Add some more string test cases

include/util.hpp

@@ -3,13 +3,6 @@
 #ifndef RGBDS_UTIL_HPP
 #define RGBDS_UTIL_HPP
 
-#include <stddef.h>
-#include <stdint.h>
-#include <vector>
-
 char const *printChar(int c);
 
-// @return The number of bytes read, or 0 if invalid data was found
-size_t readUTF8Char(std::vector<int32_t> *dest, char const *src);
-
 #endif // RGBDS_UTIL_HPP

src/asm/charmap.cpp

@@ -11,6 +11,7 @@
 #include <unordered_map>
 #include <utility>
 
+#include "extern/utf8decoder.hpp"
 #include "helpers.hpp"
 #include "util.hpp"
 
@@ -183,6 +184,26 @@ std::vector<int32_t> charmap_Convert(std::string const &input) {
 	return output;
 }
 
+static size_t readUTF8Char(std::vector<int32_t> *dest, char const *src) {
+	uint32_t state = 0, codepoint;
+	size_t i = 0;
+
+	for (;;) {
+		if (decode(&state, &codepoint, src[i]) == 1) {
+			return 0;
+		}
+
+		if (dest) {
+			dest->push_back(src[i]);
+		}
+		i++;
+
+		if (state == 0) {
+			return i;
+		}
+	}
+}
+
 size_t charmap_ConvertNext(std::string_view &input, std::vector<int32_t> *output) {
 	// The goal is to match the longest mapping possible.
 	// For that, advance through the trie with each character read.

src/util.cpp

@@ -5,9 +5,6 @@
 #include <ctype.h>
 #include <stdint.h>
 #include <stdio.h>
-#include <vector>
-
-#include "extern/utf8decoder.hpp"
 
 char const *printChar(int c) {
 	// "'A'" + '\0': 4 bytes
@@ -53,23 +50,3 @@ char const *printChar(int c) {
 	buf[4] = '\0';
 	return buf;
 }
-
-size_t readUTF8Char(std::vector<int32_t> *dest, char const *src) {
-	uint32_t state = 0, codepoint;
-	size_t i = 0;
-
-	for (;;) {
-		if (decode(&state, &codepoint, src[i]) == 1) {
-			return 0;
-		}
-
-		if (dest) {
-			dest->push_back(src[i]);
-		}
-		i++;
-
-		if (state == 0) {
-			return i;
-		}
-	}
-}

test/asm/invalid-format.asm

@@ -15,3 +15,4 @@ println STRFMT("%q16s", "hello")
 
 println STRFMT("%#d", 42)
 println STRFMT("%.5d", 42)
+println STRFMT("%q5d", 42)

test/asm/invalid-format.err

@@ -26,4 +26,6 @@ error: invalid-format.asm(16):
     Formatting type 'd' with exact flag '#'
 error: invalid-format.asm(17):
     Formatting type 'd' with fractional width
-error: Assembly aborted (14 errors)!
+error: invalid-format.asm(18):
+    Formatting type 'd' with fractional precision
+error: Assembly aborted (15 errors)!

test/asm/invalid-format.out

@@ -11,3 +11,4 @@ hello
 hello
 42
 42
+42

test/asm/invalid-utf-8-strings.asm

@@ -21,3 +21,17 @@ println "\"{invalid}\" == \"{copy}\" ({d:n})"
 DEF mid1 EQUS STRSUB("{invalid}", 5, 2)
 DEF mid2 EQUS STRSUB("{invalid}", 9, 1)
 println "\"{mid2}{mid1}\""
+
+; characters:
+; 1: U+0041 A
+; 2: U+0020 space
+; 3: invalid byte 0xFE
+; 4: invalid byte 0x81
+; 5: invalid byte 0xFF
+; 6: U+0020 space
+; 7: U+0042 B
+REDEF invalid EQUS "A þÿ B"
+
+DEF n = strlen("{invalid}")
+DEF r = charlen("{invalid}")
+println "\"{#s:invalid}\": {d:n} != {d:r}"

test/asm/invalid-utf-8-strings.err

@@ -42,4 +42,12 @@ error: invalid-utf-8-strings.asm(22):
     STRSUB: Invalid UTF-8 byte 0xA3
 error: invalid-utf-8-strings.asm(22):
     STRSUB: Invalid UTF-8 byte 0xA4
-error: Assembly aborted (22 errors)!
+error: invalid-utf-8-strings.asm(35):
+    STRLEN: Invalid UTF-8 byte 0xFE
+error: invalid-utf-8-strings.asm(35):
+    STRLEN: Invalid UTF-8 byte 0x81
+error: invalid-utf-8-strings.asm(35):
+    STRLEN: Invalid UTF-8 byte 0xFF
+error: invalid-utf-8-strings.asm(36):
+    Input string is not valid UTF-8
+error: Assembly aborted (26 errors)!

test/asm/invalid-utf-8-strings.out

@@ -1,2 +1,3 @@
 "aäb漢,a��b��!" == "aäb漢,a��b��!" (12)
 "b,a"
+"A ��� B": 7 != 2

test/asm/string-formatting.asm

@@ -4,11 +4,13 @@ def f equ -123.0456
 def pi equ 3.14159
 def s equs "hello"
 def t equs "\"\\t\" is '\t'"
+def u equs "\t\r\0\n"
 
 	println "<{ -6d:n}> <{+06u:n}> <{5x:n}> <{#16b:n}>"
 	println "<{u:m}> <{+3d:m}> <{#016o:m}>"
 	println "<{f:pi}> <{06.f:f}> <{.10f:f}>"
 	println "\"{#-20s:t}\", \"{#20s:t}\", \"{20s:t}\""
+	println "{#s:u}"
 
 macro foo
 	println "\1 <{\1}>"

test/asm/string-formatting.out

@@ -2,4 +2,5 @@
 <4294967254> <-42> <&000037777777726>
 <3.14159> <-00123> <-123.0455932617>
 "\"\\t\" is '\t'     ", "     \"\\t\" is '\t'", "         "\t" is '	'"
+\t\r\0\n
 -6d:n <300   >