Add basic Hack support

See hacklang.org

Changes to PHP and C++ (and corresponding tests) to make sure that
ambiguities are covered.

Will add XHP support in a separate pull request.

refs rouge-ruby#392

lib/rouge/lexers/cpp.rb

@@ -73,6 +73,11 @@ def self.reserved
         rule /\s*(?=>)/m, Text, :pop!
         mixin :whitespace
       end
+
+      # high priority for filename matches
+      def self.analyze_text(*)
+        0.3
+      end
     end
   end
 end

lib/rouge/lexers/hack.rb

@@ -0,0 +1,52 @@
+# -*- coding: utf-8 -*- #
+
+module Rouge
+  module Lexers
+    load_lexer 'php.rb'
+
+    class Hack < PHP
+      title 'Hack'
+      desc 'The Hack programming language (hacklang.org)'
+      tag 'hack'
+      aliases 'hack', 'hh'
+      filenames '*.php', '*.hh'
+
+      def self.analyze_text(text)
+        return 1 if /<\?hh/ =~ text
+        return 0.5 if text.shebang?('hhvm')
+        return 0.2 if /async function/ =~ text
+        return 0.2 if /\): Awaitable</ =~ text
+        return 0.2 if /(vec|dict|keyset)\[/ =~ text
+        return 0.2 if /(Map|Set|Vec)\s*{/ =~ text
+        # Never succeed on a filename-only match:
+        #  - PHP wins for .php
+        #  - C++ wins for .hh
+        -1
+      end
+
+      def self.keywords
+        @hh_keywords ||= super.merge Set.new %w(
+          type newtype enum
+          as super
+          async await Awaitable
+          vec dict keyset
+          void int string bool float double
+          arraykey num Stringish
+        )
+      end
+
+      prepend :template do
+        rule /<\?hh(\s*\/\/\s*(strict|decl|partial))?$/, Comment::Preproc, :php
+      end
+
+      prepend :php do
+        rule %r((/\*\s*)(HH_(?:IGNORE_ERROR|FIXME)\[\d+\])([^*]*)(\*/)) do
+          groups Comment::Preproc, Comment::Preproc, Comment::Multiline, Comment::Preproc
+        end
+
+        rule %r(// UNSAFE(?:_EXPR|_BLOCK)?), Comment::Preproc
+        rule %r(/\*\s*UNSAFE_EXPR\s*\*/), Comment::Preproc
+      end
+    end
+  end
+end

lib/rouge/lexers/php.rb

@@ -1,8 +1,7 @@
 # -*- coding: utf-8 -*- #
 
 module Rouge
-  module Lexers
-    class PHP < TemplateLexer
+  module Lexers class PHP < TemplateLexer
       title "PHP"
       desc "The PHP scripting language (php.net)"
       tag 'php'
@@ -77,7 +76,8 @@ def self.keywords
 
       def self.analyze_text(text)
         return 1 if text.shebang?('php')
-        return 0.3 if /<\?(?!xml)/ =~ text
+        return 1 if /^<\?php/ =~ text
+        return 0.3 if /<\?(?!xml|hh)/ =~ text
         0
       end
 

spec/lexers/cpp_spec.rb

@@ -14,9 +14,11 @@
 
       assert_guess :filename => 'foo.hpp'
       assert_guess :filename => 'foo.h++'
-      assert_guess :filename => 'foo.hh'
       assert_guess :filename => 'foo.hxx'
 
+      # Disambiguate with hacklang.org
+      assert_guess :filename => 'foo.hh', :source => 'foo'
+
       # Arduino stuff
       assert_guess :filename => 'foo.pde'
       assert_guess :filename => 'foo.ino'

spec/lexers/hack_spec.rb

@@ -0,0 +1,20 @@
+# -*- coding: utf-8 -*- #
+
+describe Rouge::Lexers::Hack do
+  let(:subject) { Rouge::Lexers::Hack.new }
+
+  describe 'guessing' do
+    include Support::Guessing
+
+
+    it 'Guesses .php and .hh files that contain Hack code' do
+      assert_guess :filename => 'foo.php', :source => '<?hh // strict'
+      assert_guess :filename => 'foo.hh', :source => '<?hh // strict'
+    end
+
+    it 'Does not guess .php or .hh files that contain non-hack code' do
+      deny_guess :filename => 'foo.php', :source => '<? foo();'
+      deny_guess :filename => 'foo.hh', :source => '#include <foo.h>'
+    end
+  end
+end

spec/lexers/php_spec.rb

@@ -0,0 +1,27 @@
+# -*- coding: utf-8 -*- #
+
+describe Rouge::Lexers::PHP do
+  let(:subject) { Rouge::Lexers::PHP.new }
+
+  describe 'guessing' do
+    include Support::Guessing
+
+    it 'Guesses files containing <?php' do
+      assert_guess :source => '<?php foo();'
+    end
+
+    it 'Guesses PHP files that do not contain Hack code' do
+      assert_guess :filename => 'foo.php', :source => '<? foo();'
+    end
+
+    it 'Guesses .php files containing <?, but not hack code' do
+      deny_guess :filenaame => 'foo.php', :source => '<?hh // strict'
+    end
+
+    it "Does not guess files containing <?hh" do
+      deny_guess :source => '<?hh foo();'
+      deny_guess :source => '<?hh // strict'
+      deny_guess :filename => '.php', :source => '<?hh foo();'
+    end
+  end
+end

spec/visual/samples/hack

@@ -0,0 +1,27 @@
+<?hh // strict
+
+// Unlike PHP, there's no top-level content (e.g. HTML) before `<?hh`, and `?>` is unsupported.
+
+namespace Foo\Bar;
+
+async function main(Vector<string> $argv): Awaitable<void> {
+  var_dump($argv->map($x ==> 'Arg: '.$x."\n"));
+}
+
+// vec: new by-value container
+// Vector: old by-ref (object semantics) container
+async function wrap(vec<string> $argv): Awaitable<void> {
+  await main(new Vector($argv));
+}
+
+function foo(int $x): string {
+  // UNSAFE
+  return $x;
+}
+
+function bar(int $x): string {
+  return /* UNSAFE_EXPR */ $x;
+}
+
+/* HH_FIXME[1002] pseudomain in strict file */
+HH\Asio\join(main());