query 0.1.4 → 0.1.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: e5043b0180a473ab3d213136dfe9ab55ccb4a6d9
-  data.tar.gz: dc0b8b1ee15dc3f4437439712904de92b838520e
+  metadata.gz: fd353429ae373984a436cf6ac00061be274feca7
+  data.tar.gz: 9d7505c017f842ea9aa22398dfd871b0431b0d72
 SHA512:
-  metadata.gz: baa8ca09bc7bfd19f1eb3e9dffa24cf0cc28aba704d8671769bc26a79c68f10c01d57df4f76d4e6558ea638fd6c211111ca21d1883675450971256fc8369bc2a
-  data.tar.gz: 804e6685b6d7d49e563318a9150eae1a655c52cb6f6ca9084c7fbb0c908fde92a44025761668abe2b7aa4b21ba31fbb0c9ebcbc24a6d3c61fd2415e5344b9fd4
+  metadata.gz: 66b91a869fedb625008905b11e6c960bab5057e6689b4268219ce66a3446fd4b09873b0d91d3b48c54514918cec7fd2ea7fe3a8e37fefd35e9e0f2197787675e
+  data.tar.gz: 3d324814ca1997e351ca2d36ade7f7a4d287e8c78b8f13020b09e6b0080793b4fc91ab98c1a394154b3633ec551eae05f99adc957dbfdcca587537ee8c32fb9b

data/.gitignore

@@ -24,3 +24,9 @@ doc/
 lib/query/.DS_Store
 
 lib/query/.DS_Store
+
+.gemtags
+
+.tags
+
+.tags_sorted_by_file

data/README.md

@@ -22,7 +22,10 @@ end
 
 # to get the top rank of host "www.abc.com.cn" by querying "abc"
 
-Query::Engine::Baidu.new.query("abc").rank("www.abc.com.cn")
+```ruby
+puts Query::Engine::Baidu.new.query("abc").rank("www.abc.com.cn")
+#[3,1,2,4] => [rank_seo, rank_top_ads, rank_right_ads, rank_bottom_ads]
+```
 
 TODO:
 查询结果不多,翻页不存在时的处理,及rspec

data/lib/query/result/baidu.rb

@@ -1,87 +1,87 @@
 module Query
-    module Result
-        class Baidu
-            include Query::Result
-            def seo_ranks
-                return @ranks unless @ranks.nil?
-                @page.search("//*[@class='result']|//*[@class='result-op']|//*[@class='result-op c-container']").map.with_index do |table,index|
-                    parse_seo(table).merge({:rank => index + 1})
-                end
-            end
+  module Result
+    class Baidu
+      include Query::Result
+      def seo_ranks
+        return @ranks unless @ranks.nil?
+        @page.search("//*[@class='result']|//*[@class='result-op']|//*[@class='result-op c-container']").map.with_index do |table,index|
+          parse_seo(table).merge({:rank => index + 1})
+        end
+      end
 
-            def ads_top
-                @page.search("//*[@class='result']/preceding-sibling::*[contains(@class,'EC_result')]").map.with_index do |div, index|
-                    parse_ad(div).merge(:rank => index + 1)
-                end
-            end
+      def ads_top
+        @page.search("//*[@class='result']/preceding-sibling::*[contains(@class,'EC_result')]").map.with_index do |div, index|
+          parse_ad(div).merge(:rank => index + 1)
+        end
+      end
 
-            def ads_bottom
-                @page.search("//*[@class='result']/following-sibling::*[contains(@class,'EC_result')]").map.with_index do |div,index|
-                    parse_ad(div).merge(:rank => index + 1)
-                end
-            end
+      def ads_bottom
+        @page.search("//*[@class='result']/following-sibling::*[contains(@class,'EC_result')]").map.with_index do |div,index|
+          parse_ad(div).merge(:rank => index + 1)
+        end
+      end
 
-            def ads_right
-                @page.search("//div[@id='ec_im_container']/div[@id]").map.with_index do |div,index|
-                    a = div.search('a').first
-                    url = div.search("*[@class='EC_url']").first.text
-                    url = "http://#{url}"
-                    {
-                        :rank => index + 1,
-                        :text => a.text.strip,
-                        :href => a['href'].strip,
-                        :host => Addressable::URI.parse(URI.encode(url)).host
-                    }
-                end
-            end
+      def ads_right
+        @page.search("//div[@id='ec_im_container']/div[@id]").map.with_index do |div,index|
+          a = div.search('a').first
+          url = div.search("*[@class='EC_url']").first.text
+          url = "http://#{url}"
+          {
+            :rank => index + 1,
+            :text => a.text.strip,
+            :href => a['href'].strip,
+            :host => Addressable::URI.parse(URI.encode(url)).host
+          }
+        end
+      end
 
