[libc-commits] [libc] [libc] wchar string conversion functions mb to wc (PR #149423)
via libc-commits
libc-commits at lists.llvm.org
Fri Jul 18 13:08:57 PDT 2025
https://github.com/sribee8 updated https://github.com/llvm/llvm-project/pull/149423
>From 5757dcecd44fd1c123d5f1ed36fb8ce1bc9054ac Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Tue, 15 Jul 2025 22:08:14 +0000
Subject: [PATCH 01/12] started mbstowcs implementation
---
libc/src/wchar/mbstowcs.cpp | 29 +++++++++++++++++++++++++++++
libc/src/wchar/mbstowcs.h | 22 ++++++++++++++++++++++
2 files changed, 51 insertions(+)
create mode 100644 libc/src/wchar/mbstowcs.cpp
create mode 100644 libc/src/wchar/mbstowcs.h
diff --git a/libc/src/wchar/mbstowcs.cpp b/libc/src/wchar/mbstowcs.cpp
new file mode 100644
index 0000000000000..8b86cb03fe7a3
--- /dev/null
+++ b/libc/src/wchar/mbstowcs.cpp
@@ -0,0 +1,29 @@
+//===-- Implementation of mbstowcs ----------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/wchar/mbstowcs.h"
+
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/common.h"
+#include "src/__support/libc_errno.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/wchar/mbstate.h"
+#include "src/__support/wchar/string_converter.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(int, mbstowcs,
+ (wchar_t *__restrict pwcs, const char *__restrict s,
+ size_t n)) {
+ static internal::mbstate internal_mbstate;
+ internal::StringConverter<char8_t> str_conv(
+ reinterpret_cast<const char8_t *>(pwcs), &internal_mbstate, n);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/wchar/mbstowcs.h b/libc/src/wchar/mbstowcs.h
new file mode 100644
index 0000000000000..20caaf9f7bcea
--- /dev/null
+++ b/libc/src/wchar/mbstowcs.h
@@ -0,0 +1,22 @@
+//===-- Implementation header for mbstowcs --------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_WCHAR_MBSTOWCS_H
+#define LLVM_LIBC_SRC_WCHAR_MBSTOWCS_H
+
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+int mbstowcs(wchar_t *__restrict pwcs, const char *__restrict s, size_t n);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_WCHAR_MBSTOWCS_H
>From c9d06fcf9b7eff66efc0bc8127a72717ee3eaf24 Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Tue, 15 Jul 2025 22:16:12 +0000
Subject: [PATCH 02/12] finished mbstowcs implementation
---
libc/src/wchar/mbstowcs.cpp | 15 ++++++++++++++-
1 file changed, 14 insertions(+), 1 deletion(-)
diff --git a/libc/src/wchar/mbstowcs.cpp b/libc/src/wchar/mbstowcs.cpp
index 8b86cb03fe7a3..e7cd3416099ae 100644
--- a/libc/src/wchar/mbstowcs.cpp
+++ b/libc/src/wchar/mbstowcs.cpp
@@ -23,7 +23,20 @@ LLVM_LIBC_FUNCTION(int, mbstowcs,
size_t n)) {
static internal::mbstate internal_mbstate;
internal::StringConverter<char8_t> str_conv(
- reinterpret_cast<const char8_t *>(pwcs), &internal_mbstate, n);
+ reinterpret_cast<const char8_t *>(s), &internal_mbstate, n);
+ int dst_idx = 0;
+ ErrorOr<char32_t> converted = str_conv.popUTF32();
+ while (converted.has_value()) {
+ if (pwcs != nullptr)
+ pwcs[dst_idx] = converted.value();
+ dst_idx++;
+ converted = str_conv.popUTF32();
+ }
+ if (converted.error() == -1) // if we hit conversion limit
+ return dst_idx;
+
+ libc_errno = converted.error();
+ return -1;
}
} // namespace LIBC_NAMESPACE_DECL
>From 26a67410ff16da7d88d4d439741ecbab49cebe80 Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Wed, 16 Jul 2025 16:38:27 +0000
Subject: [PATCH 03/12] implemented mbsrtowcs
---
libc/src/wchar/mbsrtowcs.cpp | 47 ++++++++++++++++++++++++++++++++++++
libc/src/wchar/mbsrtowcs.h | 24 ++++++++++++++++++
libc/src/wchar/mbstowcs.cpp | 2 +-
libc/src/wchar/mbstowcs.h | 2 +-
4 files changed, 73 insertions(+), 2 deletions(-)
create mode 100644 libc/src/wchar/mbsrtowcs.cpp
create mode 100644 libc/src/wchar/mbsrtowcs.h
diff --git a/libc/src/wchar/mbsrtowcs.cpp b/libc/src/wchar/mbsrtowcs.cpp
new file mode 100644
index 0000000000000..4407daa293f34
--- /dev/null
+++ b/libc/src/wchar/mbsrtowcs.cpp
@@ -0,0 +1,47 @@
+//===-- Implementation of mbsrtowcs ---------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/wchar/mbsrtowcs.h"
+
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/common.h"
+#include "src/__support/libc_errno.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/wchar/mbstate.h"
+#include "src/__support/wchar/string_converter.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(size_t, mbsrtowcs,
+ (wchar_t *__restrict dst, const char **__restrict src,
+ size_t len, mbstate_t *__restrict ps)) {
+ static internal::mbstate internal_mbstate;
+ internal::StringConverter<char8_t> str_conv(
+ reinterpret_cast<const char8_t *>(src),
+ ps == nullptr ? &internal_mbstate
+ : reinterpret_cast<internal::mbstate *>(ps),
+ len);
+
+ int dst_idx = 0;
+ ErrorOr<char32_t> converted = str_conv.popUTF32();
+ while (converted.has_value()) {
+ dst[dst_idx] = converted.value();
+ dst_idx++;
+ converted = str_conv.popUTF32();
+ }
+
+ src += str_conv.getSourceIndex();
+ if (converted.error() == -1) // if we hit conversion limit
+ return dst_idx;
+
+ libc_errno = converted.error();
+ return -1;
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/wchar/mbsrtowcs.h b/libc/src/wchar/mbsrtowcs.h
new file mode 100644
index 0000000000000..f8d4cc26e63ae
--- /dev/null
+++ b/libc/src/wchar/mbsrtowcs.h
@@ -0,0 +1,24 @@
+//===-- Implementation header for mbsrtowcs -------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_WCHAR_MBSRTOWCS_H
+#define LLVM_LIBC_SRC_WCHAR_MBSRTOWCS_H
+
+#include "hdr/types/mbstate_t.h"
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+size_t mbsrtowcs(wchar_t *__restrict dst, const char **__restrict src,
+ size_t len, mbstate_t *__restrict ps);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_WCHAR_MBSRTOWCS_H
diff --git a/libc/src/wchar/mbstowcs.cpp b/libc/src/wchar/mbstowcs.cpp
index e7cd3416099ae..3b43bcf753793 100644
--- a/libc/src/wchar/mbstowcs.cpp
+++ b/libc/src/wchar/mbstowcs.cpp
@@ -18,7 +18,7 @@
namespace LIBC_NAMESPACE_DECL {
-LLVM_LIBC_FUNCTION(int, mbstowcs,
+LLVM_LIBC_FUNCTION(size_t, mbstowcs,
(wchar_t *__restrict pwcs, const char *__restrict s,
size_t n)) {
static internal::mbstate internal_mbstate;
diff --git a/libc/src/wchar/mbstowcs.h b/libc/src/wchar/mbstowcs.h
index 20caaf9f7bcea..7d08a838b2324 100644
--- a/libc/src/wchar/mbstowcs.h
+++ b/libc/src/wchar/mbstowcs.h
@@ -15,7 +15,7 @@
namespace LIBC_NAMESPACE_DECL {
-int mbstowcs(wchar_t *__restrict pwcs, const char *__restrict s, size_t n);
+size_t mbstowcs(wchar_t *__restrict pwcs, const char *__restrict s, size_t n);
} // namespace LIBC_NAMESPACE_DECL
>From 0694e910d92162fea1dc0b72bae82f066472d08f Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Thu, 17 Jul 2025 16:46:53 +0000
Subject: [PATCH 04/12] added tests for mbstowcs
---
libc/config/linux/x86_64/entrypoints.txt | 2 +
libc/include/wchar.yaml | 17 ++++
libc/src/wchar/CMakeLists.txt | 32 +++++++
libc/src/wchar/mbstowcs.cpp | 6 ++
libc/test/src/wchar/CMakeLists.txt | 13 +++
libc/test/src/wchar/mbstowcs_test.cpp | 117 +++++++++++++++++++++++
6 files changed, 187 insertions(+)
create mode 100644 libc/test/src/wchar/mbstowcs_test.cpp
diff --git a/libc/config/linux/x86_64/entrypoints.txt b/libc/config/linux/x86_64/entrypoints.txt
index 9223911f04a93..e37370fdf47f6 100644
--- a/libc/config/linux/x86_64/entrypoints.txt
+++ b/libc/config/linux/x86_64/entrypoints.txt
@@ -1261,6 +1261,8 @@ if(LLVM_LIBC_FULL_BUILD)
# wchar.h entrypoints
libc.src.wchar.mbrtowc
libc.src.wchar.mbtowc
+ libc.src.wchar.mbstowcs
+ libc.src.wchar.mbsrtowcs
libc.src.wchar.wcrtomb
libc.src.wchar.wctomb
)
diff --git a/libc/include/wchar.yaml b/libc/include/wchar.yaml
index 123d3440aeec3..3d63d9476e002 100644
--- a/libc/include/wchar.yaml
+++ b/libc/include/wchar.yaml
@@ -53,6 +53,23 @@ functions:
- type: wchar_t *__restrict
- type: const char *__restrict
- type: size_t
+ - name: mbstowcs
+ standards:
+ - stdc
+ return_type: size_t
+ arguments:
+ - type: wchar_t *__restrict
+ - type: const char *__restrict
+ - type: size_t
+ - name: mbsrtowcs
+ standards:
+ - stdc
+ return_type: size_t
+ arguments:
+ - type: wchar_t *__restrict
+ - type: const char **__restrict
+ - type: size_t
+ - type: mbstate_t *__restrict
- name: wmemset
standards:
- stdc
diff --git a/libc/src/wchar/CMakeLists.txt b/libc/src/wchar/CMakeLists.txt
index 7ace1a6ca66ba..a70486950d664 100644
--- a/libc/src/wchar/CMakeLists.txt
+++ b/libc/src/wchar/CMakeLists.txt
@@ -159,6 +159,38 @@ add_entrypoint_object(
libc.src.__support.wchar.mbstate
)
+add_entrypoint_object(
+ mbstowcs
+ SRCS
+ mbstowcs.cpp
+ HDRS
+ mbstowcs.h
+ DEPENDS
+ libc.hdr.types.size_t
+ libc.hdr.types.wchar_t
+ libc.src.__support.common
+ libc.src.__support.macros.config
+ libc.src.__support.libc_errno
+ libc.src.__support.wchar.string_converter
+ libc.src.__support.wchar.mbstate
+)
+
+add_entrypoint_object(
+ mbsrtowcs
+ SRCS
+ mbsrtowcs.cpp
+ HDRS
+ mbsrtowcs.h
+ DEPENDS
+ libc.hdr.types.size_t
+ libc.hdr.types.wchar_t
+ libc.src.__support.common
+ libc.src.__support.macros.config
+ libc.src.__support.libc_errno
+ libc.src.__support.wchar.string_converter
+ libc.src.__support.wchar.mbstate
+)
+
add_entrypoint_object(
wmemset
SRCS
diff --git a/libc/src/wchar/mbstowcs.cpp b/libc/src/wchar/mbstowcs.cpp
index 3b43bcf753793..1bba5cb4893e7 100644
--- a/libc/src/wchar/mbstowcs.cpp
+++ b/libc/src/wchar/mbstowcs.cpp
@@ -21,14 +21,20 @@ namespace LIBC_NAMESPACE_DECL {
LLVM_LIBC_FUNCTION(size_t, mbstowcs,
(wchar_t *__restrict pwcs, const char *__restrict s,
size_t n)) {
+ n = pwcs == nullptr ? SIZE_MAX : n;
static internal::mbstate internal_mbstate;
internal::StringConverter<char8_t> str_conv(
reinterpret_cast<const char8_t *>(s), &internal_mbstate, n);
int dst_idx = 0;
+
ErrorOr<char32_t> converted = str_conv.popUTF32();
+
while (converted.has_value()) {
if (pwcs != nullptr)
pwcs[dst_idx] = converted.value();
+ // if it is null terminator, do not count in return value
+ if (converted.value() == L'\0')
+ return dst_idx;
dst_idx++;
converted = str_conv.popUTF32();
}
diff --git a/libc/test/src/wchar/CMakeLists.txt b/libc/test/src/wchar/CMakeLists.txt
index 176cf7c3487cd..ba1a59211c778 100644
--- a/libc/test/src/wchar/CMakeLists.txt
+++ b/libc/test/src/wchar/CMakeLists.txt
@@ -64,6 +64,19 @@ add_libc_test(
libc.test.UnitTest.ErrnoCheckingTest
)
+add_libc_test(
+ mbstowcs_test
+ SUITE
+ libc_wchar_unittests
