baidu 0.2.5 → 0.2.6

data/lib/baidu.rb

@@ -1,4 +1,5 @@
 #coding:UTF-8
+require 'rubygems'
 require 'mechanize'
 require 'json'
 require 'uri'
@@ -24,6 +25,22 @@ class Baidu
         return JSON.parse m[0]
     end
 
+    def extend(words,level=3,sleeptime=1)
+        level = level.to_i - 1
+        words = [words] unless words.respond_to? 'each'
+        
+        extensions = Array.new
+        words.each do |word|
+            self.query(word)
+            extensions += related_keywords
+            extensions += suggestions(word)
+            sleep sleeptime
+        end
+        extensions.uniq!
+        return extensions if level < 1
+        return extensions + extend(extensions,level)
+    end
+
     def popular?(wd)
         return @a.get("http://index.baidu.com/main/word.php?word=#{URI.encode(wd.encode("GBK"))}").body.include?"boxFlash"
     end
@@ -37,7 +54,11 @@ class Baidu
         q << "rn=#{@perpage}"
         queryStr = q.join("&")
         uri = URI.encode((BaseUri + queryStr).encode('GBK'))
-        @page = @a.get uri
+        begin 
+            @page = @a.get uri
+        rescue Net::HTTP::Persistent::Error
+            warn "#{uri}timeout"
+        end
         clean
         @number = self.how_many
         @maxpage = (@number / @perpage.to_f).round
@@ -118,6 +139,7 @@ class Baidu
         div.each do |keyword|
             keywords << keyword.text
         end
+        @data[:realated_keywords] = keywords
         return keywords
         #m = /href="[^"]+">([^<]+)<\/a>/.match(related.content)
     end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: baidu
 version: !ruby/object:Gem::Version
-  version: 0.2.5
+  version: 0.2.6
   prerelease: 
 platform: ruby
 authors:
@@ -39,7 +39,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 1.8.17
+rubygems_version: 1.8.21
 signing_key: 
 specification_version: 3
 summary: to get keyword ranking,related queries and popularity from baidu.com. this