-            def count
-                @count ||= @page.search("//span[@class='nums']").map{|num|num.content.gsub(/\D/,'').to_i unless num.nil?}.first
-            end
+      def count
+        @count ||= @page.search("//span[@class='nums']").map{|num|num.content.gsub(/\D/,'').to_i unless num.nil?}.first
+      end
 
-            def related_keywords
-                @related_keywords ||= @page.search("//div[@id=\"rs\"]//tr//a").map{|keyword| keyword.text}
-            end
+      def related_keywords
+        @related_keywords ||= @page.search("//div[@id=\"rs\"]//tr//a").map{|keyword| keyword.text}
+      end
 
-            def has_result?
-                submit = @page.search('//a[text()="提交网址"]').first
-                return false if submit and submit['href'].include?'sitesubmit'
-                return true
-            end
+      def has_result?
+        submit = @page.search('//a[text()="提交网址"]').first
+        return false if submit and submit['href'].include?'sitesubmit'
+        return true
+      end
 
-            def next_url
-                @page.search("//a[text()='下一页>']").first['href']
-            end
+      def next_url
+        @page.search("//a[text()='下一页>']").first['href']
+      end
 
-            private
-            def parse_ad(div)
-                #@todo  should be :
-                #title = div.xpath("*[contains(@class,'ec_title')]",MyFilter.new).first
-                title = div.xpath("//*[contains(@class,'ec_title')]",MyFilter.new).first
-                url = %w( span[@class='ec_url']  a[@class='EC_url'] ).map do |xpath|
-                    node = div.search(xpath).first
-                    node.text if node
-                end.compact.first
-                url = "http://" + url
-                {
-                    :text => title.text,
-                    :href => title['href'],
-                    :host => Addressable::URI.parse(URI.encode(url)).host
-                }
-            end
+      private
+      def parse_ad(div)
+        #@todo  should be :
+        #title = div.xpath("*[contains(@class,'ec_title')]",MyFilter.new).first
+        title = div.xpath("//*[contains(@class,'ec_title')]",MyFilter.new).first
+        url = %w( span[@class='ec_url']  a[@class='EC_url'] ).map do |xpath|
+          node = div.search(xpath).first
+          node.text if node
+        end.compact.first
+        url = "http://" + url
+        {
+          :text => title.text,
+          :href => title['href'],
+          :host => Addressable::URI.parse(URI.encode(url)).host
+        }
+      end
 
-            def parse_seo(table)
-                url = %w( span[@class="g"]  span[@class="c-showurl"] div[@class="op_zhidao_showurl"]).map do |xpath|
-                    span = table.search(xpath).first
-                    span.text.sub(/\d{4}-\d{1,2}-\d{1,2}/,'').strip if span
-                end.compact.first
-                host = Addressable::URI.parse(URI.encode("http://#{url}")).host
-                {
-                    :text => table.search("h3").first.text.strip,
-                    :href => table.search('a').first['href'].strip,
-                    :host => host
-                }
-            end
-        end
+      def parse_seo(table)
+        url = %w( span[@class="g"]  span[@class="c-showurl"] div[@class="op_zhidao_showurl"]).map do |xpath|
+          span = table.search(xpath).first
+          span.text.sub(/\d{4}-\d{1,2}-\d{1,2}/,'').strip if span
+        end.compact.first
+        host = Addressable::URI.parse(URI.encode("http://#{url}")).host
+        {
+          :text => table.search("h3").first.text.strip,
+          :href => table.search('a').first['href'].strip,
+          :host => host
+        }
+      end
     end
-end
+  end
+end

data/lib/query/result/qihu.rb

@@ -34,6 +34,7 @@ module Query
       def ads_right
         @page.search("//ul[@id='rightbox']/li").map.with_index do |li,index|
           a = li.search('a').first
+          next unless a['_cs']
           href = CGI.parse(URI(a['_cs']).query)['aurl'].first
           host = Addressable::URI.parse(URI.encode(href)).host
           {
@@ -42,7 +43,7 @@ module Query
             :href => href,
             :host => host
           }
-        end
+        end.compact
       end
 
       def related_keywords

data/lib/query/version.rb

@@ -1,3 +1,3 @@
 module Query
-  VERSION = "0.1.4"
+  VERSION = "0.1.7"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: query
 version: !ruby/object:Gem::Version
-  version: 0.1.4
+  version: 0.1.7
 platform: ruby
 authors:
 - seoaqua
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-02-07 00:00:00.000000000 Z
+date: 2014-02-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -146,7 +146,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.2.1
+rubygems_version: 2.2.2
 signing_key: 
 specification_version: 4
 summary: I dont have time to write the document yet. Usage is almost within rspec