Merge commit 'f26aca0aa1b9485634314d0d1710d2a0bb67776d'

src/ucd-tools/.gitignore

@@ -1,4 +1,5 @@
 .*.swp
+*~
 
 # intermediate files:
 

src/ucd-tools/CHANGELOG.md

@@ -7,11 +7,14 @@ These are eSpeak NG specific modifications to the `ucd-tools` project:
 *  `data/espeak-ng` data files for eSpeak NG extended data.
 *  espeak-ng PropList property lookup as part of the `ucd_property` API.
 
-## 9.0.0.1 - (In Progress)
+## 10.0.0 - 2017-06-25
 
 *  Add `iswblank` and `iswxdigit` compatibility.
 *  Improve ctype compatibility.
-*  PropList property lookup.
+*  PropList and emoji-data property lookup.
+*  Support building with a C89 compiler.
+*  Update to Unicode Character Data 10.0.0.
+*  Unicode Emoji 5.0.
 
 ## 9.0.0 - 2016-12-28
 

src/ucd-tools/Makefile.am

@@ -55,34 +55,34 @@ EXTRA_DIST += ChangeLog
 
 ############################# Unicode Data ####################################
 
-EMOJI_VERSION=4.0
+EMOJI_VERSION=5.0
 UCD_VERSION=@UCD_VERSION@
 UCD_ROOTDIR=data/ucd
 UCD_SRCDIR=http://www.unicode.org/Public
 
 data/emoji/emoji-data.txt:
 	mkdir -pv data/emoji
-	curl ${UCD_SRCDIR}/emoji/${EMOJI_VERSION}/emoji-data.txt > $@
+	curl ${UCD_SRCDIR}/emoji/${EMOJI_VERSION}/emoji-data.txt -o $@
 
 data/ucd/PropList.txt:
 	mkdir -pv data/ucd
-	curl ${UCD_SRCDIR}/${UCD_VERSION}/ucd/PropList.txt > $@
+	curl ${UCD_SRCDIR}/${UCD_VERSION}/ucd/PropList.txt -o $@
 
 data/ucd/DerivedCoreProperties.txt:
 	mkdir -pv data/ucd
-	curl ${UCD_SRCDIR}/${UCD_VERSION}/ucd/DerivedCoreProperties.txt > $@
+	curl ${UCD_SRCDIR}/${UCD_VERSION}/ucd/DerivedCoreProperties.txt -o $@
 
 data/ucd/PropertyValueAliases.txt:
 	mkdir -pv data/ucd
-	curl ${UCD_SRCDIR}/${UCD_VERSION}/ucd/PropertyValueAliases.txt > $@
+	curl ${UCD_SRCDIR}/${UCD_VERSION}/ucd/PropertyValueAliases.txt -o $@
 
 data/ucd/Scripts.txt:
 	mkdir -pv data/ucd
-	curl ${UCD_SRCDIR}/${UCD_VERSION}/ucd/Scripts.txt > $@
+	curl ${UCD_SRCDIR}/${UCD_VERSION}/ucd/Scripts.txt -o $@
 
 data/ucd/UnicodeData.txt:
 	mkdir -pv data/ucd
-	curl ${UCD_SRCDIR}/${UCD_VERSION}/ucd/UnicodeData.txt > $@
+	curl ${UCD_SRCDIR}/${UCD_VERSION}/ucd/UnicodeData.txt -o $@
 
 ############################# documentation ###################################
 

src/ucd-tools/configure.ac

@@ -24,6 +24,7 @@ dnl library checks.
 dnl ================================================================
 
 AC_CHECK_HEADERS([stddef.h])   dnl C89
+AC_CHECK_FUNCS([iswblank])     dnl C99
 
 AC_TYPE_UINT8_T
 AC_TYPE_UINT32_T

src/ucd-tools/src/case.c

@@ -18,14 +18,15 @@
  * along with ucd-tools.  If not, see <http://www.gnu.org/licenses/>.
  */
 
-// NOTE: This file is automatically generated from the UnicodeData.txt file in
-// the Unicode Character database by the ucd-tools/tools/categories.py script.
+/* NOTE: This file is automatically generated from the UnicodeData.txt file in
+ * the Unicode Character database by the ucd-tools/tools/categories.py script.
+ */
 
 #include "ucd/ucd.h"
 
 #include <stddef.h>
 
-// Unicode Character Data 9.0.0
+/* Unicode Character Data 9.0.0 */
 
 struct case_conversion_entry
 {

src/ucd-tools/src/ctype.c

@@ -69,16 +69,16 @@ int ucd_isblank(codepoint_t c)
 	switch (ucd_lookup_category(c))
 	{
 	case UCD_CATEGORY_Zs:
-		switch (c) // Exclude characters with the <noBreak> DispositionType
+		switch (c) /* Exclude characters with the <noBreak> DispositionType */
 		{
-		case 0x00A0: // U+00A0 : NO-BREAK SPACE
-		case 0x2007: // U+2007 : FIGURE SPACE
-		case 0x202F: // U+202F : NARROW NO-BREAK SPACE
+		case 0x00A0: /* U+00A0 : NO-BREAK SPACE */
+		case 0x2007: /* U+2007 : FIGURE SPACE */
+		case 0x202F: /* U+202F : NARROW NO-BREAK SPACE */
 			return 0;
 		}
 		return 1;
 	case UCD_CATEGORY_Cc:
-		return c == 0x09; // U+0009 : CHARACTER TABULATION
+		return c == 0x09; /* U+0009 : CHARACTER TABULATION */
 	default:
 		return 0;
 	}
@@ -91,7 +91,7 @@ int ucd_iscntrl(codepoint_t c)
 
 int ucd_isdigit(codepoint_t c)
 {
-	return (c >= 0x30 && c <= 0x39); // [0-9]
+	return (c >= 0x30 && c <= 0x39); /* [0-9] */
 }
 
