created_at => datetime)
+    doc.values.store(:created_at, datetime).should == datetime
+    doc.values.fetch(:created_at).should be_close(datetime, 0.0001) # miliseconds are not stored
+    doc.to_xapian_document.values.first.value.should == datetime.to_s
+  end
+
+  it "should store fields defined as Date as a string" do
+    xdb = XapianDb.new(:fields => { :created_on => { :type => Date, :store => true }})
+    date = Date.today
+    doc = xdb.documents.new(:created_on => date)
+    doc.values.store(:created_on, date).should == date
+    doc.values.fetch(:created_on).should == date
+    doc.to_xapian_document.values.first.value.should == date.to_s
+  end
+
+  it "should count the stored values when size is called" do
+    doc = XapianDoc.new(nil)
+    lambda { doc.values[:city] = "London" }.should change(doc.values, :size).by(1)
+  end
+
+  it "should delete values from the Xapian::Document" do
+    doc = XapianDoc.new(nil)
+    doc.values[:city] = "Leeds"
+    lambda { doc.values.delete(:city) }.should change(doc.values, :size).by(-1)
+    doc.values[:city] = "London"
+    doc.values.delete(:city).should == "London"
+  end
+
+end
+
+

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: xapian-fu
 version: !ruby/object:Gem::Version 
-  version: "0.2"
+  version: 1.0.1
 platform: ruby
 authors: 
 - John Leach
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2009-06-20 00:00:00 +01:00
+date: 2009-09-09 00:00:00 +01:00
 default_executable: 
 dependencies: []
 
@@ -24,9 +24,28 @@ extra_rdoc_files:
 - LICENSE
 files: 
 - lib/xapian_fu.rb
-- lib/xapian_fu
+- lib/xapian_fu/result_set.rb
+- lib/xapian_fu/stopper_factory.rb
+- lib/xapian_fu/stopwords/swedish.txt
+- lib/xapian_fu/stopwords/french.txt
+- lib/xapian_fu/stopwords/english.txt
+- lib/xapian_fu/stopwords/dutch.txt
+- lib/xapian_fu/stopwords/italian.txt
+- lib/xapian_fu/stopwords/norwegian.txt
+- lib/xapian_fu/stopwords/finnish.txt
+- lib/xapian_fu/stopwords/hungarian.txt
+- lib/xapian_fu/stopwords/README
+- lib/xapian_fu/stopwords/update.rb
+- lib/xapian_fu/stopwords/portuguese.txt
+- lib/xapian_fu/stopwords/german.txt
+- lib/xapian_fu/stopwords/danish.txt
+- lib/xapian_fu/stopwords/spanish.txt
+- lib/xapian_fu/stopwords/russian.txt
 - lib/xapian_fu/xapian_db.rb
 - lib/xapian_fu/xapian_doc.rb
+- lib/xapian_fu/xapian_documents_accessor.rb
+- lib/xapian_fu/query_parser.rb
+- lib/xapian_fu/xapian_doc_value_accessor.rb
 - examples/ar_spider.rb
 - examples/query.rb
 - examples/spider.rb
@@ -35,6 +54,8 @@ files:
 - LICENSE
 has_rdoc: true
 homepage: http://github.com/johnl/xapian-fu/tree/master
+licenses: []
+
 post_install_message: 
 rdoc_options: 
 - --title
@@ -59,11 +80,14 @@ required_rubygems_version: !ruby/object:Gem::Requirement
 requirements: []
 
 rubyforge_project: xapian-fu
-rubygems_version: 1.3.1
+rubygems_version: 1.3.5
 signing_key: 
-specification_version: 2
+specification_version: 3
 summary: A Ruby interface to the Xapian search engine
 test_files: 
+- spec/stopper_factory_spec.rb
 - spec/xapian_doc_spec.rb
 - spec/xapian_db_spec.rb
+- spec/query_parser_spec.rb
+- spec/xapian_doc_value_accessor_spec.rb
 - spec/spec.opts