fixed edge case with null dest; added tests

Author: uzairnawaz

libc/src/__support/wchar/wcsrtombs.cpp

@@ -25,29 +25,32 @@ ErrorOr<size_t> wcsrtombs(char *__restrict dst, const wchar_t **__restrict src,
                           size_t len, mbstate *__restrict ps) {
   static_assert(sizeof(wchar_t) == 4);
 
-  if (src == nullptr || dst == nullptr)
+  if (src == nullptr)
     return Error(-1);
+  
+  // ignore len parameter when theres no destination string
+  if (dst == nullptr)
+    len = SIZE_MAX;
 
   size_t bytes_written = 0;
-  const wchar_t *wc_ptr = *src;
   while (bytes_written < len) {
     char buf[4];
-    auto result = internal::wcrtomb(dst + bytes_written, *wc_ptr, ps,
-                                    len - bytes_written);
+    auto result =
+        internal::wcrtomb(dst + bytes_written, **src, ps, len - bytes_written);
     if (!result.has_value())
       return result; // forward the error
 
     if (result.value() == -1) // couldn't complete the conversion
       return len;
 
     // terminate the loop after converting the null wide character
-    if (*wc_ptr == L'\0') {
+    if (**src == L'\0') {
       *src = '\0';
       return bytes_written;
     }
 
     bytes_written += result.value();
-    wc_ptr++;
+    (*src)++;
   }
 
   return bytes_written;

libc/src/wchar/wcsrtombs.cpp

@@ -25,9 +25,6 @@ LLVM_LIBC_FUNCTION(size_t, wcsrtombs,
   static internal::mbstate internal_mbstate;
 
   LIBC_CRASH_ON_NULLPTR(src);
-  char buf[len];
-  if (dst == nullptr)
-    dst = buf;
 