 int ucd_isgraph(codepoint_t c)
@@ -174,23 +174,23 @@ int ucd_isspace(codepoint_t c)
 	case UCD_CATEGORY_Zp:
 		return 1;
 	case UCD_CATEGORY_Zs:
-		switch (c) // Exclude characters with the <noBreak> DispositionType
+		switch (c) /* Exclude characters with the <noBreak> DispositionType */
 		{
-		case 0x00A0: // U+00A0 : NO-BREAK SPACE
-		case 0x2007: // U+2007 : FIGURE SPACE
-		case 0x202F: // U+202F : NARROW NO-BREAK SPACE
+		case 0x00A0: /* U+00A0 : NO-BREAK SPACE */
+		case 0x2007: /* U+2007 : FIGURE SPACE */
+		case 0x202F: /* U+202F : NARROW NO-BREAK SPACE */
 			return 0;
 		}
 		return 1;
 	case UCD_CATEGORY_Cc:
-		switch (c) // Include control characters marked as White_Space
+		switch (c) /* Include control characters marked as White_Space */
 		{
-		case 0x09: // U+0009 : CHARACTER TABULATION
-		case 0x0A: // U+000A : LINE FEED
-		case 0x0B: // U+000B : LINE TABULATION
-		case 0x0C: // U+000C : FORM FEED
-		case 0x0D: // U+000D : CARRIAGE RETURN
-		case 0x85: // U+0085 : NEXT LINE
+		case 0x09: /* U+0009 : CHARACTER TABULATION */
+		case 0x0A: /* U+000A : LINE FEED */
+		case 0x0B: /* U+000B : LINE TABULATION */
+		case 0x0C: /* U+000C : FORM FEED */
+		case 0x0D: /* U+000D : CARRIAGE RETURN */
+		case 0x85: /* U+0085 : NEXT LINE */
 			return 1;
 		}
 	default:
@@ -217,7 +217,7 @@ int ucd_isupper(codepoint_t c)
 
 int ucd_isxdigit(codepoint_t c)
 {
-	return (c >= 0x30 && c <= 0x39)  // [0-9]
-	    || (c >= 0x41 && c <= 0x46)  // [A-Z]
-	    || (c >= 0x61 && c <= 0x66); // [a-z]
+	return (c >= 0x30 && c <= 0x39)  /* [0-9] */
+	    || (c >= 0x41 && c <= 0x46)  /* [A-Z] */
+	    || (c >= 0x61 && c <= 0x66); /* [a-z] */
 }

src/ucd-tools/src/include/ucd/ucd.h

@@ -176,6 +176,7 @@ typedef enum ucd_script_
 	UCD_SCRIPT_Geok, /**< @brief Khutsuri Script */
 	UCD_SCRIPT_Geor, /**< @brief Geirgian Script */
 	UCD_SCRIPT_Glag, /**< @brief Glagolitic Script */
+	UCD_SCRIPT_Gonm, /**< @brief Masaram Gondi */
 	UCD_SCRIPT_Goth, /**< @brief Gothic Script */
 	UCD_SCRIPT_Gran, /**< @brief Grantha Script */
 	UCD_SCRIPT_Grek, /**< @brief Greek Script */
@@ -273,6 +274,7 @@ typedef enum ucd_script_
 	UCD_SCRIPT_Sind, /**< @brief Sindhi Script */
 	UCD_SCRIPT_Sinh, /**< @brief Sinhala Script */
 	UCD_SCRIPT_Sora, /**< @brief Sora Sompeng Script */
+	UCD_SCRIPT_Soyo, /**< @brief Soyombo */
 	UCD_SCRIPT_Sund, /**< @brief Sundanese Script */
 	UCD_SCRIPT_Sylo, /**< @brief Syloti Nagri Script */
 	UCD_SCRIPT_Syrc, /**< @brief Syriac Script */
@@ -302,6 +304,7 @@ typedef enum ucd_script_
 	UCD_SCRIPT_Xpeo, /**< @brief Old Persian Script */
 	UCD_SCRIPT_Xsux, /**< @brief Cuneiform Script */
 	UCD_SCRIPT_Yiii, /**< @brief Yi Script */
+	UCD_SCRIPT_Zanb, /**< @brief Zanabazar Square */
 	UCD_SCRIPT_Zinh, /**< @brief Inherited Script */
 	UCD_SCRIPT_Zmth, /**< @brief Mathematical Notation */
 	UCD_SCRIPT_Zsym, /**< @brief Symbols */
@@ -366,6 +369,8 @@ typedef uint64_t ucd_property;
 #define UCD_PROPERTY_EMOJI_PRESENTATION                 0x0000000400000000ull /**< @brief Emoji_Presentation */
 #define UCD_PROPERTY_EMOJI_MODIFIER                     0x0000000800000000ull /**< @brief Emoji_Modifier */
 #define UCD_PROPERTY_EMOJI_MODIFIER_BASE                0x0000001000000000ull /**< @brief Emoji_Modifier_Base */
+#define UCD_PROPERTY_REGIONAL_INDICATOR                 0x0000002000000000ull /**< @brief Regional_Indicator */
+#define UCD_PROPERTY_EMOJI_COMPONENT                    0x0000004000000000ull /**< @brief Emoji_Component */
 
 // eSpeak NG extended properties:
 #define ESPEAKNG_PROPERTY_INVERTED_TERMINAL_PUNCTUATION 0x0010000000000000ull /**< @brief Inverted_Terminal_Punctuation */
@@ -679,6 +684,7 @@ namespace ucd
 		Geok = UCD_SCRIPT_Geok, /**< @brief Khutsuri Script */
 		Geor = UCD_SCRIPT_Geor, /**< @brief Geirgian Script */
 		Glag = UCD_SCRIPT_Glag, /**< @brief Glagolitic Script */
+		Gonm = UCD_SCRIPT_Gonm, /**< @brief Masaram Gondi */
 		Goth = UCD_SCRIPT_Goth, /**< @brief Gothic Script */
 		Gran = UCD_SCRIPT_Gran, /**< @brief Grantha Script */
 		Grek = UCD_SCRIPT_Grek, /**< @brief Greek Script */
@@ -776,6 +782,7 @@ namespace ucd
 		Sind = UCD_SCRIPT_Sind, /**< @brief Sindhi Script */
 		Sinh = UCD_SCRIPT_Sinh, /**< @brief Sinhala Script */
 		Sora = UCD_SCRIPT_Sora, /**< @brief Sora Sompeng Script */
+		Soyo = UCD_SCRIPT_Soyo, /**< @brief Soyombo */
 		Sund = UCD_SCRIPT_Sund, /**< @brief Sundanese Script */
 		Sylo = UCD_SCRIPT_Sylo, /**< @brief Syloti Nagri Script */
 		Syrc = UCD_SCRIPT_Syrc, /**< @brief Syriac Script */
@@ -805,6 +812,7 @@ namespace ucd
 		Xpeo = UCD_SCRIPT_Xpeo, /**< @brief Old Persian Script */
 		Xsux = UCD_SCRIPT_Xsux, /**< @brief Cuneiform Script */
 		Yiii = UCD_SCRIPT_Yiii, /**< @brief Yi Script */
+		Zanb = UCD_SCRIPT_Zanb, /**< @brief Zanabazar Square */
 		Zinh = UCD_SCRIPT_Zinh, /**< @brief Inherited Script */
 		Zmth = UCD_SCRIPT_Zmth, /**< @brief Mathematical Notation */
 		Zsym = UCD_SCRIPT_Zsym, /**< @brief Symbols */
@@ -876,6 +884,8 @@ namespace ucd
 		Emoji_Presentation = UCD_PROPERTY_EMOJI_PRESENTATION, /**< @brief Emoji_Presentation */
 		Emoji_Modifier = UCD_PROPERTY_EMOJI_MODIFIER, /**< @brief Emoji_Modifier */
 		Emoji_Modifier_Base = UCD_PROPERTY_EMOJI_MODIFIER_BASE, /**< @brief Emoji_Modifier_Base */
+		Regional_Indicator = UCD_PROPERTY_REGIONAL_INDICATOR, /**< @brief Regional_Indicator */
+		Emoji_Component = UCD_PROPERTY_EMOJI_COMPONENT, /**< @brief Emoji_Component */
 	};
 
