Merge branch 'romeovs-master'

* romeovs-master:
  make flex work with webkit
  fix correct index
  fix tables to work everywhere, add comments
  clean up complexity notation and deprecated, better caption handling
  add haskell logos
  get the thing working for haskell
  start adding haskell scraper

Author: Thibaut

assets/stylesheets/pages/_haskell.scss

@@ -0,0 +1,121 @@
+._icon-haskell:before {
+  background-image: image-url('/icons/docs/haskell/16.png');
+  background-size: cover;
+  background-repeat: no-repeat;
+}
+
+.empty-table .empty {
+  display: none;
+}
+
+.arguments td.src {
+    background: #faf9e2;
+    width: 30%;
+}
+
+th.src,
+td.src {
+  font-family: $monoFont;
+  font-weight: normal;
+  font-style: normal;
+  background: #f8f8f8;
+}
+
+caption {
+  font-weight: bold;
+  text-align: left;
+  font-style: italic;
+  font-size: 1.1em;
+}
+
+// remove margin in descript listing
+dd > pre {
+  @extend %pre;
+  margin: 0;
+  background: #faf9e2;
+  border-color: #dddaaa #dddaaa #d7d7a9;
+}
+
+// warnings are red
+.warning {
+    @extend %note;
+    @extend %note-red;
+}
+
+
+// complexity classes are blue boxes
+.with-complexity {
+  display: flex;
+  display: -webkit-flex;
+
+  justify-content: space-between;
+  -webkit-justify-content: space-between;
+
+  align-items: flex-start;
+  -webkit-align-items: flex-start;
+
+  align-content: stretch;
+  -webkit-align-content: stretch;
+}
+
+.complexity {
+    @extend %note;
+    @extend %note-blue;
+    margin: 0;
+    margin-left: 1em;
+    margin-bottom: 0.75em;
+    font-style: italic;
+    white-space: nowrap;
+
+    flex-shrink: 0;
+    -webkit-flex-shrink: 0;
+
+    order: 2;
+    -webkit-order: 2;
+}
+
+.complexity + span {
+  order: 1;
+    -webkit-order: 1;
+}
+
+// add box type to "since: ..."
+.added {
+    @extend %note;
+    @extend %note-gold;
+}
+
+.added-cell {
+    @extend %note-gold;
+}
+
+.fields h3 {
+  display: none;
+}
+
+// separate types more
+.src {
+    margin-top: 2.5em;
+}
+
+h1 + .top .src,
+h2 + .top .src,
+h3 + .top .src,
+.caption + .top .src {
+  margin-top: 0;
+}
+
+// but not for first type
+h1 + .top,
+h2 + .top,
+h3 + .top,
+h4 + .top {
+  margin-top: 0;
+}
+
+// change color of example code
+.example {
+    border: 1px solid;
+    background: #faf9e2;
+    border-color: #dddaaa #dddaaa #d7d7a9;
+}

