react-native-quick-crypto 1.0.18 → 1.1.0

package/deps/simdutf/include/simdutf/scalar/utf16_to_latin1/valid_utf16_to_latin1.h

@@ -0,0 +1,40 @@
+#ifndef SIMDUTF_VALID_UTF16_TO_LATIN1_H
+#define SIMDUTF_VALID_UTF16_TO_LATIN1_H
+
+namespace simdutf {
+namespace scalar {
+namespace {
+namespace utf16_to_latin1 {
+
+template <endianness big_endian, class InputIterator, class OutputIterator>
+simdutf_constexpr23 inline size_t
+convert_valid_impl(InputIterator data, size_t len,
+                   OutputIterator latin_output) {
+  static_assert(
+      std::is_same<typename std::decay<decltype(*data)>::type, uint16_t>::value,
+      "must decay to uint16_t");
+  size_t pos = 0;
+  const auto start = latin_output;
+  uint16_t word = 0;
+
+  while (pos < len) {
+    word = !match_system(big_endian) ? u16_swap_bytes(data[pos]) : data[pos];
+    *latin_output++ = char(word);
+    pos++;
+  }
+
+  return latin_output - start;
+}
+
+template <endianness big_endian>
+simdutf_really_inline size_t convert_valid(const char16_t *buf, size_t len,
+                                           char *latin_output) {
+  return convert_valid_impl<big_endian>(reinterpret_cast<const uint16_t *>(buf),
+                                        len, latin_output);
+}
+} // namespace utf16_to_latin1
+} // unnamed namespace
+} // namespace scalar
+} // namespace simdutf
+
+#endif

package/deps/simdutf/include/simdutf/scalar/utf16_to_utf32/utf16_to_utf32.h

@@ -0,0 +1,86 @@
+#ifndef SIMDUTF_UTF16_TO_UTF32_H
+#define SIMDUTF_UTF16_TO_UTF32_H
+
+namespace simdutf {
+namespace scalar {
+namespace {
+namespace utf16_to_utf32 {
+
+template <endianness big_endian>
+simdutf_constexpr23 size_t convert(const char16_t *data, size_t len,
+                                   char32_t *utf32_output) {
+  size_t pos = 0;
+  char32_t *start{utf32_output};
+  while (pos < len) {
+    uint16_t word =
+        !match_system(big_endian) ? u16_swap_bytes(data[pos]) : data[pos];
+    if ((word & 0xF800) != 0xD800) {
+      // No surrogate pair, extend 16-bit word to 32-bit word
+      *utf32_output++ = char32_t(word);
+      pos++;
+    } else {
+      // must be a surrogate pair
+      uint16_t diff = uint16_t(word - 0xD800);
+      if (diff > 0x3FF) {
+        return 0;
+      }
+      if (pos + 1 >= len) {
+        return 0;
+      } // minimal bound checking
+      uint16_t next_word = !match_system(big_endian)
+                               ? u16_swap_bytes(data[pos + 1])
+                               : data[pos + 1];
+      uint16_t diff2 = uint16_t(next_word - 0xDC00);
+      if (diff2 > 0x3FF) {
+        return 0;
+      }
+      uint32_t value = (diff << 10) + diff2 + 0x10000;
+      *utf32_output++ = char32_t(value);
+      pos += 2;
+    }
+  }
+  return utf32_output - start;
+}
+
+template <endianness big_endian>
+simdutf_constexpr23 result convert_with_errors(const char16_t *data, size_t len,
+                                               char32_t *utf32_output) {
+  size_t pos = 0;
+  char32_t *start{utf32_output};
+  while (pos < len) {
+    uint16_t word =
+        !match_system(big_endian) ? u16_swap_bytes(data[pos]) : data[pos];
+    if ((word & 0xF800) != 0xD800) {
+      // No surrogate pair, extend 16-bit word to 32-bit word
+      *utf32_output++ = char32_t(word);
+      pos++;
+    } else {
+      // must be a surrogate pair
+      uint16_t diff = uint16_t(word - 0xD800);
+      if (diff > 0x3FF) {
+        return result(error_code::SURROGATE, pos);
+      }
+      if (pos + 1 >= len) {
+        return result(error_code::SURROGATE, pos);
+      } // minimal bound checking
+      uint16_t next_word = !match_system(big_endian)
+                               ? u16_swap_bytes(data[pos + 1])
+                               : data[pos + 1];
+      uint16_t diff2 = uint16_t(next_word - 0xDC00);
+      if (diff2 > 0x3FF) {
+        return result(error_code::SURROGATE, pos);
+      }
+      uint32_t value = (diff << 10) + diff2 + 0x10000;
+      *utf32_output++ = char32_t(value);
+      pos += 2;
+    }
+  }
+  return result(error_code::SUCCESS, utf32_output - start);
+}
+
+} // namespace utf16_to_utf32
+} // unnamed namespace
+} // namespace scalar
+} // namespace simdutf
+
+#endif

package/deps/simdutf/include/simdutf/scalar/utf16_to_utf32/valid_utf16_to_utf32.h