 	/** @brief Return the properties of the specified codepoint.

src/ucd-tools/src/proplist.c

@@ -78,9 +78,6 @@ static ucd_property properties_Cn(codepoint_t c)
 	case 0x2000:
 		if (c == 0x2065)                return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
 		break;
-	case 0x2300:
-		if (c == 0x23FF)                return UCD_PROPERTY_PATTERN_SYNTAX;
-		break;
 	case 0x2400:
 		if (c >= 0x2427 && c <= 0x243F) return UCD_PROPERTY_PATTERN_SYNTAX;
 		if (c >= 0x244B && c <= 0x245F) return UCD_PROPERTY_PATTERN_SYNTAX;
@@ -90,7 +87,7 @@ static ucd_property properties_Cn(codepoint_t c)
 		if (c >= 0x2B96 && c <= 0x2B97) return UCD_PROPERTY_PATTERN_SYNTAX;
 		if (c >= 0x2BBA && c <= 0x2BBC) return UCD_PROPERTY_PATTERN_SYNTAX;
 		if (c == 0x2BC9)                return UCD_PROPERTY_PATTERN_SYNTAX;
-		if (c >= 0x2BD2 && c <= 0x2BEB) return UCD_PROPERTY_PATTERN_SYNTAX;
+		if (c >= 0x2BD3 && c <= 0x2BEB) return UCD_PROPERTY_PATTERN_SYNTAX;
 		if (c >= 0x2BF0 && c <= 0x2BFF) return UCD_PROPERTY_PATTERN_SYNTAX;
 		break;
 	case 0x2E00:
@@ -148,11 +145,11 @@ static ucd_property properties_Ll(codepoint_t c)
 		if (c == 0x029D)                return UCD_PROPERTY_SOFT_DOTTED;
 		break;
 	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 */
 		if (c == 0x03D5)                return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x03F0 && c <= 0x03F1) return UCD_PROPERTY_OTHER_MATH;
 		if (c == 0x03F3)                return UCD_PROPERTY_SOFT_DOTTED;
-		if (c >= 0x03F4 && c <= 0x03F5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x03F4 && c <= 0x03F5) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		break;
 	case 0x0400:
 		if (c == 0x0456)                return UCD_PROPERTY_SOFT_DOTTED;
@@ -166,12 +163,12 @@ static ucd_property properties_Ll(codepoint_t c)
 		if (c == 0x1ECB)                return UCD_PROPERTY_SOFT_DOTTED;
 		break;
 	case 0x2100:
-		if (c >= 0x210A && c <= 0x2113) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x212F && c <= 0x2131) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x2133 && c <= 0x2134) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x210A && c <= 0x2113) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x212F && c <= 0x2131) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x2133 && c <= 0x2134) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c == 0x2139)                return UCD_PROPERTY_EMOJI;
-		if (c >= 0x213C && c <= 0x213F) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x2145 && c <= 0x2147) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x213C && c <= 0x213F) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x2145 && c <= 0x2147) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x2148 && c <= 0x2149) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
 		break;
 	case 0xFF00:
@@ -179,45 +176,45 @@ static ucd_property properties_Ll(codepoint_t c)
 		break;
 	case 0x01D400:
 		if (c >= 0x01D422 && c <= 0x01D423) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
-		if (c >= 0x01D400 && c <= 0x01D454) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D400 && c <= 0x01D454) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D456 && c <= 0x01D457) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
 		if (c >= 0x01D48A && c <= 0x01D48B) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
-		if (c >= 0x01D458 && c <= 0x01D49C) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x01D4AE && c <= 0x01D4B9) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D458 && c <= 0x01D49C) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x01D4AE && c <= 0x01D4B9) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c == 0x01D4BB)                  return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x01D4BE && c <= 0x01D4BF) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
 		if (c >= 0x01D4BD && c <= 0x01D4C3) return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x01D4F2 && c <= 0x01D4F3) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
-		if (c >= 0x01D4C5)                  return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D4C5)                  return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		break;
 	case 0x01D500:
-		if                  (c <= 0x01D505) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if                  (c <= 0x01D505) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D526 && c <= 0x01D527) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
-		if (c >= 0x01D51E && c <= 0x01D539) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D51E && c <= 0x01D539) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D55A && c <= 0x01D55B) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
 		if (c >= 0x01D58E && c <= 0x01D58F) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
 		if (c >= 0x01D5C2 && c <= 0x01D5C3) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
 		if (c >= 0x01D5F6 && c <= 0x01D5F7) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
-		if (c >= 0x01D552)                  return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D552)                  return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		break;
 	case 0x01D600:
 		if (c >= 0x01D62A && c <= 0x01D62B) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
 		if (c >= 0x01D65E && c <= 0x01D65F) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
 		if (c >= 0x01D692 && c <= 0x01D693) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
-		if                  (c <= 0x01D6A5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if                  (c <= 0x01D6A5) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D6C2 && c <= 0x01D6DA) return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x01D6DC && c <= 0x01D6FA) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D6DC && c <= 0x01D6FA) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D6FC)                  return UCD_PROPERTY_OTHER_MATH;
 		break;
 	case 0x01D700:
 		if                  (c <= 0x01D714) return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x01D716 && c <= 0x01D734) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D716 && c <= 0x01D734) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D736 && c <= 0x01D74E) return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x01D750 && c <= 0x01D76E) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D750 && c <= 0x01D76E) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D770 && c <= 0x01D788) return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x01D78A && c <= 0x01D7A8) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D78A && c <= 0x01D7A8) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D7AA && c <= 0x01D7C2) return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x01D7C4 && c <= 0x01D7CB) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D7C4 && c <= 0x01D7CB) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		break;