+ SRCS
+ mbstowcs_test.cpp
+ DEPENDS
+ libc.src.__support.libc_errno
+ libc.src.wchar.mbstowcs
+ libc.hdr.types.wchar_t
+ libc.test.UnitTest.ErrnoCheckingTest
+)
+
add_libc_test(
wctob_test
SUITE
diff --git a/libc/test/src/wchar/mbstowcs_test.cpp b/libc/test/src/wchar/mbstowcs_test.cpp
new file mode 100644
index 0000000000000..6ae31678e45c8
--- /dev/null
+++ b/libc/test/src/wchar/mbstowcs_test.cpp
@@ -0,0 +1,117 @@
+//===-- Unittests for mbstowcs --------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "hdr/types/wchar_t.h"
+#include "src/__support/libc_errno.h"
+#include "src/wchar/mbstowcs.h"
+#include "test/UnitTest/ErrnoCheckingTest.h"
+#include "test/UnitTest/Test.h"
+
+using LlvmLibcMBSToWCSTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
+
+TEST_F(LlvmLibcMBSToWCSTest, OneByteOneChar) {
+ const char *ch = "A";
+ wchar_t dest[2];
+ size_t n = LIBC_NAMESPACE::mbstowcs(dest, ch, 1);
+ ASSERT_EQ(static_cast<char>(*dest), 'A');
+ ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_ERRNO_SUCCESS();
+
+ n = LIBC_NAMESPACE::mbstowcs(dest + 1, ch + 1, 1);
+ ASSERT_EQ(static_cast<char>(dest[1]), '\0');
+ // Should not include null terminator
+ ASSERT_EQ(static_cast<int>(n), 0);
+ ASSERT_ERRNO_SUCCESS();
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, FourByteOneChar) {
+ const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
+ wchar_t dest[2];
+ size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_TRUE(dest[1] == L'\0');
+ // Should not count null terminator in number
+ ASSERT_EQ(static_cast<int>(n), 1);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, MultiByteTwoCharacters) {
+ // Two laughing cat emojis "😹😹"
+ const char *src = "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ wchar_t dest[3];
+ size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+ ASSERT_TRUE(dest[2] == L'\0');
+ // Should not count null terminator in number
+ ASSERT_EQ(static_cast<int>(n), 2);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, MixedNumberOfBytes) {
+ // 'A', sigma symbol 'Σ', recycling symbol '♻', laughing cat emoji '😹'
+ const char *src = "A\xce\xa3\xe2\x99\xbb\xf0\x9f\x98\xb9";
+ wchar_t dest[5];
+ size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 5);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<char>(dest[0]), 'A');
+ ASSERT_EQ(static_cast<int>(dest[1]), 931);
+ ASSERT_EQ(static_cast<int>(dest[2]), 9851);
+ ASSERT_EQ(static_cast<int>(dest[3]), 128569);
+ ASSERT_TRUE(dest[4] == L'\0');
+ // Should not count null terminator in number
+ ASSERT_EQ(static_cast<int>(n), 4);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, ReadLessThanStringLength) {
+ // Four laughing cat emojis "😹😹😹😹"
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ wchar_t dest[5] = {L'a', L'b', L'c', L'd', L'e'};
+ size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
+ ASSERT_ERRNO_SUCCESS();
+ // Should have read 3 emojis
+ ASSERT_EQ(static_cast<int>(n), 3);
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[2]), 128569);
+ ASSERT_TRUE(dest[3] == L'd');
+ ASSERT_TRUE(dest[4] == L'e');
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, InvalidFirstByte) {
+ // 0x80 is invalid first byte of mb character
+ const char *src =
+ "\x80\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ wchar_t dest[3];
+ size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
+ // Should return error and set errno
+ ASSERT_EQ(static_cast<int>(n), -1);
+ ASSERT_ERRNO_EQ(EILSEQ);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, InvalidMiddleByte) {
+ // The 7th byte is invalid for a 4 byte character
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ wchar_t dest[3];
+ size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 5);
+ // Should return error and set errno
+ ASSERT_EQ(static_cast<int>(n), -1);
+ ASSERT_ERRNO_EQ(EILSEQ);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, NullDestination) {
+ // Four laughing cat emojis "😹😹😹😹"
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ size_t n = LIBC_NAMESPACE::mbstowcs(nullptr, src, 2);
+ ASSERT_ERRNO_SUCCESS();
+ // Null destination should ignore len and read till end of string
+ ASSERT_EQ(static_cast<int>(n), 4);
+}
>From 192135610909f9b7d0d88cbfdee0e61e3368bbc9 Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Thu, 17 Jul 2025 17:11:51 +0000
Subject: [PATCH 05/12] added final test
---
libc/test/src/wchar/mbstowcs_test.cpp | 17 +++++++++++++++++
1 file changed, 17 insertions(+)
diff --git a/libc/test/src/wchar/mbstowcs_test.cpp b/libc/test/src/wchar/mbstowcs_test.cpp
index 6ae31678e45c8..1f6633bdd2766 100644
--- a/libc/test/src/wchar/mbstowcs_test.cpp
+++ b/libc/test/src/wchar/mbstowcs_test.cpp
@@ -115,3 +115,20 @@ TEST_F(LlvmLibcMBSToWCSTest, NullDestination) {
// Null destination should ignore len and read till end of string
ASSERT_EQ(static_cast<int>(n), 4);
}
+
+TEST_F(LlvmLibcMBSToWCSTest, ErrnoChecks) {
+ // Two laughing cat emojis and invalid 3rd mb char (3rd byte of it)
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9";
+ wchar_t dest[5];
+ // First two bytes are valid --> should not set errno
+ size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(n), 2);
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+ // Trying to read the 3rd byte should set errno
+ n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
+ ASSERT_ERRNO_EQ(EILSEQ);
+ ASSERT_EQ(static_cast<int>(n), -1);
+}
>From 6bc7497d1aef5dc0a6fc96dcfc600db4ae90b8c9 Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Thu, 17 Jul 2025 18:21:11 +0000
Subject: [PATCH 06/12] added internal mbsnrtowcs
---
libc/src/__support/wchar/CMakeLists.txt | 18 +++
libc/src/__support/wchar/mbsnrtowcs.cpp | 57 ++++++++++
libc/src/__support/wchar/mbsnrtowcs.h | 29 +++++
libc/src/wchar/CMakeLists.txt | 2 +-
libc/src/wchar/mbsrtowcs.cpp | 9 +-
libc/src/wchar/mbstowcs.cpp | 27 ++---
libc/test/src/wchar/CMakeLists.txt | 16 +++
libc/test/src/wchar/mbsrtowcs_test.cpp | 139 ++++++++++++++++++++++++
8 files changed, 275 insertions(+), 22 deletions(-)
create mode 100644 libc/src/__support/wchar/mbsnrtowcs.cpp
create mode 100644 libc/src/__support/wchar/mbsnrtowcs.h
create mode 100644 libc/test/src/wchar/mbsrtowcs_test.cpp
diff --git a/libc/src/__support/wchar/CMakeLists.txt b/libc/src/__support/wchar/CMakeLists.txt
index 802441d37fe92..fa5139a8b68b4 100644
--- a/libc/src/__support/wchar/CMakeLists.txt
+++ b/libc/src/__support/wchar/CMakeLists.txt
@@ -68,3 +68,21 @@ add_object_library(
.character_converter
.mbstate
)
+
+add_object_library(
+ mbsnrtowcs
+ HDRS
+ mbsnrtowcs.h
+ SRCS
+ mbsnrtowcs.cpp
+ DEPENDS
+ libc.hdr.errno_macros
+ libc.hdr.types.wchar_t
+ libc.hdr.types.size_t
+ libc.src.__support.common
+ libc.src.__support.error_or
+ libc.src.__support.macros.config
+ .character_converter
+ .mbstate
+ .string_converter
+)
diff --git a/libc/src/__support/wchar/mbsnrtowcs.cpp b/libc/src/__support/wchar/mbsnrtowcs.cpp
new file mode 100644
index 0000000000000..106a596e59303
--- /dev/null
+++ b/libc/src/__support/wchar/mbsnrtowcs.cpp
@@ -0,0 +1,57 @@
+//===-- Implementation for mbsnrtowcs function ------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/__support/wchar/mbsnrtowcs.h"
+#include "hdr/errno_macros.h"
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/common.h"
+#include "src/__support/error_or.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/wchar/character_converter.h"
+#include "src/__support/wchar/mbstate.h"
+#include "src/__support/wchar/string_converter.h"
+
+namespace LIBC_NAMESPACE_DECL {
+namespace internal {
+
+ErrorOr<size_t> mbsnrtowcs(wchar_t *__restrict dst, const char **__restrict src,
+ size_t nmc, size_t len, mbstate *__restrict ps) {
+ if (*src == nullptr)
+ return 0;
+ // Checking if mbstate is valid
+ CharacterConverter char_conv(ps);
+ if (!char_conv.isValidState())
+ return Error(EINVAL);
+
+ StringConverter<char8_t> str_conv(reinterpret_cast<const char8_t *>(*src), ps,
+ len, nmc);
+ size_t dst_idx = 0;
+ ErrorOr<char32_t> converted = str_conv.popUTF32();
+ while (converted.has_value()) {
+ if (dst != nullptr)
+ dst[dst_idx] = converted.value();
+ // null terminator should not be counted in return value
+ if (converted.value() == L'\0') {
+ src = nullptr;
+ return dst_idx;
+ }
+ dst_idx++;
+ converted = str_conv.popUTF32();
+ }
+
+ *src += str_conv.getSourceIndex();
+ if (converted.error() == -1) // if we hit conversion limit
+ return dst_idx;
+
+ return Error(converted.error());
+}
+
+} // namespace internal
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/__support/wchar/mbsnrtowcs.h b/libc/src/__support/wchar/mbsnrtowcs.h
new file mode 100644
index 0000000000000..8ce497325fdad
--- /dev/null
+++ b/libc/src/__support/wchar/mbsnrtowcs.h
@@ -0,0 +1,29 @@
+//===-- Implementation header for mbsnrtowcs function -----------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS
+#define LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS
+
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/common.h"
+#include "src/__support/error_or.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/wchar/mbstate.h"
+
+namespace LIBC_NAMESPACE_DECL {
+namespace internal {
+
+ErrorOr<size_t> mbsnrtowcs(wchar_t *__restrict dst, const char **__restrict src,
+ size_t nmc, size_t len, mbstate *__restrict ps);
+
+} // namespace internal
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS
diff --git a/libc/src/wchar/CMakeLists.txt b/libc/src/wchar/CMakeLists.txt
index a70486950d664..2613f6daf111d 100644
--- a/libc/src/wchar/CMakeLists.txt
+++ b/libc/src/wchar/CMakeLists.txt
@@ -171,8 +171,8 @@ add_entrypoint_object(
libc.src.__support.common
libc.src.__support.macros.config
libc.src.__support.libc_errno
- libc.src.__support.wchar.string_converter
libc.src.__support.wchar.mbstate
+ libc.src.__support.wchar.mbsnrtowcs
)
add_entrypoint_object(
diff --git a/libc/src/wchar/mbsrtowcs.cpp b/libc/src/wchar/mbsrtowcs.cpp
index 4407daa293f34..fa00a26692d7c 100644
--- a/libc/src/wchar/mbsrtowcs.cpp
+++ b/libc/src/wchar/mbsrtowcs.cpp
@@ -23,7 +23,7 @@ LLVM_LIBC_FUNCTION(size_t, mbsrtowcs,
size_t len, mbstate_t *__restrict ps)) {
static internal::mbstate internal_mbstate;
internal::StringConverter<char8_t> str_conv(
- reinterpret_cast<const char8_t *>(src),
+ reinterpret_cast<const char8_t *>(*src),
ps == nullptr ? &internal_mbstate
