From b699ed6b2f2e4ba24ab29d9738e7f17382258dd2 Mon Sep 17 00:00:00 2001
From: Hyeonung Baek <wiserlix@gmail.com>
Date: Fri, 28 Jan 2022 01:52:57 +0900
Subject: [PATCH] refactor lex

---
 lex.ml  | 51 ++++++++++++++++++++++++++++++---------------------
 main.ml | 14 +++++++-------
 2 files changed, 37 insertions(+), 28 deletions(-)

diff --git a/lex.ml b/lex.ml
index 6827430..3c0a252 100644
--- a/lex.ml
+++ b/lex.ml
@@ -1,9 +1,10 @@
 open Ast.Value
 
-type tokens = Token.t Seq.t
+exception Invalid_character of char
+exception Expected of char
 
-exception Token_not_found
-exception Unclosed_quote
+let invalid_char c = raise @@ Invalid_character c
+let expected c = raise @@ Expected c
 
 let either f g c =
   f c || g c
@@ -29,6 +30,12 @@ let is_ident_start =
 let is_ident =
   either is_ident_start is_digit
 
+let expect_char c seq =
+  match seq () with
+  | Seq.Nil -> expected c
+  | Seq.Cons (x, seq) ->
+    if x = c then seq else expected c
+
 let expect_token str tok seq =
   let rec aux ts seq =
     match ts (), seq () with
@@ -45,7 +52,7 @@ let find_token seq =
     (fun (s, t) -> expect_token s t seq)
 
 (* same as take_while f seq, drop_while f seq *)
-let rec partition_while f seq : 'a Seq.t * 'a Seq.t =
+let rec partition_while f seq =
   match seq () with
   | Seq.Nil -> Seq.empty, seq
   | Seq.Cons (x, seq) ->
@@ -55,44 +62,46 @@ let rec partition_while f seq : 'a Seq.t * 'a Seq.t =
     else
       Seq.(empty, cons x seq)
 
-let tokenize (str : string) : tokens =
+let tokenize str =
   let seq = String.to_seq str in
   let rec aux seq =
     let open Token in
+    let open Seq in
     match seq () with
-    | Seq.Nil -> Seq.empty
-    | Seq.Cons (x, seq) ->
+    | Nil -> empty
+    | Cons (x, seq) ->
+      (* skip whitespace *)
       if is_whitespace x then
-        aux seq (* skip whitespace *)
+        aux seq
 
+      (* string *)
       else if x = '"' then
         let str, seq = partition_while ((<>) '"') seq in
         let str = String (String.of_seq str) in
-        begin match seq () with
-          | Seq.Nil -> raise Unclosed_quote
-          | Seq.Cons (x, seq) ->
-            if x = '"' then Seq.cons (Value str) (aux seq)
-            else raise Unclosed_quote
-        end
+        let seq = expect_char '"' seq in
+        cons (Value str) (aux seq)
 
+      (* number (int, float) *)
       else if is_digit x then
         let n, seq = partition_while is_num seq in
-        let n = String.of_seq @@ Seq.cons x n in
+        let n = String.of_seq @@ cons x n in
         let n =
           if String.contains n '.' (* float *)
           then Float (float_of_string n)
           else Int (int_of_string n)
         in
-        Seq.cons (Value n) (aux seq)
+        cons (Value n) (aux seq)
 
+      (* idents *)
       else if is_ident_start x then
         let id, seq = partition_while is_ident seq in
-        let id = String.of_seq @@ Seq.cons x id in
-        Seq.cons (Ident id) (aux seq)
+        let id = String.of_seq @@ cons x id in
+        cons (Ident id) (aux seq)
 
+      (* tokens *)
       else
-        match find_token @@ Seq.cons x seq with
-        | None -> raise Token_not_found
-        | Some (t, seq) -> Seq.cons t (aux seq)
+        match find_token @@ cons x seq with
+        | None -> invalid_char x
+        | Some (t, seq) -> cons t (aux seq)
   in
   aux seq
diff --git a/main.ml b/main.ml
index f5ede32..16866f0 100644
--- a/main.ml
+++ b/main.ml
@@ -1,13 +1,11 @@
 open Printf
 
-exception Reset_line (* used to indicate ^C is pressed *)
-
 let version = "%%VERSION%%"
 
 let error_to_string e =
   try raise e with
-  | Lex.Token_not_found -> sprintf "invalid token"
-  | Lex.Unclosed_quote -> sprintf "string not closed"
+  | Lex.Invalid_character c -> sprintf "invalid character %c" c
+  | Lex.Expected c -> sprintf "expected %c" c
   | Parser.Expected t -> sprintf "expected %s" t
   | Parser.Unexpected_token t -> sprintf "unexpected token \"%s\"" t
   | Ast.Invalid_type t -> sprintf "invalid type %s" (Ast.Type.to_string t)
@@ -38,16 +36,18 @@ let rep vars : unit =
     Hashtbl.replace vars "ans" v;
     printf "%s\n" @@ Ast.Value.to_string v
 
+exception Reset_line (* used to indicate ^C is pressed *)
+
 let init_repl () =
   Hashtbl.replace vars "ans" (Ast.Value.Int 0);
   (* treat Ctrl-C as to reset line *)
-  let sigintf _ = raise Reset_line in
-  Sys.(set_signal sigint (Signal_handle sigintf))
+  let reset_line _ = raise Reset_line in
+  Sys.(set_signal sigint (Signal_handle reset_line))
 
 (* simple REPL with error handling *)
 let rec repl vars : unit =
   try rep vars; repl vars with
-  | Exit | End_of_file -> ()
+  | Exit | End_of_file (* Ctrl-D *) -> ()
   | Reset_line -> printf "\n"; repl vars
   | e -> print_error e; repl vars