Better error messages in the lexer

7 files changed, 148 insertions, 36 deletions

diff --git a/src/c8asm/common.h b/src/c8asm/common.h
index f5899f0..49af0f7 100644
--- a/src/c8asm/common.h
+++ b/src/c8asm/common.h
@@ -6,5 +6,6 @@
 extern size_t filesize;
 extern const char *filename;
 extern const char8_t *baseptr;
+extern struct u8view filebuf;
 
 #endif /* !AHOY_C8ASM_COMMON_H */
diff --git a/src/c8asm/lexer.c b/src/c8asm/lexer.c
index 04fb8ad..3d2a3bf 100644
--- a/src/c8asm/lexer.c
+++ b/src/c8asm/lexer.c
@@ -10,12 +10,15 @@
 #define ISDIGIT(n)   ((n) >= '0' && (n) <= '9')
 #define U8MOV(sv, n) ((sv)->p += (n), (sv)->len -= (n))
 
-#define E_BASE         "integer with invalid base specifier ‘%.*s’"
-#define E_EXTRA        "unknown extraneous character ‘%.*s’"
-#define E_IDENTCHAR    "illegal character in identifier ‘%.*s’"
+#define die_at_pos_with_code(HL, OFF, ...) \
+	die_at_pos_with_code(filename, filebuf, (HL), (OFF), __VA_ARGS__)
+
+#define E_BASE         "integer with invalid base specifier"
+#define E_EXTRA        "unknown extraneous character"
+#define E_IDENTCCHAR   "illegal character in identifier"
 #define E_IDENTLOST    "local label missing identifier"
-#define E_IDENTSCHAR   "illegal first character in identifier ‘%.*s’"
-#define E_UNTERMINATED "unterminated string literal ‘%.*s%.*s’"
+#define E_IDENTSCHAR   "illegal first character in identifier"
+#define E_UNTERMINATED "unterminated string literal"
 #define E_UTF8         "invalid UTF-8 byte near ‘%02X’"
 
 #define EOLS     U"\n\v\f\r\x85\u2028\u2029"
@@ -35,10 +38,11 @@ tokrepr(tokkind k)
 }
 
 struct tokens
-lexfile(struct u8view sv)
+lexfile(void)
 {
 	const char8_t *s;
 	struct tokens toks;
+	struct u8view sv = filebuf;
 
 	if (s = u8chk(sv.p, sv.len))
 		die_with_off(filename, s - sv.p, E_UTF8, *s);
@@ -67,8 +71,8 @@ lexfile(struct u8view sv)
 void
 lexline(struct tokens *toks, struct u8view *sv)
 {
-#define die_with_off(...) \
-	die_with_off(filename, sv->p - baseptr - w, __VA_ARGS__);
+#define _die_at_pos_with_code(HL, ...) \
+	die_at_pos_with_code((HL), sv->p - baseptr - w, __VA_ARGS__)
 
 	struct token tok;
 
@@ -113,7 +117,7 @@ lexline(struct tokens *toks, struct u8view *sv)
 					break;
 				default:
 					if (!ISDIGIT(ch))
-						die_with_off(E_BASE, w, sv->p - w);
+						_die_at_pos_with_code(tok.sv, E_BASE);
 				}
 			}
 
@@ -124,14 +128,12 @@ out:
 		} else if (ch == '.' || ch == '_' || rprop_is_xids(ch)) {
 			tok.kind = T_IDENT;
 			if (ch == '.') {
-				if (!sv->len)
-					die_with_off(E_IDENTLOST);
-
 				tok.sv.len += w = u8next(&ch, &sv->p, &sv->len);
-				if (rprop_is_pat_ws(ch))
-					die_with_off(E_IDENTLOST);
+				if (!w || rprop_is_pat_ws(ch))
+					_die_at_pos_with_code(tok.sv, E_IDENTLOST);
 				if (ch != '_' && !rprop_is_xids(ch)) {
-					die_with_off(E_IDENTSCHAR, w, sv->p - w);
+					U8MOV(&tok.sv, 1);
+					_die_at_pos_with_code(tok.sv, E_IDENTSCHAR);
 				}
 			}
 
@@ -140,8 +142,13 @@ out:
 					U8MOV(sv, -w);
 					break;
 				}
-				if (!rprop_is_xidc(ch))
-					die_with_off(E_IDENTCHAR, w, sv->p - w);
+				if (!rprop_is_xidc(ch)) {
+					struct u8view hl = {
+						.p = sv->p - w,
+						.len = w,
+					};
+					_die_at_pos_with_code(hl, E_IDENTCCHAR);
+				}
 
 				tok.sv.len += w;
 			}
@@ -152,22 +159,24 @@ out:
 				if (ch == '"')
 					goto found;
 			}
-			die_with_off(E_UNTERMINATED, (int)MIN(tok.sv.len, 20), tok.sv.p,
-			             tok.sv.len > 20 ? (int)lengthof(u8"…") - 1 : 0, u8"…");
+			_die_at_pos_with_code(tok.sv, E_UNTERMINATED);
 found:
 		} else if (ch == ':') {
 			tok.kind = T_COLON;
 		} else if (ch == ';') {
 			goto end;
 		} else {
-			die_with_off(E_EXTRA, w, sv->p - w);
+			struct u8view hl = {.p = sv->p - w, .len = w};
+			_die_at_pos_with_code(hl, E_EXTRA);
 		}
 
