Merge pull request #928 from jmerle/cobol

Add GnuCOBOL documentation

assets/javascripts/templates/pages/about_tmpl.coffee

@@ -306,6 +306,11 @@ credits = [
     '2012-2018 Scott Chacon and others',
     'MIT',
     'https://raw.githubusercontent.com/git/git-scm.com/master/MIT-LICENSE.txt'
+  ], [
+    'GnuCOBOL',
+    'Free Software Foundation',
+    'GFDL',
+    'https://www.gnu.org/licenses/fdl-1.3.en.html'
   ], [
     'Go',
     'Google, Inc.',

lib/docs/filters/gnu_cobol/clean_html.rb

@@ -0,0 +1,74 @@
+module Docs
+  class GnuCobol
+    class CleanHtmlFilter < Filter
+      def call
+        # Replace the title
+        at_css('.settitle').content = 'GnuCOBOL'
+
+        # Remove the Table of Contents
+        # It's huge and the DevDocs sidebar is basically a direct copy
+        css('.contents, .contents-heading').remove
+
+        # Remove the changelog
+        at_css('p').remove
+        at_css('ol').remove
+
+        # Remove horizontal lines
+        css('hr').remove
+
+        # Remove acronym tags but keep the content
+        css('acronym').each {|node| node.name = 'span'}
+
+        # Remove everything after Appendix B
+        # This includes the license text, the document changelog, the compiler changelog and the footnote
+        current_element = at_css('a[name="Appendix-C-_002d-GNU-Free-Documentation-License"]').previous
+        until current_element.nil?
+          next_element = current_element.next
+          current_element.remove
+          current_element = next_element
+        end
+
+        # Make headers bigger
+        css('h4').each {|node| node.name = 'h3'}
+        css('h3.unnumberedsec').each {|node| node.name = 'h2'}
+
+        # Remove the newlines
+        # All paragraphs are inside <p> tags already anyways
+        css('br').remove
+
+        # The original document contains sub-headers surrounded by equal signs
+        # Convert those to actual header elements
+        css('div[align="center"]').each do |node|
+          if node.content.include?('=' * 50)
+            previous = node.previous_element
+            if !previous.nil? && previous.name == 'div' && previous['align'] == 'center'
+              previous.name = 'h4'
+            end
+
+            node.remove
+          end
+        end
+
+        # Remove align="center" attributes
+        css('[align="center"]').remove_attribute('align')
+
+        # Convert tt tags into inline code blocks and remove any surrounding quotes
+        css('tt').each do |node|
+          node.name = 'code'
+
+          previous_node = node.previous
+          if !previous_node.nil? && previous_node.text?
+            previous_node.content = previous_node.content.sub(/([^"]?")\Z/, '')
+          end
+
+          next_node = node.next
+          if !next_node.nil? && next_node.text?
+            next_node.content = next_node.content.sub(/\A("[^"]?)/, '')
+          end
+        end
+
+        doc
+      end
+    end
+  end
+end

lib/docs/filters/gnu_cobol/entries.rb

@@ -0,0 +1,50 @@
+module Docs
+  class GnuCobol
+    class EntriesFilter < Docs::EntriesFilter
+      # The entire reference is one big page, so get_name and get_type are not necessary
+
+      def additional_entries
+        entries = []
+
+        css('.contents > ul > li:not(:last-child)').each do |node|
+          parent = node.at_css('a')
+
+          entries << create_entry(parent, parent)
+
+          node.css('ul a').each do |link|
+            entries << create_entry(parent, link)
+          end
+        end
+
+        entries.compact
+      end
+
+      def create_entry(parent_link, current_link)
+        name = current_link.content
+        id = current_link['href'][1..-1]
+        type = parent_link.content
+
+        # The navigation link don't actually navigate to the correct header
+        # Instead, it references an `a` tag above it
+        # The `a` tag it is referencing is removed by a filter further down the pipeline
+        # This adds the id to the correct header element
+        target_node = at_css("a[name='#{id}']")
+        target_node.next_element.next_element['id'] = id
+
+        if name.start_with?('Appendix')
+          type = 'Appendices'
+        end
+
+        # Everything after Appendix B is removed by the clean_html filter
+        ignored_names = [
+          'Appendix C - GNU Free Documentation License',
+          'Appendix D - Summary of Document Changes',
+          'Appendix E - Summary of Compiler Changes since 2009 and version v1-1',
+          'Index'
+        ]
+
+        ignored_names.include?(name) ? nil : [name, id, type]
+      end
+    end
+  end
+end

lib/docs/scrapers/gnu_cobol.rb

@@ -0,0 +1,26 @@
+module Docs
+  class GnuCobol < UrlScraper
+    self.name = 'GnuCOBOL'
+    self.slug = 'gnu_cobol'
+    self.type = 'simple'
+    self.release = '2.2'
+    self.base_url = 'https://open-cobol.sourceforge.io/HTML/gnucobpg.html'
+    self.links = {
+      home: 'https://sourceforge.net/projects/open-cobol/',
+      code: 'https://sourceforge.net/p/open-cobol/code/HEAD/tree/trunk/'
+    }
+
+    html_filters.push 'gnu_cobol/entries', 'gnu_cobol/clean_html'
+
+    options[:attribution] = <<-HTML
+      Copyright &copy; 2000, 2001, 2002, 2007, 2008 Free Software Foundation, Inc.<br>
+      Licensed under the GNU Free Documentation License.
+    HTML
+
+    def get_latest_version(opts)
+      doc = fetch_doc('https://open-cobol.sourceforge.io/HTML/gnucobpg.html', opts)
+      title = doc.at_css('h1').content
+      title.scan(/([0-9.]+)/)[0][0]
+    end
+  end
+end

public/icons/docs/gnu_cobol/SOURCE

@@ -0,0 +1 @@
+https://sourceforge.net/p/open-cobol/icon