@@ -0,0 +1,44 @@
+#ifndef SIMDUTF_VALID_UTF16_TO_UTF32_H
+#define SIMDUTF_VALID_UTF16_TO_UTF32_H
+
+namespace simdutf {
+namespace scalar {
+namespace {
+namespace utf16_to_utf32 {
+
+template <endianness big_endian>
+simdutf_constexpr23 size_t convert_valid(const char16_t *data, size_t len,
+                                         char32_t *utf32_output) {
+  size_t pos = 0;
+  char32_t *start{utf32_output};
+  while (pos < len) {
+    uint16_t word =
+        !match_system(big_endian) ? u16_swap_bytes(data[pos]) : data[pos];
+    if ((word & 0xF800) != 0xD800) {
+      // No surrogate pair, extend 16-bit word to 32-bit word
+      *utf32_output++ = char32_t(word);
+      pos++;
+    } else {
+      // must be a surrogate pair
+      uint16_t diff = uint16_t(word - 0xD800);
+      if (pos + 1 >= len) {
+        return 0;
+      } // minimal bound checking
+      uint16_t next_word = !match_system(big_endian)
+                               ? u16_swap_bytes(data[pos + 1])
+                               : data[pos + 1];
+      uint16_t diff2 = uint16_t(next_word - 0xDC00);
+      uint32_t value = (diff << 10) + diff2 + 0x10000;
+      *utf32_output++ = char32_t(value);
+      pos += 2;
+    }
+  }
+  return utf32_output - start;
+}
+
+} // namespace utf16_to_utf32
+} // unnamed namespace
+} // namespace scalar
+} // namespace simdutf
+
+#endif

