helm/DEVEL/pxp/pxp/tools/ucs2_to_utf8/lexer.mll

   1 {
   2 (******************************************************)
   3 (*    Claudio Sacerdoti Coen <sacerdot@cs.unibo.it>   *)
   4 (*                   14/05/2000                       *)
   5 (******************************************************)
   6
   7 open Parser
   8
   9 let comment_depth = ref 0;;
  10
  11 let charint_of_lexeme l =
  12  String.set l 0 '0' ;
  13  int_of_string l
  14 ;;
  15 }
  16
  17 let digit = ['0'-'9']|['A'-'F']
  18
  19 rule token =
  20  parse
  21     [' ' '\t' '\n']                           { token lexbuf }
  22   | "let"                                     { LET }
  23   | (['a'-'z']|'_')(['a'-'z']|['A'-'Z']|'_'|['0'-'9']|'\'')*
  24                                               { IDENT (Lexing.lexeme lexbuf) }
  25   | '='                                       { EQ }
  26   | ";;"                                      { END_OF_LET }
  27   | "|"                                       { PIPE }
  28   | '['                                       { LBRACKET }
  29   | ']'                                       { RBRACKET }
  30   | '-'                                       { RANGE }
  31   | "(*"                                      { incr comment_depth ;
  32                                                 comment lexbuf
  33                                               }
  34   | "#x" digit digit digit digit              { CHAR (charint_of_lexeme (Lexing.lexeme lexbuf)) }
  35   | eof                                       { EOF }
  36
  37 and comment =
  38  parse
  39     "(*" { incr comment_depth ; comment lexbuf }
  40   | "*)" { decr comment_depth ;
  41            if !comment_depth = 0 then token lexbuf else comment lexbuf
  42          }
  43   | _    { comment lexbuf }