 	}
 	return 0;
@@ -332,7 +329,7 @@ static ucd_property properties_Lm(codepoint_t c)
 		break;
 	case 0x016F00:
 		if (c >= 0x016F93 && c <= 0x016F9F) return UCD_PROPERTY_DIACRITIC;
-		if (c == 0x016FE0)                  return UCD_PROPERTY_EXTENDER;
+		if (c >= 0x016FE0 && c <= 0x016FE1) return UCD_PROPERTY_EXTENDER;
 		break;
 	}
 	return 0;
@@ -407,19 +404,21 @@ static ucd_property properties_Lo_ideographic(codepoint_t c)
 	{
 	case 0x000000:
 		if (c >= 0x3400 && c <= 0x4DB5) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
-		if (c >= 0x4E00 && c <= 0x9FD5) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
+		if (c >= 0x4E00 && c <= 0x9FEA) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
 		if (c >= 0xF900 && c <= 0xFA6D) return UCD_PROPERTY_IDEOGRAPHIC;
 		if (c >= 0xFA70 && c <= 0xFAD9) return UCD_PROPERTY_IDEOGRAPHIC;
 		break;
 	case 0x010000:
 		if (c >= 0x017000 && c <= 0x0187EC) return UCD_PROPERTY_IDEOGRAPHIC;
 		if (c >= 0x018800 && c <= 0x018AF2) return UCD_PROPERTY_IDEOGRAPHIC;
+		if (c >= 0x01B170 && c <= 0x01B2FB) return UCD_PROPERTY_IDEOGRAPHIC;
 		break;
 	case 0x020000:
 		if (c >= 0x020000 && c <= 0x02A6D6) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
 		if (c >= 0x02A700 && c <= 0x02B734) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
 		if (c >= 0x02B740 && c <= 0x02B81D) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
 		if (c >= 0x02B820 && c <= 0x02CEA1) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
+		if (c >= 0x02CEB0 && c <= 0x02EBE0) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
 		if (c >= 0x02F800 && c <= 0x02FA1D) return UCD_PROPERTY_IDEOGRAPHIC;
 		break;
 	}
@@ -434,8 +433,8 @@ static ucd_property properties_Lu(codepoint_t c)
 		if (c >= 0x0041 && c <= 0x0046) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT;
 		break;
 	case 0x0300:
-		if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x03F4 && c <= 0x03F5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x03F4 && c <= 0x03F5) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		break;
 	case 0xFF00:
 		if (c >= 0xFF21 && c <= 0xFF26) return UCD_PROPERTY_HEX_DIGIT;
@@ -443,49 +442,49 @@ static ucd_property properties_Lu(codepoint_t c)
 	case 0x2100:
 		if (c == 0x2102)                return UCD_PROPERTY_OTHER_MATH;
 		if (c == 0x2107)                return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x210A && c <= 0x2113) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x210A && c <= 0x2113) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c == 0x2115)                return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x2119 && c <= 0x211D) return UCD_PROPERTY_OTHER_MATH;
 		if (c == 0x2124)                return UCD_PROPERTY_OTHER_MATH;
 		if (c == 0x2128)                return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x212C && c <= 0x212D) return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x212F && c <= 0x2131) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x2133 && c <= 0x2134) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x213C && c <= 0x213F) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x2145 && c <= 0x2149) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x212F && c <= 0x2131) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x2133 && c <= 0x2134) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x213C && c <= 0x213F) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x2145 && c <= 0x2149) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		break;
 	case 0x01D400:
-		if (c >= 0x01D400 && c <= 0x01D454) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x01D456 && c <= 0x01D49C) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D400 && c <= 0x01D454) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x01D456 && c <= 0x01D49C) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D49E && c <= 0x01D49F) return UCD_PROPERTY_OTHER_MATH;
 		if (c == 0x01D4A2)                  return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x01D4A5 && c <= 0x01D4A6) return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x01D4A9 && c <= 0x01D4AC) return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x01D4AE && c <= 0x01D4B9) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x01D4C5)                  return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D4AE && c <= 0x01D4B9) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x01D4C5)                  return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		break;
 	case 0x01D500:
-		if                  (c <= 0x01D505) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if                  (c <= 0x01D505) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D507 && c <= 0x01D50A) return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x01D50D && c <= 0x01D514) return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x01D516 && c <= 0x01D51C) return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x01D51E && c <= 0x01D539) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D51E && c <= 0x01D539) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D53B && c <= 0x01D53E) return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x01D540 && c <= 0x01D544) return UCD_PROPERTY_OTHER_MATH;
 		if (c == 0x01D546)                  return UCD_PROPERTY_OTHER_MATH;
 		if (c >= 0x01D54A && c <= 0x01D550) return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x01D552)                  return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D552)                  return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		break;
 	case 0x01D600:
-		if                  (c <= 0x01D6A5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if                  (c <= 0x01D6A5) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		if (c >= 0x01D6A8 && c <= 0x01D6C0) return UCD_PROPERTY_OTHER_MATH;
-		if (c >= 0x01D6DC && c <= 0x01D6FA) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D6DC && c <= 0x01D6FA) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		break;
 	case 0x01D700:
-		if (c >= 0x01D716 && c <= 0x01D734) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x01D750 && c <= 0x01D76E) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x01D78A && c <= 0x01D7A8) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
-		if (c >= 0x01D7C4 && c <= 0x01D7CB) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
