UCD_PROPERTY_TERMINAL_PUNCTUATION = 0x00000080, /**< @brief Terminal_Punctuation PropList */ | UCD_PROPERTY_TERMINAL_PUNCTUATION = 0x00000080, /**< @brief Terminal_Punctuation PropList */ | ||||
UCD_PROPERTY_OTHER_MATH = 0x00000100, /**< @brief Other_Math PropList */ | UCD_PROPERTY_OTHER_MATH = 0x00000100, /**< @brief Other_Math PropList */ | ||||
UCD_PROPERTY_HEX_DIGIT = 0x00000200, /**< @brief Hex_Digit PropList */ | UCD_PROPERTY_HEX_DIGIT = 0x00000200, /**< @brief Hex_Digit PropList */ | ||||
UCD_PROPERTY_ASCII_HEX_DIGIT = 0x00000400, /**< @brief ASCII_Hex_Digit PropList */ | |||||
} ucd_property; | } ucd_property; | ||||
/** @brief Return the properties of the specified codepoint. | /** @brief Return the properties of the specified codepoint. | ||||
Terminal_Punctuation = UCD_PROPERTY_TERMINAL_PUNCTUATION, /**< @brief Terminal_Punctuation PropList */ | Terminal_Punctuation = UCD_PROPERTY_TERMINAL_PUNCTUATION, /**< @brief Terminal_Punctuation PropList */ | ||||
Other_Math = UCD_PROPERTY_OTHER_MATH, /**< @brief Other_Math PropList */ | Other_Math = UCD_PROPERTY_OTHER_MATH, /**< @brief Other_Math PropList */ | ||||
Hex_Digit = UCD_PROPERTY_HEX_DIGIT, /**< @brief Hex_Digit PropList */ | Hex_Digit = UCD_PROPERTY_HEX_DIGIT, /**< @brief Hex_Digit PropList */ | ||||
ASCII_Hex_Digit = UCD_PROPERTY_ASCII_HEX_DIGIT, /**< @brief ASCII_Hex_Digit PropList */ | |||||
}; | }; | ||||
/** @brief Return the properties of the specified codepoint. | /** @brief Return the properties of the specified codepoint. |
switch (c & 0xFFFFFF00) | switch (c & 0xFFFFFF00) | ||||
{ | { | ||||
case 0x0000: | case 0x0000: | ||||
if (c >= 0x0061 && c <= 0x0066) return UCD_PROPERTY_HEX_DIGIT; | |||||
if (c >= 0x0061 && c <= 0x0066) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT; | |||||
break; | break; | ||||
case 0x0300: | case 0x0300: | ||||
if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu | if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu | ||||
switch (c & 0xFFFFFF00) | switch (c & 0xFFFFFF00) | ||||
{ | { | ||||
case 0x0000: | case 0x0000: | ||||
if (c >= 0x0041 && c <= 0x0046) return UCD_PROPERTY_HEX_DIGIT; | |||||
if (c >= 0x0041 && c <= 0x0046) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT; | |||||
break; | break; | ||||
case 0x0300: | case 0x0300: | ||||
if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu | if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu | ||||
switch (c & 0xFFFFFF00) | switch (c & 0xFFFFFF00) | ||||
{ | { | ||||
case 0x0000: | case 0x0000: | ||||
if (c >= 0x0030 && c <= 0x0039) return UCD_PROPERTY_HEX_DIGIT; | |||||
if (c >= 0x0030 && c <= 0x0039) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT; | |||||
break; | break; | ||||
case 0xFF00: | case 0xFF00: | ||||
if (c >= 0xFF10 && c <= 0xFF19) return UCD_PROPERTY_HEX_DIGIT; | if (c >= 0xFF10 && c <= 0xFF19) return UCD_PROPERTY_HEX_DIGIT; |
return None | return None | ||||
def properties(data): | def properties(data): | ||||
props = 0 | |||||
props += 1 * data.get('White_Space', 0) | |||||
props += 2 * (decomposition_type(data, '<noBreak>') != None) | |||||
props += 4 * data.get('Bidi_Control', 0) | |||||
props += 8 * data.get('Join_Control', 0) | |||||
props += 16 * data.get('Dash', 0) | |||||
props += 32 * data.get('Hyphen', 0) | |||||
props += 64 * data.get('Quotation_Mark', 0) | |||||
props += 128 * data.get('Terminal_Punctuation', 0) | |||||
props += 256 * data.get('Other_Math', 0) | |||||
props += 512 * data.get('Hex_Digit', 0) | |||||
props = 0 | |||||
props += 1 * data.get('White_Space', 0) | |||||
props += 2 * (decomposition_type(data, '<noBreak>') != None) | |||||
props += 4 * data.get('Bidi_Control', 0) | |||||
props += 8 * data.get('Join_Control', 0) | |||||
props += 16 * data.get('Dash', 0) | |||||
props += 32 * data.get('Hyphen', 0) | |||||
props += 64 * data.get('Quotation_Mark', 0) | |||||
props += 128 * data.get('Terminal_Punctuation', 0) | |||||
props += 256 * data.get('Other_Math', 0) | |||||
props += 512 * data.get('Hex_Digit', 0) | |||||
props += 1024 * data.get('ASCII_Hex_Digit', 0) | |||||
return props | return props | ||||
if __name__ == '__main__': | if __name__ == '__main__': |