: reinterpret_cast<internal::mbstate *>(ps),
len);
@@ -32,11 +32,16 @@ LLVM_LIBC_FUNCTION(size_t, mbsrtowcs,
ErrorOr<char32_t> converted = str_conv.popUTF32();
while (converted.has_value()) {
dst[dst_idx] = converted.value();
+ // null terminator should not be counted in return value
+ if (converted.value() == L'\0') {
+ src = nullptr;
+ return dst_idx;
+ }
dst_idx++;
converted = str_conv.popUTF32();
}
- src += str_conv.getSourceIndex();
+ *src += str_conv.getSourceIndex();
if (converted.error() == -1) // if we hit conversion limit
return dst_idx;
diff --git a/libc/src/wchar/mbstowcs.cpp b/libc/src/wchar/mbstowcs.cpp
index 1bba5cb4893e7..66046903a8c6f 100644
--- a/libc/src/wchar/mbstowcs.cpp
+++ b/libc/src/wchar/mbstowcs.cpp
@@ -13,8 +13,8 @@
#include "src/__support/common.h"
#include "src/__support/libc_errno.h"
#include "src/__support/macros/config.h"
+#include "src/__support/wchar/mbsnrtowcs.h"
#include "src/__support/wchar/mbstate.h"
-#include "src/__support/wchar/string_converter.h"
namespace LIBC_NAMESPACE_DECL {
@@ -23,26 +23,15 @@ LLVM_LIBC_FUNCTION(size_t, mbstowcs,
size_t n)) {
n = pwcs == nullptr ? SIZE_MAX : n;
static internal::mbstate internal_mbstate;
- internal::StringConverter<char8_t> str_conv(
- reinterpret_cast<const char8_t *>(s), &internal_mbstate, n);
- int dst_idx = 0;
+ const char *temp = s;
+ auto ret = internal::mbsnrtowcs(pwcs, &temp, SIZE_MAX, n, &internal_mbstate);
- ErrorOr<char32_t> converted = str_conv.popUTF32();
-
- while (converted.has_value()) {
- if (pwcs != nullptr)
- pwcs[dst_idx] = converted.value();
- // if it is null terminator, do not count in return value
- if (converted.value() == L'\0')
- return dst_idx;
- dst_idx++;
- converted = str_conv.popUTF32();
+ if (!ret.has_value()) {
+ // Encoding failure
+ libc_errno = ret.error();
+ return -1;
}
- if (converted.error() == -1) // if we hit conversion limit
- return dst_idx;
-
- libc_errno = converted.error();
- return -1;
+ return ret.value();
}
} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/test/src/wchar/CMakeLists.txt b/libc/test/src/wchar/CMakeLists.txt
index ba1a59211c778..9ed924eae34c4 100644
--- a/libc/test/src/wchar/CMakeLists.txt
+++ b/libc/test/src/wchar/CMakeLists.txt
@@ -77,6 +77,22 @@ add_libc_test(
libc.test.UnitTest.ErrnoCheckingTest
)
+add_libc_test(
+ mbsrtowcs_test
+ SUITE
+ libc_wchar_unittests
+ SRCS
+ mbsrtowcs_test.cpp
+ DEPENDS
+ libc.src.__support.libc_errno
+ libc.src.__support.wchar.mbstate
+ libc.src.string.memset
+ libc.src.wchar.mbsrtowcs
+ libc.hdr.types.mbstate_t
+ libc.hdr.types.wchar_t
+ libc.test.UnitTest.ErrnoCheckingTest
+)
+
add_libc_test(
wctob_test
SUITE
diff --git a/libc/test/src/wchar/mbsrtowcs_test.cpp b/libc/test/src/wchar/mbsrtowcs_test.cpp
new file mode 100644
index 0000000000000..2f6e6e6a71b31
--- /dev/null
+++ b/libc/test/src/wchar/mbsrtowcs_test.cpp
@@ -0,0 +1,139 @@
+//===-- Unittests for mbsetowcs -------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#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/mbsrtowcs.h"
+#include "test/UnitTest/ErrnoCheckingTest.h"
+#include "test/UnitTest/Test.h"
+
+using LlvmLibcMBSRToWCSTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
+
+TEST_F(LlvmLibcMBSRToWCSTest, OneByteOneChar) {
+ const char *ch = "A";
+ wchar_t dest[2];
+ mbstate_t *mb;
+ LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &ch, 1, mb);
+ ASSERT_EQ(static_cast<char>(*dest), 'A');
+ ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_ERRNO_SUCCESS();
+
+ n = LIBC_NAMESPACE::mbsrtowcs(dest + 1, &ch, 1, mb);
+ ASSERT_EQ(static_cast<char>(dest[1]), '\0');
+ // Should not include null terminator
+ ASSERT_EQ(static_cast<int>(n), 0);
+ ASSERT_ERRNO_SUCCESS();
+}
+
+// TEST_F(LlvmLibcMBSRToWCSTest, FourByteOneChar) {
+// const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
+// wchar_t dest[2];
+// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
+// ASSERT_ERRNO_SUCCESS();
+// ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+// ASSERT_TRUE(dest[1] == L'\0');
+// // Should not count null terminator in number
+// ASSERT_EQ(static_cast<int>(n), 1);
+// }
+
+// TEST_F(LlvmLibcMBSRToWCSTest, MultiByteTwoCharacters) {
+// // Two laughing cat emojis "😹😹"
+// const char *src = "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+// wchar_t dest[3];
+// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
+// ASSERT_ERRNO_SUCCESS();
+// ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+// ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+// ASSERT_TRUE(dest[2] == L'\0');
+// // Should not count null terminator in number
+// ASSERT_EQ(static_cast<int>(n), 2);
+// }
+
+// TEST_F(LlvmLibcMBSRToWCSTest, MixedNumberOfBytes) {
+// // 'A', sigma symbol 'Σ', recycling symbol '♻', laughing cat emoji '😹'
+// const char *src = "A\xce\xa3\xe2\x99\xbb\xf0\x9f\x98\xb9";
+// wchar_t dest[5];
+// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 5);
+// ASSERT_ERRNO_SUCCESS();
+// ASSERT_EQ(static_cast<char>(dest[0]), 'A');
+// ASSERT_EQ(static_cast<int>(dest[1]), 931);
+// ASSERT_EQ(static_cast<int>(dest[2]), 9851);
+// ASSERT_EQ(static_cast<int>(dest[3]), 128569);
+// ASSERT_TRUE(dest[4] == L'\0');
+// // Should not count null terminator in number
+// ASSERT_EQ(static_cast<int>(n), 4);
+// }
+
+// TEST_F(LlvmLibcMBSRToWCSTest, ReadLessThanStringLength) {
+// // Four laughing cat emojis "😹😹😹😹"
+// const char *src =
+// "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+// wchar_t dest[5] = {L'a', L'b', L'c', L'd', L'e'};
+// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
+// ASSERT_ERRNO_SUCCESS();
+// // Should have read 3 emojis
+// ASSERT_EQ(static_cast<int>(n), 3);
+// ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+// ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+// ASSERT_EQ(static_cast<int>(dest[2]), 128569);
+// ASSERT_TRUE(dest[3] == L'd');
+// ASSERT_TRUE(dest[4] == L'e');
+// }
+
+// TEST_F(LlvmLibcMBSRToWCSTest, InvalidFirstByte) {
+// // 0x80 is invalid first byte of mb character
+// const char *src =
+// "\x80\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+// wchar_t dest[3];
+// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
+// // Should return error and set errno
+// ASSERT_EQ(static_cast<int>(n), -1);
+// ASSERT_ERRNO_EQ(EILSEQ);
+// }
+
+// TEST_F(LlvmLibcMBSRToWCSTest, InvalidMiddleByte) {
+// // The 7th byte is invalid for a 4 byte character
+// const char *src =
+// "\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+// wchar_t dest[3];
+// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 5);
+// // Should return error and set errno
+// ASSERT_EQ(static_cast<int>(n), -1);
+// ASSERT_ERRNO_EQ(EILSEQ);
+// }
+
+// TEST_F(LlvmLibcMBSRToWCSTest, NullDestination) {
+// // Four laughing cat emojis "😹😹😹😹"
+// const char *src =
+// "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+// size_t n = LIBC_NAMESPACE::mbstowcs(nullptr, src, 2);
+// ASSERT_ERRNO_SUCCESS();
+// // Null destination should ignore len and read till end of string
+// ASSERT_EQ(static_cast<int>(n), 4);
+// }
+
+// TEST_F(LlvmLibcMBSRToWCSTest, ErrnoChecks) {
+// // Two laughing cat emojis and invalid 3rd mb char (3rd byte of it)
+// const char *src =
+// "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9";
+// wchar_t dest[5];
+// // First two bytes are valid --> should not set errno
+// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
+// ASSERT_ERRNO_SUCCESS();
+// ASSERT_EQ(static_cast<int>(n), 2);
+// ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+// ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+// // Trying to read the 3rd byte should set errno
+// n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
+// ASSERT_ERRNO_EQ(EILSEQ);
+// ASSERT_EQ(static_cast<int>(n), -1);
+// }
>From 40aa6eeee39d8e89118d98a0587c3e10e9f8a575 Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Thu, 17 Jul 2025 23:24:11 +0000
Subject: [PATCH 07/12] implemented mbsnrtowcs and tests
---
libc/config/linux/x86_64/entrypoints.txt | 1 +
libc/include/wchar.yaml | 10 +
libc/src/__support/wchar/mbsnrtowcs.cpp | 8 +-
libc/src/wchar/CMakeLists.txt | 20 +-
libc/src/wchar/mbsnrtowcs.cpp | 39 ++++
libc/src/wchar/mbsnrtowcs.h | 24 +++
libc/src/wchar/mbsrtowcs.cpp | 35 ++--
libc/src/wchar/mbstowcs.cpp | 1 +
libc/test/src/wchar/CMakeLists.txt | 16 ++
libc/test/src/wchar/mbsnrtowcs_test.cpp | 203 ++++++++++++++++++++
libc/test/src/wchar/mbsrtowcs_test.cpp | 229 +++++++++++++----------
libc/test/src/wchar/mbstowcs_test.cpp | 30 ++-
12 files changed, 491 insertions(+), 125 deletions(-)
create mode 100644 libc/src/wchar/mbsnrtowcs.cpp
create mode 100644 libc/src/wchar/mbsnrtowcs.h
create mode 100644 libc/test/src/wchar/mbsnrtowcs_test.cpp
diff --git a/libc/config/linux/x86_64/entrypoints.txt b/libc/config/linux/x86_64/entrypoints.txt
index e37370fdf47f6..e0a02abef001c 100644
--- a/libc/config/linux/x86_64/entrypoints.txt
+++ b/libc/config/linux/x86_64/entrypoints.txt
@@ -1263,6 +1263,7 @@ if(LLVM_LIBC_FULL_BUILD)
libc.src.wchar.mbtowc
libc.src.wchar.mbstowcs
libc.src.wchar.mbsrtowcs
+ libc.src.wchar.mbsnrtowcs
libc.src.wchar.wcrtomb
libc.src.wchar.wctomb
)
diff --git a/libc/include/wchar.yaml b/libc/include/wchar.yaml
index 3d63d9476e002..d3166d60c19de 100644
--- a/libc/include/wchar.yaml
+++ b/libc/include/wchar.yaml
@@ -70,6 +70,16 @@ functions:
- type: const char **__restrict
- type: size_t
- type: mbstate_t *__restrict
+ - name: mbsnrtowcs
+ standards:
+ - stdc
+ return_type: size_t
+ arguments:
+ - type: wchar_t *__restrict
+ - type: const char **__restrict
+ - type: size_t
+ - type: size_t
+ - type: mbstate_t *__restrict
- name: wmemset
standards:
- stdc
diff --git a/libc/src/__support/wchar/mbsnrtowcs.cpp b/libc/src/__support/wchar/mbsnrtowcs.cpp
index 106a596e59303..9eb2b5423af87 100644
--- a/libc/src/__support/wchar/mbsnrtowcs.cpp
+++ b/libc/src/__support/wchar/mbsnrtowcs.cpp
@@ -38,16 +38,18 @@ ErrorOr<size_t> mbsnrtowcs(wchar_t *__restrict dst, const char **__restrict src,
dst[dst_idx] = converted.value();
// null terminator should not be counted in return value
if (converted.value() == L'\0') {
- src = nullptr;
+ if (dst != nullptr)
+ *src = nullptr;
return dst_idx;
}
dst_idx++;
converted = str_conv.popUTF32();
}
- *src += str_conv.getSourceIndex();
- if (converted.error() == -1) // if we hit conversion limit
