Reverse UTF-8 symbol removal for Windows.

rizinorg · Jul 25, 2022 · 0705e11 · 0705e11
1 parent d64f53f
commit 0705e11
Show file tree

Hide file tree

Showing 3 changed files with 5 additions and 34 deletions.
diff --git a/librz/asm/arch/hexagon/hexagon_arch.h b/librz/asm/arch/hexagon/hexagon_arch.h
@@ -43,31 +43,18 @@ typedef struct {
 #define HEX_PKT_UNK         "?   "
 #define HEX_PKT_SINGLE      "[   "
 #define HEX_PKT_SINGLE_UTF8 "[   "
-#if __WINDOWS__
-// regex matching doesn't work on Windows with UTF-8 characters
-#define HEX_PKT_FIRST_UTF8 "/   "
-#define HEX_PKT_MID_UTF8   "|   "
-#define HEX_PKT_LAST_UTF8  "\\   "
-#else
 #define HEX_PKT_FIRST_UTF8 "┌   "
 #define HEX_PKT_MID_UTF8   "│   "
 #define HEX_PKT_LAST_UTF8  "└   "
-#endif
 #define HEX_PKT_FIRST_SDK   "{   "
 #define HEX_PKT_SDK_PADDING "    "
 #define HEX_PKT_LAST_SDK    " }"
 #define HEX_PKT_FIRST       "/   "
 #define HEX_PKT_MID         "|   "
 #define HEX_PKT_LAST        "\\   "
-#if __WINDOWS__
-#define HEX_PKT_ELOOP_01_UTF8 "     < endloop01"
-#define HEX_PKT_ELOOP_1_UTF8  "     < endloop1"
-#define HEX_PKT_ELOOP_0_UTF8  "     < endloop0"
-#else
 #define HEX_PKT_ELOOP_01_UTF8 "     ∎ endloop01"
 #define HEX_PKT_ELOOP_1_UTF8  "     ∎ endloop1"
 #define HEX_PKT_ELOOP_0_UTF8  "     ∎ endloop0"
-#endif
 #define HEX_PKT_ELOOP_01     "     < endloop01"
 #define HEX_PKT_ELOOP_1      "     < endloop1"
 #define HEX_PKT_ELOOP_0      "     < endloop0"

diff --git a/librz/asm/p/asm_hexagon.c b/librz/asm/p/asm_hexagon.c
@@ -30,27 +30,15 @@ static RZ_OWN RzPVector /* RzAsmTokenPattern */ *get_token_patterns() {
 	RzAsmTokenPattern *pat = RZ_NEW0(RzAsmTokenPattern);
 	pat->type = RZ_ASM_TOKEN_META;
 	pat->pattern = strdup(
-#if __WINDOWS__
-		// regex matching doesn't work on Windows with UTF-8 characters
-		"(^[\\[\\?\\/\\|\\\\\\{])|"
-		"[<\\}][ :](endloop[01]{1,2})"
-#else
 		"(^[\\[\\?\\/\\|\\\\\\{])|(┌)|(│)|(└)|" // Packet prefix
 		"((∎)|[<\\}])([ :])(endloop[01]{1,2})" // Endloop markers
-#endif
-	);
-	rz_pvector_push(pvec, pat);
-
-	pat = RZ_NEW0(RzAsmTokenPattern);
-	pat->type = RZ_ASM_TOKEN_META;
-	pat->pattern = strdup(
-		"(#{1,2})|(\\}$)" // Immediate prefix, Closing packet bracket
 	);
 	rz_pvector_push(pvec, pat);
 
 	pat = RZ_NEW0(RzAsmTokenPattern);
 	pat->type = RZ_ASM_TOKEN_META;
 	pat->pattern = strdup(
+		"(#{1,2})|(\\}$)|" // Immediate prefix, Closing packet bracket
 		"\\.new|:n?t|:raw|<err>" // .new and jump hints
 	);
 	rz_pvector_push(pvec, pat);

