Merge pull request #1104 from Qwinci/binary-printf

options/ansi: Handle binary specifiers in format strings
managarm · Aug 20, 2024 · 295a609 · 295a609
2 parents 2eded87 + 9b3c2cd
commit 295a609
Show file tree

Hide file tree

Showing 5 changed files with 76 additions and 1 deletion.
diff --git a/options/ansi/generic/stdio-stubs.cpp b/options/ansi/generic/stdio-stubs.cpp
@@ -54,7 +54,7 @@ struct PrintfAgent {
 		case 'p': case 's':
 			frg::do_printf_chars(*_formatter, t, opts, szmod, _vsp);
 			break;
-		case 'd': case 'i': case 'o': case 'x': case 'X': case 'u':
+		case 'd': case 'i': case 'o': case 'x': case 'X': case 'b': case 'B': case 'u':
 			frg::do_printf_ints(*_formatter, t, opts, szmod, _vsp);
 			break;
 		case 'f': case 'F': case 'g': case 'G': case 'e': case 'E':
@@ -596,6 +596,34 @@ static int do_scanf(H &handler, const char *fmt, __builtin_va_list args) {
 					store_int(dest, type, res);
 				break;
 			}
+			case 'b': {
+				unsigned long long res = 0;
+				char c = handler.look_ahead();
+				int count = 0;
+				EOF_CHECK(c == '\0');
+				if (c == '0') {
+					handler.consume();
+					c = handler.look_ahead();
+					if (c == 'b' || c == 'B') {
+						handler.consume();
+						c = handler.look_ahead();
+					}
+				}
+				while (true) {
+					if (c == '0' || c == '1') {
+						handler.consume();
+						res = res * 2 + (c - '0');
+					} else {
+						break;
+					}
+					count++;
+					c = handler.look_ahead();
+				}
+				NOMATCH_CHECK(count == 0);
+				if (dest)
+					store_int(dest, type, res);
+				break;
+			}
 			case 's': {
 				char c = handler.look_ahead();
 				EOF_CHECK(c == '\0');

diff --git a/options/internal/include/mlibc/strtol.hpp b/options/internal/include/mlibc/strtol.hpp
@@ -83,6 +83,7 @@ Return stringToInteger(const Char *__restrict nptr, Char **__restrict endptr, in
 
 	bool hasOctalPrefix = s[0] == widen<Char>('0');
 	bool hasHexPrefix = hasOctalPrefix && (s[1] == widen<Char>('x') || s[1] == widen<Char>('X'));
+	bool hasBinPrefix = hasOctalPrefix && (s[1] == widen<Char>('b') || s[1] == widen<Char>('B'));
 
 	// There's two tricky cases we need to keep in mind here:
 	//   1. We should interpret "0x5" as hex 5 rather than octal 0.
@@ -91,6 +92,9 @@ Return stringToInteger(const Char *__restrict nptr, Char **__restrict endptr, in
 	if ((base == 0 || base == 16) && hasHexPrefix && char_detail<Char>::isHexDigit(s[2])) {
 		s += 2;
 		base = 16;
+	} else if ((base == 0 || base == 2) && hasBinPrefix) {
+		s += 2;
+		base = 2;
 	} else if ((base == 0 || base == 8) && hasOctalPrefix) {
 		base = 8;
 	} else if (base == 0) { 

diff --git a/tests/ansi/sprintf.c b/tests/ansi/sprintf.c
@@ -110,13 +110,27 @@ int main() {
 	assert(!strcmp(buf, "0XC"));
 	sprintf(buf, "%#o", 12);
 	assert(!strcmp(buf, "014"));
+
 	sprintf(buf, "%#x", 0);
 	assert(!strcmp(buf, "0"));
 	sprintf(buf, "%#X", 0);
 	assert(!strcmp(buf, "0"));
 	sprintf(buf, "%#o", 0);
 	assert(!strcmp(buf, "0"));
 
+	// Disable -Wformat here because the compiler might not know about the b specifier.
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wformat"
+	sprintf(buf, "%#b", 12);
+	assert(!strcmp(buf, "0b1100"));
+	sprintf(buf, "%#B", 12);
+	assert(!strcmp(buf, "0B1100"));
+	sprintf(buf, "%#b", 0);
+	assert(!strcmp(buf, "0"));
+	sprintf(buf, "%#B", 0);
+	assert(!strcmp(buf, "0"));
+#pragma GCC diagnostic pop
+
 	// Test 'd' with different size mods to see
 	// if they work
 	sprintf(buf, "%d", 12);
@@ -177,6 +191,25 @@ int main() {
 	sprintf(buf, "%hho", (unsigned char) 12);
 	assert(!strcmp(buf, "14"));
 
+	// Disable -Wformat here because the compiler might not know about the b specifier.
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wformat"
+	// Test 'b' with different size mods to see
+	// if they work
+	sprintf(buf, "%b", 12);
+	assert(!strcmp(buf, "1100"));
+	sprintf(buf, "%lb", 12L);
+	assert(!strcmp(buf, "1100"));
+	sprintf(buf, "%llb", 12LL);
+	assert(!strcmp(buf, "1100"));
+	sprintf(buf, "%zb", (size_t)12);
+	assert(!strcmp(buf, "1100"));
+	sprintf(buf, "%hb", (unsigned short) 12);
+	assert(!strcmp(buf, "1100"));
+	sprintf(buf, "%hhb", (unsigned char) 12);
+	assert(!strcmp(buf, "1100"));
+#pragma GCC diagnostic pop
+
 	// Test %n$ syntax.
 	sprintf(buf, "%1$d", 12);
 	assert(!strcmp(buf, "12"));

diff --git a/tests/ansi/sscanf.c b/tests/ansi/sscanf.c
@@ -24,6 +24,10 @@ struct format_test_cases {
 	{"%u", "0420", 420, T_UINT, 1},
 	{"%o", "0420", 0420, T_UINT, 1},
 	{"%x", "0xCB7", 0xCB7, T_UINT, 1},
+#ifndef USE_HOST_LIBC
+	{"%b", "0b1011", 0b1011, T_UINT, 1},
+	{"%b", "0B1011", 0b1011, T_UINT, 1},
+#endif
 	{"%%", "%", 0, T_NONE, 0},
 	{"%c", "         I am not a fan of this solution.", ' ', T_CHAR, 1},
 	{" %c", "           CBT (capybara therapy)", 'C', T_CHAR, 1},

diff --git a/tests/ansi/strtol.c b/tests/ansi/strtol.c
@@ -48,6 +48,12 @@ int main () {
 	DO_TEST("-1101110100110100100000", -3624224, -1, strtol, 2);
 	DO_TEST("0x6fffff", 0x6fffff, -1, strtol, 0);
 	DO_TEST("0666", 0666, -1, strtol, 0);
+#ifndef USE_HOST_LIBC
+	DO_TEST("1100", 0b1100, -1, strtol, 2);
+	DO_TEST("0b1100", 0b1100, -1, strtol, 0);
+	DO_TEST("0B1100", 0b1100, -1, strtol, 0);
+	DO_TEST("0b1zzz", 1, 3, strtol, 0);
+#endif
 	DO_TEST("0xzzz", 0, 1, strtol, 0);
 	DO_TEST("0yzzz", 0, 1, strtol, 0);
 	DO_TEST("00xzz", 0, 2, strtol, 0);