[libc] Error fixes for mbrtowc and wcrtomb (#145785)
Invalid mbstate_t should set errno to EINVAL. Changed Error return for the internal functions and added tests for the public functions. Co-authored-by: Sriya Pratipati <sriyap@google.com>
This commit is contained in:
@@ -31,10 +31,12 @@ add_libc_test(
|
||||
mbrtowc_test.cpp
|
||||
DEPENDS
|
||||
libc.src.__support.libc_errno
|
||||
libc.src.__support.wchar.mbstate
|
||||
libc.src.string.memset
|
||||
libc.src.wchar.mbrtowc
|
||||
libc.hdr.types.mbstate_t
|
||||
libc.hdr.types.wchar_t
|
||||
libc.test.UnitTest.ErrnoCheckingTest
|
||||
)
|
||||
|
||||
add_libc_test(
|
||||
@@ -72,6 +74,8 @@ add_libc_test(
|
||||
libc.hdr.types.wchar_t
|
||||
libc.hdr.types.mbstate_t
|
||||
libc.src.__support.libc_errno
|
||||
libc.src.__support.wchar.mbstate
|
||||
libc.test.UnitTest.ErrnoCheckingTest
|
||||
)
|
||||
|
||||
add_libc_test(
|
||||
|
||||
@@ -9,11 +9,15 @@
|
||||
#include "hdr/types/mbstate_t.h"
|
||||
#include "hdr/types/wchar_t.h"
|
||||
#include "src/__support/libc_errno.h"
|
||||
#include "src/__support/wchar/mbstate.h"
|
||||
#include "src/string/memset.h"
|
||||
#include "src/wchar/mbrtowc.h"
|
||||
#include "test/UnitTest/ErrnoCheckingTest.h"
|
||||
#include "test/UnitTest/Test.h"
|
||||
|
||||
TEST(LlvmLibcMBRToWC, OneByte) {
|
||||
using LlvmLibcMBRToWCTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
|
||||
|
||||
TEST_F(LlvmLibcMBRToWCTest, OneByte) {
|
||||
const char *ch = "A";
|
||||
wchar_t dest[2];
|
||||
// Testing if it works with nullptr mbstate_t
|
||||
@@ -21,13 +25,15 @@ TEST(LlvmLibcMBRToWC, OneByte) {
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
|
||||
ASSERT_EQ(static_cast<char>(*dest), 'A');
|
||||
ASSERT_EQ(static_cast<int>(n), 1);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
|
||||
// Should fail since we have not read enough
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest, ch, 0, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), -2);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcMBRToWC, TwoByte) {
|
||||
TEST_F(LlvmLibcMBRToWCTest, TwoByte) {
|
||||
const char ch[2] = {static_cast<char>(0xC2),
|
||||
static_cast<char>(0x8E)}; // car symbol
|
||||
wchar_t dest[2];
|
||||
@@ -36,6 +42,7 @@ TEST(LlvmLibcMBRToWC, TwoByte) {
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 2, mb);
|
||||
ASSERT_EQ(static_cast<int>(*dest), 142);
|
||||
ASSERT_EQ(static_cast<int>(n), 2);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
|
||||
// Should fail since we have not read enough
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
|
||||
@@ -44,9 +51,10 @@ TEST(LlvmLibcMBRToWC, TwoByte) {
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest, ch + 1, 1, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), 1);
|
||||
ASSERT_EQ(static_cast<int>(*dest), 142);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcMBRToWC, ThreeByte) {
|
||||
TEST_F(LlvmLibcMBRToWCTest, ThreeByte) {
|
||||
const char ch[3] = {static_cast<char>(0xE2), static_cast<char>(0x88),
|
||||
static_cast<char>(0x91)}; // ∑ sigma symbol
|
||||
wchar_t dest[2];
|
||||
@@ -55,17 +63,20 @@ TEST(LlvmLibcMBRToWC, ThreeByte) {
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 3, mb);