+		if (c >= 0x01D716 && c <= 0x01D734) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x01D750 && c <= 0x01D76E) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x01D78A && c <= 0x01D7A8) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
+		if (c >= 0x01D7C4 && c <= 0x01D7CB) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
 		break;
 	}
 	return 0;
@@ -613,6 +612,7 @@ static ucd_property properties_Mc(codepoint_t c)
 		if (c >= 0x1C34 && c <= 0x1C35) return UCD_PROPERTY_OTHER_ALPHABETIC;
 		if (c == 0x1CE1)                return UCD_PROPERTY_DIACRITIC;
 		if (c >= 0x1CF2 && c <= 0x1CF3) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c == 0x1CF7)                return UCD_PROPERTY_DIACRITIC;
 		break;
 	case 0x3000:
 		if (c >= 0x302E && c <= 0x302F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
@@ -709,6 +709,12 @@ static ucd_property properties_Mc(codepoint_t c)
 		if (c >= 0x011720 && c <= 0x011721) return UCD_PROPERTY_OTHER_ALPHABETIC;
 		if (c == 0x011726)                  return UCD_PROPERTY_OTHER_ALPHABETIC;
 		break;
+	case 0x011A00:
+		if (c >= 0x011A07 && c <= 0x011A08) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c == 0x011A39)                  return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c >= 0x011A57 && c <= 0x011A58) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c == 0x011A97)                  return UCD_PROPERTY_OTHER_ALPHABETIC;
+		break;
 	case 0x011C00:
 		if (c == 0x011C2F)                  return UCD_PROPERTY_OTHER_ALPHABETIC;
 		if (c == 0x011C3E)                  return UCD_PROPERTY_OTHER_ALPHABETIC;
@@ -818,6 +824,8 @@ static ucd_property properties_Mn(codepoint_t c)
 		if (c >= 0x0AC7 && c <= 0x0AC8) return UCD_PROPERTY_OTHER_ALPHABETIC;
 		if (c == 0x0ACD)                return UCD_PROPERTY_DIACRITIC;
 		if (c >= 0x0AE2 && c <= 0x0AE3) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c >= 0x0AFA && c <= 0x0AFC) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c >= 0x0AFD && c <= 0x0AFF) return UCD_PROPERTY_DIACRITIC;
 		break;
 	case 0x0B00:
 		if (c == 0x0B01)                return UCD_PROPERTY_OTHER_ALPHABETIC;
@@ -848,7 +856,8 @@ static ucd_property properties_Mn(codepoint_t c)
 		if (c >= 0x0CE2 && c <= 0x0CE3) return UCD_PROPERTY_OTHER_ALPHABETIC;
 		break;
 	case 0x0D00:
-		if (c == 0x0D01)                return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c >= 0x0D00 && c <= 0x0D01) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c >= 0x0D3B && c <= 0x0D3C) return UCD_PROPERTY_DIACRITIC;
 		if (c >= 0x0D41 && c <= 0x0D44) return UCD_PROPERTY_OTHER_ALPHABETIC;
 		if (c == 0x0D4D)                return UCD_PROPERTY_DIACRITIC;
 		if (c >= 0x0D62 && c <= 0x0D63) return UCD_PROPERTY_OTHER_ALPHABETIC;
@@ -964,7 +973,7 @@ static ucd_property properties_Mn(codepoint_t c)
 	case 0x1D00:
 		if (c >= 0x1DC4 && c <= 0x1DCF) return UCD_PROPERTY_DIACRITIC;
 		if (c >= 0x1DE7 && c <= 0x1DF4) return UCD_PROPERTY_OTHER_ALPHABETIC;
-		if (c == 0x1DF5)                return UCD_PROPERTY_DIACRITIC;
+		if (c >= 0x1DF5 && c <= 0x1DF9) return UCD_PROPERTY_DIACRITIC;
 		if (c >= 0x1DFD && c <= 0x1DFF) return UCD_PROPERTY_DIACRITIC;
 		break;
 	case 0x2000:
@@ -1111,6 +1120,16 @@ static ucd_property properties_Mn(codepoint_t c)
 		if (c >= 0x011727 && c <= 0x01172A) return UCD_PROPERTY_OTHER_ALPHABETIC;
 		if (c == 0x01172B)                  return UCD_PROPERTY_DIACRITIC;
 		break;
+	case 0x011A00:
+		if (c >= 0x011A01 && c <= 0x011A0A) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c == 0x011A34)                  return UCD_PROPERTY_DIACRITIC;
+		if (c >= 0x011A35 && c <= 0x011A3E) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c == 0x011A47)                  return UCD_PROPERTY_DIACRITIC;
+		if (c >= 0x011A51 && c <= 0x011A5B) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c >= 0x011A8A && c <= 0x011A96) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c == 0x011A98)                  return UCD_PROPERTY_EXTENDER;
+		if (c == 0x011A99)                  return UCD_PROPERTY_DIACRITIC;
+		break;
 	case 0x011C00:
 		if (c >= 0x011C30 && c <= 0x011C36) return UCD_PROPERTY_OTHER_ALPHABETIC;
 		if (c >= 0x011C38 && c <= 0x011C3D) return UCD_PROPERTY_OTHER_ALPHABETIC;
@@ -1120,6 +1139,16 @@ static ucd_property properties_Mn(codepoint_t c)
 		if (c >= 0x011CB2 && c <= 0x011CB3) return UCD_PROPERTY_OTHER_ALPHABETIC;
 		if (c >= 0x011CB5 && c <= 0x011CB6) return UCD_PROPERTY_OTHER_ALPHABETIC;
 		break;
+	case 0x011D00:
+		if (c >= 0x011D31 && c <= 0x011D36) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c == 0x011D3A)                  return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c >= 0x011D3C && c <= 0x011D3D) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c >= 0x011D3F && c <= 0x011D41) return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c == 0x011D42)                  return UCD_PROPERTY_DIACRITIC;
+		if (c == 0x011D43)                  return UCD_PROPERTY_OTHER_ALPHABETIC;
+		if (c >= 0x011D44 && c <= 0x011D45) return UCD_PROPERTY_DIACRITIC;
+		if (c == 0x011D47)                  return UCD_PROPERTY_OTHER_ALPHABETIC;
+		break;
 	case 0x016A00:
 		if (c >= 0x016AF0 && c <= 0x016AF4) return UCD_PROPERTY_DIACRITIC;
 		break;
@@ -1165,7 +1194,7 @@ static ucd_property properties_Nd(codepoint_t c)
 	switch (c & 0xFFFFFF00)
 	{
 	case 0x0000:
-		if (c >= 0x0030 && c <= 0x0039) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT | UCD_PROPERTY_EMOJI;
+		if (c >= 0x0030 && c <= 0x0039) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_COMPONENT;
 		break;
 	case 0xFF00:
 		if (c >= 0xFF10 && c <= 0xFF19) return UCD_PROPERTY_HEX_DIGIT;
@@ -1279,10 +1308,10 @@ static ucd_property properties_Pe(codepoint_t c)
 		break;
 	case 0x2700:
 		if (c == 0x27C6)                return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
-		if (c >= 0x27E6 && c <= 0x27EF) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX; // Pe|Ps
+		if (c >= 0x27E6 && c <= 0x27EF) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX; /* Pe|Ps */
 		return UCD_PROPERTY_PATTERN_SYNTAX;
 	case 0x2900:
-		return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX; // Pe|Ps
+		return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX; /* Pe|Ps */
 	case 0x2E00:
 		return UCD_PROPERTY_PATTERN_SYNTAX;
 	case 0x3000:
@@ -1337,9 +1366,9 @@ static ucd_property properties_Po(codepoint_t c)
 	case 0x0000:
 		if (c == 0x0021)                return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX | ESPEAKNG_PROPERTY_EXCLAMATION_MARK;
 		if (c == 0x0022)                return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
-		if (c == 0x0023)                return UCD_PROPERTY_EMOJI | UCD_PROPERTY_PATTERN_SYNTAX;
