Add support for ANTLR

lib/rouge/demos/antlr

@@ -0,0 +1,19 @@
+grammar GrammarName;
+options { anOption = optionValue; }
+program: item_list item? EOF;
+param_list /* multi-line
+comment */ : name (',' name)* ;
+terminated_statement:
+  IF '(' expr ')' newline_opt terminated_statement (
+    ELSE newline_opt terminated_statement
+  )? # if_stmt
+  | WHILE '(' expr ')' newline_opt terminated_statement #while_stmt
+  | FOR '(' simple_statement_opt ';' expr_opt ';' simple_statement_opt ')' newline_opt
+    terminated_statement # for_stmt;
+COMMENT: '#' .*? NEWLINE -> channel(HIDDEN);
+ESC_NEWLINE: '\\' NEWLINE -> skip;
+STRING: '"' (~["\\\r\n] | ESCAPE_SEQUENCE)* '"';
+mode WORD_MODE;
+WORD: [A-Za-z_] [A-Za-z_0-9]*; // single line comment
+fragment EXPONENT_PART: [eE] [+\-]? DIGIT_SEQUENCE;
+fragment HEX_CONSTANT: '0' [xX] [0-9A-Fa-f]+;

lib/rouge/lexers/antlr.rb

@@ -0,0 +1,114 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Lexers
+    class ANTLR < RegexLexer
+      title "ANTLR"
+      desc "ANother Tool for Language Recognition"
+      tag 'antlr'
+      filenames '*.g4'
+      def self.keywords
+        @keywords ||= Set.new %w(
+          catch channel channels finally fragment grammar import lexer locals
+          mode more options parser popMode private protected public pushMode
+          returns skip throws tokens type
+        )
+      end
+      get_label = true
+      identifier = %r/[A-Za-z][a-zA-Z0-9_]*/
+      integer = %r/0|[1-9][0-9]*/
+      lowercase_name = %r/[a-z][a-zA-Z0-9_]*/
+      uppercase_name = %r/[A-Z][a-zA-Z0-9_]*/
+      state :whitespace do
+        rule %r/\s+/, Text
+      end
+      state :comment_and_whitespace do
+        mixin :whitespace
+        rule %r(//.*?$), Comment::Single
+        rule %r(/\*.*?\*/)m, Comment::Multiline
+      end
+      state :escape_sequence do
+        rule %r/\\./, Str::Escape
+      end
+      state :string do
+        mixin :escape_sequence
+        rule %r/'/, Str, :pop!
+        rule %r/[^\\'\n]+/, Str
+      end
+      state :options_spec do
+        mixin :comment_and_whitespace
+        rule %r/{/, Punctuation
+        rule %r/}/, Punctuation, :pop!
+        rule %r/=/, Operator, :option_value
+        rule identifier, Name::Attribute
+      end
+      state :option_value do
+        mixin :comment_and_whitespace
+        rule %r/;/, Punctuation, :pop!
+        rule %r/./, Punctuation
+        rule %r/'/, Str, :string
+        rule %r/{/, Punctuation, :action_block
+        rule identifier, Name::Constant
+        rule integer, Num::Integer
+      end
+      state :action_block do
+        mixin :escape_sequence
+        mixin :whitespace
+        rule %r/[^\\{}\s]+/, Name::Function
+        rule %r/{/, Punctuation, :action_block
+        rule %r/}/, Punctuation, :pop!
+      end
+      state :arg_action_block do
+        mixin :escape_sequence
+        mixin :whitespace
+        rule %r/[^\\\[\]]+/, Str
+        rule %r/\]/, Str, :pop!
+      end
+      state :label do
+        mixin :comment_and_whitespace
+        rule lowercase_name, Name::Label, :pop!
+      end
+      state :root do
+        mixin :comment_and_whitespace
+        rule %r/'/, Str, :string
+        rule %r/[@<>=~\-+?*]/, Operator
+        rule %r/[|,.()]/, Punctuation
+        rule %r/{/, Punctuation, :action_block
+        rule %r/\[/, Str, :arg_action_block
+        rule %r/#/, Name::Label, :label
+        rule integer, Num::Integer
+        rule %r/:/ do
+          token Punctuation
+          get_label = false
+        end
+        rule %r/;/ do
+          token Punctuation
+          get_label = true
+        end
+        rule uppercase_name do
+          if get_label
+            token Name::Label
+          else
+            token Name::Class
+          end
+        end
+        rule lowercase_name do |m|
+          if self.class.keywords.include? m[0]
+            token Keyword
+            case m[0]
+            when 'options'
+              push :options_spec
+            when 'throws'
+              get_label = false
+            end
+          elsif get_label
+            token Name::Label
+          else
+            token Name::Variable
+          end
+        end
+      end
+    end
+  end
+end

spec/lexers/antlr_spec.rb

@@ -0,0 +1,15 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+describe Rouge::Lexers::ANTLR do
+  let(:subject) { Rouge::Lexers::ANTLR.new }
+
+  describe 'guessing' do
+    include Support::Guessing
+
+    it 'guesses by filename' do
+      assert_guess :filename => 'foo.g4'
+    end
+
+  end
+end

spec/visual/samples/antlr

@@ -0,0 +1,37 @@
+grammar MyGrammar;
+options {
+  language = Ruby;
+  output = AST;
+  backtrack = true;
+}
+@header {
+  require 'strscan'
+}
+@lexer::members {
+  def scan_number
+    # implementation here
+  end
+}
+mode COMMAND_MODE;
+WORD: [a-zA-Z]+;
+NUMBER: [0-9]+;
+parse: expr+; // single line comment
+expr: WORD (PLUS | MINUS) expr | NUMBER;
+expr
+  returns[int value]
+  @init {
+  $value = 0
+}:
+  WORD { $value = lookup_word($WORD.text) } /* multi-line
+comment */
+  | NUMBER { $value = $NUMBER.text.to_i };
+WORD:
+  [a-zA-Z]+ {
+       # option example: sets the token type
+       $type = $options.myToken
+       # lexer command example: calls the method defined in @lexer::members
+       scan_number
+     };
+COMMAND: '/' -> pushMode(COMMAND_MODE);
+mode COMMAND_MODE;
+COMMAND_MODE_COMMAND: ~[\r\n]+ -> type(COMMAND);