baidu 0.2.6 → 1.0.0

data/lib/baidu.rb

@@ -4,27 +4,23 @@ require 'mechanize'
 require 'json'
 require 'uri'
 class Baidu
-    attr_accessor :perpage,:pagenumber,:debug
-    attr_reader :page,:wd,:data
     BaseUri = 'http://www.baidu.com/s?'
+    PerPage = 100
+
     def initialize
         @a = Mechanize.new {|agent| agent.user_agent_alias = 'Linux Mozilla'}
         @a.idle_timeout = 2
         @a.max_history = 1
-        @perpage = 100
         @page = nil
-        @debug = false
-        @data = Hash.new
-        #@baseuri = "http://www.baidu.com/s?rn=#{@perpage}&wd="
     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
 
+=begin
     def extend(words,level=3,sleeptime=1)
         level = level.to_i - 1
         words = [words] unless words.respond_to? 'each'
@@ -40,32 +36,29 @@ class Baidu
         return extensions if level < 1
         return extensions + extend(extensions,level)
     end
+=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
-        @data.clear
         q = Array.new
         q << "wd=#{wd}"
-        q << "rn=#{@perpage}"
+        q << "rn=#{PerPage}"
         queryStr = q.join("&")
-        uri = URI.encode((BaseUri + queryStr).encode('GBK'))
-        begin 
+        #uri = URI.encode((BaseUri + queryStr).encode('GBK'))
+        uri = URI.encode((BaseUri + queryStr))
+        begin
             @page = @a.get uri
+            BaiduResult.new(@page)
         rescue Net::HTTP::Persistent::Error
             warn "#{uri}timeout"
+            return false
         end
-        clean
-        @number = self.how_many
-        @maxpage = (@number / @perpage.to_f).round
-        @currpage =0
 =begin
         query = "#{query}"
-        @uri = @baseuri+URI.encode(query.encode('GBK'))
+        @uri = BaseUri+URI.encode(query.encode('GBK'))
         @page = @a.get @uri
         self.clean
         @number = self.how_many
@@ -75,96 +68,63 @@ class Baidu
 =end
     end
 
+=begin
+    def maxpage
+        @maxpage ||= (how_many / PerPage.to_f).round
+    end
+=end
+
     #site:xxx.yyy.com
     def how_many_pages(host)
-        return @data['how_many']if @data.has_key?'how_many'
-        query("site:#{host}")
-        return how_many
+        query("site:#{host}").how_many
     end
 
     #domain:xxx.yyy.com/path/file.html
     def how_many_links(uri)
-        return @data['how_many']if @data.has_key?'how_many'
-        query("domain:\"#{uri}\"")
-        return how_many
+        query("domain:\"#{uri}\"").how_many
     end
 
     #site:xxx.yyy.com inurl:zzz
     def how_many_pages_with(host,string)
-        return @data['how_many']if @data.has_key?'how_many'
-        query("site:#{host} inurl:#{string}")
-        return how_many
+        query("site:#{host} inurl:#{string}").how_many
     end
-    ########################################################################################################################
-    #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
-        ranks.each_with_index do |uri,index|
-            if URI.parse(URI.encode(uri).host)
-                @data << {:rank=>{host=>index+1}}
-                return index+1
-            end
-        end
+
 =begin
-        @page.search("//table[@class=\"result\"]").each do |table|
-            href = @page.search("//table[@id=\"#{table['id']}\"]//a").first['href']
-            begin
-                return table['id'] if host==URI.parse(URI.encode(href)).host
-            rescue URI::InvalidURIError
-                puts "invalid uri:#{href}" if @debug
-            end
-        end
-        return false
-=end
+    private
+    def clean
+        @page.body.force_encoding('GBK')
+        @page.body.encode!('UTF-8',:invalid => :replace, :undef => :replace, :replace => "")
+        @page.body.gsub! ("[\U0080-\U2C77]+") #mechanize will be confuzed without removing the few characters
     end
+=end
+end
 
-    def ranks#(keyword=false)
-        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
-        return ranks
+class BaiduResult
+    def initialize(page)
+        raise ArgumentError 'should be Mechanize::Page' unless page.class == Mechanize::Page
+        @page = page
     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
+    
+    def ranks
+        @ranks ||= @page.search("//table[@class=\"result\"]").map{|table|@page.search("//table[@id=\"#{table['id']}\"]//a").first['href'] }
+    end
+    
+    #look up a word and get the rank of a uri with $host
+    def rank(host)#on base of ranks
+        ranks.each_with_index do |uri,index|
+            index+1 if URI.parse(URI.encode(uri)).host == host
         end
-        @data[:realated_keywords] = keywords
-        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
-        numSpan = @page.search("//span[@class='nums']").first
-        return false if numSpan.nil?
-        return numSpan.content.gsub(/\D/,'').to_i
-        #return false if @page.search("//span[@class='nums']").first.nil?
-        #return @page.search("//span[@class='nums']").first.content.gsub(/\D/,'').to_i
+        @how_many ||= @page.search("//span[@class='nums']").map{|num|num.content.gsub(/\D/,'').to_i unless num.nil?}.first
     end
 
-    def next
-        nextbtn = @page.link_with(:text=>/下一页/)
-        return false if (nextbtn.nil? or @currpage >= @maxpage)
-        @page = @a.click(nextbtn)
-        self.clean
-        return true
+    def related_keywords
+        @related_keywords ||= @page.search("//div[@id=\"rs\"]//tr//a").map{|keyword| keyword.text}
     end
-    private
-    def clean
-        @page.body.force_encoding('GBK')
-        @page.body.encode!('UTF-8',:invalid => :replace, :undef => :replace, :replace => "")
-        @page.body.gsub! ("[\U0080-\U2C77]+") #mechanize will be confuzed without removing the few characters
+    
+    def next
+        @page = BaiduResult.new(Mechanize.new.click(@page.link_with(:text=>/下一页/))) unless @page.link_with(:text=>/下一页/).nil?
     end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: baidu
 version: !ruby/object:Gem::Version
-  version: 0.2.6
+  version: 1.0.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,17 +9,17 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2011-11-11 00:00:00.000000000 Z
+date: 2012-06-13 00:00:00.000000000 Z
 dependencies: []
 description: to get keyword ranking,related queries and popularity from baidu.com.
-  this is built by a newbie, so please be careful
+  this is built by a newbie, so please be careful. welcome to check my homepage, http://seoaqua.com
 email: seoaqua@qq.com
 executables: []
 extensions: []
 extra_rdoc_files: []
 files:
 - lib/baidu.rb
-homepage: http://seoaqua.com
+homepage: https://github.com/seoaqua/ruby-baidu
 licenses: []
 post_install_message: 
 rdoc_options: []
@@ -43,5 +43,5 @@ rubygems_version: 1.8.21
 signing_key: 
 specification_version: 3
 summary: to get keyword ranking,related queries and popularity from baidu.com. this
-  is built by a newbie, so please be careful
+  is built by a newbie, so please be careful. welcome to check my homepage, http://seoaqua.com
 test_files: []