baidu 0.2.4 → 0.2.5

data/lib/baidu.rb

@@ -1,5 +1,7 @@
 #coding:UTF-8
 require 'mechanize'
+require 'json'
+require 'uri'
 class Baidu
     attr_accessor :perpage,:pagenumber,:debug
     attr_reader :page,:wd,:data
@@ -16,6 +18,16 @@ class Baidu
     end
 
     public
+    def suggestions(wd)
+        json = @a.get("http://suggestion.baidu.com/su?wd=#{URI.encode(wd)}&cb=callback").body.force_encoding('GBK').encode("UTF-8")
+        m = /\[([^\]]*)\]/.match json
+        return JSON.parse m[0]
+    end
+
+    def popular?(wd)
+        return @a.get("http://index.baidu.com/main/word.php?word=#{URI.encode(wd.encode("GBK"))}").body.include?"boxFlash"
+    end
+    
     def query(wd)
         @data.clear
         @wd = wd
@@ -25,11 +37,7 @@ class Baidu
         q << "rn=#{@perpage}"
         queryStr = q.join("&")
         uri = URI.encode((BaseUri + queryStr).encode('GBK'))
-        begin
         @page = @a.get uri
-        rescue SocketError => e
-            puts e
-        end
         clean
         @number = self.how_many
         @maxpage = (@number / @perpage.to_f).round
@@ -69,10 +77,10 @@ class Baidu
     ########################################################################################################################
     #look up a word and get the rank of a uri with $host
     def rank(host)#on base of ranks
-        return @data['rank'][host] if @data.has_key?'rank' and @data['rank'].has_key?host
+        return @data[:rank][host] if @data.has_key?:rank and @data[:rank].has_key?host
         ranks.each_with_index do |uri,index|
             if URI.parse(URI.encode(uri).host)
-                @data << {'rank'=>{host=>index+1}}
+                @data << {:rank=>{host=>index+1}}
                 return index+1
             end
         end
@@ -90,17 +98,30 @@ class Baidu
     end
 
     def ranks#(keyword=false)
-        return @data['ranks'] if @data.has_key?'ranks'
+        return @data[:ranks] if @data.has_key?:ranks
         raise StandardError,'wrong with @page' unless @page.instance_of? Mechanize::Page
         #self.query(keyword) if keyword
         ranks = Array.new
         @page.search("//table[@class=\"result\"]").each do |table|
             ranks << @page.search("//table[@id=\"#{table['id']}\"]//a").first['href']
         end
-        @data['ranks'] = ranks
+        @data[:ranks] = ranks
         return ranks
     end
 
+    def related_keywords
+        return @data[:realated_keywords] if @data.has_key?:realated_keywords
+        raise StandardError,'wrong with @page' unless @page.instance_of? Mechanize::Page
+        keywords = Array.new
+        div = @page.search("//div[@id=\"rs\"]//tr//a")
+        return false if div.nil?
+        div.each do |keyword|
+            keywords << keyword.text
+        end
+        return keywords
+        #m = /href="[^"]+">([^<]+)<\/a>/.match(related.content)
+    end
+
     def how_many
         return @data['how_many'] if @data.has_key?'how_many'
         raise StandardError,'wrong with @page' unless @page.instance_of? Mechanize::Page

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: baidu
 version: !ruby/object:Gem::Version
-  version: 0.2.4
+  version: 0.2.5
   prerelease: 
 platform: ruby
 authors:
@@ -11,8 +11,8 @@ bindir: bin
 cert_chain: []
 date: 2011-11-11 00:00:00.000000000 Z
 dependencies: []
-description: to get data from www.baidu.com. this is built by a newbie, so please
-  be careful
+description: to get keyword ranking,related queries and popularity from baidu.com.
+  this is built by a newbie, so please be careful
 email: seoaqua@qq.com
 executables: []
 extensions: []
@@ -42,5 +42,6 @@ rubyforge_project:
 rubygems_version: 1.8.17
 signing_key: 
 specification_version: 3
-summary: to get data from www.baidu.com. this is built by a newbie, so please be careful
+summary: to get keyword ranking,related queries and popularity from baidu.com. this
+  is built by a newbie, so please be careful
 test_files: []