package/deps/simdutf/include/simdutf/scalar/utf16_to_utf8/utf16_to_utf8.h

@@ -0,0 +1,295 @@
+#ifndef SIMDUTF_UTF16_TO_UTF8_H
+#define SIMDUTF_UTF16_TO_UTF8_H
+
+namespace simdutf {
+namespace scalar {
+namespace {
+namespace utf16_to_utf8 {
+
+template <endianness big_endian, typename InputPtr, typename OutputPtr>
+#if SIMDUTF_CPLUSPLUS20
+  requires simdutf::detail::indexes_into_utf16<InputPtr>
+// FIXME constrain output as well
+#endif
+simdutf_constexpr23 size_t convert(InputPtr data, size_t len,
+                                   OutputPtr utf8_output) {
+  size_t pos = 0;
+  const auto start = utf8_output;
+  while (pos < len) {
+#if SIMDUTF_CPLUSPLUS23
+    if !consteval
+#endif
+    {
+      // try to convert the next block of 8 bytes
+      if (pos + 4 <= len) { // if it is safe to read 8 more bytes, check that
+                            // they are ascii
+        uint64_t v;
+        ::memcpy(&v, data + pos, sizeof(uint64_t));
+        if simdutf_constexpr (!match_system(big_endian)) {
+          v = (v >> 8) | (v << (64 - 8));
+        }
+        if ((v & 0xFF80FF80FF80FF80) == 0) {
+          size_t final_pos = pos + 4;
+          while (pos < final_pos) {
+            *utf8_output++ = !match_system(big_endian)
+                                 ? char(u16_swap_bytes(data[pos]))
+                                 : char(data[pos]);
+            pos++;
+          }
+          continue;
+        }
+      }
+    }
+    uint16_t word =
+        !match_system(big_endian) ? u16_swap_bytes(data[pos]) : data[pos];
+    if ((word & 0xFF80) == 0) {
+      // will generate one UTF-8 bytes
+      *utf8_output++ = char(word);
+      pos++;
+    } else if ((word & 0xF800) == 0) {
+      // will generate two UTF-8 bytes
+      // we have 0b110XXXXX 0b10XXXXXX
+      *utf8_output++ = char((word >> 6) | 0b11000000);
+      *utf8_output++ = char((word & 0b111111) | 0b10000000);
+      pos++;
+    } else if ((word & 0xF800) != 0xD800) {
+      // will generate three UTF-8 bytes
+      // we have 0b1110XXXX 0b10XXXXXX 0b10XXXXXX
+      *utf8_output++ = char((word >> 12) | 0b11100000);
+      *utf8_output++ = char(((word >> 6) & 0b111111) | 0b10000000);
+      *utf8_output++ = char((word & 0b111111) | 0b10000000);
+      pos++;
+    } else {
+      // must be a surrogate pair
+      if (pos + 1 >= len) {
+        return 0;
+      }
+      uint16_t diff = uint16_t(word - 0xD800);
+      if (diff > 0x3FF) {
+        return 0;
+      }
+      uint16_t next_word = !match_system(big_endian)
+                               ? u16_swap_bytes(data[pos + 1])
+                               : data[pos + 1];
+      uint16_t diff2 = uint16_t(next_word - 0xDC00);
+      if (diff2 > 0x3FF) {
+        return 0;
+      }
+      uint32_t value = (diff << 10) + diff2 + 0x10000;
+      // will generate four UTF-8 bytes
+      // we have 0b11110XXX 0b10XXXXXX 0b10XXXXXX 0b10XXXXXX
+      *utf8_output++ = char((value >> 18) | 0b11110000);
+      *utf8_output++ = char(((value >> 12) & 0b111111) | 0b10000000);
+      *utf8_output++ = char(((value >> 6) & 0b111111) | 0b10000000);
+      *utf8_output++ = char((value & 0b111111) | 0b10000000);
+      pos += 2;
+    }
+  }
+  return utf8_output - start;
+}
+
+template <endianness big_endian, bool check_output = false, typename InputPtr,
+          typename OutputPtr>
+#if SIMDUTF_CPLUSPLUS20
+  requires(simdutf::detail::indexes_into_utf16<InputPtr> &&
+           simdutf::detail::index_assignable_from_char<OutputPtr>)
+#endif
+simdutf_constexpr23 full_result convert_with_errors(InputPtr data, size_t len,
+                                                    OutputPtr utf8_output,
+                                                    size_t utf8_len = 0) {
+  if (check_output && utf8_len == 0) {
+    return full_result(error_code::OUTPUT_BUFFER_TOO_SMALL, 0, 0);
+  }
+
+  size_t pos = 0;
+  auto start = utf8_output;
+  auto end = utf8_output + utf8_len;
+
+  while (pos < len) {
+#if SIMDUTF_CPLUSPLUS23
+    if !consteval
+#endif
+    {
+      // try to convert the next block of 8 bytes
+      if (pos + 4 <= len) { // if it is safe to read 8 more bytes, check that
+                            // they are ascii
+        uint64_t v;
+        ::memcpy(&v, data + pos, sizeof(uint64_t));
+        if simdutf_constexpr (!match_system(big_endian))
+          v = (v >> 8) | (v << (64 - 8));
+        if ((v & 0xFF80FF80FF80FF80) == 0) {
+          size_t final_pos = pos + 4;
+          while (pos < final_pos) {
+            if (check_output && size_t(end - utf8_output) < 1) {
+              return full_result(error_code::OUTPUT_BUFFER_TOO_SMALL, pos,
+                                 utf8_output - start);
+            }
+            *utf8_output++ = !match_system(big_endian)
+                                 ? char(u16_swap_bytes(data[pos]))
+                                 : char(data[pos]);
+            pos++;
+          }
+          continue;
+        }
+      }
+    }
+
+    uint16_t word =
+        !match_system(big_endian) ? u16_swap_bytes(data[pos]) : data[pos];
+    if ((word & 0xFF80) == 0) {
+      // will generate one UTF-8 bytes
+      if (check_output && size_t(end - utf8_output) < 1) {
+        return full_result(error_code::OUTPUT_BUFFER_TOO_SMALL, pos,
+                           utf8_output - start);
+      }
+      *utf8_output++ = char(word);
+      pos++;
+    } else if ((word & 0xF800) == 0) {
+      // will generate two UTF-8 bytes
+      // we have 0b110XXXXX 0b10XXXXXX
+      if (check_output && size_t(end - utf8_output) < 2) {
+        return full_result(error_code::OUTPUT_BUFFER_TOO_SMALL, pos,
