feat: better error messages

src/bin/build.ml

@@ -5,7 +5,6 @@ Licensed under the Apache License, Version 2.0 as described in the file LICENSE.
 Authors: Júnior Nascimento
 *)
 open Lang
-open Sedlexing
 
 type compiler_options = {
   input_files : string list;
@@ -41,22 +40,19 @@ let read_until_eof () =
   read_loop()
 
 let parse_from_stdin json_output =
-  let str = read_until_eof () in
-  let buf = Sedlexing.Utf8.from_string str in
-  try
-    let ast = MenhirLib.Convert.Simplified.traditional2revised Grammar.program (Lexer.provider buf) in
+  let source_code = read_until_eof () in
+  let parsed = Parser.parse_from_source "stdin" source_code in
+
+  match parsed with
+  | Ok ast ->
     if json_output then
       failwith "TODO implement json output"
     else
       print_endline (Ast.show_program ast);
       `Ok()
-  with
-    | Lexer.Invalid_token msg ->
-      Printf.eprintf "Lexical error: %s at position %d\n" msg (lexeme_start buf);
-      exit 1
-    | Grammar.Error ->
-      Printf.eprintf "Parse error starting at %d, ending at %d\n" (lexeme_start buf) (lexeme_end buf);
-      exit 1
+  | Error error ->
+    Errors.print_compiler_error Format.err_formatter error source_code;
+    exit 1
 
 let compile options () =
     Printf.printf "Compiling files: %s\n" (String.concat ", " options.input_files);
@@ -71,7 +67,7 @@ let compile options () =
 
 let process options =
   match options with
-  | { features = true; _ } ->  print_features ()
+  | { features = true; _ } -> print_features ()
   | { from_stdin = true; json_output; _ } -> parse_from_stdin json_output
   | { input_files = []; _ } ->
       prerr_endline "Error: No input files provided.";

src/lib/ast.ml

@@ -4,20 +4,7 @@ Licensed under the Apache License, Version 2.0 as described in the file LICENSE.
 
 Authors: Davi William, Sofia Rodrigues
 *)
-
-(* A point in the source code defined using line and column. *)
-type point = {
-  line: int;
-  column: int
-}
-[@@deriving show]
-
-(* A position represents a location in the source code. *)
-type position = {
-  start: point;
-  end': point;
-}
-[@@deriving show]
+open Location
 
 (* An identifier is a string with a position in the source code, usually representing a name. *)
 type identifier = string * position

src/lib/ast.mli

@@ -4,20 +4,7 @@ Licensed under the Apache License, Version 2.0 as described in the file LICENSE.
 
 Authors: Davi William, Sofia Rodrigues
 *)
-
-(* A point in the source code defined using line and column. *)
-type point = {
-  line: int;
-  column: int
-}
-[@@deriving show]
-
-(* A position represents a location in the source code. *)
-type position = {
-  start: point;
-  end': point;
-}
-[@@deriving show]
+open Location
 
 (* An identifier is a string with a position in the source code, usually representing a name. *)
 type identifier = string * position

src/lib/errors.ml

@@ -0,0 +1,53 @@
+(*
+Copyright 2025 Lemonadic. All rights reserved.
+Licensed under the Apache License, Version 2.0 as described in the file LICENSE.
+
+Authors: Sofia Rodrigues
+*)
+open Location
+
+type compiler_error = {
+  id : int;
+  message : string;
+  file : string;
+  location : position;
+  hints : (string * position) list;
+  additional_info : (string * string) list;
+}
+
+(* ANSI escape codes for color formatting *)
+let bold = "\027[1m"
+let reset = "\027[0m"
+let red = "\027[31m"
+let white = "\027[37m"
+
+(* Pretty prints an error message with a more beautiful layout. *)
+let print_compiler_error fmt error source =
+  let pp_range fmt { start_pos; end_pos } = Format.fprintf fmt "%d:%d~%d:%d" start_pos.line start_pos.column end_pos.line end_pos.column in
+
+  Format.fprintf fmt "\n %s%sERROR E%04d%s: %s%s\n" bold red error.id white error.message reset;
+
+  Format.fprintf fmt "  --> %s:" error.file;
+  pp_range fmt error.location;
+  Format.fprintf fmt "\n\n";
+
+  (* Split source code into lines and highlight relevant lines *)
+  let relevant_lines = expand_positions (error.location :: (List.map (fun (_, pos) -> pos) error.hints)) in
+  let sorted_lines = List.sort_uniq compare relevant_lines in
+  let source_lines = String.split_on_char '\n' source |> Array.of_list in
+
+  (* Print the relevant lines of code with error highlighting *)
+  List.iter (fun line_num ->
+    if line_num > 0 && line_num <= Array.length source_lines then
+      let line = source_lines.(line_num - 1) in
+      Format.fprintf fmt "%4d | %s\n" line_num line;
+
+      if line_num = error.location.start_pos.line then
+        let start_col = error.location.start_pos.column in
+        let end_col = error.location.end_pos.column in
+        let error_length = end_col - start_col in
+        let caret_line = String.make (start_col) ' ' ^ String.make error_length '^' in
+        Format.fprintf fmt "     | %s%s%s%s\n" red bold caret_line reset
+  ) sorted_lines;
+
+  Format.fprintf fmt "\n";

src/lib/errors.mli

@@ -0,0 +1,19 @@
+(*
+Copyright 2025 Lemonadic. All rights reserved.
+Licensed under the Apache License, Version 2.0 as described in the file LICENSE.
+
+Authors: Sofia Rodrigues
+*)
+open Location
+
+type compiler_error = {
+  id : int;
+  message : string;
+  file : string;
+  location : position;
+  hints : (string * position) list;
+  additional_info : (string * string) list;
+}
+
+(* Pretty prints an error message. *)
+val print_compiler_error : Format.formatter -> compiler_error -> string -> unit

src/lib/grammar.mly

@@ -6,12 +6,7 @@ Authors: Davi William, Sofia Rodrigues
 *)
 %{
 open Ast
-
-(* Creates an AST position using the Menhir internal position *)
-let loc startpos endpos = {
-  start = { column = startpos.Lexing.pos_cnum - startpos.Lexing.pos_bol; line = startpos.Lexing.pos_lnum };
-  end' = { column = endpos.Lexing.pos_cnum - endpos.Lexing.pos_bol; line = endpos.Lexing.pos_lnum }
-}
+open Location
 
