Fix Openjdk additional entries and style

Author: MasterEnoc

lib/docs/filters/openjdk/clean_html.rb

@@ -5,14 +5,14 @@ module Docs
   class Openjdk
     class CleanHtmlFilter < Filter
       def call
-        css('.topNav', '.subNav', '.bottomNav', '.legalCopy', 'noscript', '.subTitle').remove
+        css('.topNav', '.subNav', '.bottomNav', '.legalCopy', 'noscript', '.subTitle', 'hr').remove
 
         # Preserve internal fragment links
         # Transform <a name="foo"><!-- --></a><bar>text</bar>
         #      into <bar id="foo">text</bar>
-        css('a[name]').each do |node|
+        css('a[name]','a[id]').each do |node|
           if node.children.all?(&:blank?)
-            node.next_element['id'] = node['name'] if node.next_element
+            node.next_element['id'] = (node['id'] || node['name'])if node.next_element
             node.remove
           end
         end
@@ -23,11 +23,9 @@ def call
           node.remove
         end
 
-        # Replace summary tables with their detail content
-        css('h3[id$=".summary"]').each do |node|
-          id = node['id'].sub('summary', 'detail')
-          detail = at_css("h3[id='#{id}']") || at_css("h3[id='#{id.remove('optional.').remove('required.')}']")
-          node.parent.children = detail.parent.children if detail
+        # remove captions in tables
+        css('table caption').each do |node|
+          node.remove
         end
 
         css('h3[id$=".summary"]', 'h3[id$=".detail"]').each do |node|

lib/docs/filters/openjdk/clean_html_new.rb

@@ -7,15 +7,7 @@ def call
           at_css('h1').content = "OpenJDK #{release} Documentation"
         end
 
-        css('.header .sub-title').remove
-
-        css('blockquote pre').each do |node|
-          node.parent.name = 'pre'
-          node.parent['class'] = 'highlight'
-          node.parent['data-language'] = 'java'
-          node.parent.content = node.content
-          node.remove
-        end
+        css('.header .sub-title', 'hr', '.table-tabs').remove
 
         # fix ul section that contains summaries or tables
         css('ul').each do |node|
@@ -24,6 +16,13 @@ def call
           end
         end
 
+        css('ul.summary-list').each do |node|
+          node.css('li').each do |subnode|
+            subnode.name = 'div'
+          end
+          node.name = 'div'
+        end
+
         # add syntax highlight to code blocks
         css('pre > code').each do |node|
           node.parent['class'] = 'lang-java'

lib/docs/filters/openjdk/entries.rb

@@ -29,16 +29,31 @@ def get_type
       def additional_entries
         # Only keep the first found entry with a unique name,
         # i.e. overloaded methods are skipped in index
-        css('a[name$=".summary"]').each_with_object({}) do |summary, entries|
-          next if summary['name'].include?('nested') || summary['name'].include?('constructor') ||
-                  summary['name'].include?('field') || summary['name'].include?('constant')
-          summary.parent.css('.memberNameLink a').each do |node|
-            name = node.parent.parent.content.strip
-            name.sub! %r{\(.+?\)}m, '()'
-            id = node['href'].remove(%r{.*#})
-            entries[name] ||= ["#{self.name}.#{name}", id]
-          end
-        end.values
+        if version == '8' || version == '8 Gui' || version == '8 Web'
+          css('a[name$=".summary"]').each_with_object({}) do |summary, entries|
+            next if summary['name'].include?('nested') || summary['name'].include?('constructor') ||
+                    summary['name'].include?('field') || summary['name'].include?('constant')
+            summary.parent.css('.memberNameLink a').each do |node|
+              name = node.parent.parent.content.strip
+              name.sub! %r{\(.+?\)}m, '()'
+              id = node['href'].remove(%r{.*#})
+              entries[name] ||= ["#{self.name}.#{name}", id]
+            end
+          end.values
+
+        else
+          css('a[id$=".summary"]').each_with_object({}) do |summary, entries|
+            next if summary['id'].include?('nested') || summary['id'].include?('constructor') ||
+                    summary['id'].include?('field') || summary['id'].include?('constant')
+            summary.parent.css('.memberNameLink a').each do |node|
+              name = node.parent.parent.content.strip
+              name.sub! %r{\(.+?\)}m, '()'
+              id = node['href'].remove(%r{.*#})
+              entries[name] ||= ["#{self.name}.#{name}", id]
+            end
+          end.values
+        end
+
       end
     end
   end

lib/docs/filters/openjdk/entries_new.rb

@@ -28,16 +28,14 @@ def get_type
       end
 
       def additional_entries
-        css('a[name$=".summary"]').each_with_object({}) do |summary, entries|
-          next if summary['name'].include?('nested') || summary['name'].include?('constructor') ||
-                  summary['name'].include?('field') || summary['name'].include?('constant')
-          summary.parent.css('.memberNameLink a').each do |node|
-            name = node.parent.parent.content.strip
-            name.sub! %r{\(.+?\)}m, '()'
-            id = node['href'].remove(%r{.*#})
-            entries[name] ||= ["#{self.name}.#{name}", id]
-          end
-        end.values
+        entries = []
+
+        css('section[id]').each do |node|
+          next if !(node['id'].match?(/\(/))
+          entries << [self.name+ '.' +node.at_css('h3').content + '()', node['id']]
+        end
+
+        entries
       end
 
     end

lib/docs/scrapers/openjdk.rb

@@ -22,7 +22,10 @@ class Openjdk < FileScraper
       /\.png/
     ]
 
-    options[:only_patterns] = [/\Ajava\./]
+    options[:only_patterns] = [
+      /\Ajava\./,
+      /\Ajdk\./
+    ]
 
     options[:attribution] = <<-HTML
       &copy; 1993, 2020, Oracle and/or its affiliates. All rights reserved.<br>
@@ -37,6 +40,7 @@ class Openjdk < FileScraper
     version '15' do
       self.release = '15.0.1'
       self.root_path = 'index.html'
+      self.base_url = 'https://docs.oracle.com/en/java/javase/15/docs/api/'
 
       html_filters.push NEWFILTERS