diff --git a/src/include/ucd/ucd.h b/src/include/ucd/ucd.h index 8356a31c..40103581 100644 --- a/src/include/ucd/ucd.h +++ b/src/include/ucd/ucd.h @@ -329,12 +329,13 @@ ucd_script ucd_lookup_script(codepoint_t c); */ typedef enum ucd_property_ { - UCD_PROPERTY_WHITE_SPACE = 0x0001, /**< @brief White_Space */ - UCD_PROPERTY_NO_BREAK = 0x0002, /**< @brief */ - UCD_PROPERTY_BIDI_CONTROL = 0x0004, /**< @brief Bidi_Control */ - UCD_PROPERTY_JOIN_CONTROL = 0x0008, /**< @brief Join_Control */ - UCD_PROPERTY_DASH = 0x0010, /**< @brief Dash */ - UCD_PROPERTY_HYPHEN = 0x0020, /**< @brief Hyphen */ + UCD_PROPERTY_WHITE_SPACE = 0x00000001, /**< @brief White_Space */ + UCD_PROPERTY_NO_BREAK = 0x00000002, /**< @brief */ + UCD_PROPERTY_BIDI_CONTROL = 0x00000004, /**< @brief Bidi_Control */ + UCD_PROPERTY_JOIN_CONTROL = 0x00000008, /**< @brief Join_Control */ + UCD_PROPERTY_DASH = 0x00000010, /**< @brief Dash */ + UCD_PROPERTY_HYPHEN = 0x00000020, /**< @brief Hyphen */ + UCD_PROPERTY_QUOTATION_MARK = 0x00000020, /**< @brief Quotation_Mark */ } ucd_property; /** @brief Return the properties of the specified codepoint. @@ -800,6 +801,7 @@ namespace ucd Join_Control = UCD_PROPERTY_JOIN_CONTROL, /**< @brief Join_Control */ Dash = UCD_PROPERTY_DASH, /**< @brief Dash */ Hyphen = UCD_PROPERTY_HYPHEN, /**< @brief Hyphen */ + Quotation_Mark = UCD_PROPERTY_QUOTATION_MARK, /**< @brief Quotation_Mark */ }; /** @brief Return the properties of the specified codepoint. diff --git a/src/proplist.c b/src/proplist.c index 69bdafe7..636c609f 100644 --- a/src/proplist.c +++ b/src/proplist.c @@ -92,10 +92,67 @@ static int properties_Pd(codepoint_t c) return 0; } +static int properties_Pe(codepoint_t c) +{ + switch (c & 0xFFFFFF00) + { + case 0x3000: + if (c == 0x300D) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0x300F) return UCD_PROPERTY_QUOTATION_MARK; + if (c >= 0x301E && c <= 0x301F) return UCD_PROPERTY_QUOTATION_MARK; + break; + case 0xFE00: + if (c == 0xFE42) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0xFE44) return UCD_PROPERTY_QUOTATION_MARK; + break; + case 0xFF00: + if (c == 0xFF63) return UCD_PROPERTY_QUOTATION_MARK; + break; + } + return 0; +} + +static int properties_Pf(codepoint_t c) +{ + switch (c & 0xFFFFFF00) + { + case 0x0000: + if (c == 0x00BB) return UCD_PROPERTY_QUOTATION_MARK; + break; + case 0x2000: + if (c == 0x2019) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0x201D) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0x203A) return UCD_PROPERTY_QUOTATION_MARK; + break; + } + return 0; +} + +static int properties_Pi(codepoint_t c) +{ + switch (c & 0xFFFFFF00) + { + case 0x0000: + if (c == 0x00AB) return UCD_PROPERTY_QUOTATION_MARK; + break; + case 0x2000: + if (c == 0x2018) return UCD_PROPERTY_QUOTATION_MARK; + if (c >= 0x201B && c <= 0x201C) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0x201F) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0x2039) return UCD_PROPERTY_QUOTATION_MARK; + break; + } + return 0; +} + static int properties_Po(codepoint_t c) { switch (c & 0xFFFFFF00) { + case 0x0000: + if (c == 0x0022) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0x0027) return UCD_PROPERTY_QUOTATION_MARK; + break; case 0x0F00: if (c == 0x0F0C) return UCD_PROPERTY_NO_BREAK; break; @@ -106,12 +163,41 @@ static int properties_Po(codepoint_t c) if (c == 0x30FB) return UCD_PROPERTY_HYPHEN; break; case 0xFF00: + if (c == 0xFF02) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0xFF07) return UCD_PROPERTY_QUOTATION_MARK; if (c == 0xFF65) return UCD_PROPERTY_HYPHEN; break; } return 0; } +static int properties_Ps(codepoint_t c) +{ + switch (c & 0xFFFFFF00) + { + case 0x2000: + if (c == 0x201A) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0x201E) return UCD_PROPERTY_QUOTATION_MARK; + break; + case 0x2E00: + if (c == 0x2E42) return UCD_PROPERTY_QUOTATION_MARK; + break; + case 0x3000: + if (c == 0x300C) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0x300E) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0x301D) return UCD_PROPERTY_QUOTATION_MARK; + break; + case 0xFE00: + if (c == 0xFE41) return UCD_PROPERTY_QUOTATION_MARK; + if (c == 0xFE43) return UCD_PROPERTY_QUOTATION_MARK; + break; + case 0xFF00: + if (c == 0xFF62) return UCD_PROPERTY_QUOTATION_MARK; + break; + } + return 0; +} + static int properties_Sm(codepoint_t c) { switch (c & 0xFFFFFF00) @@ -146,7 +232,11 @@ ucd_property ucd_properties(codepoint_t c, ucd_category category) case UCD_CATEGORY_Cc: return properties_Cc(c); case UCD_CATEGORY_Cf: return properties_Cf(c); case UCD_CATEGORY_Pd: return properties_Pd(c); + case UCD_CATEGORY_Pe: return properties_Pe(c); + case UCD_CATEGORY_Pf: return properties_Pf(c); + case UCD_CATEGORY_Pi: return properties_Pi(c); case UCD_CATEGORY_Po: return properties_Po(c); + case UCD_CATEGORY_Ps: return properties_Ps(c); case UCD_CATEGORY_Sm: return properties_Sm(c); case UCD_CATEGORY_Zl: return UCD_PROPERTY_WHITE_SPACE; case UCD_CATEGORY_Zp: return UCD_PROPERTY_WHITE_SPACE;