Finish Apache Pig scraper

assets/javascripts/news.json

@@ -1,5 +1,8 @@
 [
   [
+    "2016-04-24",
+    "New documentation: <a href=\"/apache_pig/\">Apache Pig</a>"
+  ], [
     "2016-04-17",
     "New documentation: <a href=\"/perl/\">Perl</a>"
   ], [

assets/javascripts/templates/pages/about_tmpl.coffee

@@ -96,8 +96,8 @@ credits = [
     'GPLv3',
     'https://raw.githubusercontent.com/ansible/ansible/devel/COPYING'
   ], [
-    'Apache HTTP Server',
+    'Apache HTTP Server<br>Apache Pig',
-    '2016 The Apache Software Foundation',
+    '2016 The Apache Software Foundation<br>Apache and the Apache feather logo are trademarks of The Apache Software Foundation.',
     'Apache',
     'https://www.apache.org/licenses/LICENSE-2.0'
   ], [

assets/stylesheets/global/_icons.scss

@@ -137,3 +137,4 @@
 ._icon-gnu_fortran:before   { background-position: -1rem -11rem; }
 ._icon-gcc:before           { background-position: -2rem -11rem; }
 ._icon-perl:before          { background-position: -3rem -11rem; }
+._icon-apache_pig:before    { background-position: -4rem -11rem; }

assets/stylesheets/pages/_simple.scss

@@ -16,6 +16,7 @@
   blockquote { @extend %note; }
 }
 
+._apache_pig,
 ._cordova,
 ._gnu,
 ._grunt,
@@ -27,7 +28,6 @@
 ._markdown,
 ._mocha,
 ._mongoose,
-._pig,
 ._sinon,
 ._typescript,
 ._webpack {

lib/docs/filters/apache_pig/clean_html.rb

@@ -1,9 +1,7 @@
 module Docs
-  class Pig
+  class ApachePig
     class CleanHtmlFilter < Filter
       def call
-        @doc = at_css('#content')
-
         css('.pdflink').remove
 
         css('a[name]').each do |node|

lib/docs/filters/apache_pig/entries.rb

@@ -0,0 +1,33 @@
+module Docs
+  class ApachePig
+    class EntriesFilter < Docs::EntriesFilter
+      def get_name
+        at_css('h1').content
+      end
+
+      def get_type
+        at_css('h1').content
+      end
+
+      def additional_entries
+        nodes = case slug
+        when 'basic', 'cmds', 'func'
+          css('h3')
+        when 'cont'
+          css('h2, #macros + div > h3')
+        when 'test'
+          css('h2, #diagnostic-ops + div > h3')
+        when 'perf'
+          css('h2, #optimization-rules + div > h3, #specialized-joins + div > h3')
+        else
+          css('h2')
+        end
+
+        nodes.each_with_object [] do |node, entries|
+          name = node.content.strip
+          entries << [name, node['id']] unless name == 'Introduction'
+        end
+      end
+    end
+  end
+end

lib/docs/filters/pig/entries.rb

@@ -1,33 +0,0 @@
-module Docs
-  class Pig
-    class EntriesFilter < Docs::EntriesFilter
-
-      def include_default_entry?
-        false
-      end
-
-      def get_type
-        at_css('h1').content
-      end
-
-      def additional_entries
-        case slug
-          when 'basic', 'cmds', 'func'
-            nodes = css('h3')
-          when 'cont'
-            nodes = css('h2, #macros + div > h3')
-          when 'test'
-            nodes = css('h2, #diagnostic-ops + div > h3')
-          when 'perf'
-            nodes = css('h2, #optimization-rules + div > h3, #specialized-joins + div > h3')
-          else
-            nodes = css('h2')
-        end
-
-        nodes.each_with_object [] do |node, entries|
-          entries << [node.content, node['id'], get_type]
-        end
-      end
-    end
-  end
-end

lib/docs/scrapers/apache_pig.rb

@@ -0,0 +1,37 @@
+module Docs
+  class ApachePig < UrlScraper
+    self.name = 'Apache Pig'
+    self.slug = 'apache_pig'
+    self.type = 'apache_pig'
+    self.links = {
+      home: 'https://pig.apache.org/'
+    }
+
+    html_filters.push 'apache_pig/clean_html', 'apache_pig/entries'
+
+    options[:container] = '#content'
+    options[:skip] = %w(pig-index.html)
+    options[:skip_patterns] = [/\Aapi/, /\Ajdiff/]
+
+    options[:attribution] = <<-HTML
+      &copy; 2007&ndash;2016 Apache Software Foundation<br>
+      Licensed under the Apache Software License version 2.0.
+    HTML
+
+    version '0.15' do
+      self.release = '0.15.0'
+      self.base_url = "https://pig.apache.org/docs/r#{release}/"
+    end
+
+    version '0.14' do
+      self.release = '0.14.0'
+      self.base_url = "https://pig.apache.org/docs/r#{release}/"
+    end
+
+    version '0.13' do
+      self.release = '0.13.0'
+      self.base_url = "https://pig.apache.org/docs/r#{release}/"
+    end
+
+  end
+end

lib/docs/scrapers/pig.rb

@@ -1,39 +0,0 @@
-module Docs
-  class Pig < UrlScraper
-    self.name = 'Pig'
-    self.type = 'pig'
-    self.links = {
-      home: 'https://pig.apache.org/'
-    }
-
-    html_filters.push 'pig/clean_html', 'pig/entries'
-
-    options[:skip] = %w( pig-index.html )
-
-    options[:skip_patterns] = [
-      /\Aapi/,
-      /\Ajdiff/
-    ]
-
-    options[:attribution] = <<-HTML
-      &copy; 2007&ndash;2016 Apache Software Foundation<br>
-      Licensed under the Apache Software License version 2.0.
-    HTML
-
-    version '0.15.0' do
-      self.release = '0.15.0'
-      self.base_url = "http://pig.apache.org/docs/r#{release}/"
-    end
-
-    version '0.14.0' do
-      self.release = '0.14.0'
-      self.base_url = "http://pig.apache.org/docs/r#{release}/"
-    end
-
-    version '0.13.0' do
-      self.release = '0.13.0'
-      self.base_url = "http://pig.apache.org/docs/r#{release}/"
-    end
-
-  end
-end