Group sequences of garbage characters (#1672)

src/asm/lexer.cpp

@@ -1657,6 +1657,12 @@ static void appendStringLiteral(std::string &str, bool raw) {
 
 static Token yylex_SKIP_TO_ENDC(); // forward declaration for yylex_NORMAL
 
+// Must stay in sync with the `switch` in `yylex_NORMAL`!
+static bool isGarbageCharacter(int c) {
+	return c != EOF && !continuesIdentifier(c)
+	       && (c == '\0' || !strchr("; \t~[](),+-*/|^=!<>:&%`\"\r\n\\", c));
+}
+
 static Token yylex_NORMAL() {
 	for (;;) {
 		int c = nextChar();
@@ -1982,10 +1988,21 @@ static Token yylex_NORMAL() {
 
 			// Do not report weird characters when capturing, it'll be done later
 			if (!lexerState->capturing) {
-				// TODO: try to group reportings
+				assume(isGarbageCharacter(c) || c == '#');
+				if (isGarbageCharacter(peek())) {
+					// At least two characters are garbage; group them into one error report
+					std::string garbage = printChar(c);
+					while (isGarbageCharacter(peek())) {
+						c = nextChar();
+						garbage += ", ";
+						garbage += printChar(c);
+					}
+					error("Unknown characters %s\n", garbage.c_str());
+				} else {
 					error("Unknown character %s\n", printChar(c));
 				}
 			}
+		}
 		lexerState->atLineStart = false;
 	}
 }

test/asm/garbage_sequence.asm

@@ -0,0 +1,3 @@
+assert 1 +# 1 == 2
+assert 2 '?* 2 == 4
+assert 3 **?''?##?? 3 == 27

test/asm/garbage_sequence.err

@@ -0,0 +1,11 @@
+error: garbage_sequence.asm(1):
+    Unknown character '#'
+error: garbage_sequence.asm(2):
+    Unknown characters ''', '?'
+error: garbage_sequence.asm(3):
+    Unknown characters '?', ''', ''', '?'
+error: garbage_sequence.asm(3):
+    Unknown character '#'
+error: garbage_sequence.asm(3):
+    Unknown characters '#', '?', '?'
+error: Assembly aborted (5 errors)!

test/asm/invalid-utf-8.err

@@ -1,5 +1,3 @@
 error: invalid-utf-8.asm(6) -> invalid-utf-8.asm::m(4):
-    Unknown character 0xCF
+    Unknown characters 0xCF, 0xD3
-error: invalid-utf-8.asm(6) -> invalid-utf-8.asm::m(4):
+error: Assembly aborted (1 error)!
-    Unknown character 0xD3
-error: Assembly aborted (2 errors)!