textquery 0.1.0 → 0.1.1

data/VERSION

@@ -1 +1 @@
-0.1.0
+0.1.1

data/examples/web.rb

@@ -0,0 +1,49 @@
+require "rubygems"
+require "sinatra"
+require "textquery"
+require "erb"
+
+$KCODE = 'u'
+
+get "/" do
+  params['textstring'] = 'some random text'
+  params['query'] = 'some AND (text AND NOT (random OR match OR word))'
+  @result = TextQuery.new(params['query']).eval(params['textstring'])
+
+  erb :test
+end
+
+post "/" do
+  @result = TextQuery.new(params['query']).eval(params['textstring'])
+  erb :test
+end
+
+__END__
+
+@@ test
+<p>Result: <strong><%= @result ? "Matched" : "No match" %></strong></p>
+
+<form action="/" method="post">
+  <label>Text</label><br />
+  <textarea name="textstring" cols="100" rows="6"><%= params['textstring'] %></textarea><br />
+
+  <label>Query</label><br />
+  <textarea name="query" cols="100" rows="2"><%= params['query'] %></textarea><br />
+
+  <br />
+  <input type="submit">
+</form>
+
+<pre style="background-color:#ccc; padding:2em;">
+Supported operators and rules:
+ * AND (spaces are implicit AND’s)
+ * OR
+ * NOT ('-' is an alias)
+ * 'quoted strings'
+
+Examples queries:
+ * 'to be' OR NOT 'to_be'
+ * -omitstring
+ * a AND b
+ * a AND (b AND NOT (c OR d))
+</pre>

data/lib/textquery/textquery.rb

@@ -1,8 +1,14 @@
 require 'treetop'
 
+# make it utf-8 compatible
+if RUBY_VERSION < '1.9'
+  require 'active_support'
+  $KCODE = 'u'
+end
+
 class WordMatch < Treetop::Runtime::SyntaxNode
-  def eval(text)
-    not text.match(/^#{query}\s|\s#{query}\s|\s#{query}$|^#{query}$/).nil?
+  def eval(text, opt)
+    not text.match("^#{query}#{opt[:delim]}|#{opt[:delim]}#{query}#{opt[:delim]}|#{opt[:delim]}#{query}$|^#{query}$").nil?
   end
 
   def query
@@ -13,28 +19,43 @@ end
 Treetop.load File.dirname(__FILE__) + "/textquery_grammar"
 
 class TextQuery
-  def initialize(query = '')
+  def initialize(query = '', options = {})
     @parser = TextQueryGrammarParser.new
     @query  = nil
 
+    update_options(options)
     parse(query) if not query.empty?
   end
 
   def parse(query)
+    query = query.mb_chars if RUBY_VERSION < '1.9'
     @query = @parser.parse(query)
     if not @query
       puts @parser.terminal_failures.join("\n")
     end
-    @query
+    self
   end
 
-  def eval(input)
+  def eval(input, options = {})
+    update_options(options) if not options.empty?
+
     if @query
-      @query.eval(input)
+      @query.eval(input, @options)
     else
       puts 'no query specified'
     end
   end
   alias :match? :eval
-  
+
+  def terminal_failures
+    @parser.terminal_failures
+  end
+
+  private
+
+  def update_options(options)
+    @options = {:delim => ' '}.merge options
+    @options[:delim] = Regexp.escape @options[:delim]
+  end
+
 end

data/lib/textquery/textquery_grammar.treetop

@@ -6,14 +6,14 @@ grammar TextQueryGrammar
 
   rule logical
     op1:value space operator:binary space op2:expression {
-      def eval(text)
-        operator.eval(op1.eval(text), op2.eval(text))
+      def eval(text, opt)
+        operator.eval(op1.eval(text, opt), op2.eval(text, opt))
       end
     }
     /
     op1:value [\s]+ op2:expression {
-        def eval(text)
-          op1.eval(text) && op2.eval(text)
+        def eval(text, opt)
+          op1.eval(text, opt) && op2.eval(text, opt)
         end
       }
   end
@@ -54,20 +54,20 @@ grammar TextQueryGrammar
 
