(* lexer.mll -*- tuareg -*- *)
{
  open Parser
  let get = Lexing.lexeme
}

(* Helpers *)

  let tab   = '\009'
  let cr    = '\013'
  let lf    = '\010'
  let eol   = cr | lf | cr lf

  let char  = ['a'-'z'] | ['A'-'Z'] | ['0'-'9']

(* Tokens *)
rule token = parse
  | eol                  { token lexbuf }
  | (' ' | tab)          { token lexbuf }
  | eof                  { EOF }
  | "<a"                 { STARTA }
  | "href"               { HREF }
  | '='                  { EQ }
  | '"'                  { QUOTE }
  | '>'                  { GT }
  | "</a>"               { ENDA }
  | "<b>"                { STARTB }
  | "<i>"                { STARTI }
  | "<em>"               { STARTEM }
  | "</b>"               { ENDB }
  | "</i>"               { ENDI }
  | "</em>"              { ENDEM }
  | char char*           { WORD (get lexbuf) }