Finish GNU Fortran scraper

assets/javascripts/news.json

@@ -1,13 +1,13 @@
 [
   [
     "2016-04-10",
-    "New documentation: <a href=\"/browser_support_tables/\">Support tables (caniuse.com)</a>"
+    "New documentations: <a href=\"/browser_support_tables/\">Support tables (caniuse.com)</a> and <a href=\"/gnu_fortran/\">GNU Fortran</a>"
   ], [
     "2016-03-27",
     "New documentation: <a href=\"/typescript/\">TypeScript</a>"
   ], [
     "2016-03-06",
-    "New documentation: <a href=\"/tensorflow/\">TensorFlow</a>, <a href=\"/haxe/\">Haxe</a> and <a href=\"/ansible/\">Ansible</a>"
+    "New documentations: <a href=\"/tensorflow/\">TensorFlow</a>, <a href=\"/haxe/\">Haxe</a> and <a href=\"/ansible/\">Ansible</a>"
   ], [
     "2016-02-28",
     "New documentations: <a href=\"/codeigniter/\">CodeIgniter</a>, <a href=\"/nginx_lua_module/\">nginx Lua Module</a> and <a href=\"/influxdata/\">InfluxData</a>"

assets/javascripts/templates/pages/about_tmpl.coffee

@@ -199,6 +199,11 @@ credits = [
     '2005-2016 Linus Torvalds and others',
     'GPLv2',
     'https://raw.githubusercontent.com/git/git/master/COPYING'
+  ], [
+    'GNU Fortran',
+    'Free Software Foundation',
+    'GFDL',
+    'https://gcc.gnu.org/onlinedocs/gcc-5.3.0/gfortran/GNU-Free-Documentation-License.html'
   ], [
     'Go',
     'Google, Inc.',

assets/stylesheets/global/_icons.scss

@@ -134,3 +134,4 @@
 ._icon-ansible:before       { background-position: -8rem -10rem; @extend %darkIconFix !optional; }
 ._icon-typescript:before    { background-position: -9rem -10rem; }
 ._icon-browser_support_tables:before { background-position: 0rem -11rem; }
+._icon-gnu_fortran:before   { background-position: -1rem -11rem; }

assets/stylesheets/pages/_base.scss

@@ -17,7 +17,7 @@
 }
 
 ._cordova,
-._fortran,
+._gnu_fortran,
 ._grunt,
 ._haxe,
 ._influxdata,

lib/docs/filters/fortran/clean_html.rb

@@ -1,18 +0,0 @@
-module Docs
-  class Fortran
-    class CleanHtmlFilter < Filter
-      def call
-        css('h2', 'h3', 'h4').each do |node|
-          node.name = 'h1'
-        end
-
-        # Move page anchor to page title
-        at_css('h1')['id'] = at_css('.node > a')['name']
-
-        css('.node', 'br').remove
-
-        doc
-      end
-    end
-  end
-end

lib/docs/filters/fortran/entries.rb

@@ -1,34 +0,0 @@
-module Docs
-  class Fortran
-    class EntriesFilter < Docs::EntriesFilter
-      REPLACE_TYPES = {
-        1 => 'Introduction',
-        2 => 'GNU Fortran Command Options',
-        3 => 'Runtime Environment Variables',
-        4 => 'Fortran 2003 and 2008 Status',
-        5 => 'Compiler Characteristics',
-        6 => 'Extensions',
-        7 => 'Mixed Language Programming',
-        8 => 'Coarray Programming',
-        9 => 'Intrinsic Procedures',
-        10 => 'Intrinsic Modules' }
-
-      def chapter_number
-        at_css('h1').content.to_i
-      end
-
-      def include_default_entry?
-        REPLACE_TYPES[chapter_number] and not at_css('ul.menu')
-      end
-
-      def get_name
-        at_css('h1').content.split(' ').drop(1).join(' ').split('—').first
-      end
-
-      def get_type
-        REPLACE_TYPES[chapter_number]
-      end
-
-    end
-  end
-end