+		if (c == 0x0023)                return UCD_PROPERTY_EMOJI | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI_COMPONENT;
 		if (c == 0x0027)                return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
-		if (c == 0x002A)                return UCD_PROPERTY_EMOJI | UCD_PROPERTY_PATTERN_SYNTAX;
+		if (c == 0x002A)                return UCD_PROPERTY_EMOJI | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI_COMPONENT;
 		if (c == 0x002C)                return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX | ESPEAKNG_PROPERTY_COMMA;
 		if (c == 0x002E)                return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX | ESPEAKNG_PROPERTY_FULL_STOP;
 		if (c == 0x003A)                return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX | ESPEAKNG_PROPERTY_COLON;
@@ -1609,6 +1638,11 @@ static ucd_property properties_Po(codepoint_t c)
 	case 0x11700:
 		if (c >= 0x01173C && c <= 0x01173E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
 		break;
+	case 0x11A00:
+		if (c >= 0x011A42 && c <= 0x011A43) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
+		if (c >= 0x011A9B && c <= 0x011A9C) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
+		if (c >= 0x011AA1 && c <= 0x011AA2) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
+		break;
 	case 0x11C00:
 		if (c >= 0x011C41 && c <= 0x011C42) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
 		if (c == 0x011C43)                  return UCD_PROPERTY_TERMINAL_PUNCTUATION;
@@ -1664,7 +1698,7 @@ static ucd_property properties_Ps(codepoint_t c)
 		break;
 	case 0x2700:
 		if (c == 0x27C5)                return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
-		if (c >= 0x27E6 && c <= 0x27EF) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX; // Pe|Ps
+		if (c >= 0x27E6 && c <= 0x27EF) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX; /* Pe|Ps */
 		return UCD_PROPERTY_PATTERN_SYNTAX;
 	case 0x2900:
 		return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
@@ -1746,7 +1780,7 @@ static ucd_property properties_Sk(codepoint_t c)
 		if (c == 0xFFE3)                return UCD_PROPERTY_DIACRITIC;
 		break;
 	case 0x01F300:
-		return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER;
+		return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER | UCD_PROPERTY_EMOJI_COMPONENT;
 	}
 	return 0;
 }
@@ -1839,7 +1873,7 @@ static ucd_property properties_So(codepoint_t c)
 		if (c == 0x23F3)                return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
 		if (c >= 0x23E9 && c <= 0x23F3) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
 		if (c >= 0x23F8 && c <= 0x23FA) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
-		if (c >= 0x23E3 && c <= 0x23FE) return UCD_PROPERTY_PATTERN_SYNTAX;
+		if (c >= 0x23E3)                return UCD_PROPERTY_PATTERN_SYNTAX;
 		break;
 	case 0x2400:
 		if (c >= 0x2400 && c <= 0x244A) return UCD_PROPERTY_PATTERN_SYNTAX;
@@ -1979,7 +2013,7 @@ static ucd_property properties_So(codepoint_t c)
 		if (c >= 0x01F170 && c <= 0x01F189) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE;
 		if (c == 0x01F18E)                  return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
 		if (c >= 0x01F191 && c <= 0x01F19A) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
-		if (c >= 0x01F1E6)                  return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
+		if (c >= 0x01F1E6)                  return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_REGIONAL_INDICATOR | UCD_PROPERTY_EMOJI_COMPONENT;
 		break;
 	case 0x01F200:
 		if (c == 0x01F201)                  return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
@@ -2074,23 +2108,29 @@ static ucd_property properties_So(codepoint_t c)
 		if (c >= 0x01F6EB && c <= 0x01F6EC) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
 		if (c == 0x01F6F0)                  return UCD_PROPERTY_EMOJI;
 		if (c == 0x01F6F3)                  return UCD_PROPERTY_EMOJI;
-		if (c >= 0x01F6F4 && c <= 0x01F6F6) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
+		if (c >= 0x01F6F4 && c <= 0x01F6F8) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
 		break;
 	case 0x01F900:
+		if (c <= 0x01F90B)                  return 0;
 		if (c >= 0x01F918 && c <= 0x01F91C) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
 		if (c >= 0x01F910 && c <= 0x01F91D) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
-		if (c == 0x01F91E)                  return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
+		if (c >= 0x01F91E && c <= 0x01F91F) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
 		if (c == 0x01F926)                  return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
-		if (c >= 0x01F920 && c <= 0x01F927) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
-		if (c == 0x01F930)                  return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
+		if (c >= 0x01F920 && c <= 0x01F92F) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
+		if (c >= 0x01F930 && c <= 0x01F932) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
 		if (c == 0x01F93B)                  return 0;
 		if (c >= 0x01F93A && c <= 0x01F93C) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
 		if (c >= 0x01F933 && c <= 0x01F93E) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
 		if (c == 0x01F946)                  return 0;
 		if (c >= 0x01F940 && c <= 0x01F94B) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
+		if (c == 0x01F94C)                  return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
 		if (c >= 0x01F950 && c <= 0x01F95E) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
+		if (c >= 0x01F95F && c <= 0x01F96B) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
 		if (c >= 0x01F980 && c <= 0x01F991) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
+		if (c >= 0x01F992 && c <= 0x01F997) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
 		if (c == 0x01F9C0)                  return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
+		if (c >= 0x01F9D1 && c <= 0x01F9DD) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
+		if (c >= 0x01F9D0 && c <= 0x01F9E6) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
 		return UCD_PROPERTY_EMOJI;
 	}
 	return 0;
@@ -2132,6 +2172,6 @@ ucd_property ucd_properties(codepoint_t c, ucd_category category)
 	case UCD_CATEGORY_Zl: return UCD_PROPERTY_WHITE_SPACE | UCD_PROPERTY_PATTERN_WHITE_SPACE;
 	case UCD_CATEGORY_Zp: return UCD_PROPERTY_WHITE_SPACE | UCD_PROPERTY_PATTERN_WHITE_SPACE | ESPEAKNG_PROPERTY_PARAGRAPH_SEPARATOR;
 	case UCD_CATEGORY_Zs: return properties_Zs(c);
-	default:              return 0; // Co Cs Ii Lt Me
+	default:              return 0; /* Co Cs Ii Lt Me */
 	};
 }