+ if (converted.error() == -1) { // if we hit conversion limit
+ *src += str_conv.getSourceIndex();
return dst_idx;
+ }
return Error(converted.error());
}
diff --git a/libc/src/wchar/CMakeLists.txt b/libc/src/wchar/CMakeLists.txt
index 2613f6daf111d..ac359064ec10c 100644
--- a/libc/src/wchar/CMakeLists.txt
+++ b/libc/src/wchar/CMakeLists.txt
@@ -187,7 +187,25 @@ add_entrypoint_object(
libc.src.__support.common
libc.src.__support.macros.config
libc.src.__support.libc_errno
- libc.src.__support.wchar.string_converter
+ libc.src.__support.wchar.mbstate
+ libc.src.__support.wchar.mbsnrtowcs
+ libc.src.__support.wchar.mbstate
+)
+
+add_entrypoint_object(
+ mbsnrtowcs
+ SRCS
+ mbsnrtowcs.cpp
+ HDRS
+ mbsnrtowcs.h
+ DEPENDS
+ libc.hdr.types.size_t
+ libc.hdr.types.wchar_t
+ libc.src.__support.common
+ libc.src.__support.macros.config
+ libc.src.__support.libc_errno
+ libc.src.__support.wchar.mbstate
+ libc.src.__support.wchar.mbsnrtowcs
libc.src.__support.wchar.mbstate
)
diff --git a/libc/src/wchar/mbsnrtowcs.cpp b/libc/src/wchar/mbsnrtowcs.cpp
new file mode 100644
index 0000000000000..28e0ff31b26c4
--- /dev/null
+++ b/libc/src/wchar/mbsnrtowcs.cpp
@@ -0,0 +1,39 @@
+//===-- Implementation of mbsnrtowcs --------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/wchar/mbsnrtowcs.h"
+
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/common.h"
+#include "src/__support/libc_errno.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/wchar/mbsnrtowcs.h"
+#include "src/__support/wchar/mbstate.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(size_t, mbsnrtowcs,
+ (wchar_t *__restrict dst, const char **__restrict src,
+ size_t nmc, size_t len, mbstate_t *__restrict ps)) {
+ static internal::mbstate internal_mbstate;
+ // If destination is null, ignore len
+ len = dst == nullptr ? SIZE_MAX : len;
+ auto ret = internal::mbsnrtowcs(
+ dst, src, nmc, len,
+ ps == nullptr ? &internal_mbstate
+ : reinterpret_cast<internal::mbstate *>(ps));
+ if (!ret.has_value()) {
+ // Encoding failure
+ libc_errno = ret.error();
+ return -1;
+ }
+ return ret.value();
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/wchar/mbsnrtowcs.h b/libc/src/wchar/mbsnrtowcs.h
new file mode 100644
index 0000000000000..5c9687b20b3fa
--- /dev/null
+++ b/libc/src/wchar/mbsnrtowcs.h
@@ -0,0 +1,24 @@
+//===-- Implementation header for mbsnrtowcs ------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_WCHAR_MBSNRTOWCS_H
+#define LLVM_LIBC_SRC_WCHAR_MBSNRTOWCS_H
+
+#include "hdr/types/mbstate_t.h"
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+size_t mbsnrtowcs(wchar_t *__restrict dst, const char **__restrict src,
+ size_t nmc, size_t len, mbstate_t *__restrict ps);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_WCHAR_MBSNRTOWCS_H
diff --git a/libc/src/wchar/mbsrtowcs.cpp b/libc/src/wchar/mbsrtowcs.cpp
index fa00a26692d7c..82ca25a3d863d 100644
--- a/libc/src/wchar/mbsrtowcs.cpp
+++ b/libc/src/wchar/mbsrtowcs.cpp
@@ -13,8 +13,8 @@
#include "src/__support/common.h"
#include "src/__support/libc_errno.h"
#include "src/__support/macros/config.h"
+#include "src/__support/wchar/mbsnrtowcs.h"
#include "src/__support/wchar/mbstate.h"
-#include "src/__support/wchar/string_converter.h"
namespace LIBC_NAMESPACE_DECL {
@@ -22,31 +22,18 @@ LLVM_LIBC_FUNCTION(size_t, mbsrtowcs,
(wchar_t *__restrict dst, const char **__restrict src,
size_t len, mbstate_t *__restrict ps)) {
static internal::mbstate internal_mbstate;
- internal::StringConverter<char8_t> str_conv(
- reinterpret_cast<const char8_t *>(*src),
+ // If destination is null, ignore len
+ len = dst == nullptr ? SIZE_MAX : len;
+ auto ret = internal::mbsnrtowcs(
+ dst, src, SIZE_MAX, len,
ps == nullptr ? &internal_mbstate
- : reinterpret_cast<internal::mbstate *>(ps),
- len);
-
- int dst_idx = 0;
- ErrorOr<char32_t> converted = str_conv.popUTF32();
- while (converted.has_value()) {
- dst[dst_idx] = converted.value();
- // null terminator should not be counted in return value
- if (converted.value() == L'\0') {
- src = nullptr;
- return dst_idx;
- }
- dst_idx++;
- converted = str_conv.popUTF32();
+ : reinterpret_cast<internal::mbstate *>(ps));
+ if (!ret.has_value()) {
+ // Encoding failure
+ libc_errno = ret.error();
+ return -1;
}
-
- *src += str_conv.getSourceIndex();
- if (converted.error() == -1) // if we hit conversion limit
- return dst_idx;
-
- libc_errno = converted.error();
- return -1;
+ return ret.value();
}
} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/wchar/mbstowcs.cpp b/libc/src/wchar/mbstowcs.cpp
index 66046903a8c6f..1a3232847fa71 100644
--- a/libc/src/wchar/mbstowcs.cpp
+++ b/libc/src/wchar/mbstowcs.cpp
@@ -21,6 +21,7 @@ namespace LIBC_NAMESPACE_DECL {
LLVM_LIBC_FUNCTION(size_t, mbstowcs,
(wchar_t *__restrict pwcs, const char *__restrict s,
size_t n)) {
+ // If destination is null, ignore n
n = pwcs == nullptr ? SIZE_MAX : n;
static internal::mbstate internal_mbstate;
const char *temp = s;
diff --git a/libc/test/src/wchar/CMakeLists.txt b/libc/test/src/wchar/CMakeLists.txt
index 9ed924eae34c4..1a8f9981fc5bb 100644
--- a/libc/test/src/wchar/CMakeLists.txt
+++ b/libc/test/src/wchar/CMakeLists.txt
@@ -93,6 +93,22 @@ add_libc_test(
libc.test.UnitTest.ErrnoCheckingTest
)
+add_libc_test(
+ mbsnrtowcs_test
+ SUITE
+ libc_wchar_unittests
+ SRCS
+ mbsnrtowcs_test.cpp
+ DEPENDS
+ libc.src.__support.libc_errno
+ libc.src.__support.wchar.mbstate
+ libc.src.string.memset
+ libc.src.wchar.mbsnrtowcs
+ libc.hdr.types.mbstate_t
+ libc.hdr.types.wchar_t
+ libc.test.UnitTest.ErrnoCheckingTest
+)
+
add_libc_test(
wctob_test
SUITE
diff --git a/libc/test/src/wchar/mbsnrtowcs_test.cpp b/libc/test/src/wchar/mbsnrtowcs_test.cpp
new file mode 100644
index 0000000000000..a042794997b9f
--- /dev/null
+++ b/libc/test/src/wchar/mbsnrtowcs_test.cpp
@@ -0,0 +1,203 @@
+//===-- Unittests for mbsetowcs -------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#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/mbsnrtowcs.h"
+#include "test/UnitTest/ErrnoCheckingTest.h"
+#include "test/UnitTest/Test.h"
+
+using LlvmLibcMBSNRToWCSTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
+
+TEST_F(LlvmLibcMBSNRToWCSTest, OneByteOneChar) {
+ const char *ch = "A";
+ const char *original = ch;
+ wchar_t dest[2];
+ mbstate_t *mb;
+ LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &ch, 1, 1, mb);
+ ASSERT_EQ(static_cast<char>(*dest), 'A');
+ ASSERT_EQ(static_cast<int>(n), 1);
+ // Should point to null terminator now
+ ASSERT_EQ(ch, original + 1);
+ ASSERT_ERRNO_SUCCESS();
+
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &ch, 1, 1, mb);
+ ASSERT_EQ(static_cast<char>(dest[1]), '\0');
+ // Should not include null terminator
+ ASSERT_EQ(static_cast<int>(n), 0);
+ // Should now be a nullptr
+ ASSERT_EQ(ch, nullptr);
+ ASSERT_ERRNO_SUCCESS();
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, FourByteOneChar) {
+ const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
+ const char *original = src;
+ wchar_t dest[2];
+ mbstate_t *mb;
+ LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+ // Not enough bytes for the full character
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 3, 2, mb);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(n), 0);
+ ASSERT_EQ(src, original + 3);
+ // Needs 2 more bytes (last byte of cat + null terminator)
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 2, 2, mb);
+ ASSERT_ERRNO_SUCCESS();
+ // Does not include null terminator
+ ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_EQ(src, nullptr);
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_TRUE(dest[1] == L'\0');
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, MixedNumberOfBytes) {
+ // 'A', sigma symbol 'Σ', recycling symbol '♻', laughing cat emoji '😹'
+ const char *src = "A\xce\xa3\xe2\x99\xbb\xf0\x9f\x98\xb9";
+ const char *original = src;
+ wchar_t dest[5];
+ mbstate_t *mb;
+ LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+
+ // Read 'A'
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 1, 1, mb);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<char>(dest[0]), 'A');
+ ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_EQ(src, original + 1);
+
+ // Read sigma 'Σ'
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &src, 2, 1, mb);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(dest[1]), 931);
+ ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_EQ(src, original + 3);
+
+ // Read recycling 'â™»'
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 2, 5, mb);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(n), 0);
+ ASSERT_EQ(src, original + 5);
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 1, 1, mb);
+ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_EQ(src, original + 6);
+ ASSERT_EQ(static_cast<int>(dest[2]), 9851);
+
+ // Read laughing cat emoji '😹'
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 3, &src, 4, 5, mb);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_EQ(src, original + 10);
+ ASSERT_EQ(static_cast<int>(dest[3]), 128569);
+
+
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 4, &src, 4, 4, nullptr);
+ ASSERT_TRUE(dest[4] == L'\0');
+ ASSERT_ERRNO_SUCCESS();
+ // Should not count null terminator in number
+ ASSERT_EQ(static_cast<int>(n), 0);
+ // Should now be a nullptr
+ ASSERT_EQ(src, nullptr);
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, ReadLessThanStringLength) {
+ // Four laughing cat emojis "😹😹😹😹"
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
+ wchar_t dest[5] = {L'a', L'b', L'c', L'd', L'e'};
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 100, 3, nullptr);
+ ASSERT_ERRNO_SUCCESS();
+ // Should have read 3 emojis
+ ASSERT_EQ(static_cast<int>(n), 3);
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[2]), 128569);
+ ASSERT_TRUE(dest[3] == L'd');
+ ASSERT_TRUE(dest[4] == L'e');
+ // Read three laughing cat emojis, 12 bytes
+ ASSERT_EQ(src, original + 12);
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, InvalidFirstByte) {
+ // 0x80 is invalid first byte of mb character
+ const char *src =
+ "\x80\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
+ wchar_t dest[3];
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 88, 88, nullptr);
+ // Should return error and set errno
+ ASSERT_EQ(static_cast<int>(n), -1);