   rule value
     '(' space expression space ')' {
-       def eval(text)
-         expression.eval(text)
+       def eval(text, opt)
+         expression.eval(text, opt)
        end
     }
     /
     operator:unary space value {
-      def eval(text)
-        operator.eval(value.eval(text))
+      def eval(text, opt)
+        operator.eval(value.eval(text, opt))
       end
     }
     /
     "'" space words space "'" {
-      def eval(text)
-        words.eval(text)
+      def eval(text, opt)
+        words.eval(text, opt)
       end
     }
     /

data/spec/textquery_spec.rb

@@ -18,18 +18,7 @@ describe TextQuery do
   end
 
   def parse(input)
-    result = @parser.parse(input)
-    unless result
-      puts @parser.terminal_failures.join("\n")
-    end
-    result
-  end
-
-  it "should accept any non space separated sequence" do
-    %w[query 123 text123 #tag $%*].each do |input|
-      @parser.parse(input).text_value.should == input
-      parse(input).eval(input).should be_true
-    end
+    @parser.parse(input)
   end
 
   it "should look for exact word boundary match" do
@@ -48,7 +37,7 @@ describe TextQuery do
     parse("a AND b").eval("a b").should be_true
     parse("a AND b").eval("a c b").should be_true
   end
-  
+
   it "should accept logical OR" do
     parse("a OR b").eval("c").should be_false
     parse("a OR b").eval("a").should be_true
@@ -90,10 +79,10 @@ describe TextQuery do
     parse("(a AND b) OR c").eval("a b c").should be_true
     parse("(a AND b) OR c").eval("a b").should be_true
     parse("(a AND b) OR c").eval("a c").should be_true
-    
+
     parse("(a AND b) OR c").eval("c").should be_true
     parse("a AND (b OR c)").eval("c").should be_false
-    
+
     # for the win...
     parse("a AND (b AND (c OR d))").eval("d a b").should be_true
   end
@@ -116,7 +105,7 @@ describe TextQuery do
     parse("a AND (b AND (c OR NOT d))").eval("a b d").should be_false
     parse("a AND (b AND (c OR NOT d))").eval("a b c").should be_true
     parse("a AND (b AND (c OR NOT d))").eval("a b e").should be_true
-    
+
     parse("a AND (b AND NOT (c OR d))").eval("a b").should be_true
     parse("a AND (b AND NOT (c OR d))").eval("a b c").should be_false
     parse("a AND (b AND NOT (c OR d))").eval("a b d").should be_false
@@ -169,6 +158,16 @@ describe TextQuery do
     q.match?("b").should be_false
     q.match?("a b cdefg").should be_true
     q.eval("a b cdefg").should be_true
+  end
+
+  it "should work on CJK text" do
+    JP = "に入れるわけにはいかないので、プラグインの出力が同一であることでもって同一性を判定する"
+
+    q = TextQuery.new("に入".mb_chars, :delim => '')
+    q.eval(JP).should be_true
+    q.eval("けにはい").should be_false
 
+    q.parse("れるわ AND が同".mb_chars).eval(JP).should be_true
+    q.parse("れるわ AND NOT す".mb_chars).eval(JP).should be_false
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: textquery
 version: !ruby/object:Gem::Version 
-  version: 0.1.0
+  version: 0.1.1
 platform: ruby
 authors: 
 - Ilya Grigorik
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2009-12-28 00:00:00 -05:00
+date: 2009-12-29 00:00:00 -05:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -34,6 +34,7 @@ files:
 - README.rdoc
 - Rakefile
 - VERSION
+- examples/web.rb
 - lib/textquery.rb
 - lib/textquery/textquery.rb
 - lib/textquery/textquery_grammar.treetop
@@ -68,3 +69,4 @@ specification_version: 3
 summary: Evaluate any text against a collection of match rules
 test_files: 
 - spec/textquery_spec.rb
+- examples/web.rb