src/ucd-tools/src/tostring.c

@@ -120,6 +120,7 @@ const char *ucd_get_script_string(ucd_script s)
 		"Geok",
 		"Geor",
 		"Glag",
+		"Gonm",
 		"Goth",
 		"Gran",
 		"Grek",
@@ -217,6 +218,7 @@ const char *ucd_get_script_string(ucd_script s)
 		"Sind",
 		"Sinh",
 		"Sora",
+		"Soyo",
 		"Sund",
 		"Sylo",
 		"Syrc",
@@ -246,6 +248,7 @@ const char *ucd_get_script_string(ucd_script s)
 		"Xpeo",
 		"Xsux",
 		"Yiii",
+		"Zanb",
 		"Zinh",
 		"Zmth",
 		"Zsym",

src/ucd-tools/tests/printcdata.c

@@ -17,6 +17,7 @@
  * along with ucd-tools.  If not, see <http://www.gnu.org/licenses/>.
  */
 
+#include "config.h"
 #include "ucd/ucd.h"
 
 #include <locale.h>
@@ -25,6 +26,13 @@
 #include <wchar.h>
 #include <wctype.h>
 
+#ifndef HAVE_ISWBLANK
+static int iswblank(wint_t c)
+{
+	return iswspace(c) && !(c >= 0x0A && c <= 0x0D);
+}
+#endif
+
 void fput_utf8c(FILE *out, codepoint_t c)
 {
 	if (c < 0x80)
@@ -86,7 +94,7 @@ void uprintf_codepoint(FILE *out, codepoint_t c, char mode)
 {
 	switch (mode)
 	{
-	case 'c': // character
+	case 'c': /* character */
 		switch (c)
 		{
 		case '\t': fputs("\\t", out);  break;
@@ -95,10 +103,10 @@ void uprintf_codepoint(FILE *out, codepoint_t c, char mode)
 		default:   fput_utf8c(out, c); break;
 		}
 		break;
-	case 'h': // hexadecimal (lower)
+	case 'h': /* hexadecimal (lower) */
 		fprintf(out, "%06x", c);
 		break;
-	case 'H': // hexadecimal (upper)
+	case 'H': /* hexadecimal (upper) */
 		fprintf(out, "%06X", c);
 		break;
 	}
@@ -108,40 +116,40 @@ void uprintf_is(FILE *out, codepoint_t c, char mode)
 {
 	switch (mode)
 	{
-	case 'A': // alpha-numeric
+	case 'A': /* alpha-numeric */
 		fputc(iswalnum(c) ? '1' : '0', out);
 		break;
-	case 'a': // alpha
+	case 'a': /* alpha */
 		fputc(iswalpha(c) ? '1' : '0', out);
 		break;
-	case 'b': // blank
+	case 'b': /* blank */
 		fputc(iswblank(c) ? '1' : '0', out);
 		break;
-	case 'c': // control
+	case 'c': /* control */
 		fputc(iswcntrl(c) ? '1' : '0', out);
 		break;
-	case 'd': // numeric
+	case 'd': /* numeric */
 		fputc(iswdigit(c) ? '1' : '0', out);
 		break;
-	case 'g': // glyph
+	case 'g': /* glyph */
 		fputc(iswgraph(c) ? '1' : '0', out);
 		break;
-	case 'l': // lower case
+	case 'l': /* lower case */
 		fputc(iswlower(c) ? '1' : '0', out);
 		break;
-	case 'P': // printable
+	case 'P': /* printable */
 		fputc(iswprint(c) ? '1' : '0', out);
 		break;
-	case 'p': // punctuation
+	case 'p': /* punctuation */
 		fputc(iswpunct(c) ? '1' : '0', out);
 		break;
-	case 's': // whitespace
+	case 's': /* whitespace */
 		fputc(iswspace(c) ? '1' : '0', out);
 		break;
-	case 'u': // upper case
+	case 'u': /* upper case */
 		fputc(iswupper(c) ? '1' : '0', out);
 		break;
-	case 'x': // xdigit
+	case 'x': /* xdigit */
 		fputc(iswxdigit(c) ? '1' : '0', out);
 		break;
 	}
@@ -154,31 +162,31 @@ void uprintf(FILE *out, codepoint_t c, const char *format)
 	case '%':
 		switch (*++format)
 		{
-		case 'c': // category
+		case 'c': /* category */
 			fputs(ucd_get_category_string(ucd_lookup_category(c)), out);
 			break;
-		case 'C': // category group
+		case 'C': /* category group */
 			fputs(ucd_get_category_group_string(ucd_lookup_category_group(c)), out);
 			break;
-		case 'p': // codepoint
+		case 'p': /* codepoint */
 			uprintf_codepoint(out, c, *++format);
 			break;
-		case 'P': // properties
+		case 'P': /* properties */
 			fprintf(out, "%016llx", ucd_properties(c, ucd_lookup_category(c)));
 			break;
-		case 'i': // is*
+		case 'i': /* is* */
 			uprintf_is(out, c, *++format);
 			break;
-		case 'L': // lowercase
+		case 'L': /* lowercase */
 			uprintf_codepoint(out, towlower(c), *++format);
 			break;
-		case 's': // script
+		case 's': /* script */
 			fputs(ucd_get_script_string(ucd_lookup_script(c)), out);
 			break;
-		case 'T': // titlecase
+		case 'T': /* titlecase */
 			uprintf_codepoint(out, ucd_totitle(c), *++format);
 			break;
-		case 'U': // uppercase
+		case 'U': /* uppercase */
 			uprintf_codepoint(out, towupper(c), *++format);
 			break;
 		}
@@ -224,7 +232,8 @@ int main(int argc, char **argv)
 {
 	FILE *in = NULL;
 	const char *format = NULL;
-	for (int argn = 1; argn != argc; ++argn)
+	int argn;
+	for (argn = 1; argn != argc; ++argn)
 	{
 		const char *arg = argv[argn];
 		if (!strcmp(arg, "--stdin") || !strcmp(arg, "-"))
@@ -250,7 +259,8 @@ int main(int argc, char **argv)
 	}
 	else
 	{
-		for (codepoint_t c = 0; c <= 0x10FFFF; ++c)
+		codepoint_t c;
+		for (c = 0; c <= 0x10FFFF; ++c)
 			uprintf(stdout, c, format ? format :
 			        "%pH %s %C %c %UH %LH %TH %id %ix %ic %is %ib %ip %iP %ig %iA %ia %iu %il %P\n");
 	}