+ ASSERT_ERRNO_EQ(EILSEQ);
+ // Should not update pointer
+ ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, InvalidMiddleByte) {
+ // The 7th byte is invalid for a 4 byte character
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
+ wchar_t dest[3];
+ mbstate_t *mb;
+ LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+ // Successfully read one character and first byte of the second character
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 5, 88, mb);
+ ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(src, original + 5);
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &src, 5, 88, mb);
+ // Should return error, set errno, and not update the pointer
+ ASSERT_EQ(static_cast<int>(n), -1);
+ ASSERT_ERRNO_EQ(EILSEQ);
+ ASSERT_EQ(src, original + 5);
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, NullDestination) {
+ // Four laughing cat emojis "😹😹😹😹"
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(nullptr, &src, 88, 88, nullptr);
+ ASSERT_ERRNO_SUCCESS();
+ // Null destination should ignore len and read till end of string
+ ASSERT_EQ(static_cast<int>(n), 4);
+ // It should also not change the src pointer
+ ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, ErrnoChecks) {
+ // Two laughing cat emojis and invalid 3rd mb char (3rd byte of it)
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
+ wchar_t dest[5];
+ mbstate_t *mb;
+ LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+ // First two bytes are valid --> should not set errno
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 80, 2, mb);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(n), 2);
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+ ASSERT_EQ(src, original + 8);
+
+ // Trying to read the 3rd byte should set errno
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 4, 2, mb);
+ ASSERT_ERRNO_EQ(EILSEQ);
+ ASSERT_EQ(static_cast<int>(n), -1);
+ // Should not move the pointer
+ ASSERT_EQ(src, original + 8);
+}
diff --git a/libc/test/src/wchar/mbsrtowcs_test.cpp b/libc/test/src/wchar/mbsrtowcs_test.cpp
index 2f6e6e6a71b31..fae6074487263 100644
--- a/libc/test/src/wchar/mbsrtowcs_test.cpp
+++ b/libc/test/src/wchar/mbsrtowcs_test.cpp
@@ -19,121 +19,158 @@ using LlvmLibcMBSRToWCSTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
TEST_F(LlvmLibcMBSRToWCSTest, OneByteOneChar) {
const char *ch = "A";
+ const char *original = ch;
wchar_t dest[2];
mbstate_t *mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &ch, 1, mb);
ASSERT_EQ(static_cast<char>(*dest), 'A');
ASSERT_EQ(static_cast<int>(n), 1);
+ // Should point to null terminator now
+ ASSERT_EQ(ch, original + 1);
ASSERT_ERRNO_SUCCESS();
n = LIBC_NAMESPACE::mbsrtowcs(dest + 1, &ch, 1, mb);
ASSERT_EQ(static_cast<char>(dest[1]), '\0');
// Should not include null terminator
ASSERT_EQ(static_cast<int>(n), 0);
+ // Should now be a nullptr
+ ASSERT_EQ(ch, nullptr);
ASSERT_ERRNO_SUCCESS();
}
-// TEST_F(LlvmLibcMBSRToWCSTest, FourByteOneChar) {
-// const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
-// wchar_t dest[2];
-// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
-// ASSERT_ERRNO_SUCCESS();
-// ASSERT_EQ(static_cast<int>(dest[0]), 128569);
-// ASSERT_TRUE(dest[1] == L'\0');
-// // Should not count null terminator in number
-// ASSERT_EQ(static_cast<int>(n), 1);
-// }
+TEST_F(LlvmLibcMBSRToWCSTest, FourByteOneChar) {
+ const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
+ wchar_t dest[2];
+ mbstate_t *mb;
+ LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 2, mb);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_TRUE(dest[1] == L'\0');
+ // Should not count null terminator in number
+ ASSERT_EQ(static_cast<int>(n), 1);
+ // Should now be a nullptr
+ ASSERT_EQ(src, nullptr);
+}
-// TEST_F(LlvmLibcMBSRToWCSTest, MultiByteTwoCharacters) {
-// // Two laughing cat emojis "😹😹"
-// const char *src = "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
-// wchar_t dest[3];
-// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
-// ASSERT_ERRNO_SUCCESS();
-// ASSERT_EQ(static_cast<int>(dest[0]), 128569);
-// ASSERT_EQ(static_cast<int>(dest[1]), 128569);
-// ASSERT_TRUE(dest[2] == L'\0');
-// // Should not count null terminator in number
-// ASSERT_EQ(static_cast<int>(n), 2);
-// }
+TEST_F(LlvmLibcMBSRToWCSTest, MultiByteTwoCharacters) {
+ // Two laughing cat emojis "😹😹"
+ const char *src = "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ wchar_t dest[3];
+ mbstate_t *mb;
+ LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 3, mb);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+ ASSERT_TRUE(dest[2] == L'\0');
+ // Should not count null terminator in number
+ ASSERT_EQ(static_cast<int>(n), 2);
+ // Should now be a nullptr
+ ASSERT_EQ(src, nullptr);
+}
-// TEST_F(LlvmLibcMBSRToWCSTest, MixedNumberOfBytes) {
-// // 'A', sigma symbol 'Σ', recycling symbol '♻', laughing cat emoji '😹'
-// const char *src = "A\xce\xa3\xe2\x99\xbb\xf0\x9f\x98\xb9";
-// wchar_t dest[5];
-// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 5);
-// ASSERT_ERRNO_SUCCESS();
-// ASSERT_EQ(static_cast<char>(dest[0]), 'A');
-// ASSERT_EQ(static_cast<int>(dest[1]), 931);
-// ASSERT_EQ(static_cast<int>(dest[2]), 9851);
-// ASSERT_EQ(static_cast<int>(dest[3]), 128569);
-// ASSERT_TRUE(dest[4] == L'\0');
-// // Should not count null terminator in number
-// ASSERT_EQ(static_cast<int>(n), 4);
-// }
+TEST_F(LlvmLibcMBSRToWCSTest, MixedNumberOfBytes) {
+ // 'A', sigma symbol 'Σ', recycling symbol '♻', laughing cat emoji '😹'
+ const char *src = "A\xce\xa3\xe2\x99\xbb\xf0\x9f\x98\xb9";
+ const char *original = src;
+ wchar_t dest[5];
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 4, nullptr);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<char>(dest[0]), 'A');
+ ASSERT_EQ(static_cast<int>(dest[1]), 931);
+ ASSERT_EQ(static_cast<int>(dest[2]), 9851);
+ ASSERT_EQ(static_cast<int>(dest[3]), 128569);
+ // Should point to null terminator (byte at 10th index)
+ ASSERT_EQ(src, original + 10);
+ ASSERT_EQ(static_cast<int>(n), 4);
+ n = LIBC_NAMESPACE::mbsrtowcs(dest + 4, &src, 4, nullptr);
+ ASSERT_TRUE(dest[4] == L'\0');
+ ASSERT_ERRNO_SUCCESS();
+ // Should not count null terminator in number
+ ASSERT_EQ(static_cast<int>(n), 0);
+ // Should now be a nullptr
+ ASSERT_EQ(src, nullptr);
+}
-// TEST_F(LlvmLibcMBSRToWCSTest, ReadLessThanStringLength) {
-// // Four laughing cat emojis "😹😹😹😹"
-// const char *src =
-// "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
-// wchar_t dest[5] = {L'a', L'b', L'c', L'd', L'e'};
-// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
-// ASSERT_ERRNO_SUCCESS();
-// // Should have read 3 emojis
-// ASSERT_EQ(static_cast<int>(n), 3);
-// ASSERT_EQ(static_cast<int>(dest[0]), 128569);
-// ASSERT_EQ(static_cast<int>(dest[1]), 128569);
-// ASSERT_EQ(static_cast<int>(dest[2]), 128569);
-// ASSERT_TRUE(dest[3] == L'd');
-// ASSERT_TRUE(dest[4] == L'e');
-// }
+TEST_F(LlvmLibcMBSRToWCSTest, ReadLessThanStringLength) {
+ // Four laughing cat emojis "😹😹😹😹"
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
+ wchar_t dest[5] = {L'a', L'b', L'c', L'd', L'e'};
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 3, nullptr);
+ ASSERT_ERRNO_SUCCESS();
+ // Should have read 3 emojis
+ ASSERT_EQ(static_cast<int>(n), 3);
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[2]), 128569);
+ ASSERT_TRUE(dest[3] == L'd');
+ ASSERT_TRUE(dest[4] == L'e');
+ // Read three laughing cat emojis, 12 bytes
+ ASSERT_EQ(src, original + 12);
+}
-// TEST_F(LlvmLibcMBSRToWCSTest, InvalidFirstByte) {
-// // 0x80 is invalid first byte of mb character
-// const char *src =
-// "\x80\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
-// wchar_t dest[3];
-// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
-// // Should return error and set errno
-// ASSERT_EQ(static_cast<int>(n), -1);
-// ASSERT_ERRNO_EQ(EILSEQ);
-// }
+TEST_F(LlvmLibcMBSRToWCSTest, InvalidFirstByte) {
+ // 0x80 is invalid first byte of mb character
+ const char *src =
+ "\x80\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
+ wchar_t dest[3];
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 3, nullptr);
+ // Should return error and set errno
+ ASSERT_EQ(static_cast<int>(n), -1);
+ ASSERT_ERRNO_EQ(EILSEQ);
+ // Should not update pointer
+ ASSERT_EQ(src, original);
+}
-// TEST_F(LlvmLibcMBSRToWCSTest, InvalidMiddleByte) {
-// // The 7th byte is invalid for a 4 byte character
-// const char *src =
-// "\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
-// wchar_t dest[3];
-// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 5);
-// // Should return error and set errno
-// ASSERT_EQ(static_cast<int>(n), -1);
-// ASSERT_ERRNO_EQ(EILSEQ);
-// }
+TEST_F(LlvmLibcMBSRToWCSTest, InvalidMiddleByte) {
+ // The 7th byte is invalid for a 4 byte character
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
+ wchar_t dest[3];
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 5, nullptr);
+ // Should return error, set errno, and not update the pointer
+ ASSERT_EQ(static_cast<int>(n), -1);
+ ASSERT_ERRNO_EQ(EILSEQ);
+ ASSERT_EQ(src, original);
+}
-// TEST_F(LlvmLibcMBSRToWCSTest, NullDestination) {
-// // Four laughing cat emojis "😹😹😹😹"
-// const char *src =
-// "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
-// size_t n = LIBC_NAMESPACE::mbstowcs(nullptr, src, 2);
-// ASSERT_ERRNO_SUCCESS();
-// // Null destination should ignore len and read till end of string
-// ASSERT_EQ(static_cast<int>(n), 4);
-// }
+TEST_F(LlvmLibcMBSRToWCSTest, NullDestination) {
+ // Four laughing cat emojis "😹😹😹😹"
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(nullptr, &src, 2, nullptr);
+ ASSERT_ERRNO_SUCCESS();
+ // Null destination should ignore len and read till end of string
+ ASSERT_EQ(static_cast<int>(n), 4);
+ // It should also not change the src pointer
+ ASSERT_EQ(src, original);
+}
-// TEST_F(LlvmLibcMBSRToWCSTest, ErrnoChecks) {
-// // Two laughing cat emojis and invalid 3rd mb char (3rd byte of it)
-// const char *src =
-// "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9";
-// wchar_t dest[5];
-// // First two bytes are valid --> should not set errno
-// size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
-// ASSERT_ERRNO_SUCCESS();
-// ASSERT_EQ(static_cast<int>(n), 2);
-// ASSERT_EQ(static_cast<int>(dest[0]), 128569);