lib/docs/filters/haskell/clean_html.rb

@@ -0,0 +1,159 @@
+module Docs
+  class Haskell
+    class CleanHtmlFilter < Filter
+      def call
+
+        # remove unwanted elements
+        css('#footer', '#package-header', '#module-header', '#synopsis', '.link', '#table-of-contents', '.package').remove
+
+        # cpations in tables are h3
+        css('table .caption').each do |node|
+          node.name = 'h3'
+        end
+
+        # turn captions into real headers
+        css('.caption').each do |node|
+          node.name = 'h1'
+        end
+
+        # section
+        css('.top > .caption').each do |node|
+          node.name = 'h2'
+        end
+
+        # subsections
+        css('.top > .subs > .caption', '.fields > .caption').each do |node|
+          node.name = 'h3'
+        end
+
+        # subsubsections
+        css('.top > .subs > .subs > .caption').each do |node|
+          node.name = 'h4'
+        end
+
+        # ...
+        css('.top > .subs > .subs > .subs > .caption').each do |node|
+          node.name = 'h5'
+        end
+
+        # ......
+        css('.top > .subs > .subs > .subs > .subs > .caption').each do |node|
+          node.name = 'h6'
+        end
+
+        # all pre's are examples
+        css('pre').each do |node|
+          node.add_css_class('example')
+        end
+
+        # turn source listing in to pre
+        css('.src').each do |node|
+          if node.name != "td"
+            node.name = 'pre'
+          end
+        end
+
+        # check if second column of table is totally empty.
+        # and remove it if it is
+        css('table').each do |table|
+          empty = true
+          table.css('td + td').each do |snd|
+            empty = empty && snd['class'] =~ /empty/
+          end
+          if empty
+            # remove empty column
+            table.css('td + td').remove
+          end
+        end
+
+        # move table captions into the tables
+        css(".caption + table").each do |table|
+          caption = table.previous
+          caption.name = "caption"
+          caption.parent = table
+        end
+
+        css(".caption + .show table").each do |table|
+          caption = table.parent.parent.css('.caption')[0]
+          caption.name = 'caption'
+          caption.parent = table
+        end
+
+        # better arguments display:
+        css('.src + .arguments table').each do |table|
+          src = table.parent.previous # the function name
+          row = doc.document.create_element('tr')
+          table.css('tr')[0].before(row)
+          src.parent = row
+          src.name = "th"
+          src['colspan'] = 2
+        end
+
+        # remove root page title
+        if root_page?
+          at_css('h1').remove
+        end
+
+        # add id to links (based on name)
+        css('a').each do |node|
+          if node['name']
+            node['id'] = node['name']
+          end
+        end
+
+        # make code in description into proper pre
+        css('dd > code').each do |node|
+          node.name = 'pre'
+        end
+
+        # add some informational boxes
+        css('em').each do |node|
+          if node.content == 'Deprecated.'
+            # Make deprecated messages red.
+            node.parent.add_css_class('warning')
+          elsif node.content =~ /O\(.*\)/
+            # this is big_O notation, but only apply the class if this is not
+            # inside running text (it must be at the start of a paragraph)
+            # from:
+            # <p><em>O(n)</em>. Koel ok</p>
+            # to:
+            # <p class="with-complexity">
+            #   <span class="complexity">O(n)</span>
+            #   <span>Koel ok</span>
+            # </p>
+            if node.previous == nil
+              node.add_css_class('complexity')                        # add css class
+              node.name="span"                                        # just make it div
+              node.next.content = node.next.content.gsub(/^. /, "")   # remove . if directly after em
+              node.content = node.content.gsub(/\.$/, "")             # remove trailing . if it's inside em
+
+              # reparent the nodes
+              cont = doc.document.create_element "p", :class => "with-complexity"
+              node.parent.previous = cont
+              par = node.parent
+              node.parent = cont
+              par.parent = cont
+              par.name = "span"
+            end
+          elsif node.content =~ /Since: .*/
+            # add box to 'Since:' annotations
+            if node.parent.parent.name == "td"
+              node.parent.parent.add_css_class('added-cell')
+            else
+              node.add_css_class('added')
+            end
+          end
+        end
+
+        doc
+      end
+    end
+  end
+end
+
+class Nokogiri::XML::Node
+  def add_css_class( *classes )
+    existing = (self['class'] || "").split(/\s+/)
+    self['class'] = existing.concat(classes).uniq.join(" ")
+  end
+end

lib/docs/filters/haskell/entries.rb

@@ -0,0 +1,56 @@
+module Docs
+  class Haskell
+    class EntriesFilter < Docs::EntriesFilter
+
+      # gets name and type in one fell swoop
+      # 
+      # eg.
+      #  Control.Monad > [Monad, Control]
+      #  Control.Concurrent.Mvar > [Concurrent.MVar, Control]
+      #  Array > [Array, nil]
+      def get_name_and_type
+        if at_css('h1') && at_css('h1').content == 'Haskell Hierarchical Libraries'
+          puts 'ok'
+          name = 'Haskell'
+          type = nil
+        else
+          # find full module identifier
+          caption = at_css('#module-header .caption')
+
+          if caption
+            # split the module path
+            parts   = caption.content.split('.')
+
+            if parts.length > 1
+              # if more than one part then the 
+              # first is the type and the rest is the name
+              type = parts[0]
+              name = parts.drop(1).join('.')
+            else
+              # if only one part, this is the name
+              name = parts[0]
+              type = nil
+            end
+          else
+            # no caption found -> no type / no name
+            name = 'no-name'
+            type = 'no-type'
+          end
+        end
+        [name, type]
+      end
+
+      # get the name
+      def get_name
+        n, t = get_name_and_type()
+        n
+      end
+
+      # get the type
+      def get_type
+        n, t = get_name_and_type()
+        t
+      end
+    end
+  end
+end

lib/docs/scrapers/haskell.rb

@@ -0,0 +1,24 @@
+module Docs
+  class Haskell < UrlScraper
+    self.name = 'Haskell'
+    self.slug = 'haskell'
+    self.type = 'haskell'
+    self.version = '7.8.2'
+    self.base_url = 'http://www.haskell.org/ghc/docs/7.8.2/html/libraries/'
+    self.initial_paths = ['/index.html']
+
+    html_filters.push 'haskell/entries'
+    html_filters.push 'haskell/clean_html'
+    html_filters.push 'title'
+
+
+    options[:container]     = '#content'
+    options[:skip_patterns] = [/src/, /index/, /haskell2010/, /ghc-/, /Cabal-/]   # skip source listings and index files
+
+    options[:attribution] = <<-HTML
+      &copy; The University Court of the University of Glasgow.<br>
+      All rights reserved. <a href="http://www.haskell.org/ghc/license">See here for more info</a>
+    HTML
+
+    end
+end