|
||||
ASSERT_EQ(static_cast<int>(*dest), 8721);
|
||||
ASSERT_EQ(static_cast<int>(n), 3);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
|
||||
// Should fail since we have not read enough
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), -2);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
// Should pass after reading two more bytes
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest, ch + 1, 2, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), 2);
|
||||
ASSERT_EQ(static_cast<int>(*dest), 8721);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcMBRToWC, FourByte) {
|
||||
TEST_F(LlvmLibcMBRToWCTest, FourByte) {
|
||||
const char ch[4] = {static_cast<char>(0xF0), static_cast<char>(0x9F),
|
||||
static_cast<char>(0xA4),
|
||||
static_cast<char>(0xA1)}; // 🤡 clown emoji
|
||||
@@ -75,27 +86,29 @@ TEST(LlvmLibcMBRToWC, FourByte) {
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 4, mb);
|
||||
ASSERT_EQ(static_cast<int>(*dest), 129313);
|
||||
ASSERT_EQ(static_cast<int>(n), 4);
|
||||
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
// Should fail since we have not read enough
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest, ch, 2, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), -2);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
// Should pass after reading two more bytes
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest, ch + 2, 2, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), 2);
|
||||
ASSERT_EQ(static_cast<int>(*dest), 129313);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcMBRToWC, InvalidByte) {
|
||||
TEST_F(LlvmLibcMBRToWCTest, InvalidByte) {
|
||||
const char ch[1] = {static_cast<char>(0x80)};
|
||||
wchar_t dest[2];
|
||||
mbstate_t *mb;
|
||||
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), -1);
|
||||
ASSERT_EQ(static_cast<int>(libc_errno), EILSEQ);
|
||||
ASSERT_ERRNO_EQ(EILSEQ);
|
||||
}
|
||||
|
||||
TEST(LlvmLibcMBRToWC, InvalidMultiByte) {
|
||||
TEST_F(LlvmLibcMBRToWCTest, InvalidMultiByte) {
|
||||
const char ch[4] = {static_cast<char>(0x80), static_cast<char>(0x00),
|
||||
static_cast<char>(0x80),
|
||||
static_cast<char>(0x00)}; // invalid sequence of bytes
|
||||
@@ -105,18 +118,19 @@ TEST(LlvmLibcMBRToWC, InvalidMultiByte) {
|
||||
// Trying to push all 4 should error
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 4, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), -1);
|
||||
ASSERT_EQ(static_cast<int>(libc_errno), EILSEQ);
|
||||
ASSERT_ERRNO_EQ(EILSEQ);
|
||||
// Trying to push just the first one should error
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), -1);
|
||||
ASSERT_EQ(static_cast<int>(libc_errno), EILSEQ);
|
||||
ASSERT_ERRNO_EQ(EILSEQ);
|
||||
// Trying to push the second and third should correspond to null wc
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest, ch + 1, 2, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), 0);
|
||||
ASSERT_TRUE(*dest == L'\0');
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcMBRToWC, InvalidLastByte) {
|
||||
TEST_F(LlvmLibcMBRToWCTest, InvalidLastByte) {
|
||||
// Last byte is invalid since it does not have correct starting sequence.