-// ASSERT_EQ(static_cast<int>(dest[1]), 128569);
-// // Trying to read the 3rd byte should set errno
-// n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
-// ASSERT_ERRNO_EQ(EILSEQ);
-// ASSERT_EQ(static_cast<int>(n), -1);
-// }
+TEST_F(LlvmLibcMBSRToWCSTest, ErrnoChecks) {
+ // Two laughing cat emojis and invalid 3rd mb char (3rd byte of it)
+ const char *src =
+ "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
+ wchar_t dest[5];
+ // First two bytes are valid --> should not set errno
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 2, nullptr);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(n), 2);
+ ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+ ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+ ASSERT_EQ(src, original + 8);
+
+ // Trying to read the 3rd byte should set errno
+ n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 2, nullptr);
+ ASSERT_ERRNO_EQ(EILSEQ);
+ ASSERT_EQ(static_cast<int>(n), -1);
+ // Should not move the pointer
+ ASSERT_EQ(src, original + 8);
+}
diff --git a/libc/test/src/wchar/mbstowcs_test.cpp b/libc/test/src/wchar/mbstowcs_test.cpp
index 1f6633bdd2766..0fc84aa7a062e 100644
--- a/libc/test/src/wchar/mbstowcs_test.cpp
+++ b/libc/test/src/wchar/mbstowcs_test.cpp
@@ -16,21 +16,27 @@ using LlvmLibcMBSToWCSTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
TEST_F(LlvmLibcMBSToWCSTest, OneByteOneChar) {
const char *ch = "A";
+ const char *original = ch;
wchar_t dest[2];
size_t n = LIBC_NAMESPACE::mbstowcs(dest, ch, 1);
ASSERT_EQ(static_cast<char>(*dest), 'A');
ASSERT_EQ(static_cast<int>(n), 1);
+ // Making sure the pointer is not getting updated
+ ASSERT_EQ(ch, original);
ASSERT_ERRNO_SUCCESS();
n = LIBC_NAMESPACE::mbstowcs(dest + 1, ch + 1, 1);
ASSERT_EQ(static_cast<char>(dest[1]), '\0');
// Should not include null terminator
ASSERT_EQ(static_cast<int>(n), 0);
+ // Making sure the pointer is not getting updated
+ ASSERT_EQ(ch, original);
ASSERT_ERRNO_SUCCESS();
}
TEST_F(LlvmLibcMBSToWCSTest, FourByteOneChar) {
const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
+ const char *original = src;
wchar_t dest[2];
size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
ASSERT_ERRNO_SUCCESS();
@@ -38,11 +44,14 @@ TEST_F(LlvmLibcMBSToWCSTest, FourByteOneChar) {
ASSERT_TRUE(dest[1] == L'\0');
// Should not count null terminator in number
ASSERT_EQ(static_cast<int>(n), 1);
+ // Making sure the pointer is not getting updated
+ ASSERT_EQ(src, original);
}
TEST_F(LlvmLibcMBSToWCSTest, MultiByteTwoCharacters) {
// Two laughing cat emojis "😹😹"
const char *src = "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
wchar_t dest[3];
size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
ASSERT_ERRNO_SUCCESS();
@@ -51,11 +60,14 @@ TEST_F(LlvmLibcMBSToWCSTest, MultiByteTwoCharacters) {
ASSERT_TRUE(dest[2] == L'\0');
// Should not count null terminator in number
ASSERT_EQ(static_cast<int>(n), 2);
+ // Making sure the pointer is not getting updated
+ ASSERT_EQ(src, original);
}
TEST_F(LlvmLibcMBSToWCSTest, MixedNumberOfBytes) {
// 'A', sigma symbol 'Σ', recycling symbol '♻', laughing cat emoji '😹'
const char *src = "A\xce\xa3\xe2\x99\xbb\xf0\x9f\x98\xb9";
+ const char *original = src;
wchar_t dest[5];
size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 5);
ASSERT_ERRNO_SUCCESS();
@@ -66,12 +78,15 @@ TEST_F(LlvmLibcMBSToWCSTest, MixedNumberOfBytes) {
ASSERT_TRUE(dest[4] == L'\0');
// Should not count null terminator in number
ASSERT_EQ(static_cast<int>(n), 4);
+ // Making sure the pointer is not getting updated
+ ASSERT_EQ(src, original);
}
TEST_F(LlvmLibcMBSToWCSTest, ReadLessThanStringLength) {
// Four laughing cat emojis "😹😹😹😹"
const char *src =
"\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
wchar_t dest[5] = {L'a', L'b', L'c', L'd', L'e'};
size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
ASSERT_ERRNO_SUCCESS();
@@ -82,6 +97,8 @@ TEST_F(LlvmLibcMBSToWCSTest, ReadLessThanStringLength) {
ASSERT_EQ(static_cast<int>(dest[2]), 128569);
ASSERT_TRUE(dest[3] == L'd');
ASSERT_TRUE(dest[4] == L'e');
+ // Making sure the pointer is not getting updated
+ ASSERT_EQ(src, original);
}
TEST_F(LlvmLibcMBSToWCSTest, InvalidFirstByte) {
@@ -99,27 +116,34 @@ TEST_F(LlvmLibcMBSToWCSTest, InvalidMiddleByte) {
// The 7th byte is invalid for a 4 byte character
const char *src =
"\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
wchar_t dest[3];
size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 5);
// Should return error and set errno
ASSERT_EQ(static_cast<int>(n), -1);
ASSERT_ERRNO_EQ(EILSEQ);
+ // Making sure the pointer is not getting updated
+ ASSERT_EQ(src, original);
}
TEST_F(LlvmLibcMBSToWCSTest, NullDestination) {
// Four laughing cat emojis "😹😹😹😹"
const char *src =
"\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
size_t n = LIBC_NAMESPACE::mbstowcs(nullptr, src, 2);
ASSERT_ERRNO_SUCCESS();
// Null destination should ignore len and read till end of string
ASSERT_EQ(static_cast<int>(n), 4);
+ // Making sure the pointer is not getting updated
+ ASSERT_EQ(src, original);
}
TEST_F(LlvmLibcMBSToWCSTest, ErrnoChecks) {
// Two laughing cat emojis and invalid 3rd mb char (3rd byte of it)
const char *src =
"\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9";
+ const char *original = src;
wchar_t dest[5];
// First two bytes are valid --> should not set errno
size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
@@ -127,8 +151,12 @@ TEST_F(LlvmLibcMBSToWCSTest, ErrnoChecks) {
ASSERT_EQ(static_cast<int>(n), 2);
ASSERT_EQ(static_cast<int>(dest[0]), 128569);
ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+ // Making sure the pointer is not getting updated
+ ASSERT_EQ(src, original);
// Trying to read the 3rd byte should set errno
- n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
+ n = LIBC_NAMESPACE::mbstowcs(dest, src + 2, 2);
ASSERT_ERRNO_EQ(EILSEQ);
ASSERT_EQ(static_cast<int>(n), -1);
+ // Making sure the pointer is not getting updated
+ ASSERT_EQ(src, original);
}
>From b3b7b2a168d794a95eb8004b33c0988cf73bc459 Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Thu, 17 Jul 2025 23:31:58 +0000
Subject: [PATCH 08/12] fixed formatting
---
libc/src/wchar/mbsnrtowcs.h | 2 +-
libc/test/src/wchar/mbsnrtowcs_test.cpp | 19 +++++++++----------
2 files changed, 10 insertions(+), 11 deletions(-)
diff --git a/libc/src/wchar/mbsnrtowcs.h b/libc/src/wchar/mbsnrtowcs.h
index 5c9687b20b3fa..0d66b952db01a 100644
--- a/libc/src/wchar/mbsnrtowcs.h
+++ b/libc/src/wchar/mbsnrtowcs.h
@@ -17,7 +17,7 @@
namespace LIBC_NAMESPACE_DECL {
size_t mbsnrtowcs(wchar_t *__restrict dst, const char **__restrict src,
- size_t nmc, size_t len, mbstate_t *__restrict ps);
+ size_t nmc, size_t len, mbstate_t *__restrict ps);
} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/test/src/wchar/mbsnrtowcs_test.cpp b/libc/test/src/wchar/mbsnrtowcs_test.cpp
index a042794997b9f..be6f4a45fac42 100644
--- a/libc/test/src/wchar/mbsnrtowcs_test.cpp
+++ b/libc/test/src/wchar/mbsnrtowcs_test.cpp
@@ -67,40 +67,39 @@ TEST_F(LlvmLibcMBSNRToWCSTest, MixedNumberOfBytes) {
wchar_t dest[5];
mbstate_t *mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
-
+
// Read 'A'
size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 1, 1, mb);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(static_cast<char>(dest[0]), 'A');
- ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_EQ(static_cast<int>(n), 1);
ASSERT_EQ(src, original + 1);
// Read sigma 'Σ'
n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &src, 2, 1, mb);
ASSERT_ERRNO_SUCCESS();
- ASSERT_EQ(static_cast<int>(dest[1]), 931);
- ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_EQ(static_cast<int>(dest[1]), 931);
+ ASSERT_EQ(static_cast<int>(n), 1);
ASSERT_EQ(src, original + 3);
// Read recycling 'â™»'
n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 2, 5, mb);
ASSERT_ERRNO_SUCCESS();
- ASSERT_EQ(static_cast<int>(n), 0);
+ ASSERT_EQ(static_cast<int>(n), 0);
ASSERT_EQ(src, original + 5);
n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 1, 1, mb);
-ASSERT_ERRNO_SUCCESS();
- ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_ERRNO_SUCCESS();
+ ASSERT_EQ(static_cast<int>(n), 1);
ASSERT_EQ(src, original + 6);
ASSERT_EQ(static_cast<int>(dest[2]), 9851);
// Read laughing cat emoji '😹'
n = LIBC_NAMESPACE::mbsnrtowcs(dest + 3, &src, 4, 5, mb);
ASSERT_ERRNO_SUCCESS();
- ASSERT_EQ(static_cast<int>(n), 1);
+ ASSERT_EQ(static_cast<int>(n), 1);
ASSERT_EQ(src, original + 10);
ASSERT_EQ(static_cast<int>(dest[3]), 128569);
-
n = LIBC_NAMESPACE::mbsnrtowcs(dest + 4, &src, 4, 4, nullptr);
ASSERT_TRUE(dest[4] == L'\0');
ASSERT_ERRNO_SUCCESS();
@@ -149,7 +148,7 @@ TEST_F(LlvmLibcMBSNRToWCSTest, InvalidMiddleByte) {
"\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
const char *original = src;
wchar_t dest[3];
- mbstate_t *mb;
+ mbstate_t *mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
// Successfully read one character and first byte of the second character
size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 5, 88, mb);
>From d8f2d6fab458d3b6899dbf38dc99aedf9e8b542d Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Fri, 18 Jul 2025 18:03:09 +0000
Subject: [PATCH 09/12] added crash on nullptr for src and moved internal
implementation to header file
---
libc/src/__support/wchar/CMakeLists.txt | 4 +-
libc/src/__support/wchar/mbsnrtowcs.cpp | 59 -------------------------
libc/src/__support/wchar/mbsnrtowcs.h | 44 +++++++++++++++---
libc/src/wchar/CMakeLists.txt | 2 -
libc/test/src/wchar/mbsnrtowcs_test.cpp | 45 +++++++++++--------
libc/test/src/wchar/mbsrtowcs_test.cpp | 23 +++++++---
libc/test/src/wchar/mbstowcs_test.cpp | 8 ++++
7 files changed, 89 insertions(+), 96 deletions(-)
delete mode 100644 libc/src/__support/wchar/mbsnrtowcs.cpp
diff --git a/libc/src/__support/wchar/CMakeLists.txt b/libc/src/__support/wchar/CMakeLists.txt
index fa5139a8b68b4..607bfe6447c3c 100644
--- a/libc/src/__support/wchar/CMakeLists.txt
+++ b/libc/src/__support/wchar/CMakeLists.txt
@@ -69,12 +69,10 @@ add_object_library(
.mbstate
)
-add_object_library(
+add_header_library(
mbsnrtowcs
HDRS
mbsnrtowcs.h
- SRCS
- mbsnrtowcs.cpp
DEPENDS
libc.hdr.errno_macros
libc.hdr.types.wchar_t
diff --git a/libc/src/__support/wchar/mbsnrtowcs.cpp b/libc/src/__support/wchar/mbsnrtowcs.cpp