+                           utf8_output - start);
+      }
+      *utf8_output++ = char((word >> 6) | 0b11000000);
+      *utf8_output++ = char((word & 0b111111) | 0b10000000);
+      pos++;
+
+    } else if ((word & 0xF800) != 0xD800) {
+      // will generate three UTF-8 bytes
+      // we have 0b1110XXXX 0b10XXXXXX 0b10XXXXXX
+      if (check_output && size_t(end - utf8_output) < 3) {
+        return full_result(error_code::OUTPUT_BUFFER_TOO_SMALL, pos,
+                           utf8_output - start);
+      }
+      *utf8_output++ = char((word >> 12) | 0b11100000);
+      *utf8_output++ = char(((word >> 6) & 0b111111) | 0b10000000);
+      *utf8_output++ = char((word & 0b111111) | 0b10000000);
+      pos++;
+    } else {
+
+      if (check_output && size_t(end - utf8_output) < 4) {
+        return full_result(error_code::OUTPUT_BUFFER_TOO_SMALL, pos,
+                           utf8_output - start);
+      }
+      // must be a surrogate pair
+      if (pos + 1 >= len) {
+        return full_result(error_code::SURROGATE, pos, utf8_output - start);
+      }
+      uint16_t diff = uint16_t(word - 0xD800);
+      if (diff > 0x3FF) {
+        return full_result(error_code::SURROGATE, pos, utf8_output - start);
+      }
+      uint16_t next_word = !match_system(big_endian)
+                               ? u16_swap_bytes(data[pos + 1])
+                               : data[pos + 1];
+      uint16_t diff2 = uint16_t(next_word - 0xDC00);
+      if (diff2 > 0x3FF) {
+        return full_result(error_code::SURROGATE, pos, utf8_output - start);
+      }
+      uint32_t value = (diff << 10) + diff2 + 0x10000;
+      // will generate four UTF-8 bytes
+      // we have 0b11110XXX 0b10XXXXXX 0b10XXXXXX 0b10XXXXXX
+      *utf8_output++ = char((value >> 18) | 0b11110000);
+      *utf8_output++ = char(((value >> 12) & 0b111111) | 0b10000000);
+      *utf8_output++ = char(((value >> 6) & 0b111111) | 0b10000000);
+      *utf8_output++ = char((value & 0b111111) | 0b10000000);
+      pos += 2;
+    }
+  }
+  return full_result(error_code::SUCCESS, pos, utf8_output - start);
+}
+
+template <endianness big_endian>
+inline result simple_convert_with_errors(const char16_t *buf, size_t len,
+                                         char *utf8_output) {
+  return convert_with_errors<big_endian, false>(buf, len, utf8_output, 0);
+}
+
+template <endianness big_endian>
+simdutf_constexpr23 size_t convert_with_replacement(const char16_t *data,
+                                                    size_t len,
+                                                    char *utf8_output) {
+  size_t pos = 0;
+  char *start = utf8_output;
+  while (pos < len) {
+#if SIMDUTF_CPLUSPLUS23
+    if !consteval
+#endif
+    {
+      // try to convert the next block of 8 bytes
+      if (pos + 4 <= len) { // if it is safe to read 8 more bytes, check that
+                            // they are ascii
+        uint64_t v;
+        ::memcpy(&v, data + pos, sizeof(uint64_t));
+        if simdutf_constexpr (!match_system(big_endian)) {
+          v = (v >> 8) | (v << (64 - 8));
+        }
+        if ((v & 0xFF80FF80FF80FF80) == 0) {
+          size_t final_pos = pos + 4;
+          while (pos < final_pos) {
+            *utf8_output++ = !match_system(big_endian)
+                                 ? char(u16_swap_bytes(data[pos]))
+                                 : char(data[pos]);
+            pos++;
+          }
+          continue;
+        }
+      }
+    }
+    uint16_t word =
+        !match_system(big_endian) ? u16_swap_bytes(data[pos]) : data[pos];
+    if ((word & 0xFF80) == 0) {
+      // will generate one UTF-8 bytes
+      *utf8_output++ = char(word);
+      pos++;
+    } else if ((word & 0xF800) == 0) {
+      // will generate two UTF-8 bytes
+      // we have 0b110XXXXX 0b10XXXXXX
+      *utf8_output++ = char((word >> 6) | 0b11000000);
+      *utf8_output++ = char((word & 0b111111) | 0b10000000);
+      pos++;
+    } else if ((word & 0xF800) != 0xD800) {
+      // will generate three UTF-8 bytes
+      // we have 0b1110XXXX 0b10XXXXXX 0b10XXXXXX
+      *utf8_output++ = char((word >> 12) | 0b11100000);
+      *utf8_output++ = char(((word >> 6) & 0b111111) | 0b10000000);
+      *utf8_output++ = char((word & 0b111111) | 0b10000000);
+      pos++;
+    } else {
+      // surrogate range
+      uint16_t diff = uint16_t(word - 0xD800);
+      if (diff <= 0x3FF && pos + 1 < len) {
+        // high surrogate, check for valid pair
+        uint16_t next_word = !match_system(big_endian)
+                                 ? u16_swap_bytes(data[pos + 1])
+                                 : data[pos + 1];
+        uint16_t diff2 = uint16_t(next_word - 0xDC00);
+        if (diff2 <= 0x3FF) {
+          // valid surrogate pair
+          uint32_t value = (diff << 10) + diff2 + 0x10000;
+          // will generate four UTF-8 bytes
+          *utf8_output++ = char((value >> 18) | 0b11110000);
+          *utf8_output++ = char(((value >> 12) & 0b111111) | 0b10000000);
+          *utf8_output++ = char(((value >> 6) & 0b111111) | 0b10000000);
+          *utf8_output++ = char((value & 0b111111) | 0b10000000);
+          pos += 2;
+          continue;
+        }
+      }
+      // unpaired surrogate: replace with U+FFFD (0xEF 0xBF 0xBD)
+      *utf8_output++ = char(0xef);
+      *utf8_output++ = char(0xbf);
+      *utf8_output++ = char(0xbd);
+      pos++;
+    }
+  }
+  return utf8_output - start;
+}
+
+} // namespace utf16_to_utf8
+} // unnamed namespace
+} // namespace scalar
+} // namespace simdutf
+
+#endif