 		/* The colon is the only token that isn’t whitespace separated */
 		if (ch != ':' && sv->len) {
 			w = u8next(&ch, &sv->p, &sv->len);
-			if (!w || !rprop_is_pat_ws(ch))
-				die_with_off(E_EXTRA, w, sv->p - w);
+			if (!w || !rprop_is_pat_ws(ch)) {
+				struct u8view hl = {.p = sv->p - w, .len = w};
+				_die_at_pos_with_code(hl, E_EXTRA);
+			}
 		}
 
 		dapush(toks, tok);
@@ -181,7 +190,7 @@ end:;
 	};
 	dapush(toks, tok);
 
-#undef die_with_off
+#undef _die_at_pos_with_code
 }
 
 bool
diff --git a/src/c8asm/lexer.h b/src/c8asm/lexer.h
index ef20cef..46692a8 100644
--- a/src/c8asm/lexer.h
+++ b/src/c8asm/lexer.h
@@ -23,6 +23,6 @@ struct tokens {
 };
 
 const char *tokrepr(tokkind);
-struct tokens lexfile(struct u8view);
+struct tokens lexfile();
 
 #endif /* !AHOY_C8ASM_LEXER_H */
diff --git a/src/c8asm/main.c b/src/c8asm/main.c
index 5ee9c9e..b286778 100644
--- a/src/c8asm/main.c
+++ b/src/c8asm/main.c
@@ -21,6 +21,7 @@ static void asmfile(int, const char *);
 size_t filesize;
 const char *filename;
 const char8_t *baseptr;
+struct u8view filebuf;
 
 int
 main(int argc, char **argv)
@@ -100,8 +101,9 @@ asmfile(int fd, const char *fn)
 
 	free(buf);
 	filesize = sb.len;
+	filebuf = u8strtou8(sb);
 	baseptr = u8strfit(&sb)->p;
-	assemble(stdout, ast = parsefile(toks = lexfile(u8strtou8(sb))));
+	assemble(stdout, ast = parsefile(toks = lexfile()));
 
 	da_foreach (&ast, node) {
 		if (node->kind == D_INSTR && node->instr.kind == I_DB)
diff --git a/src/c8asm/parser.c b/src/c8asm/parser.c
index e0f3660..5a68fad 100644
--- a/src/c8asm/parser.c
+++ b/src/c8asm/parser.c
@@ -206,14 +206,20 @@ struct token
 reqnext(const char *want, tokkind msk)
 {
 	struct token t;
+
 	if (i >= tokens->len)
 		die_with_off(baseptr + filesize - 1, E_EARLY, want);
 
 	if ((t = tokens->buf[i++]).kind & msk)
 		return t;
-	if (t.kind == T_EOL)
-		die_with_off(t.sv.p, E_EXPECTED2, want, tokrepr(t.kind));
-	die_with_off(t.sv.p, E_EXPECTED, want, tokrepr(t.kind), U8_PRI_ARGS(t.sv));
+	if (t.kind == T_EOL) {
+		die_at_pos_with_code(filename, filebuf, (struct u8view){},
+		                     t.sv.p - baseptr, E_EXPECTED2, want,
+		                     tokrepr(t.kind));
+	}
+
+	die_at_pos_with_code(filename, filebuf, t.sv, t.sv.p - baseptr, E_EXPECTED,
+	                     want, tokrepr(t.kind), U8_PRI_ARGS(t.sv));
 }
 
 #define I(...) ((struct dir){.kind = D_INSTR, .instr = (__VA_ARGS__)})
diff --git a/src/common/cerr.c b/src/common/cerr.c
index 8570247..819df4a 100644
--- a/src/common/cerr.c
+++ b/src/common/cerr.c
@@ -1,19 +1,33 @@
 #include <errno.h>
 #include <stdarg.h>
+#include <stddef.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
 
+#include <mbstring.h>
+
 #include "cerr.h"
+#include "macros.h"
+#include "mbstring.h"
+
+#define TAB_AS_SPC "        "
+
+#define EOLS8  u8"\n\v\f\r\x85\u2028\u2029"
+#define EOLS32 U"\n\v\f\r\x85\u2028\u2029"
 
 #define SGR_BOLD "\33[1m"
 #define SGR_DONE "\33[0m"
+#define SGR_WARN "\33[1;35m"
+#define SGR_ERR  "\33[1;31m"
+
+int sizelen(size_t);
 
 static bool color;
 static const char *progname;
 
