From a28874145cce8e2863f3e9f67dc1452d536dab91 Mon Sep 17 00:00:00 2001
From: Max Wash <mbswash@gmail.com>
Date: Thu, 8 May 2025 20:31:55 +0100
Subject: [PATCH] lang: lex: add diagnostic support

---
 lang/include/ivy/lang/lex.h |  2 ++
 lang/lex.c                  | 32 ++++++++++++++++++++++++++++++++
 lang/lex.h                  |  1 +
 3 files changed, 35 insertions(+)

diff --git a/lang/include/ivy/lang/lex.h b/lang/include/ivy/lang/lex.h
index f12f095..2f8cca3 100644
--- a/lang/include/ivy/lang/lex.h
+++ b/lang/include/ivy/lang/lex.h
@@ -136,12 +136,14 @@ struct ivy_token {
 struct ivy_lexer_symbol_node;
 struct ivy_lexer_state;
 struct ivy_lexer;
+struct ivy_diag_ctx;
 
 IVY_API enum ivy_status ivy_lexer_create(struct ivy_lexer **lex);
 IVY_API void ivy_lexer_destroy(struct ivy_lexer *lex);
 
 IVY_API void ivy_lexer_set_source(
 	struct ivy_lexer *lex, struct ivy_line_source *src);
+IVY_API void ivy_lexer_set_diag_ctx(struct ivy_lexer *lex, struct ivy_diag_ctx *ctx);
 IVY_API enum ivy_status ivy_lexer_get_status(struct ivy_lexer *lex);
 
 IVY_API struct ivy_token *ivy_lexer_peek(struct ivy_lexer *lex);
diff --git a/lang/lex.c b/lang/lex.c
index 925929d..a5b4d77 100644
--- a/lang/lex.c
+++ b/lang/lex.c
@@ -6,6 +6,8 @@
 #include <blue/object/number.h>
 #include <blue/object/string.h>
 #include <ctype.h>
+#include <ivy/diag.h>
+#include <ivy/lang/diag.h>
 #include <ivy/lang/lex.h>
 #include <stdbool.h>
 #include <stdio.h>
@@ -338,6 +340,11 @@ void ivy_lexer_set_source(struct ivy_lexer *lex, struct ivy_line_source *src)
 	lex->lex_source = src;
 }
 
+void ivy_lexer_set_diag_ctx(struct ivy_lexer *lex, struct ivy_diag_ctx *ctx)
+{
+	lex->lex_diag_ctx = ctx;
+}
+
 enum ivy_status ivy_lexer_get_status(struct ivy_lexer *lex)
 {
 	return lex->lex_status;
@@ -728,6 +735,11 @@ static enum ivy_status read_atom(struct ivy_lexer *lex)
 		advance(lex);
 	}
 
+	if (b_string_get_size(str, B_STRLEN_NORMAL) == 0) {
+		/* TODO report empty atom error */
+		return IVY_ERR_BAD_SYNTAX;
+	}
+
 	char *s = b_string_steal(str);
 	return push_atom(lex, s);
 }
@@ -998,6 +1010,25 @@ static enum ivy_status read_ident(struct ivy_lexer *lex)
 	return push_token(lex, tok);
 }
 
+static void report_unrecognised_char(struct ivy_lexer *lex, int c)
+{
+	struct ivy_diag *diag = ivy_diag_ctx_create_diag(
+		lex->lex_diag_ctx, IVY_LANG_E_UNRECOGNISED_SYMBOL);
+
+	ivy_diag_set_location(diag, lex->lex_cursor_row, lex->lex_cursor_col);
+	ivy_diag_push_msg(diag, IVY_LANG_MSG_UNKNOWN_SYMBOL_ENCOUNTERED);
+
+	const struct ivy_diag_highlight hl[] = {
+		IVY_DIAG_HL(
+			ERROR, lex->lex_cursor_row, lex->lex_cursor_col,
+			lex->lex_cursor_row, lex->lex_cursor_col),
+	};
+	const size_t nr_hl = sizeof hl / sizeof hl[0];
+
+	ivy_diag_push_snippet(
+		diag, lex->lex_cursor_row, lex->lex_cursor_row, NULL, 0, hl, nr_hl);
+}
+
 static enum ivy_status pump_tokens(struct ivy_lexer *lex)
 {
 	struct lexer_state *state = get_lexer_state(lex);
@@ -1059,6 +1090,7 @@ static enum ivy_status pump_tokens(struct ivy_lexer *lex)
 		return read_number(lex);
 	}
 
+	report_unrecognised_char(lex, c);
 	return IVY_ERR_BAD_SYNTAX;
 }
 
diff --git a/lang/lex.h b/lang/lex.h
index 0a91498..3aa951a 100644
--- a/lang/lex.h
+++ b/lang/lex.h
@@ -10,6 +10,7 @@
 
 struct ivy_lexer {
 	struct ivy_lexer_symbol_node *lex_sym_tree;
+	struct ivy_diag_ctx *lex_diag_ctx;
 	struct ivy_line_source *lex_source;
 	b_dict *lex_keywords;
 	enum ivy_status lex_status;