package/deps/simdutf/include/simdutf/scalar/utf16_to_utf8/valid_utf16_to_utf8.h

@@ -0,0 +1,91 @@
+#ifndef SIMDUTF_VALID_UTF16_TO_UTF8_H
+#define SIMDUTF_VALID_UTF16_TO_UTF8_H
+
+namespace simdutf {
+namespace scalar {
+namespace {
+namespace utf16_to_utf8 {
+
+template <endianness big_endian, typename InputPtr, typename OutputPtr>
+#if SIMDUTF_CPLUSPLUS20
+  requires(simdutf::detail::indexes_into_utf16<InputPtr> &&
+           simdutf::detail::index_assignable_from_char<OutputPtr>)
+#endif
+simdutf_constexpr23 size_t convert_valid(InputPtr data, size_t len,
+                                         OutputPtr utf8_output) {
+  size_t pos = 0;
+  auto start = utf8_output;
+  while (pos < len) {
+#if SIMDUTF_CPLUSPLUS23
+    if !consteval
+#endif
+    {
+      // try to convert the next block of 4 ASCII characters
+      if (pos + 4 <= len) { // if it is safe to read 8 more bytes, check that
+                            // they are ascii
+        uint64_t v;
+        ::memcpy(&v, data + pos, sizeof(uint64_t));
+        if simdutf_constexpr (!match_system(big_endian)) {
+          v = (v >> 8) | (v << (64 - 8));
+        }
+        if ((v & 0xFF80FF80FF80FF80) == 0) {
+          size_t final_pos = pos + 4;
+          while (pos < final_pos) {
+            *utf8_output++ = !match_system(big_endian)
+                                 ? char(u16_swap_bytes(data[pos]))
+                                 : char(data[pos]);
+            pos++;
+          }
+          continue;
+        }
+      }
+    }
+
+    uint16_t word =
+        !match_system(big_endian) ? u16_swap_bytes(data[pos]) : data[pos];
+    if ((word & 0xFF80) == 0) {
+      // will generate one UTF-8 bytes
+      *utf8_output++ = char(word);
+      pos++;
+    } else if ((word & 0xF800) == 0) {
+      // will generate two UTF-8 bytes
+      // we have 0b110XXXXX 0b10XXXXXX
+      *utf8_output++ = char((word >> 6) | 0b11000000);
+      *utf8_output++ = char((word & 0b111111) | 0b10000000);
+      pos++;
+    } else if ((word & 0xF800) != 0xD800) {
+      // will generate three UTF-8 bytes
+      // we have 0b1110XXXX 0b10XXXXXX 0b10XXXXXX
+      *utf8_output++ = char((word >> 12) | 0b11100000);
+      *utf8_output++ = char(((word >> 6) & 0b111111) | 0b10000000);
+      *utf8_output++ = char((word & 0b111111) | 0b10000000);
+      pos++;
+    } else {
+      // must be a surrogate pair
+      uint16_t diff = uint16_t(word - 0xD800);
+      if (pos + 1 >= len) {
+        return 0;
+      } // minimal bound checking
+      uint16_t next_word = !match_system(big_endian)
+                               ? u16_swap_bytes(data[pos + 1])
+                               : data[pos + 1];
+      uint16_t diff2 = uint16_t(next_word - 0xDC00);
+      uint32_t value = (diff << 10) + diff2 + 0x10000;
+      // will generate four UTF-8 bytes
+      // we have 0b11110XXX 0b10XXXXXX 0b10XXXXXX 0b10XXXXXX
+      *utf8_output++ = char((value >> 18) | 0b11110000);
+      *utf8_output++ = char(((value >> 12) & 0b111111) | 0b10000000);
+      *utf8_output++ = char(((value >> 6) & 0b111111) | 0b10000000);
+      *utf8_output++ = char((value & 0b111111) | 0b10000000);
+      pos += 2;
+    }
+  }
+  return utf8_output - start;
+}
+
+} // namespace utf16_to_utf8
+} // unnamed namespace
+} // namespace scalar
+} // namespace simdutf
+
+#endif

