Add Mason lexer (#838)

This commit adds a lexer for the Mason framework.

lib/rouge/demos/mason

@@ -0,0 +1,22 @@
+<%doc>
+    This is a mason component.
+    # This is a comment.
+</%doc>
+
+<%args>
+    $color         # this argument is required!
+    $size => 20    # default size
+    $country => undef   # this argument is optional, default value is 'undef'
+    @items => (1, 2, 'something else')
+    %pairs => (name => "John", age => 29)
+</%args>
+
+% # A random block of Perl code
+<%perl>
+    my @people = ('mary' 'john' 'pete' 'david');
+</%perl>
+
+% # Note how each line of code begins with the mandatory %
+% foreach my $person (@people) {
+    Name: <% $person %>
+% }

lib/rouge/guessers/disambiguation.rb

@@ -86,7 +86,11 @@ def match?(filename)
         next Mathematica if contains?('(*')
         next Mathematica if contains?(':=')
 
+        next Mason if matches?(/<%(def|method|text|doc|args|flags|attr|init|once|shared|perl|cleanup|filter)([^>]*)(>)/)
+
         next Matlab if matches?(/^\s*?%/)
+
+        next Mason if matches? %r!(</?%|<&)!
       end
 
       disambiguate '*.php' do

lib/rouge/lexers/mason.rb

@@ -0,0 +1,116 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Lexers
+    class Mason < TemplateLexer
+      title 'Mason'
+      desc 'The HTML::Mason framework (https://metacpan.org/pod/HTML::Mason)'
+      tag 'mason'
+      filenames '*.mi', '*.mc', '*.mas', '*.m', '*.mhtml', '*.mcomp', 'autohandler', 'dhandler'
+      mimetypes 'text/x-mason', 'application/x-mason'
+
+      def initialize(*)
+        super
+        @perl = Perl.new
+      end
+
+      def self.detect?(text)
+        return false if text.doctype?(/((?:ht|x)ml)/)
+        return true if text.doctype?
+      end
+
+      # Note: If you add a tag in the lines below, you also need to modify "disambiguate '*.m'" in file disambiguation.rb
+      TEXT_BLOCKS = %w(text doc)
+      PERL_BLOCKS = %w(args flags attr init once shared perl cleanup filter strings general_media shared_vars)
+      COMPONENTS = %w(def method)
+
+      state :root do
+        mixin :mason_tags
+      end
+
+      state :mason_tags do
+        rule %r/\s+/, Text::Whitespace
+
+        rule %r/<%(#{TEXT_BLOCKS.join('|')})>/oi, Comment::Preproc, :text_block
+
+        rule %r/<%(#{PERL_BLOCKS.join('|')})>/oi, Comment::Preproc, :perl_block
+
+        rule %r/(<%(#{COMPONENTS.join('|')}))([^>]*)(>)/oi do |m|
+          token Comment::Preproc, m[1]
+          token Name, m[3]
+          token Comment::Preproc, m[4]
+          push :component_block
+        end
+
+        # perl line
+        rule %r/^(%)(.*)$/ do |m|
+          token Comment::Preproc, m[1]
+          delegate @perl, m[2]
+        end
+
+        # start of component call
+        rule %r/<%/, Comment::Preproc, :component_call
+
+         # start of component with content
+        rule %r/<&\|/ do
+          token Comment::Preproc
+          push :component_with_content
+          push :component_sub
+        end
+
+        # start of component substitution
+        rule %r/<&/, Comment::Preproc, :component_sub
+
+        # fallback to HTML until a mason tag is encountered
+        rule(/(.+?)(?=(<\/?&|<\/?%|^%|^#))/m) { delegate parent }
+
+        # if we get here, there's no more mason tags, so we parse the rest of the doc as HTML
+        rule(/.+/m) { delegate parent }
+      end
+
+      state :perl_block do
+        rule %r/<\/%(#{PERL_BLOCKS.join('|')})>/oi, Comment::Preproc, :pop!
+        rule %r/\s+/, Text::Whitespace
+        rule %r/^(#.*)$/, Comment
+        rule(/(.*?[^"])(?=<\/%)/m) { delegate @perl }
+      end
+
+      state :text_block do
+        rule %r/<\/%(#{TEXT_BLOCKS.join('|')})>/oi, Comment::Preproc, :pop!
+        rule %r/\s+/, Text::Whitespace
+        rule %r/^(#.*)$/, Comment
+        rule %r/(.*?[^"])(?=<\/%)/m, Comment
+      end
+
+      state :component_block do
+        rule %r/<\/%(#{COMPONENTS.join('|')})>/oi, Comment::Preproc, :pop!
+        rule %r/\s+/, Text::Whitespace
+        rule %r/^(#.*)$/, Comment
+        mixin :mason_tags
+      end
+
+      state :component_with_content do
+        rule %r/<\/&>/ do 
+          token Comment::Preproc
+          pop!
+        end
+
+        mixin :mason_tags
+      end
+
+      state :component_sub do
+        rule %r/&>/, Comment::Preproc, :pop!
+
+        rule(/(.*?)(?=&>)/m) { delegate @perl }
+      end
+
+      state :component_call do
+        rule %r/%>/, Comment::Preproc, :pop!
+
+        rule(/(.*?)(?=%>)/m) { delegate @perl }
+      end
+    end
+  end
+end
+

spec/lexers/mason_spec.rb

@@ -0,0 +1,43 @@
+# -*- coding: utf-8 -*- #
+
+describe Rouge::Lexers::Mason do
+  let(:subject) { Rouge::Lexers::Mason.new }
+
+  describe 'guessing' do
+    include Support::Guessing
+
+    it 'guesses by filename' do
+      assert_guess :filename => 'foo.mas'
+      assert_guess :filename => 'foo.mi'
+      assert_guess :filename => 'foo.mc'
+      assert_guess :filename => 'foo.mhtml'
+      assert_guess :filename => 'foo.mcomp'
+      assert_guess :filename => 'autohandler'
+      assert_guess :filename => 'dhandler'
+    end
+
+    it 'guesses by mimetype' do
+      assert_guess :mimetype => 'text/x-mason'
+      assert_guess :mimetype => 'application/x-mason'
+    end
+
+    it 'guesses by source' do
+      assert_guess :filename => 'foo.m', :source => <<-eos
+      <%doc>
+      This is a mason component.
+  </%doc>
+  
+  eos
+
+    end
+
+    it 'guesses by source when using component calls' do
+      assert_guess :filename => 'foo.m', :source => <<-eos
+      <& 'SELF:component' &>
+  
+  eos
+
+    end
+
+  end
+end

spec/lexers/matlab_spec.rb

@@ -15,6 +15,24 @@
       assert_guess :mimetype => 'text/x-matlab'
       assert_guess :mimetype => 'application/x-matlab'
     end
+
+    it 'guesses by source' do
+      assert_guess :filename => 'foo.m', :source => <<-eos
+      function ImageGaborPhase(TFType,pkt,ell,titlestr)
+      % ImageGaborPhase -- Time-Frequency Display with congruent rectangles
+      
+      eos
+    end
+
+    it 'guesses by source' do
+      assert_guess :filename => 'foo.m', :source => <<-eos
+      methods
+        function obj = GaussianBlur(sigma, img_size, edges)
+            %GAUSSIANBLUR Create Gaussian Blur filter object
+        end
+      end
+      eos
+    end
   end
 end