95 static int DecodeUtf8Char(uint8 *data, size_t length, uint32 *codepoint)

96 {

97  if (length < 1) return 0;

98  uint32 value = *data;

99  data++;

100  if ((value & 0x80) == 0) {

101  *codepoint = value;

102  return 1;

103  }

104  int size;

105  uint32 min_value;

106  if ((value & 0xE0) == 0xC0) {

107  size = 2;

108  min_value = 0x80;

109  value &= 0x1F;

110  } else if ((value & 0xF0) == 0xE0) {

111  size = 3;

112  min_value = 0x800;

113  value &= 0x0F;

114  } else if ((value & 0xF8) == 0xF0) {

115  size = 4;

116  min_value = 0x10000;

117  value &= 0x07;

118  } else {

119  return 0;

120  }

121

122  if (length < static_cast<size_t>(size)) return 0;

123  for (int n = 1; n < size; n++) {

124  uint8 val = *data;

125  data++;

126  if ((val & 0xC0) != 0x80) return 0;

127  value = (value << 6) | (val & 0x3F);

128  }

129  if (value < min_value || (value >= 0xD800 && value <= 0xDFFF) || value > 0x10FFFF) return 0;

130  *codepoint = value;

131  return size;

132 }