lib/docs/filters/gnu_fortran/clean_html.rb

@@ -0,0 +1,39 @@
+module Docs
+  class GnuFortran
+    class CleanHtmlFilter < Filter
+      def call
+        heading = at_css('h1, h2, h3, h4, h5')
+        heading_level = heading.name[/h(\d)/, 1].to_i
+
+        css('h2, h3, h4, h5, h6').each do |node|
+          node.name = node.name.sub(/\d/) { |i| i.to_i - (heading_level - 1) }
+        end
+
+        css('.node > a[name]').each do |node|
+          node.parent.next_element['id'] = node['name']
+        end
+
+        css('a[name]').each do |node|
+          node['id'] = node['name']
+        end
+
+        css('samp > span:first-child:last-child').each do |node|
+          node.parent.name = 'code'
+          node.before(node.children).remove
+        end
+
+        css('pre').each do |node|
+          node.inner_html = node.inner_html.strip_heredoc.strip
+        end
+
+        css('dt > em', 'acronym', 'dfn').each do |node|
+          node.before(node.children).remove
+        end
+
+        css('.node', 'br').remove
+
+        doc
+      end
+    end
+  end
+end

lib/docs/filters/gnu_fortran/entries.rb

@@ -0,0 +1,41 @@
+module Docs
+  class GnuFortran
+    class EntriesFilter < Docs::EntriesFilter
+      TYPE_BY_CHAPTER = { }
+
+      def initialize(*)
+        super
+        detect_chapters if root_page?
+      end
+
+      def get_name
+        at_css('h1').content.split(' ').drop(1).join(' ').split('—').first
+      end
+
+      def get_type
+        "#{chapter_number}. #{TYPE_BY_CHAPTER[chapter_number]}"
+      end
+
+      def include_default_entry?
+        !at_css('ul.menu')
+      end
+
+      private
+
+      def detect_chapters
+        css('.contents > ul > li > a').each do |node|
+          index = node.content.strip.to_i
+          next unless index > 0
+          name = node.content.split(' ').drop(1).join(' ')
+          name.remove! 'GNU Fortran '
+          name.remove! %r{:.*}
+          TYPE_BY_CHAPTER[index] = name # YOLO
+        end
+      end
+
+      def chapter_number
+        at_css('h1').content.to_i
+      end
+    end
+  end
+end

lib/docs/scrapers/fortran.rb

@@ -1,32 +0,0 @@
-module Docs
-  class Fortran < FileScraper
-    self.name = 'GNU Fortran'
-    self.slug = 'fortran'
-    self.type = 'fortran'
-    self.release = '5.3.0'
-    self.base_url = "https://gcc.gnu.org/onlinedocs/gcc-#{release}/gfortran/"
-    self.dir = ''
-    self.root_path = 'index.html'
-
-    self.links = {
-      home: 'https://gcc.gnu.org/fortran/'
-    }
-
-    html_filters.push 'fortran/clean_html', 'fortran/entries'
-
-    options[:skip_patterns] = [
-      /Funding/,
-      /Projects/,
-      /Copying/,
-      /License/,
-      /Proposed/,
-      /Contribut/,
-      /Index/
-    ]
-
-    options[:attribution] = <<-HTML
-      &copy; Free Software Foundation<br>
-      Licensed under the GNU Free Documentation License version 1.3.
-    HTML
-  end
-end

lib/docs/scrapers/gnu_fortran.rb

@@ -0,0 +1,40 @@
+module Docs
+  class GnuFortran < FileScraper
+    self.name = 'GNU Fortran'
+    self.slug = 'gnu_fortran'
+    self.type = 'gnu_fortran'
+    self.dir = '/Users/Thibaut/DevDocs/Docs/gfortran'
+    self.root_path = 'index.html'
+
+    self.links = {
+      home: 'https://gcc.gnu.org/fortran/'
+    }
+
+    html_filters.push 'gnu_fortran/clean_html', 'gnu_fortran/entries'
+
+    options[:skip_patterns] = [
+      /Funding/,
+      /Projects/,
+      /Copying/,
+      /License/,
+      /Proposed/,
+      /Contribut/,
+      /Index/
+    ]
+
+    options[:attribution] = <<-HTML
+      &copy; Free Software Foundation<br>
+      Licensed under the GNU Free Documentation License, Version 1.3.
+    HTML
+
+    version '5' do
+      self.release = '5.3.0'
+      self.base_url = "https://gcc.gnu.org/onlinedocs/gcc-#{release}/gfortran/"
+    end
+
+    version '4' do
+      self.release = '4.9.3'
+      self.base_url = "https://gcc.gnu.org/onlinedocs/gcc-#{release}/gfortran/"
+    end
+  end
+end