Add Dash support from PropList.txt.

src/include/ucd/ucd.h

@@ -333,6 +333,7 @@ typedef enum ucd_property_
 	UCD_PROPERTY_NO_BREAK     = 0x0002, /**< @brief <noBreak> */
 	UCD_PROPERTY_BIDI_CONTROL = 0x0004, /**< @brief Bidi_Control */
 	UCD_PROPERTY_JOIN_CONTROL = 0x0008, /**< @brief Join_Control */
+	UCD_PROPERTY_DASH         = 0x0010, /**< @brief Dash */
 } ucd_property;
 
 /** @brief Return the properties of the specified codepoint.
@@ -796,6 +797,7 @@ namespace ucd
 		noBreak = UCD_PROPERTY_NO_BREAK, /**< @brief <noBreak> */
 		Bidi_Control = UCD_PROPERTY_BIDI_CONTROL, /**< @brief Bidi_Control */
 		Join_Control = UCD_PROPERTY_JOIN_CONTROL, /**< @brief Join_Control */
+		Dash = UCD_PROPERTY_DASH, /**< @brief Dash */
 	};
 
 	/** @brief Return the properties of the specified codepoint.

src/proplist.c

@@ -20,6 +20,80 @@
 
 #include "ucd/ucd.h"
 
+static int properties_Pd(codepoint_t c)
+{
+	switch (c & 0xFFFFFF00)
+	{
+	case 0x0000:
+		if (c == 0x002D)                return UCD_PROPERTY_DASH;
+		break;
+	case 0x0500:
+		if (c == 0x058A)                return UCD_PROPERTY_DASH;
+		if (c == 0x05BE)                return UCD_PROPERTY_DASH;
+		break;
+	case 0x1400:
+		if (c == 0x1400)                return UCD_PROPERTY_DASH;
+		break;
+	case 0x1800:
+		if (c == 0x1806)                return UCD_PROPERTY_DASH;
+		break;
+	case 0x2000:
+		if (c == 0x2010)                return UCD_PROPERTY_DASH;
+		if (c == 0x2011)                return UCD_PROPERTY_DASH | UCD_PROPERTY_NO_BREAK;
+		if (c >= 0x2012 && c <= 0x2015) return UCD_PROPERTY_DASH;
+		break;
+	case 0x2E00:
+		if (c == 0x2E17)                return UCD_PROPERTY_DASH;
+		if (c == 0x2E1A)                return UCD_PROPERTY_DASH;
+		if (c >= 0x2E3A && c <= 0x2E3B) return UCD_PROPERTY_DASH;
+		if (c == 0x2E40)                return UCD_PROPERTY_DASH;
+		break;
+	case 0x3000:
+		if (c == 0x301C)                return UCD_PROPERTY_DASH;
+		if (c == 0x3030)                return UCD_PROPERTY_DASH;
+		if (c == 0x30A0)                return UCD_PROPERTY_DASH;
+		break;
+	case 0xFE00:
+		if (c >= 0xFE31 && c <= 0xFE32) return UCD_PROPERTY_DASH;
+		if (c == 0xFE58)                return UCD_PROPERTY_DASH;
+		if (c == 0xFE63)                return UCD_PROPERTY_DASH;
+		break;
+	case 0xFF00:
+		if (c == 0xFF0D)                return UCD_PROPERTY_DASH;
+		break;
+	}
+	return 0;
+}
+
+static int properties_Po(codepoint_t c)
+{
+	switch (c & 0xFFFFFF00)
+	{
+	case 0x0F00:
+		if (c == 0x0F0C)                return UCD_PROPERTY_NO_BREAK;
+		break;
+	case 0x2000:
+		if (c == 0x2053)                return UCD_PROPERTY_DASH;
+		break;
+	}
+	return 0;
+}
+
+static int properties_Sm(codepoint_t c)
+{
+	switch (c & 0xFFFFFF00)
+	{
+	case 0x2000:
+		if (c == 0x207B)                return UCD_PROPERTY_DASH;
+		if (c == 0x208B)                return UCD_PROPERTY_DASH;
+		break;
+	case 0x2200:
+		if (c == 0x2212)                return UCD_PROPERTY_DASH;
+		break;
+	}
+	return 0;
+}
+
 ucd_property ucd_properties(codepoint_t c, ucd_category category)
 {
 	switch (category)
@@ -43,14 +117,9 @@ ucd_property ucd_properties(codepoint_t c, ucd_category category)
 		if (c >= 0x202A && c <= 0x202E) return UCD_PROPERTY_BIDI_CONTROL;
 		if (c >= 0x2066 && c <= 0x2069) return UCD_PROPERTY_BIDI_CONTROL;
 		return 0;
-	case UCD_CATEGORY_Pd:
-		if (c == 0x2011) // NON-BREAKING HYPHEN
-			return UCD_PROPERTY_NO_BREAK;
-		return 0;
-	case UCD_CATEGORY_Po:
-		if (c == 0x0F0C) // TIBETAN MARK DELIMITER TSHEG BSTAR
-			return UCD_PROPERTY_NO_BREAK;
-		return 0;
+	case UCD_CATEGORY_Pd: return properties_Pd(c);
+	case UCD_CATEGORY_Po: return properties_Po(c);
+	case UCD_CATEGORY_Sm: return properties_Sm(c);
 	case UCD_CATEGORY_Zl:
 	case UCD_CATEGORY_Zp:
 		return UCD_PROPERTY_WHITE_SPACE;