examples/OCaml-Kaleidoscope/Chapter6/lexer.ml

   1 (*===----------------------------------------------------------------------===
   2  * Lexer
   3  *===----------------------------------------------------------------------===*)
   4
   5 let rec lex = parser
   6   (* Skip any whitespace. *)
   7   | [< ' (' ' | '\n' | '\r' | '\t'); stream >] -> lex stream
   8
   9   (* identifier: [a-zA-Z][a-zA-Z0-9] *)
  10   | [< ' ('A' .. 'Z' | 'a' .. 'z' as c); stream >] ->
  11       let buffer = Buffer.create 1 in
  12       Buffer.add_char buffer c;
  13       lex_ident buffer stream
  14
  15   (* number: [0-9.]+ *)
  16   | [< ' ('0' .. '9' as c); stream >] ->
  17       let buffer = Buffer.create 1 in
  18       Buffer.add_char buffer c;
  19       lex_number buffer stream
  20
  21   (* Comment until end of line. *)
  22   | [< ' ('#'); stream >] ->
  23       lex_comment stream
  24
  25   (* Otherwise, just return the character as its ascii value. *)
  26   | [< 'c; stream >] ->
  27       [< 'Token.Kwd c; lex stream >]
  28
  29   (* end of stream. *)
  30   | [< >] -> [< >]
  31
  32 and lex_number buffer = parser
  33   | [< ' ('0' .. '9' | '.' as c); stream >] ->
  34       Buffer.add_char buffer c;
  35       lex_number buffer stream
  36   | [< stream=lex >] ->
  37       [< 'Token.Number (float_of_string (Buffer.contents buffer)); stream >]
  38
  39 and lex_ident buffer = parser
  40   | [< ' ('A' .. 'Z' | 'a' .. 'z' | '0' .. '9' as c); stream >] ->
  41       Buffer.add_char buffer c;
  42       lex_ident buffer stream
  43   | [< stream=lex >] ->
  44       match Buffer.contents buffer with
  45       | "def" -> [< 'Token.Def; stream >]
  46       | "extern" -> [< 'Token.Extern; stream >]
  47       | "if" -> [< 'Token.If; stream >]
  48       | "then" -> [< 'Token.Then; stream >]
  49       | "else" -> [< 'Token.Else; stream >]
  50       | "for" -> [< 'Token.For; stream >]
  51       | "in" -> [< 'Token.In; stream >]
  52       | "binary" -> [< 'Token.Binary; stream >]
  53       | "unary" -> [< 'Token.Unary; stream >]
  54       | id -> [< 'Token.Ident id; stream >]
  55
  56 and lex_comment = parser
  57   | [< ' ('\n'); stream=lex >] -> stream
  58   | [< 'c; e=lex_comment >] -> e
  59   | [< >] -> [< >]