git.oblomov.eu Git - ohcount/blob - ext/ohcount_native/glots/c_monoglot.rb

   1 require 'monoglot'
   2
   3 module Ohcount
   4         # A generalized Monoglot that can generate the state machine for any C-like language.
   5         #
   6         # C-like languages support:
   7         #   - Single-quoted strings
   8         #   - Double-quoted strings
   9         #   - Single-line comments
  10         #   - Block comments
  11         #
  12         # When you instantiate a new CMonoglot, you specify the name of the language and provide
  13         # regular expressions to define the delimiters for the various string and comment types.
  14         class CMonoglot < Monoglot
  15
  16                 # Instantiate a new monoglot to generate a C-like language parser. Provide the following:
  17                 #
  18                 # * +language+ - the unique name of the language to be parsed.
  19                 # * +line_comment+ - one or more regular expressions that signify the start of a single-line comment
  20                 # * +block_comment+ - one or more pairs of regular expressions that signify the beginning and end of
  21                 #   a block comment
  22                 # * +double_quote+ - set to true if the language supports double-quote characters to indicate string literals
  23                 # * +single_quote+ - set to true if the language supports single-quote characters to indicate string literals
  24                 # * +options+ - an optional hash of additional options.
  25                 #
  26                 # Options include:
  27                 #
  28                 # * +no_escape_dquote+ - By default, the generated state machine will support \" as an escaped double-quote character
  29                 #   within a double-quoted string literal. To turn off this feature, specify <tt>:no_escape_dquote => true</tt>.
  30                 #
  31                 def initialize(language, line_comment, block_comment, double_quote = true, single_quote = false, options = {})
  32                         @name = options[:polyglot_name] || language
  33
  34                         # spit out states
  35                         @states = [
  36                                 State.new(language, :code, :code),
  37                                 State.new(language, :dquote_string, :code),
  38                                 State.new(language, :squote_string, :code),
  39                                 State.new(language, :line_comment, :comment),
  40                                 State.new(language, :block_comment, :comment)
  41                         ]
  42
  43                         @transitions = []
  44                         if line_comment
  45                                 [ line_comment ].flatten.each_with_index do |line_comment_token, index|
  46                                         @transitions << Transition.new(language, line_comment_token, :code, :line_comment, :to, false, index)
  47                                 end
  48                                 @transitions << Transition.new(language, '\n', :line_comment, :return, :from, false)
  49                         end
  50
  51                         if block_comment && !block_comment.empty?
  52                                 block_comment = [ block_comment ] unless block_comment[0].is_a?(Array)
  53                                 block_comment.each_with_index do |block_comment_delimiters, index|
  54                                         @transitions << Transition.new(language, block_comment_delimiters[0], :code, :block_comment, :to, false, index)
  55                                         @transitions << Transition.new(language, block_comment_delimiters[1], :block_comment, :return, :from, false, index)
  56                                 end
  57                         end
  58
  59                         if single_quote
  60                                 @transitions << Transition.new(language, "'", :code, :squote_string, :to, false)
  61                                 @transitions << Transition.new(language, e('\\\\'), :squote_string, :squote_string, :from, true, "ESC_SLASH")
  62                                 @transitions << Transition.new(language, e("\\'"), :squote_string, :squote_string, :from, true, "ESC")
  63                                 @transitions << Transition.new(language, "'", :squote_string, :return, :from, false)
  64                         end
  65
  66                         if double_quote
  67                                 @transitions << Transition.new(language, e('"'), :code, :dquote_string, :to, false)
  68                                 unless options[:no_escape_dquote]
  69                                         @transitions << Transition.new(language, e('\\\\'), :dquote_string, :dquote_string, :to, true, "ESC_SLASH")
  70                                         @transitions << Transition.new(language, e('\\"'), :dquote_string, :dquote_string, :to, true, "ESC")
  71                                 end
  72                                 @transitions << Transition.new(language, e('"'), :dquote_string, :return, :from, false)
  73                         end
  74                 end
  75
  76         end
  77 end