Update TensorFlow documentation (0.12)

assets/javascripts/templates/pages/about_tmpl.coffee

@@ -562,8 +562,8 @@ credits = [
   ], [
     'TensorFlow',
     '2015 The TensorFlow Authors',
-    'Apache',
-    'https://raw.githubusercontent.com/tensorflow/tensorflow/master/LICENSE'
+    'CC BY',
+    'https://creativecommons.org/licenses/by/3.0/'
   ], [
     'Twig',
     '2009-2016 The Twig Team',

lib/docs/core/filter.rb

@@ -36,6 +36,10 @@ module Docs
       context[:root_path]
     end
 
+    def version
+      context[:version]
+    end
+
     def subpath
       @subpath ||= subpath_to(current_url)
     end

lib/docs/core/scraper.rb

@@ -115,7 +115,8 @@ module Docs
     def options
       @options ||= self.class.options.deep_dup.tap do |options|
         options.merge! base_url: base_url, root_url: root_url,
-                       root_path: root_path, initial_paths: initial_paths
+                       root_path: root_path, initial_paths: initial_paths,
+                       version: self.class.version
 
         if root_path?
           (options[:skip] ||= []).concat ['', '/']

lib/docs/filters/tensorflow/clean_html.rb

@@ -2,27 +2,43 @@ module Docs
   class Tensorflow
     class CleanHtmlFilter < Filter
       def call
-        css('hr').remove
+        @doc = at_css('.devsite-article-inner')
 
-        css('pre > code').each do |node|
-          node.parent['class'] = node['class']
-          node.parent.content = node.content
+        css('hr', '.devsite-nav', '.devsite-content-footer', '.devsite-article-body > br').remove
+
+        css('.devsite-article-body', 'blockquote > blockquote', 'th > h2', 'th > h3').each do |node|
+          node.before(node.children).remove
         end
 
-        css('pre').each do |node|
-          node.inner_html = node.inner_html.strip_heredoc
+        css('code[class] > pre').each do |node|
+          node = node.parent
+          node.content = node.content
+          node.name = 'pre'
+        end
+
+        css('blockquote > pre:only-child', 'p > pre:only-child').each do |node|
+          next if node.previous.try(:content).present? || node.next.try(:content).present?
+          node.parent.before(node).remove
+        end
 
-          next unless node['class']
+        css('pre').each do |node|
+          node.content = node.content.strip_heredoc
 
-          if node['class'] =~ /lang-c++/i
+          if node['class'] && node['class'] =~ /lang-c++/i
             node['data-language'] = 'cpp'
-          elsif node['class'] =~ /lang-python/i
+          elsif node['class'] && node['class'] =~ /lang-python/i
             node['data-language'] = 'python'
+          else
+            node['data-language'] = version == 'Python' ? 'python' : 'cpp'
           end
         end
 
-        css('b').each do |node|
-          node.before(node.children).remove
+        css('code').each do |node|
+          node.inner_html = node.inner_html.gsub(/\s+/, ' ')
+        end
+
+        css('> code', '> b').each do |node|
+          node.replace("<p>#{node.to_html}</p>")
         end
 
         doc

lib/docs/filters/tensorflow/entries.rb

@@ -5,30 +5,56 @@ module Docs
         name = at_css('h1').content.strip
         name.remove! 'class '
         name.remove! 'struct '
+        name.remove! %r{\.\z}
+        name.sub! 'tf.contrib', 'contrib' unless version == 'Guide'
         name
       end
 
       def get_type
-        if subpath.start_with?('tutorials')
-          'Tutorials'
-        elsif subpath.start_with?('how_tos')
-          'How-Tos'
+        if version == 'Guide'
+          type = subpath.start_with?('tutorials') ? 'Tutorials' : 'How-Tos'
+
+          if node = at_css('.devsite-nav-item.devsite-nav-active')
+            node = node.previous_element until !node || node['class'].include?('devsite-nav-item-heading')
+            type << ": #{node.content}" if node
+          end
+
+          type
+        elsif version == 'C++'
+          name.remove 'tensorflow::'
         else
-          type = name.dup
-          type.remove! %r{\ \(.*\)}
-          type.remove! 'tensorflow::'
+          node = at_css('.devsite-nav-item.devsite-nav-active')
+          node = node.ancestors('.devsite-nav-item').first.at_css('.devsite-nav-title')
+          type = node.content
+          type.remove! %r{\.\z}
+          type.prepend 'Contrib: ' if type.sub!(' (contrib)', '')
           type
         end
       end
 
       def additional_entries
-        return [] if subpath.start_with?('tutorials') || subpath.start_with?('how_tos')
+        return [] if version == 'Guide'
 
-        css('h2 code', 'h3 code', 'h4 code', 'h5 code').map do |node|
-          name = node.content
-          name.sub! %r{\(.*}, '()'
-          name = name.split(' ').last
-          [name, node.parent['id']]
+        if version == 'C++'
+          names = Set.new
+          css('table.constructors td:first-child code a:first-child',
+              'table.methods      td:first-child code a:first-child',
+              'table.properties   td:first-child code a:first-child').each_with_object [] do |node, entries|
+            name = node.content
+            name.prepend "#{self.name}::"
+            name << '()' unless node.ancestors('.properties').present?
+            next unless names.add?(name)
+            id = node['href'].remove('#')
+            entries << [name, id]
+          end
+        else
+          css('h2 code', 'h3 code', 'h4 code', 'h5 code').each_with_object [] do |node, entries|
+            name = node.content
+            name.sub! %r{\(.*}, '()'
+            next if name.include?(' || ')
+            name = name.split(' ').last
+            entries << [name, node.parent['id']]
+          end
         end
       end
     end

lib/docs/scrapers/tensorflow.rb

@@ -11,33 +11,34 @@ module Docs
 
     html_filters.push 'tensorflow/entries', 'tensorflow/clean_html'
 
-    options[:container] = '#content'
+    options[:container] = '.devsite-main-content'
 
     options[:fix_urls] = ->(url) do
-      url.sub! %r{\Ahttps://www.tensorflow.org/versions(.+)/([^\.\#]+)(#.*)?\z}, 'https://www.tensorflow.org/versions\1/\2.html\3'
+      url.sub! 'how_tos/../tutorials', 'tutorials'
       url
     end
 
     options[:attribution] = <<-HTML
       &copy; 2015 The TensorFlow Authors. All rights reserved.<br>
-      Licensed under the Apache 2.0 License.
+      Licensed under the Creative Commons Attribution License 3.0.<br>
+      Code samples licensed under the Apache 2.0 License.
     HTML
 
     version 'Python' do
-      self.base_url = 'https://www.tensorflow.org/versions/r0.11/api_docs/python/'
-      self.release = '0.11'
+      self.base_url = 'https://www.tensorflow.org/api_docs/python/'
+      self.release = '0.12'
     end
 
     version 'C++' do
-      self.base_url = 'https://www.tensorflow.org/versions/r0.11/api_docs/cc/'
-      self.release = '0.11'
+      self.base_url = 'https://www.tensorflow.org/api_docs/cc/'
+      self.release = '0.12'
     end
 
     version 'Guide' do
-      self.base_url = 'https://www.tensorflow.org/versions/r0.11/'
-      self.release = '0.11'
-      self.root_path = 'tutorials/index.html'
-      self.initial_paths = %w(how_tos/index.html)
+      self.base_url = 'https://www.tensorflow.org/'
+      self.release = '0.12'
+      self.root_path = 'tutorials/'
+      self.initial_paths = %w(how_tos/)
 
       options[:only_patterns] = [/\Atutorials/, /\Ahow_tos/]
     end