 %}
 
@@ -36,7 +31,7 @@ let loc startpos endpos = {
 
 (* Macro for creating some rule with a position in a tuple. *)
 localized(x):
-  | d = x { (d, (loc $startpos $endpos)) }
+  | d = x { (d, (mk_ast_position $loc)) }
 
 (* The entrypoint of the entire parser, it parses a sequence of top level definitons. *)
 program:

src/lib/lexer.ml

@@ -63,7 +63,6 @@ let provider buf () =
   token, start, stop
 
 (* `from_string` converts an input string into a sequence of tokens using the provided lexer. *)
-
 let from_string f string =
   provider (from_string string)
   |> MenhirLib.Convert.Simplified.traditional2revised f

src/lib/location.ml

@@ -0,0 +1,51 @@
+(*
+Copyright 2024 Lemonadic. All rights reserved.
+Licensed under the Apache License, Version 2.0 as described in the file LICENSE.
+
+Authors: Sofia Rodrigues
+*)
+
+module IntSet = Set.Make(struct type t = int let compare = compare end)
+
+(* A point in the source code defined using line and column. *)
+type point = {
+  line: int;
+  column: int
+}
+[@@deriving show]
+
+(* A position represents a location in the source code. *)
+type position = {
+  start_pos: point;
+  end_pos: point;
+}
+[@@deriving show]
+
+(** Expands a list of ranges into a set of selected line numbers. *)
+let expand_positions ranges =
+  let add_if_valid acc line = if line > 0 then IntSet.add line acc else acc in
+
+  let determine_lines l1 l2 =
+    match l2 - l1 with
+    | 0 -> [l1 - 1; l1; l1 + 1]
+    | _ -> [l1 - 1; l1; l2; l2 + 1] in
+
+  let process_range acc {start_pos; end_pos} =
+    determine_lines start_pos.line end_pos.line
+    |> List.fold_left add_if_valid acc in
+
+  ranges
+  |> List.fold_left process_range IntSet.empty
+  |> IntSet.to_list
+
+(* Creates an AST position using the Menhir internal position *)
+let mk_ast_position (startpos, endpos) = {
+  start_pos = {
+    column = startpos.Lexing.pos_cnum - startpos.Lexing.pos_bol;
+    line = startpos.Lexing.pos_lnum
+  };
+  end_pos = {
+    column = endpos.Lexing.pos_cnum - endpos.Lexing.pos_bol;
+    line = endpos.Lexing.pos_lnum
+  }
+}

src/lib/location.mli

@@ -0,0 +1,26 @@
+(*
+Copyright 2024 Lemonadic. All rights reserved.
+Licensed under the Apache License, Version 2.0 as described in the file LICENSE.
+
+Authors: Sofia Rodrigues
+*)
+
+(* A point in the source code defined using line and column. *)
+type point = {
+  line: int;
+  column: int
+}
+[@@deriving show]
+
+(* A position represents a location in the source code. *)
+type position = {
+  start_pos: point;
+  end_pos: point;
+}
+[@@deriving show]
+
+(** Expands a list of ranges into a set of selected line numbers. *)
+val expand_positions : position list -> int list
+
+(* Creates an AST position using the Menhir internal position *)
+val mk_ast_position : Lexing.position * Lexing.position -> position

src/lib/parser.ml

@@ -0,0 +1,33 @@
+(*
+Copyright 2024 Lemonadic. All rights reserved.
+Licensed under the Apache License, Version 2.0 as described in the file LICENSE.
+
+Authors: Sofia Rodrigues
+*)
+open Errors
+
+let handle_error file buf id message hints =
+  {
+    id;
+    message;
+    file;
+    location = Location.mk_ast_position (Sedlexing.lexing_positions buf);
+    hints;
+    additional_info = [];
+  }
+
+let parse file (buf : Sedlexing.lexbuf) =
+  try
+    let ast = MenhirLib.Convert.Simplified.traditional2revised Grammar.program (Lexer.provider buf) in
+    Ok ast
+  with
+    | Lexer.Invalid_token msg -> Error (handle_error file buf 1 ("Lexical error: " ^ msg) [])
+    | Grammar.Error -> Error (handle_error file buf 2 "Parse error" [])
+
+(* Parses a source string from the given string. *)
+let parse_from_source file source =
+  parse file (Sedlexing.Utf8.from_string source)
+
+(* Parses input from a channel. *)
+let parse_from_channel file channel =
+  parse file (Sedlexing.Utf8.from_channel channel)

src/lib/parser.mli

@@ -0,0 +1,13 @@
+(*
+Copyright 2025 Lemonadic. All rights reserved.
+Licensed under the Apache License, Version 2.0 as described in the file LICENSE.
+
+Authors: Sofia Rodrigues
+*)
+open Errors
+
+(* Parses a source string from the given string. *)
+val parse_from_channel : string -> in_channel -> (Ast.program, compiler_error) result
+
+(* Parses input from a channel. *)
+val parse_from_source : string -> string -> (Ast.program, compiler_error) result