   auto result = internal::wcsrtombs(
       dst, src, len,

libc/test/src/wchar/wcsrtombs_test.cpp

@@ -0,0 +1,186 @@
+//===-- Unittests for wcsrtombs ------------------------------------------===//
+//
+// 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/string/memset.h"
+#include "src/wchar/wcsrtombs.h"
+#include "test/UnitTest/Test.h"
+
+TEST(LlvmLibcWCSRToMBSTest, SingleCharacterOneByte) {
+  mbstate_t state;
+  LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
+  const wchar_t *wcs = L"U";
+  const wchar_t *wcs_start = wcs;
+  char mbs[] = {0, 0};
+  size_t cnt = LIBC_NAMESPACE::wcsrtombs(mbs, &wcs, 2, &state);
+  ASSERT_EQ(cnt, static_cast<size_t>(1));
+  ASSERT_EQ(mbs[0], 'U');
+  ASSERT_EQ(mbs[1], '\0');
+  ASSERT_EQ(wcs, wcs_start + 1);
+}
+
+TEST(LlvmLibcWCSRToMBSTest, MultipleCompleteConversions) {
+  mbstate_t state;
+  LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
+
+  // utf32: 0xff -> utf8: 0xc3 0xbf
+  // utf32: 0xac15 -> utf8: 0xea 0xb0 0x95
+  const wchar_t *wcs = L"\xFF\xAC15";
+  const wchar_t *wcs_start = wcs;
+
+  // init with dummy value of 1 so that we can check when null byte written
+  char mbs[7] = {1, 1, 1, 1, 1, 1, 1};
+  char expected[6] = {0xC3, 0xBF, 0xEA, 0xB0, 0x95, 0x00};
+
+  size_t cnt1 = LIBC_NAMESPACE::wcsrtombs(mbs, &wcs, 2, &state);
+  ASSERT_EQ(cnt1, static_cast<size_t>(2));
+  ASSERT_EQ(wcs, wcs_start + 1);
+  ASSERT_EQ(mbs[0], expected[0]);
+  ASSERT_EQ(mbs[1], expected[1]);
+  ASSERT_EQ(mbs[2], '\x01'); // not modified
+
+  size_t cnt2 = LIBC_NAMESPACE::wcsrtombs(mbs + cnt1, &wcs, 3, &state);
+  ASSERT_EQ(cnt2, static_cast<size_t>(3));
+  ASSERT_EQ(wcs, wcs_start + 2);
+  ASSERT_EQ(mbs[0], expected[0]);
+  ASSERT_EQ(mbs[1], expected[1]);
+  ASSERT_EQ(mbs[2], expected[2]);
+  ASSERT_EQ(mbs[3], expected[3]);
+  ASSERT_EQ(mbs[4], expected[4]);
+  ASSERT_EQ(mbs[5], '\x01'); // null byte not yet written
+
+  // all that is left in the string is the null terminator
+  size_t cnt3 = LIBC_NAMESPACE::wcsrtombs(mbs + cnt1 + cnt2, &wcs, 50, &state);
+  ASSERT_EQ(cnt3, static_cast<size_t>(0));
+  ASSERT_EQ(wcs, nullptr);
+  ASSERT_EQ(mbs[0], expected[0]);
+  ASSERT_EQ(mbs[1], expected[1]);
+  ASSERT_EQ(mbs[2], expected[2]);
+  ASSERT_EQ(mbs[3], expected[3]);
+  ASSERT_EQ(mbs[4], expected[4]);
+  ASSERT_EQ(mbs[5], expected[5]);
+  ASSERT_EQ(mbs[6], '\x01'); // should not write beyond null terminator
+}
+
+TEST(LlvmLibcWCSRToMBSTest, MultiplePartialConversions) {
+  mbstate_t state;
+  LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
+
+  // utf32: 0xff -> utf8: 0xc3 0xbf
+  // utf32: 0xac15 -> utf8: 0xea 0xb0 0x95
+  const wchar_t *wcs = L"\xFF\xAC15";
+  const wchar_t *wcs_start = wcs;
+
+  // init with dummy value of 1 so that we can check when null byte written
+  char mbs[7] = {1, 1, 1, 1, 1, 1, 1};
+  char expected[6] = {0xC3, 0xBF, 0xEA, 0xB0, 0x95, 0x00};
+  size_t written = 0;
+  size_t count = 0;
+
+  count = LIBC_NAMESPACE::wcsrtombs(mbs, &wcs, 1, &state);
+  written += count;
+  ASSERT_EQ(count, static_cast<size_t>(1));
+  ASSERT_EQ(wcs, wcs_start);
+  ASSERT_EQ(mbs[0], expected[0]);
+  ASSERT_EQ(mbs[1], '\x01');
+
+  count = LIBC_NAMESPACE::wcsrtombs(mbs + written, &wcs, 2, &state);
+  written += count;
+  ASSERT_EQ(count, static_cast<size_t>(2));
+  ASSERT_EQ(wcs, wcs_start + 1);
+  ASSERT_EQ(mbs[0], expected[0]);
+  ASSERT_EQ(mbs[1], expected[1]);
+  ASSERT_EQ(mbs[2], expected[2]);
+  ASSERT_EQ(mbs[3], '\x01');
+
+  count = LIBC_NAMESPACE::wcsrtombs(mbs + written, &wcs, 3, &state);
+  written += count;
+  ASSERT_EQ(count, static_cast<size_t>(2));
+  ASSERT_EQ(wcs, nullptr);
+  ASSERT_EQ(mbs[0], expected[0]);
+  ASSERT_EQ(mbs[1], expected[1]);
+  ASSERT_EQ(mbs[2], expected[2]);
+  ASSERT_EQ(mbs[3], expected[3]);
+  ASSERT_EQ(mbs[4], expected[4]);
+  ASSERT_EQ(mbs[5], expected[5]);
+  ASSERT_EQ(mbs[6], '\x01');
+}
+
+TEST(LlvmLibcWCSRToMBSTest, NullDestination) {
+  mbstate_t state;
+  LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
+
+  // utf32: 0x1f921 -> utf8: 0xf0 0x9f 0xa4 0xa1
+  // utf32: 0xac15 -> utf8: 0xea 0xb0 0x95
+  const wchar_t *wcs = L"\x1F921\xAC15";
+
+  // null destination means the conversion isnt stored, but all the side effects
+  // still occur. the len parameter is also ignored
+  size_t count = LIBC_NAMESPACE::wcsrtombs(nullptr, &wcs, 3, &state);
+  ASSERT_EQ(count, static_cast<size_t>(7));
+  ASSERT_EQ(wcs, nullptr);
+}
+
+TEST(LlvmLibcWCSRToMBSTest, NullState) {
+  // same as MultiplePartialConversions test except without an explicit
+  // mbstate_t
+
+  const wchar_t *wcs = L"\xFF\xAC15";
+  const wchar_t *wcs_start = wcs;
+
+  // init with dummy value of 1 so that we can check when null byte written
+  char mbs[7] = {1, 1, 1, 1, 1, 1, 1};
+  char expected[6] = {0xC3, 0xBF, 0xEA, 0xB0, 0x95, 0x00};
+  size_t written = 0;
+  size_t count = 0;
+
+  count = LIBC_NAMESPACE::wcsrtombs(mbs, &wcs, 1, nullptr);
+  written += count;
+  ASSERT_EQ(count, static_cast<size_t>(1));
+  ASSERT_EQ(wcs, wcs_start);
+  ASSERT_EQ(mbs[0], expected[0]);
+  ASSERT_EQ(mbs[1], '\x01');
+
+  count = LIBC_NAMESPACE::wcsrtombs(mbs + written, &wcs, 2, nullptr);
+  written += count;
+  ASSERT_EQ(count, static_cast<size_t>(2));
+  ASSERT_EQ(wcs, wcs_start + 1);
+  ASSERT_EQ(mbs[0], expected[0]);
+  ASSERT_EQ(mbs[1], expected[1]);
+  ASSERT_EQ(mbs[2], expected[2]);
+  ASSERT_EQ(mbs[3], '\x01');
+
+  count = LIBC_NAMESPACE::wcsrtombs(mbs + written, &wcs, 3, nullptr);
+  written += count;
+  ASSERT_EQ(count, static_cast<size_t>(2));
+  ASSERT_EQ(wcs, nullptr);
+  ASSERT_EQ(mbs[0], expected[0]);
+  ASSERT_EQ(mbs[1], expected[1]);
+  ASSERT_EQ(mbs[2], expected[2]);
+  ASSERT_EQ(mbs[3], expected[3]);
+  ASSERT_EQ(mbs[4], expected[4]);
+  ASSERT_EQ(mbs[5], expected[5]);
+  ASSERT_EQ(mbs[6], '\x01');
+}
+
+TEST(LlvmLibcWCSRToMBSTest, InvalidWchar) {
+  mbstate_t state;
+  LIBC_NAMESPACE::memset(&state, 0, sizeof(mbstate_t));
+
+  const wchar_t *wcs = L"\xFF\xAC15\x12FFFF";
+  char mbs[15];
+  // convert the valid wchar
+  size_t count = LIBC_NAMESPACE::wcsrtombs(mbs, &wcs, 7, &state);
+  ASSERT_EQ(count, static_cast<size_t>(7));
+
+  count = LIBC_NAMESPACE::wcsrtombs(mbs + count, &wcs, 7, &state); // invalid
+  ASSERT_EQ(count, static_cast<size_t>(-1));
+  ASSERT_EQ(static_cast<int>(libc_errno), EILSEQ);
+}