deleted file mode 100644
index 9eb2b5423af87..0000000000000
--- a/libc/src/__support/wchar/mbsnrtowcs.cpp
+++ /dev/null
@@ -1,59 +0,0 @@
-//===-- Implementation for mbsnrtowcs function ------------------*- C++ -*-===//
-//
-// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-//
-//===----------------------------------------------------------------------===//
-
-#include "src/__support/wchar/mbsnrtowcs.h"
-#include "hdr/errno_macros.h"
-#include "hdr/types/size_t.h"
-#include "hdr/types/wchar_t.h"
-#include "src/__support/common.h"
-#include "src/__support/error_or.h"
-#include "src/__support/macros/config.h"
-#include "src/__support/wchar/character_converter.h"
-#include "src/__support/wchar/mbstate.h"
-#include "src/__support/wchar/string_converter.h"
-
-namespace LIBC_NAMESPACE_DECL {
-namespace internal {
-
-ErrorOr<size_t> mbsnrtowcs(wchar_t *__restrict dst, const char **__restrict src,
- size_t nmc, size_t len, mbstate *__restrict ps) {
- if (*src == nullptr)
- return 0;
- // Checking if mbstate is valid
- CharacterConverter char_conv(ps);
- if (!char_conv.isValidState())
- return Error(EINVAL);
-
- StringConverter<char8_t> str_conv(reinterpret_cast<const char8_t *>(*src), ps,
- len, nmc);
- size_t dst_idx = 0;
- ErrorOr<char32_t> converted = str_conv.popUTF32();
- while (converted.has_value()) {
- if (dst != nullptr)
- dst[dst_idx] = converted.value();
- // null terminator should not be counted in return value
- if (converted.value() == L'\0') {
- if (dst != nullptr)
- *src = nullptr;
- return dst_idx;
- }
- dst_idx++;
- converted = str_conv.popUTF32();
- }
-
- if (converted.error() == -1) { // if we hit conversion limit
- *src += str_conv.getSourceIndex();
- return dst_idx;
- }
-
- return Error(converted.error());
-}
-
-} // namespace internal
-
-} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/__support/wchar/mbsnrtowcs.h b/libc/src/__support/wchar/mbsnrtowcs.h
index 8ce497325fdad..3aedd08aefea5 100644
--- a/libc/src/__support/wchar/mbsnrtowcs.h
+++ b/libc/src/__support/wchar/mbsnrtowcs.h
@@ -1,4 +1,4 @@
-//===-- Implementation header for mbsnrtowcs function -----------*- C++ -*-===//
+//===-- Implementation for mbsnrtowcs function ------------------*- C++ -*-===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
@@ -6,24 +6,54 @@
//
//===----------------------------------------------------------------------===//
-#ifndef LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS
-#define LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS
-
+#include "hdr/errno_macros.h"
#include "hdr/types/size_t.h"
#include "hdr/types/wchar_t.h"
#include "src/__support/common.h"
#include "src/__support/error_or.h"
#include "src/__support/macros/config.h"
+#include "src/__support/macros/null_check.h"
+#include "src/__support/wchar/character_converter.h"
#include "src/__support/wchar/mbstate.h"
+#include "src/__support/wchar/string_converter.h"
namespace LIBC_NAMESPACE_DECL {
namespace internal {
ErrorOr<size_t> mbsnrtowcs(wchar_t *__restrict dst, const char **__restrict src,
- size_t nmc, size_t len, mbstate *__restrict ps);
+ size_t nmc, size_t len, mbstate *__restrict ps) {
+ LIBC_CRASH_ON_NULLPTR(src);
+ // Checking if mbstate is valid
+ CharacterConverter char_conv(ps);
+ if (!char_conv.isValidState())
+ return Error(EINVAL);
+
+ StringConverter<char8_t> str_conv(reinterpret_cast<const char8_t *>(*src), ps,
+ len, nmc);
+ size_t dst_idx = 0;
+ ErrorOr<char32_t> converted = str_conv.popUTF32();
+ while (converted.has_value()) {
+ if (dst != nullptr)
+ dst[dst_idx] = converted.value();
+ // null terminator should not be counted in return value
+ if (converted.value() == L'\0') {
+ if (dst != nullptr)
+ *src = nullptr;
+ return dst_idx;
+ }
+ dst_idx++;
+ converted = str_conv.popUTF32();
+ }
+
+ if (converted.error() == -1) { // if we hit conversion limit
+ if (dst != nullptr)
+ *src += str_conv.getSourceIndex();
+ return dst_idx;
+ }
+
+ return Error(converted.error());
+}
} // namespace internal
} // namespace LIBC_NAMESPACE_DECL
-
-#endif // LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS
diff --git a/libc/src/wchar/CMakeLists.txt b/libc/src/wchar/CMakeLists.txt
index ac359064ec10c..6df24a810bc6d 100644
--- a/libc/src/wchar/CMakeLists.txt
+++ b/libc/src/wchar/CMakeLists.txt
@@ -189,7 +189,6 @@ add_entrypoint_object(
libc.src.__support.libc_errno
libc.src.__support.wchar.mbstate
libc.src.__support.wchar.mbsnrtowcs
- libc.src.__support.wchar.mbstate
)
add_entrypoint_object(
@@ -206,7 +205,6 @@ add_entrypoint_object(
libc.src.__support.libc_errno
libc.src.__support.wchar.mbstate
libc.src.__support.wchar.mbsnrtowcs
- libc.src.__support.wchar.mbstate
)
add_entrypoint_object(
diff --git a/libc/test/src/wchar/mbsnrtowcs_test.cpp b/libc/test/src/wchar/mbsnrtowcs_test.cpp
index be6f4a45fac42..d1232e36a5880 100644
--- a/libc/test/src/wchar/mbsnrtowcs_test.cpp
+++ b/libc/test/src/wchar/mbsnrtowcs_test.cpp
@@ -21,16 +21,16 @@ TEST_F(LlvmLibcMBSNRToWCSTest, OneByteOneChar) {
const char *ch = "A";
const char *original = ch;
wchar_t dest[2];
- mbstate_t *mb;
+ mbstate_t mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
- size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &ch, 1, 1, mb);
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &ch, 1, 1, &mb);
ASSERT_EQ(static_cast<char>(*dest), 'A');
ASSERT_EQ(static_cast<int>(n), 1);
// Should point to null terminator now
ASSERT_EQ(ch, original + 1);
ASSERT_ERRNO_SUCCESS();
- n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &ch, 1, 1, mb);
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &ch, 1, 1, &mb);
ASSERT_EQ(static_cast<char>(dest[1]), '\0');
// Should not include null terminator
ASSERT_EQ(static_cast<int>(n), 0);
@@ -43,15 +43,15 @@ TEST_F(LlvmLibcMBSNRToWCSTest, FourByteOneChar) {
const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
const char *original = src;
wchar_t dest[2];
- mbstate_t *mb;
+ mbstate_t mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
// Not enough bytes for the full character
- size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 3, 2, mb);
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 3, 2, &mb);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(static_cast<int>(n), 0);
ASSERT_EQ(src, original + 3);
// Needs 2 more bytes (last byte of cat + null terminator)
- n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 2, 2, mb);
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 2, 2, &mb);
ASSERT_ERRNO_SUCCESS();
// Does not include null terminator
ASSERT_EQ(static_cast<int>(n), 1);
@@ -65,36 +65,36 @@ TEST_F(LlvmLibcMBSNRToWCSTest, MixedNumberOfBytes) {
const char *src = "A\xce\xa3\xe2\x99\xbb\xf0\x9f\x98\xb9";
const char *original = src;
wchar_t dest[5];
- mbstate_t *mb;
+ mbstate_t mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
// Read 'A'
- size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 1, 1, mb);
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 1, 1, &mb);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(static_cast<char>(dest[0]), 'A');
ASSERT_EQ(static_cast<int>(n), 1);
ASSERT_EQ(src, original + 1);
// Read sigma 'Σ'
- n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &src, 2, 1, mb);
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &src, 2, 1, &mb);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(static_cast<int>(dest[1]), 931);
ASSERT_EQ(static_cast<int>(n), 1);
ASSERT_EQ(src, original + 3);
// Read recycling 'â™»'
- n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 2, 5, mb);
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 2, 5, &mb);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(static_cast<int>(n), 0);
ASSERT_EQ(src, original + 5);
- n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 1, 1, mb);
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 1, 1, &mb);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(static_cast<int>(n), 1);
ASSERT_EQ(src, original + 6);
ASSERT_EQ(static_cast<int>(dest[2]), 9851);
// Read laughing cat emoji '😹'
- n = LIBC_NAMESPACE::mbsnrtowcs(dest + 3, &src, 4, 5, mb);
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 3, &src, 4, 5, &mb);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(static_cast<int>(n), 1);
ASSERT_EQ(src, original + 10);
@@ -148,16 +148,16 @@ TEST_F(LlvmLibcMBSNRToWCSTest, InvalidMiddleByte) {
"\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
const char *original = src;
wchar_t dest[3];
- mbstate_t *mb;
+ mbstate_t mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
// Successfully read one character and first byte of the second character
- size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 5, 88, mb);
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 5, 88, &mb);
ASSERT_EQ(static_cast<int>(n), 1);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(src, original + 5);
ASSERT_EQ(static_cast<int>(dest[0]), 128569);
- n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &src, 5, 88, mb);
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &src, 5, 88, &mb);
// Should return error, set errno, and not update the pointer
ASSERT_EQ(static_cast<int>(n), -1);
ASSERT_ERRNO_EQ(EILSEQ);
@@ -183,10 +183,10 @@ TEST_F(LlvmLibcMBSNRToWCSTest, ErrnoChecks) {
"\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9";
const char *original = src;
wchar_t dest[5];
- mbstate_t *mb;
+ mbstate_t mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
// First two bytes are valid --> should not set errno
- size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 80, 2, mb);
+ size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 80, 2, &mb);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(static_cast<int>(n), 2);
ASSERT_EQ(static_cast<int>(dest[0]), 128569);
@@ -194,9 +194,18 @@ TEST_F(LlvmLibcMBSNRToWCSTest, ErrnoChecks) {
ASSERT_EQ(src, original + 8);
// Trying to read the 3rd byte should set errno
- n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 4, 2, mb);
+ n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 4, 2, &mb);
ASSERT_ERRNO_EQ(EILSEQ);
ASSERT_EQ(static_cast<int>(n), -1);
// Should not move the pointer
ASSERT_EQ(src, original + 8);
}
+
+#if defined(LIBC_ADD_NULL_CHECKS) && !defined(LIBC_HAS_SANITIZER)
+TEST(LlvmLibcMBSNRToWCSTest, NullptrCrash) {
+ // Passing in a nullptr should crash the program.