diff --git a/test/unit/test_tokens.c b/test/unit/test_tokens.c
@@ -538,22 +538,18 @@ static bool test_rz_colorize_custom_hexagon_2(void) {
 	RzStrBuf *colored_asm;
 	RzStrBuf *expected;
 	char err_msg[2048];
+	// ?   memd(R0++#0x8) = R19:18
+	// ┌   R7:6 = valignb(R13:12,R11:10,P2)
+	// │   P0 = cmp.gtu(R4,##0x1)
+	// │   R11:10 = memd(R1++#0x8)
 	// └   memd(R0++#0x8) = R7:6     ∎ endloop0
 	ut8 buf[] = "\x08\xd2\xc0\xab\x46\x8c\x0a\xc2\x20\x40\x84\x75\x2a\x40\xc1\x9b\x08\xc6\xc0\xab";
 	const char *expected_str[] = {
-#if __WINDOWS__
-		"\x1b[90m[\x1b[0m\x1b[37m   \x1b[0m\x1b[37mmemd\x1b[0m\x1b[37m(\x1b[0m\x1b[36mR0\x1b[0m\x1b[37m+\x1b[0m\x1b[37m+\x1b[0m\x1b[90m#\x1b[0m\x1b[33m0x8\x1b[0m\x1b[37m)\x1b[0m\x1b[37m \x1b[0m\x1b[37m=\x1b[0m\x1b[37m \x1b[0m\x1b[36mR19:18\x1b[0m",
-		"\x1b[90m/\x1b[0m\x1b[37m   \x1b[0m\x1b[36mR7:6\x1b[0m\x1b[37m \x1b[0m\x1b[37m=\x1b[0m\x1b[37m \x1b[0m\x1b[37mvalignb\x1b[0m\x1b[37m(\x1b[0m\x1b[36mR13:12\x1b[0m\x1b[37m,\x1b[0m\x1b[36mR11:10\x1b[0m\x1b[37m,\x1b[0m\x1b[36mP2\x1b[0m\x1b[37m)\x1b[0m",
-		"\x1b[90m|\x1b[0m\x1b[37m   \x1b[0m\x1b[36mP0\x1b[0m\x1b[37m \x1b[0m\x1b[37m=\x1b[0m\x1b[37m \x1b[0m\x1b[37mcmp\x1b[0m\x1b[37m.\x1b[0m\x1b[37mgtu\x1b[0m\x1b[37m(\x1b[0m\x1b[36mR4\x1b[0m\x1b[37m,\x1b[0m\x1b[90m##\x1b[0m\x1b[33m0x1\x1b[0m\x1b[37m)\x1b[0m",
-		"\x1b[90m|\x1b[0m\x1b[37m   \x1b[0m\x1b[36mR11:10\x1b[0m\x1b[37m \x1b[0m\x1b[37m=\x1b[0m\x1b[37m \x1b[0m\x1b[37mmemd\x1b[0m\x1b[37m(\x1b[0m\x1b[36mR1\x1b[0m\x1b[37m+\x1b[0m\x1b[37m+\x1b[0m\x1b[90m#\x1b[0m\x1b[33m0x8\x1b[0m\x1b[37m)\x1b[0m",
-		"\x1b[90m\\\x1b[0m\x1b[37m   \x1b[0m\x1b[32mmemd\x1b[0m\x1b[37m(\x1b[0m\x1b[36mR0\x1b[0m\x1b[37m+\x1b[0m\x1b[37m+\x1b[0m\x1b[90m#\x1b[0m\x1b[33m0x8\x1b[0m\x1b[37m)\x1b[0m\x1b[37m \x1b[0m\x1b[37m=\x1b[0m\x1b[37m \x1b[0m\x1b[36mR7:6\x1b[0m\x1b[37m     \x1b[0m\x1b[90m< endloop0\x1b[0m"
-#else
 		"\x1b[90m[\x1b[0m\x1b[37m   \x1b[0m\x1b[37mmemd\x1b[0m\x1b[37m(\x1b[0m\x1b[36mR0\x1b[0m\x1b[37m+\x1b[0m\x1b[37m+\x1b[0m\x1b[90m#\x1b[0m\x1b[33m0x8\x1b[0m\x1b[37m)\x1b[0m\x1b[37m \x1b[0m\x1b[37m=\x1b[0m\x1b[37m \x1b[0m\x1b[36mR19:18\x1b[0m",
 		"\x1b[90m┌\x1b[0m\x1b[37m   \x1b[0m\x1b[36mR7:6\x1b[0m\x1b[37m \x1b[0m\x1b[37m=\x1b[0m\x1b[37m \x1b[0m\x1b[37mvalignb\x1b[0m\x1b[37m(\x1b[0m\x1b[36mR13:12\x1b[0m\x1b[37m,\x1b[0m\x1b[36mR11:10\x1b[0m\x1b[37m,\x1b[0m\x1b[36mP2\x1b[0m\x1b[37m)\x1b[0m",
 		"\x1b[90m│\x1b[0m\x1b[37m   \x1b[0m\x1b[36mP0\x1b[0m\x1b[37m \x1b[0m\x1b[37m=\x1b[0m\x1b[37m \x1b[0m\x1b[37mcmp\x1b[0m\x1b[37m.\x1b[0m\x1b[37mgtu\x1b[0m\x1b[37m(\x1b[0m\x1b[36mR4\x1b[0m\x1b[37m,\x1b[0m\x1b[90m##\x1b[0m\x1b[33m0x1\x1b[0m\x1b[37m)\x1b[0m",
 		"\x1b[90m│\x1b[0m\x1b[37m   \x1b[0m\x1b[36mR11:10\x1b[0m\x1b[37m \x1b[0m\x1b[37m=\x1b[0m\x1b[37m \x1b[0m\x1b[37mmemd\x1b[0m\x1b[37m(\x1b[0m\x1b[36mR1\x1b[0m\x1b[37m+\x1b[0m\x1b[37m+\x1b[0m\x1b[90m#\x1b[0m\x1b[33m0x8\x1b[0m\x1b[37m)\x1b[0m",
 		"\x1b[90m└\x1b[0m\x1b[37m   \x1b[0m\x1b[32mmemd\x1b[0m\x1b[37m(\x1b[0m\x1b[36mR0\x1b[0m\x1b[37m+\x1b[0m\x1b[37m+\x1b[0m\x1b[90m#\x1b[0m\x1b[33m0x8\x1b[0m\x1b[37m)\x1b[0m\x1b[37m \x1b[0m\x1b[37m=\x1b[0m\x1b[37m \x1b[0m\x1b[36mR7:6\x1b[0m\x1b[37m     \x1b[0m\x1b[90m∎ endloop0\x1b[0m"
-#endif
 	};
 
 	for (int i = 0; i < 0x14; i += 4) {