|
||||
// 0xC0 --> 11000000 starting sequence should be 10xxxxxx
|
||||
const char ch[4] = {static_cast<char>(0xF1), static_cast<char>(0x80),
|
||||
@@ -127,10 +141,10 @@ TEST(LlvmLibcMBRToWC, InvalidLastByte) {
|
||||
// Trying to push all 4 should error
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 4, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), -1);
|
||||
ASSERT_EQ(static_cast<int>(libc_errno), EILSEQ);
|
||||
ASSERT_ERRNO_EQ(EILSEQ);
|
||||
}
|
||||
|
||||
TEST(LlvmLibcMBRToWC, ValidTwoByteWithExtraRead) {
|
||||
TEST_F(LlvmLibcMBRToWCTest, ValidTwoByteWithExtraRead) {
|
||||
const char ch[3] = {static_cast<char>(0xC2), static_cast<char>(0x8E),
|
||||
static_cast<char>(0x80)};
|
||||
wchar_t dest[2];
|
||||
@@ -140,9 +154,10 @@ TEST(LlvmLibcMBRToWC, ValidTwoByteWithExtraRead) {
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 3, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), 2);
|
||||
ASSERT_EQ(static_cast<int>(*dest), 142);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcMBRToWC, TwoValidTwoBytes) {
|
||||
TEST_F(LlvmLibcMBRToWCTest, TwoValidTwoBytes) {
|
||||
const char ch[4] = {static_cast<char>(0xC2), static_cast<char>(0x8E),
|
||||
static_cast<char>(0xC7), static_cast<char>(0x8C)};
|
||||
wchar_t dest[2];
|
||||
@@ -152,12 +167,14 @@ TEST(LlvmLibcMBRToWC, TwoValidTwoBytes) {
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 2, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), 2);
|
||||
ASSERT_EQ(static_cast<int>(*dest), 142);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest + 1, ch + 2, 2, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), 2);
|
||||
ASSERT_EQ(static_cast<int>(*(dest + 1)), 460);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcMBRToWC, NullString) {
|
||||
TEST_F(LlvmLibcMBRToWCTest, NullString) {
|
||||
wchar_t dest[2] = {L'O', L'K'};
|
||||
mbstate_t *mb;
|
||||
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
|
||||
@@ -165,8 +182,24 @@ TEST(LlvmLibcMBRToWC, NullString) {
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, nullptr, 2, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), 0);
|
||||
ASSERT_TRUE(dest[0] == L'O');
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
// reading a null terminator should return 0
|
||||
const char *ch = "\0";
|
||||
n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), 0);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST_F(LlvmLibcMBRToWCTest, InvalidMBState) {
|
||||
const char ch[4] = {static_cast<char>(0xC2), static_cast<char>(0x8E),
|
||||
static_cast<char>(0xC7), static_cast<char>(0x8C)};
|
||||
wchar_t dest[2] = {L'O', L'K'};
|
||||
mbstate_t *mb;
|
||||
LIBC_NAMESPACE::internal::mbstate inv;
|
||||
inv.total_bytes = 6;
|
||||
mb = reinterpret_cast<mbstate_t *>(&inv);
|
||||
// invalid mbstate should error
|
||||
size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 2, mb);
|
||||
ASSERT_EQ(static_cast<int>(n), -1);
|
||||
ASSERT_ERRNO_EQ(EINVAL);
|
||||
}
|
||||
|
||||
@@ -9,11 +9,15 @@
|
||||
#include "hdr/types/mbstate_t.h"
|
||||
#include "hdr/types/wchar_t.h"
|
||||
#include "src/__support/libc_errno.h"
|
||||
#include "src/__support/wchar/mbstate.h"
|
||||
#include "src/string/memset.h"
|
||||
#include "src/wchar/wcrtomb.h"
|
||||
#include "test/UnitTest/ErrnoCheckingTest.h"
|
||||
#include "test/UnitTest/Test.h"
|
||||
|
||||
TEST(LlvmLibcWCRToMBTest, OneByte) {
|
||||
using LlvmLibcWCRToMBTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
|
||||
|
||||
TEST_F(LlvmLibcWCRToMBTest, OneByte) {
|
||||
mbstate_t state;
|
||||
LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
|
||||
wchar_t wc = L'U';
|
||||
@@ -21,9 +25,10 @@ TEST(LlvmLibcWCRToMBTest, OneByte) {
|
||||
size_t cnt = LIBC_NAMESPACE::wcrtomb(mb, wc, &state);
|
||||
ASSERT_EQ(cnt, static_cast<size_t>(1));
|
||||
ASSERT_EQ(mb[0], 'U');
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcWCRToMBTest, TwoByte) {
|
||||
TEST_F(LlvmLibcWCRToMBTest, TwoByte) {
|
||||
mbstate_t state;
|
||||
LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
|
||||
// testing utf32: 0xff -> utf8: 0xc3 0xbf
|
||||
@@ -33,9 +38,10 @@ TEST(LlvmLibcWCRToMBTest, TwoByte) {
|
||||
ASSERT_EQ(cnt, static_cast<size_t>(2));
|
||||
ASSERT_EQ(mb[0], static_cast<char>(0xc3));
|
||||
ASSERT_EQ(mb[1], static_cast<char>(0xbf));
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcWCRToMBTest, ThreeByte) {
|
||||
TEST_F(LlvmLibcWCRToMBTest, ThreeByte) {
|
||||
mbstate_t state;
|
||||
LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
|
||||
// testing utf32: 0xac15 -> utf8: 0xea 0xb0 0x95
|
||||
@@ -46,9 +52,10 @@ TEST(LlvmLibcWCRToMBTest, ThreeByte) {
|
||||
ASSERT_EQ(mb[0], static_cast<char>(0xea));
|
||||
ASSERT_EQ(mb[1], static_cast<char>(0xb0));
|
||||
ASSERT_EQ(mb[2], static_cast<char>(0x95));
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcWCRToMBTest, FourByte) {
|
||||
TEST_F(LlvmLibcWCRToMBTest, FourByte) {
|
||||
mbstate_t state;
|
||||
LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
|
||||
// testing utf32: 0x1f921 -> utf8: 0xf0 0x9f 0xa4 0xa1
|
||||
@@ -60,9 +67,10 @@ TEST(LlvmLibcWCRToMBTest, FourByte) {
|
||||
ASSERT_EQ(mb[1], static_cast<char>(0x9f));
|
||||
ASSERT_EQ(mb[2], static_cast<char>(0xa4));
|
||||
ASSERT_EQ(mb[3], static_cast<char>(0xa1));
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
}
|
||||
|
||||
TEST(LlvmLibcWCRToMBTest, NullString) {
|
||||
TEST_F(LlvmLibcWCRToMBTest, NullString) {
|
||||
mbstate_t state;
|
||||
LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
|
||||
wchar_t wc = L'A';
|
||||
@@ -70,24 +78,39 @@ TEST(LlvmLibcWCRToMBTest, NullString) {
|
||||
|
||||
// should be equivalent to the call wcrtomb(buf, L'\0', state)
|
||||
size_t cnt1 = LIBC_NAMESPACE::wcrtomb(nullptr, wc, &state);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
size_t cnt2 = LIBC_NAMESPACE::wcrtomb(mb, L'\0', &state);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
|
||||
ASSERT_EQ(cnt1, cnt2);
|
||||
}
|
||||
|
||||
TEST(LlvmLibcWCRToMBTest, NullState) {
|
||||
TEST_F(LlvmLibcWCRToMBTest, NullState) {
|
||||
wchar_t wc = L'A';
|
||||
char mb[4];
|
||||
size_t cnt = LIBC_NAMESPACE::wcrtomb(mb, wc, nullptr);
|
||||
ASSERT_ERRNO_SUCCESS();
|
||||
ASSERT_EQ(cnt, static_cast<size_t>(1));
|
||||
}
|
||||
|
||||
TEST(LlvmLibcWCRToMBTest, InvalidWchar) {
|
||||
TEST_F(LlvmLibcWCRToMBTest, InvalidWchar) {
|
||||
mbstate_t state;
|
||||
LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
|
||||
wchar_t wc = 0x12ffff;
|
||||
char mb[4];
|
||||
size_t cnt = LIBC_NAMESPACE::wcrtomb(mb, wc, &state);
|
||||
ASSERT_EQ(cnt, static_cast<size_t>(-1));
|
||||
ASSERT_EQ(static_cast<int>(libc_errno), EILSEQ);
|
||||
ASSERT_ERRNO_EQ(EILSEQ);
|
||||
}
|
||||
|
||||
TEST_F(LlvmLibcWCRToMBTest, InvalidMBState) {
|
||||
mbstate_t *state;
|
||||
LIBC_NAMESPACE::internal::mbstate inv;
|
||||
inv.total_bytes = 6;
|
||||
state = reinterpret_cast<mbstate_t *>(&inv);
|
||||
wchar_t wc = L'A';
|
||||
char mb[4];
|
||||
size_t cnt = LIBC_NAMESPACE::wcrtomb(mb, wc, state);
|
||||
ASSERT_EQ(cnt, static_cast<size_t>(-1));
|
||||
ASSERT_ERRNO_EQ(EINVAL);
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user