-static const char *_bold, *_done;
+static const char *_bold, *_done, *_warn, *_err;
 
 void
 cerrinit(const char *s)
@@ -23,17 +37,16 @@ cerrinit(const char *s)
 
 	if (isatty(STDOUT_FILENO)) {
 		const char *ev = getenv("NO_COLOR");
-		if (!ev || !*ev)
-			color = true;
+		color = !ev || !*ev;
 	}
 
 	if (color) {
 		_bold = SGR_BOLD;
 		_done = SGR_DONE;
-	} else {
-		_bold = "";
-		_done = "";
-	}
+		_warn = SGR_WARN;
+		_err = SGR_ERR;
+	} else
+		_bold = _done = _warn = _err = "";
 }
 
 void
@@ -71,10 +84,84 @@ die_with_off(const char *file, size_t off, const char *fmt, ...)
 	va_list ap;
 
 	va_start(ap, fmt);
-	fprintf(stderr, "%s%s:%s:%zu:%s ", _bold, progname, file, off, _done);
+	fprintf(stderr, "%s%s: %s:%zu:%s %serror:%s ", _bold, progname, file, off,
+	        _done, _err, _done);
+	vfprintf(stderr, fmt, ap);
+	fputc('\n', stderr);
+	va_end(ap);
+
+	exit(EXIT_FAILURE);
+}
+
+void
+die_at_pos_with_code(const char *file, struct u8view sv, struct u8view hl,
+                     size_t off, const char *fmt, ...)
+{
+	int w;
+	rune _;
+	size_t lb;
+	va_list ap;
+	const char8_t *prv, *end;
+
+	va_start(ap, fmt);
+	fprintf(stderr, "%s%s: %s:%zu:%s %serror:%s ", _bold, progname, file, off,
+	        _done, _err, _done);
 	vfprintf(stderr, fmt, ap);
 	fputc('\n', stderr);
 	va_end(ap);
 
+	for (lb = 0, prv = end = sv.p; end <= sv.p + off; lb++) {
+		prv = end;
+		end += u8cbspn(end, sv.p + sv.len - end, EOLS32, lengthof(EOLS32) - 1);
+		end += u8tor_uc(&_, end);
+	}
+
+	u8prev(&_, &end, prv);
+	w = sizelen(lb);
+	w = MAX(w, 4);
+
+	fprintf(stderr, " %*zu │ ", w, lb);
+
+	/* The following is really ugly, but it works! */
+	if (hl.p) {
+		ptrdiff_t w2 = hl.p - prv;
+		struct u8view pfx = {prv, w2};
+
+		for (ptrdiff_t i = 0; i < w2; i++) {
+			if (prv[i] == '\t')
+				fputs(TAB_AS_SPC, stderr);
+			else
+				fputc(prv[i], stderr);
+		}
+		fprintf(stderr, "%s%.*s%s%.*s\n", _err, U8_PRI_ARGS(hl), _done,
+		        (int)(end - (prv + w2 + hl.len)), prv + w2 + hl.len);
+		fprintf(stderr, " %*c │ ", w, ' ');
+
+		while (u8next(&_, &pfx.p, &pfx.len)) {
+			if (_ == '\t')
+				fputs(TAB_AS_SPC, stderr);
+			else
+				fputc(' ', stderr);
+		}
+
+		fprintf(stderr, "%s^", _err);
+		for (u8next(&_, &hl.p, &hl.len); u8next(&_, &hl.p, &hl.len);
+		     fputc('~', stderr))
+			;
+		fprintf(stderr, "%s\n", _done);
+	} else {
+		fprintf(stderr, "%.*s\n", (int)(end - prv), prv);
+		fprintf(stderr, " %*c │\n", w, ' ');
+	}
+
 	exit(EXIT_FAILURE);
 }
+
+int
+sizelen(size_t x)
+{
+	int n;
+	for (n = 0; x; x /= 10, n++)
+		;
+	return n;
+}
diff --git a/src/common/cerr.h b/src/common/cerr.h
index 869c9e2..5476f96 100644
--- a/src/common/cerr.h
+++ b/src/common/cerr.h
@@ -3,6 +3,8 @@
 
 #include <stddef.h>
 
+#include <mbstring.h>
+
 void cerrinit(const char *);
 
 [[noreturn, gnu::nonnull, gnu::format(printf, 1, 2)]]
@@ -14,4 +16,9 @@ void diex(const char *, ...);
 [[noreturn, gnu::nonnull, gnu::format(printf, 3, 4)]]
 void die_with_off(const char *, size_t, const char *, ...);
 
+[[noreturn, gnu::nonnull, gnu::format(printf, 5, 6)]]
+void
+die_at_pos_with_code(const char *, struct u8view, struct u8view, size_t,
+                     const char *, ...);
+
 #endif /* !AHOY_COMMON_CERR_H */