compiler/src/lib/tiger/tiger_lexer.mll

   1 {
   2   open Tiger_parser
   3
   4   let comment_level = ref 0
   5   let string_buf    = Buffer.create 100
   6 }
   7
   8 let alpha = ['a'-'z' 'A'-'Z']
   9 let num = ['0'-'9']
  10 let newline = '\n' | '\r' | "\n\r"
  11
  12 rule token = parse
  13   | eof {EOF}
  14
  15   (* Track line number *)
  16   | newline {
  17       Lexing.new_line lexbuf;
  18       token lexbuf
  19   }
  20
  21   (* Comment *)
  22   | "/*" {
  23       incr comment_level;
  24       comment lexbuf
  25   }
  26
  27   | ":=" {ASSIGN}
  28   | "<=" {LE}
  29   | ">=" {GE}
  30   | "<>" {NEQ}
  31   | '&'  {AND}
  32   | '('  {LPAREN}
  33   | ')'  {RPAREN}
  34   | '*'  {TIMES}
  35   | '+'  {PLUS}
  36   | '-'  {MINUS}
  37   | '/'  {DIVIDE}
  38   | ','  {COMMA}
  39   | '.'  {DOT}
  40   | ':'  {COLON}
  41   | ';'  {SEMICOLON}
  42   | '>'  {GT}
  43   | '<'  {LT}
  44   | '='  {EQ}
  45   | '['  {LBRACK}
  46   | ']'  {RBRACK}
  47   | '{'  {LBRACE}
  48   | '}'  {RBRACE}
  49   | '|'  {OR}
  50
  51   (* String literal *)
  52   | '"' {
  53       string_literal lexbuf
  54   }
  55
  56   (* Drop whitespace *)
  57   | [' ' '\t'] {
  58       token lexbuf
  59   }
  60
  61   | (num+ as int) {
  62     INT (int_of_string int)
  63   }
  64
  65   | (alpha (alpha | num | '_')* as id) {
  66       match id with
  67       | "array"    -> ARRAY
  68       | "break"    -> BREAK
  69       | "do"       -> DO
  70       | "else"     -> ELSE
  71       | "end"      -> END
  72       | "for"      -> FOR
  73       | "function" -> FUNCTION
  74       | "if"       -> IF
  75       | "in"       -> IN
  76       | "let"      -> LET
  77       | "nil"      -> NIL
  78       | "of"       -> OF
  79       | "then"     -> THEN
  80       | "to"       -> TO
  81       | "type"     -> TYPE
  82       | "var"      -> VAR
  83       | "while"    -> WHILE
  84       | _          -> (ID id)
  85   }
  86 and string_literal = parse
  87   (* Keep escaped quote marks as part of the string literal *)
  88   | '\\' '"' {
  89       Buffer.add_char string_buf '"';
  90       string_literal lexbuf
  91   }
  92
  93   | '"' {
  94       let string = Buffer.contents string_buf in
  95       Buffer.reset string_buf;
  96       STRING string
  97   }
  98
  99
 100   | (_ as c) {
 101       Buffer.add_char string_buf c;
 102       string_literal lexbuf
 103   }
 104 and comment = parse
 105   (* TODO: Error: unterminated comment? or we don't care? *)
 106   | eof {EOF}
 107
 108   (* Track line number *)
 109   | newline {
 110       Lexing.new_line lexbuf;
 111       comment lexbuf
 112   }
 113
 114   | "/*" {
 115       incr comment_level;
 116       comment lexbuf
 117   }
 118
 119   | "*/" {
 120       decr comment_level;
 121       match !comment_level with
 122       | 0            -> token lexbuf
 123       | n when n > 0 -> comment lexbuf
 124       | _            -> assert false
 125   }
 126
 127   (* Drop comment contents *)
 128   | _ {
 129       comment lexbuf
 130   }