baiduserp 2.0.0 → 2.0.1

data/lib/baiduserp/helper.rb

@@ -5,7 +5,7 @@ module Baiduserp
       def get_content_safe(noko)
         return nil if noko.nil?
         return nil if noko.empty?
-        noko.first.content
+        noko.first.content.strip
       end
 
 

data/lib/baiduserp/parser/ads_left.rb

@@ -1,5 +1,35 @@
 module Baiduserp::Parser::Ads_Left
   def self.parse(file)
-    
+    result = []
+    file[:doc].search('div.ec_pp_f').each do |div|
+      id = div['id'].to_i
+      next unless id >= 3000
+      r = {rank: id}
+
+      r[:title] = Baiduserp::Helper.get_content_safe(div.search('div.ec_title'))
+
+      r[:content] = Baiduserp::Helper.get_content_safe(div.search('div.ec_desc'))
+
+      r[:site] = Baiduserp::Helper.get_content_safe(div.search('span.ec_url'))
+
+      result << r
+    end
+
+    if result.empty?
+      file[:doc].search('table.ec_pp_f').each_with_index do |table,i|
+        r = {rank: i + 1}
+
+        r[:title] = Baiduserp::Helper.get_content_safe(table.search('td.EC_header/a'))
+
+        r[:content] = Baiduserp::Helper.get_content_safe(table.search('a.EC_desc'))
+
+        r[:site] = Baiduserp::Helper.get_content_safe(table.search('a.EC_url'))
+
+        result << r
+
+      end
+    end
+
+    result
   end
 end

data/lib/baiduserp/parser/ads_right.rb

@@ -1,5 +1,19 @@
 module Baiduserp::Parser::Ads_Right
   def self.parse(file)
-    
+    result = []
+    file[:doc].search('div.EC_im').each do |div|
+      r = {}
+
+      r[:rank] = div['id'].sub('bdfs','').to_i + 1
+
+      r[:title] = Baiduserp::Helper.get_content_safe(div.search('a.EC_t'))
+
+      r[:content] = Baiduserp::Helper.get_content_safe(div.search('a.EC_desc/font'))
+
+      r[:site] = Baiduserp::Helper.get_content_safe(div.search('font.EC_url'))
+
+      result << r
+    end
+    result
   end
 end

data/lib/baiduserp/parser/organic.rb

@@ -4,15 +4,15 @@ module Baiduserp::Parser::Organic
     file[:doc].search("//table").each do |table|
       id = table['id'].to_i
       next unless id > 0
-      r = Hash.new
+      r = {:rank => id}
 
-      url = table.search("h3/a").first['href']
+      url = table.search('h3/a').first['href']
       url = Baiduserp::Client.get(url).headers['location'] if url.include?('http://www.baidu.com/link?')
       r[:url] = url
 
       r[:title] = Baiduserp::Helper.get_content_safe(table.search('h3'))
 
-      r[:content] = Baiduserp::Helper.get_content_safe(table.search("div[@class='c-abstract']"))
+      r[:content] = Baiduserp::Helper.get_content_safe(table.search('div.c-abstract'))
 
       r[:mu] = table['mu']
 

data/lib/baiduserp/version.rb

@@ -1,3 +1,3 @@
 module Baiduserp
-  VERSION = "2.0.0"
+  VERSION = "2.0.1"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: baiduserp
 version: !ruby/object:Gem::Version
-  version: 2.0.0
+  version: 2.0.1
   prerelease: 
 platform: ruby
 authors: