Add pygame documentation

lib/docs/filters/pygame/clean_html.rb

@@ -0,0 +1,104 @@
+module Docs
+  class Pygame
+    class CleanHtmlFilter < Filter
+      def call
+
+        @doc = at_css '.body'
+
+        if root_page?
+          # remove unneeded stuff
+          at_css('.modindex-jumpbox').remove
+          css('[role="navigation"],.pcap, .cap, .footer').each do |node|
+            node.remove
+          end
+          # table -> list
+          list = at_css('table')
+          list.replace(list.children)
+          list.name = 'ul'
+          css('tr').each do |row|
+            row.name = 'li'
+            row['class'] = ''
+          end
+          return doc
+        end
+
+        # remove unwanted stuff
+        # .headerlink => ¶ after links
+        # .toc => table of content
+        # .tooltip-content => tooltips after links to functions
+        if toremove = css('table.toc.docutils, .headerlink, .tooltip-content')
+          toremove.each do |node|
+            node.remove
+          end
+        end
+
+        # Remove wrapper .section
+        section = at_css('.section')
+        definition = at_css('.definition')
+        definition['id'] = section['id']
+        section.replace(section.children)
+
+        # Format code for it be highlighted
+        css('.highlight-default.notranslate').each do |node|
+          pre = node.at_css('pre')
+          node.replace(pre)
+          # gets rid of the already existing syntax highlighting
+          pre.inner_html = pre.inner_text
+          pre['class'] = 'language-python'
+          pre['data-language'] = "python"
+        end
+
+        # change descriptions of functions/attributes to blockquote
+        css('.line-block').each do |node|
+          node.name = 'blockquote'
+        end
+
+        # change functions
+        css('.definition').each do |d|
+
+          # the header is the function/attribute name. It might look something like
+          # this:
+          #   pygame.image.load()
+          # It'll end up being something like this:
+          #   pygame.image.load(filename) -> Surface
+          #   pygame.image.load(fileobj, namehint="") -> Surface
+
+          header = d.at_css('dt.title')
+          if d['class'].include?('class') or d['class'].include?('module')
+            header.name = 'h1'
+            @section = header.content.strip
+          else
+            header.name = 'h3'
+          end
+          # save the original header
+          initial_header = header.content.strip
+          # save the real name for the entries
+          header['data-name'] = initial_header
+          # empty the header
+          if header.name == 'h3'
+            header.inner_html = ''
+          end
+          # to replace it with the signatures
+          next_el = header.next_element
+          signatures = next_el.css('.signature')
+          signatures.each do |sig|
+            sig.name = 'code'
+            if header.name == 'h3'
+              sig.parent = header
+              # the signature don't contain pygame.module. I think it's better
+              # to display them, as it avoids confusion with methods (have a
+              # look at the pygame.Rect page)
+              if initial_header.start_with?(@section)
+                sig.content = @section + '.' + sig.text
+              end
+              # seperate the signatures on different lines.
+              header.add_child "<br>"
+            end
+          end
+        end
+
+        doc
+      end
+    end
+  end
+end

lib/docs/filters/pygame/entries.rb

@@ -0,0 +1,35 @@
+module Docs
+  class Pygame
+    class EntriesFilter < Docs::EntriesFilter
+      def get_name
+        return 'pygame'
+      end
+
+      def get_type
+        at_css('h1').content
+      end
+
+      def include_default_entry?
+        return false
+      end
+
+      def additional_entries
+        return ['pygame'] if root_page?
+
+        entries = []
+        css('h1,h2,h3').each do |node|
+          parentclass = node.parent['class']
+          name = node['id']
+          if not name
+            name = node['data-name']
+          elsif parentclass.include?('function') or parentclass.include?('method')
+            name += '()'
+          end
+          name = name.sub('pygame.', '')
+          entries << [name, node['id'], nil]
+        end
+        entries
+      end
+    end
+  end
+end

lib/docs/scrapers/pygame.rb

@@ -0,0 +1,26 @@
+module Docs
+  class Pygame < UrlScraper
+
+    self.type = 'simple'
+
+    self.release = 'v1.9.4.dev0'
+
+    self.initial_paths = ['py-modindex.html']
+    self.base_url = 'https://www.pygame.org/docs/'
+    self.root_path = 'py-modindex.html'
+    self.initial_paths = []
+
+    self.links = {
+      home: 'https://www.pygame.org/',
+      code: 'https://github.com/pygame/pygame'
+    }
+
+    html_filters.push 'pygame/clean_html', 'pygame/entries'
+
+    options[:only_patterns] = [/ref\//]
+
+    options[:attribution] = <<-HTML
+         &copy; Pygame Developpers.
+    HTML
+  end
+end