Retrieving components of a regexp with ocamllex

 Is it possible to read and « join » two components of a certain data structure in the lexical analysis, or must that be postponed to the parser phase ?
   With the example file below, when typing parse_all "PAIR(xyz,765)";; I am expecting to obtain
[PAIR("xyz", "765")]  but the lecture only sees a sequence of chars :

# parse_all "PAIR(`xyz`,`765`)";;
- : token list =
[CHAR 'P'; CHAR 'A'; CHAR 'I'; CHAR 'R'; CHAR '('; CHAR '`'; CHAR 'x';
 CHAR 'y'; CHAR 'z'; CHAR '`'; CHAR ','; CHAR '`'; CHAR '7'; CHAR '6';
 CHAR '5'; CHAR '`'; CHAR ')']



Contents of the mll file :

{
  type token =
    | CHAR of char
    | PAIR of string*string    
}


let string_interior= ['a'-'z' 'A'-'Z' '0'-'9' '_' '+' '-' ':' '@']* 

rule parse_just_one_item = parse
  | "PAIR" [' ']* "("  [' ']*
    '`' (string_interior as s1) '`' [' ']* 
    '`' (string_interior as s2) '`' [' ']* ')'
    {PAIR(s1,s2)}
  | [' ' '\t' ]    (* eat up whitespace *)
       { parse_just_one_item lexbuf }
  | _ as c
      { 
      CHAR c
    }
  | eof
      { raise End_of_file }

{

let accu=ref([])
  
let memorize x=(accu:=x::(!accu))

let parse_all_silently s =
          let _=(accu:=[]) in
          try
            let lexbuf = Lexing.from_string s in
            while true do
              let result = parse_just_one_item lexbuf in
                memorize result; 
            done
          with End_of_file ->
            ()

let parse_all s=let _=parse_all_silently s in List.rev(!accu)
  
}