lib/mruby/presym.rb

   1 module MRuby
   2   class Presym
   3     OPERATORS = {
   4       "!" => "not",
   5       "%" => "mod",
   6       "&" => "and",
   7       "*" => "mul",
   8       "+" => "add",
   9       "-" => "sub",
  10       "/" => "div",
  11       "<" => "lt",
  12       ">" => "gt",
  13       "^" => "xor",
  14       "`" => "tick",
  15       "|" => "or",
  16       "~" => "neg",
  17       "!=" => "neq",
  18       "!~" => "nmatch",
  19       "&&" => "andand",
  20       "**" => "pow",
  21       "+@" => "plus",
  22       "-@" => "minus",
  23       "<<" => "lshift",
  24       "<=" => "le",
  25       "==" => "eq",
  26       "=~" => "match",
  27       ">=" => "ge",
  28       ">>" => "rshift",
  29       "[]" => "aref",
  30       "||" => "oror",
  31       "<=>" => "cmp",
  32       "===" => "eqq",
  33       "[]=" => "aset",
  34     }.freeze
  35
  36     SYMBOL_TO_MACRO = {
  37     #      Symbol      =>      Macro
  38     # [prefix, suffix] => [prefix, suffix]
  39       ["@@"  , ""    ] => ["CV"  , ""    ],
  40       ["@"   , ""    ] => ["IV"  , ""    ],
  41       [""    , "!"   ] => [""    , "_B"  ],
  42       [""    , "?"   ] => [""    , "_Q"  ],
  43       [""    , "="   ] => [""    , "_E"  ],
  44       [""    , ""    ] => [""    , ""    ],
  45     }.freeze
  46
  47     C_STR_LITERAL_RE = /"(?:[^\\\"]|\\.)*"/
  48
  49     def initialize(build)
  50       @build = build
  51     end
  52
  53     def scan(paths)
  54       presym_hash = {}
  55       paths.each {|path| read_preprocessed(presym_hash, path)}
  56       presym_hash.keys.sort_by!{|sym| [c_literal_size(sym), sym]}
  57     end
  58
  59     def read_list
  60       File.readlines(list_path, mode: "r:binary").each(&:chomp!)
  61     end
  62
  63     def write_list(presyms)
  64       _pp "GEN", list_path.relative_path
  65       File.binwrite(list_path, presyms.join("\n") << "\n")
  66     end
  67
  68     def write_id_header(presyms)
  69       prefix_re = Regexp.union(*SYMBOL_TO_MACRO.keys.map(&:first).uniq)
  70       suffix_re = Regexp.union(*SYMBOL_TO_MACRO.keys.map(&:last).uniq)
  71       sym_re = /\A(#{prefix_re})?([\w&&\D]\w*)(#{suffix_re})?\z/o
  72       _pp "GEN", id_header_path.relative_path
  73       File.open(id_header_path, "w:binary") do |f|
  74         f.puts "enum mruby_presym {"
  75         presyms.each.with_index(1) do |sym, num|
  76           if sym_re =~ sym && (affixes = SYMBOL_TO_MACRO[[$1, $3]])
  77             f.puts "  MRB_#{affixes * 'SYM'}__#{$2} = #{num},"
  78           elsif name = OPERATORS[sym]
  79             f.puts "  MRB_OPSYM__#{name} = #{num},"
  80           end
  81         end
  82         f.puts "};"
  83         f.puts
  84         f.puts "#define MRB_PRESYM_MAX #{presyms.size}"
  85       end
  86     end
  87
  88     def write_table_header(presyms)
  89       _pp "GEN", table_header_path.relative_path
  90       File.open(table_header_path, "w:binary") do |f|
  91         f.puts "static const uint16_t presym_length_table[] = {"
  92         presyms.each{|sym| f.puts "  #{sym.bytesize},\t/* #{sym} */"}
  93         f.puts "};"
  94         f.puts
  95         f.puts "static const char * const presym_name_table[] = {"
  96         presyms.each{|sym| f.puts %|  "#{sym}",|}
  97         f.puts "};"
  98       end
  99     end
 100
 101     def list_path
 102       @list_path ||= "#{@build.build_dir}/presym".freeze
 103     end
 104
 105     def header_dir;
 106       @header_dir ||= "#{@build.build_dir}/include/mruby/presym".freeze
 107     end
 108
 109     def id_header_path
 110       @id_header_path ||= "#{header_dir}/id.h".freeze
 111     end
 112
 113     def table_header_path
 114       @table_header_path ||= "#{header_dir}/table.h".freeze
 115     end
 116
 117     private
 118
 119     def read_preprocessed(presym_hash, path)
 120       File.binread(path).scan(/<@! (.*?) !@>/) do |part,|
 121         literals = part.scan(C_STR_LITERAL_RE)
 122         presym_hash[literals.map{|l| l[1..-2]}.join] = true unless literals.empty?
 123       end
 124     end
 125
 126     def c_literal_size(literal_without_quote)
 127       literal_without_quote.size  # TODO: consider escape sequence
 128     end
 129   end
 130 end