package/deps/simdutf/include/simdutf/scalar/utf32.h

@@ -0,0 +1,82 @@
+#ifndef SIMDUTF_UTF32_H
+#define SIMDUTF_UTF32_H
+
+namespace simdutf {
+namespace scalar {
+namespace utf32 {
+
+template <typename InputPtr>
+#if SIMDUTF_CPLUSPLUS20
+  requires simdutf::detail::indexes_into_uint32<InputPtr>
+#endif
+simdutf_warn_unused simdutf_constexpr23 bool validate(InputPtr data,
+                                                      size_t len) noexcept {
+  uint64_t pos = 0;
+  for (; pos < len; pos++) {
+    uint32_t word = data[pos];
+    if (word > 0x10FFFF || (word >= 0xD800 && word <= 0xDFFF)) {
+      return false;
+    }
+  }
+  return true;
+}
+
+simdutf_warn_unused simdutf_really_inline bool validate(const char32_t *buf,
+                                                        size_t len) noexcept {
+  return validate(reinterpret_cast<const uint32_t *>(buf), len);
+}
+
+template <typename InputPtr>
+#if SIMDUTF_CPLUSPLUS20
+  requires simdutf::detail::indexes_into_uint32<InputPtr>
+#endif
+simdutf_warn_unused simdutf_constexpr23 result
+validate_with_errors(InputPtr data, size_t len) noexcept {
+  size_t pos = 0;
+  for (; pos < len; pos++) {
+    uint32_t word = data[pos];
+    if (word > 0x10FFFF) {
+      return result(error_code::TOO_LARGE, pos);
+    }
+    if (word >= 0xD800 && word <= 0xDFFF) {
+      return result(error_code::SURROGATE, pos);
+    }
+  }
+  return result(error_code::SUCCESS, pos);
+}
+
+simdutf_warn_unused simdutf_really_inline result
+validate_with_errors(const char32_t *buf, size_t len) noexcept {
+  return validate_with_errors(reinterpret_cast<const uint32_t *>(buf), len);
+}
+
+inline simdutf_constexpr23 size_t utf8_length_from_utf32(const char32_t *p,
+                                                         size_t len) {
+  // We are not BOM aware.
+  size_t counter{0};
+  for (size_t i = 0; i < len; i++) {
+    // credit: @ttsugriy  for the vectorizable approach
+    counter++;                                     // ASCII
+    counter += static_cast<size_t>(p[i] > 0x7F);   // two-byte
+    counter += static_cast<size_t>(p[i] > 0x7FF);  // three-byte
+    counter += static_cast<size_t>(p[i] > 0xFFFF); // four-bytes
+  }
+  return counter;
+}
+
+inline simdutf_warn_unused simdutf_constexpr23 size_t
+utf16_length_from_utf32(const char32_t *p, size_t len) {
+  // We are not BOM aware.
+  size_t counter{0};
+  for (size_t i = 0; i < len; i++) {
+    counter++;                                     // non-surrogate word
+    counter += static_cast<size_t>(p[i] > 0xFFFF); // surrogate pair
+  }
+  return counter;
+}
+
+} // namespace utf32
+} // namespace scalar
+} // namespace simdutf
+
+#endif