+ EXPECT_DEATH(
+ [] { LIBC_NAMESPACE::mbsnrtowcs(nullptr, nullptr, 1, 1, nullptr); },
+ WITH_SIGNAL(-1));
+}
+#endif // LIBC_HAS_ADDRESS_SANITIZER
diff --git a/libc/test/src/wchar/mbsrtowcs_test.cpp b/libc/test/src/wchar/mbsrtowcs_test.cpp
index fae6074487263..00cb1d5cbadcd 100644
--- a/libc/test/src/wchar/mbsrtowcs_test.cpp
+++ b/libc/test/src/wchar/mbsrtowcs_test.cpp
@@ -21,16 +21,16 @@ TEST_F(LlvmLibcMBSRToWCSTest, OneByteOneChar) {
const char *ch = "A";
const char *original = ch;
wchar_t dest[2];
- mbstate_t *mb;
+ mbstate_t mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
- size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &ch, 1, mb);
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &ch, 1, &mb);
ASSERT_EQ(static_cast<char>(*dest), 'A');
ASSERT_EQ(static_cast<int>(n), 1);
// Should point to null terminator now
ASSERT_EQ(ch, original + 1);
ASSERT_ERRNO_SUCCESS();
- n = LIBC_NAMESPACE::mbsrtowcs(dest + 1, &ch, 1, mb);
+ n = LIBC_NAMESPACE::mbsrtowcs(dest + 1, &ch, 1, &mb);
ASSERT_EQ(static_cast<char>(dest[1]), '\0');
// Should not include null terminator
ASSERT_EQ(static_cast<int>(n), 0);
@@ -42,9 +42,9 @@ TEST_F(LlvmLibcMBSRToWCSTest, OneByteOneChar) {
TEST_F(LlvmLibcMBSRToWCSTest, FourByteOneChar) {
const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
wchar_t dest[2];
- mbstate_t *mb;
+ mbstate_t mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
- size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 2, mb);
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 2, &mb);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(static_cast<int>(dest[0]), 128569);
ASSERT_TRUE(dest[1] == L'\0');
@@ -58,9 +58,9 @@ TEST_F(LlvmLibcMBSRToWCSTest, MultiByteTwoCharacters) {
// Two laughing cat emojis "😹😹"
const char *src = "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
wchar_t dest[3];
- mbstate_t *mb;
+ mbstate_t mb;
LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
- size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 3, mb);
+ size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 3, &mb);
ASSERT_ERRNO_SUCCESS();
ASSERT_EQ(static_cast<int>(dest[0]), 128569);
ASSERT_EQ(static_cast<int>(dest[1]), 128569);
@@ -174,3 +174,12 @@ TEST_F(LlvmLibcMBSRToWCSTest, ErrnoChecks) {
// Should not move the pointer
ASSERT_EQ(src, original + 8);
}
+
+#if defined(LIBC_ADD_NULL_CHECKS) && !defined(LIBC_HAS_SANITIZER)
+TEST(LlvmLibcMBSRToWCSTest, NullptrCrash) {
+ // Passing in a nullptr should crash the program.
+ EXPECT_DEATH(
+ [] { LIBC_NAMESPACE::mbsrtowcs(nullptr, nullptr, 1, nullptr); },
+ WITH_SIGNAL(-1));
+}
+#endif // LIBC_HAS_ADDRESS_SANITIZER
diff --git a/libc/test/src/wchar/mbstowcs_test.cpp b/libc/test/src/wchar/mbstowcs_test.cpp
index 0fc84aa7a062e..db03bc3e38643 100644
--- a/libc/test/src/wchar/mbstowcs_test.cpp
+++ b/libc/test/src/wchar/mbstowcs_test.cpp
@@ -160,3 +160,11 @@ TEST_F(LlvmLibcMBSToWCSTest, ErrnoChecks) {
// Making sure the pointer is not getting updated
ASSERT_EQ(src, original);
}
+
+#if defined(LIBC_ADD_NULL_CHECKS) && !defined(LIBC_HAS_SANITIZER)
+TEST(LlvmLibcMBSToWCSTest, NullptrCrash) {
+ // Passing in a nullptr should crash the program.
+ EXPECT_DEATH([] { LIBC_NAMESPACE::mbstowcs(nullptr, nullptr, 1); },
+ WITH_SIGNAL(-1));
+}
+#endif // LIBC_HAS_ADDRESS_SANITIZER
>From 18e2f267b4145237bf07c38f89f2f9104f2d360e Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Fri, 18 Jul 2025 18:07:28 +0000
Subject: [PATCH 10/12] fixed formatting
---
libc/test/src/wchar/mbsrtowcs_test.cpp | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/libc/test/src/wchar/mbsrtowcs_test.cpp b/libc/test/src/wchar/mbsrtowcs_test.cpp
index 00cb1d5cbadcd..73b650fe85484 100644
--- a/libc/test/src/wchar/mbsrtowcs_test.cpp
+++ b/libc/test/src/wchar/mbsrtowcs_test.cpp
@@ -178,8 +178,7 @@ TEST_F(LlvmLibcMBSRToWCSTest, ErrnoChecks) {
#if defined(LIBC_ADD_NULL_CHECKS) && !defined(LIBC_HAS_SANITIZER)
TEST(LlvmLibcMBSRToWCSTest, NullptrCrash) {
// Passing in a nullptr should crash the program.
- EXPECT_DEATH(
- [] { LIBC_NAMESPACE::mbsrtowcs(nullptr, nullptr, 1, nullptr); },
- WITH_SIGNAL(-1));
+ EXPECT_DEATH([] { LIBC_NAMESPACE::mbsrtowcs(nullptr, nullptr, 1, nullptr); },
+ WITH_SIGNAL(-1));
}
#endif // LIBC_HAS_ADDRESS_SANITIZER
>From 84ad6a9cf18680f0e2fc2dbee5786402099e1636 Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Fri, 18 Jul 2025 19:34:18 +0000
Subject: [PATCH 11/12] header file changes
---
libc/src/__support/wchar/CMakeLists.txt | 1 +
libc/src/__support/wchar/mbsnrtowcs.h | 11 +++++++++--
2 files changed, 10 insertions(+), 2 deletions(-)
diff --git a/libc/src/__support/wchar/CMakeLists.txt b/libc/src/__support/wchar/CMakeLists.txt
index 607bfe6447c3c..9e050247124d4 100644
--- a/libc/src/__support/wchar/CMakeLists.txt
+++ b/libc/src/__support/wchar/CMakeLists.txt
@@ -80,6 +80,7 @@ add_header_library(
libc.src.__support.common
libc.src.__support.error_or
libc.src.__support.macros.config
+ libc.src.__support.nacros.null_check
.character_converter
.mbstate
.string_converter
diff --git a/libc/src/__support/wchar/mbsnrtowcs.h b/libc/src/__support/wchar/mbsnrtowcs.h
index 3aedd08aefea5..54e315210d95c 100644
--- a/libc/src/__support/wchar/mbsnrtowcs.h
+++ b/libc/src/__support/wchar/mbsnrtowcs.h
@@ -6,6 +6,9 @@
//
//===----------------------------------------------------------------------===//
+#ifndef LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS_H
+#define LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS_H
+
#include "hdr/errno_macros.h"
#include "hdr/types/size_t.h"
#include "hdr/types/wchar_t.h"
@@ -20,8 +23,10 @@
namespace LIBC_NAMESPACE_DECL {
namespace internal {
-ErrorOr<size_t> mbsnrtowcs(wchar_t *__restrict dst, const char **__restrict src,
- size_t nmc, size_t len, mbstate *__restrict ps) {
+LIBC_INLINE static ErrorOr<size_t> mbsnrtowcs(wchar_t *__restrict dst,
+ const char **__restrict src,
+ size_t nmc, size_t len,
+ mbstate *__restrict ps) {
LIBC_CRASH_ON_NULLPTR(src);
// Checking if mbstate is valid
CharacterConverter char_conv(ps);
@@ -57,3 +62,5 @@ ErrorOr<size_t> mbsnrtowcs(wchar_t *__restrict dst, const char **__restrict src,
} // namespace internal
} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS_H
>From 6595702442265f55e2a40dc76b3db1902f7c0290 Mon Sep 17 00:00:00 2001
From: Sriya Pratipati <sriyap at google.com>
Date: Fri, 18 Jul 2025 20:08:24 +0000
Subject: [PATCH 12/12] fixed spacing in cmake file
---
libc/src/__support/wchar/CMakeLists.txt | 20 ++++++++++----------
1 file changed, 10 insertions(+), 10 deletions(-)
diff --git a/libc/src/__support/wchar/CMakeLists.txt b/libc/src/__support/wchar/CMakeLists.txt
index 9e050247124d4..9c5815d1040ab 100644
--- a/libc/src/__support/wchar/CMakeLists.txt
+++ b/libc/src/__support/wchar/CMakeLists.txt
@@ -74,14 +74,14 @@ add_header_library(
HDRS
mbsnrtowcs.h
DEPENDS
- libc.hdr.errno_macros
- libc.hdr.types.wchar_t
- libc.hdr.types.size_t
- libc.src.__support.common
- libc.src.__support.error_or
- libc.src.__support.macros.config
- libc.src.__support.nacros.null_check
- .character_converter
- .mbstate
- .string_converter
+ libc.hdr.errno_macros
+ libc.hdr.types.wchar_t
+ libc.hdr.types.size_t
+ libc.src.__support.common
+ libc.src.__support.error_or
+ libc.src.__support.macros.config
+ libc.src.__support.nacros.null_check
+ .character_converter
+ .mbstate
+ .string_converter
)
More information about the libc-commits
mailing list