src/ucd-tools/tests/printucddata.c

@@ -83,7 +83,7 @@ void uprintf_codepoint(FILE *out, codepoint_t c, char mode)
 {
 	switch (mode)
 	{
-	case 'c': // character
+	case 'c': /* character */
 		switch (c)
 		{
 		case '\t': fputs("\\t", out);  break;
@@ -92,10 +92,10 @@ void uprintf_codepoint(FILE *out, codepoint_t c, char mode)
 		default:   fput_utf8c(out, c); break;
 		}
 		break;
-	case 'h': // hexadecimal (lower)
+	case 'h': /* hexadecimal (lower) */
 		fprintf(out, "%06x", c);
 		break;
-	case 'H': // hexadecimal (upper)
+	case 'H': /* hexadecimal (upper) */
 		fprintf(out, "%06X", c);
 		break;
 	}
@@ -105,40 +105,40 @@ void uprintf_is(FILE *out, codepoint_t c, char mode)
 {
 	switch (mode)
 	{
-	case 'A': // alpha-numeric
+	case 'A': /* alpha-numeric */
 		fputc(ucd_isalnum(c) ? '1' : '0', out);
 		break;
-	case 'a': // alpha
+	case 'a': /* alpha */
 		fputc(ucd_isalpha(c) ? '1' : '0', out);
 		break;
-	case 'b': // blank
+	case 'b': /* blank */
 		fputc(ucd_isblank(c) ? '1' : '0', out);
 		break;
-	case 'c': // control
+	case 'c': /* control */
 		fputc(ucd_iscntrl(c) ? '1' : '0', out);
 		break;
-	case 'd': // numeric
+	case 'd': /* numeric */
 		fputc(ucd_isdigit(c) ? '1' : '0', out);
 		break;
-	case 'g': // glyph
+	case 'g': /* glyph */
 		fputc(ucd_isgraph(c) ? '1' : '0', out);
 		break;
-	case 'l': // lower case
+	case 'l': /* lower case */
 		fputc(ucd_islower(c) ? '1' : '0', out);
 		break;
-	case 'P': // printable
+	case 'P': /* printable */
 		fputc(ucd_isprint(c) ? '1' : '0', out);
 		break;
-	case 'p': // punctuation
+	case 'p': /* punctuation */
 		fputc(ucd_ispunct(c) ? '1' : '0', out);
 		break;
-	case 's': // whitespace
+	case 's': /* whitespace */
 		fputc(ucd_isspace(c) ? '1' : '0', out);
 		break;
-	case 'u': // upper case
+	case 'u': /* upper case */
 		fputc(ucd_isupper(c) ? '1' : '0', out);
 		break;
-	case 'x': // xdigit
+	case 'x': /* xdigit */
 		fputc(ucd_isxdigit(c) ? '1' : '0', out);
 		break;
 	}
@@ -151,31 +151,31 @@ void uprintf(FILE *out, codepoint_t c, const char *format)
 	case '%':
 		switch (*++format)
 		{
-		case 'c': // category
+		case 'c': /* category */
 			fputs(ucd_get_category_string(ucd_lookup_category(c)), out);
 			break;
-		case 'C': // category group
+		case 'C': /* category group */
 			fputs(ucd_get_category_group_string(ucd_lookup_category_group(c)), out);
 			break;
-		case 'p': // codepoint
+		case 'p': /* codepoint */
 			uprintf_codepoint(out, c, *++format);
 			break;
-		case 'P': // properties
+		case 'P': /* properties */
 			fprintf(out, "%016llx", ucd_properties(c, ucd_lookup_category(c)));
 			break;
-		case 'i': // is*
+		case 'i': /* is* */
 			uprintf_is(out, c, *++format);
 			break;
-		case 'L': // lowercase
+		case 'L': /* lowercase */
 			uprintf_codepoint(out, ucd_tolower(c), *++format);
 			break;
-		case 's': // script
+		case 's': /* script */
 			fputs(ucd_get_script_string(ucd_lookup_script(c)), out);
 			break;
-		case 'T': // titlecase
+		case 'T': /* titlecase */
 			uprintf_codepoint(out, ucd_totitle(c), *++format);
 			break;
-		case 'U': // uppercase
+		case 'U': /* uppercase */
 			uprintf_codepoint(out, ucd_toupper(c), *++format);
 			break;
 		}
@@ -221,7 +221,8 @@ int main(int argc, char **argv)
 {
 	FILE *in = NULL;
 	const char *format = NULL;
-	for (int argn = 1; argn != argc; ++argn)
+	int argn;
+	for (argn = 1; argn != argc; ++argn)
 	{
 		const char *arg = argv[argn];
 		if (!strcmp(arg, "--stdin") || !strcmp(arg, "-"))
@@ -245,7 +246,8 @@ int main(int argc, char **argv)
 	}
 	else
 	{
-		for (codepoint_t c = 0; c <= 0x10FFFF; ++c)
+		codepoint_t c;
+		for (c = 0; c <= 0x10FFFF; ++c)
 			uprintf(stdout, c, format ? format :
 			        "%pH %s %C %c %UH %LH %TH %id %ix %ic %is %ib %ip %iP %ig %iA %ia %iu %il %P\n");
 	}

src/ucd-tools/tools/case.py

@@ -51,14 +51,15 @@ if __name__ == '__main__':
  * along with ucd-tools.  If not, see <http://www.gnu.org/licenses/>.
  */
 
-// NOTE: This file is automatically generated from the UnicodeData.txt file in
-// the Unicode Character database by the ucd-tools/tools/categories.py script.
+/* NOTE: This file is automatically generated from the UnicodeData.txt file in
+ * the Unicode Character database by the ucd-tools/tools/categories.py script.
+ */
 
 #include "ucd/ucd.h"
 
 #include <stddef.h>
 
-// Unicode Character Data %s
+/* Unicode Character Data %s */
 
 struct case_conversion_entry
 {

src/ucd-tools/tools/categories.py

@@ -110,8 +110,9 @@ if __name__ == '__main__':
  * along with ucd-tools.  If not, see <http://www.gnu.org/licenses/>.
  */
 
-// NOTE: This file is automatically generated from the UnicodeData.txt file in
-// the Unicode Character database by the ucd-tools/tools/categories.py script.
+/* NOTE: This file is automatically generated from the UnicodeData.txt file in
+ * the Unicode Character database by the ucd-tools/tools/categories.py script.
+ */
 
 #include "ucd/ucd.h"
 
@@ -149,7 +150,7 @@ if __name__ == '__main__':
 #define Zs UCD_CATEGORY_Zs
 #define Ii UCD_CATEGORY_Ii
 
-// Unicode Character Data %s
+/* Unicode Character Data %s */
 """ % ucd_version)
 
 	for category in special_categories:
@@ -187,7 +188,7 @@ if __name__ == '__main__':
 			sys.stdout.write('{\n')
 			for codepoint, table in sorted(category_tables[table_index].items()):
 				if isinstance(table, str):
-					sys.stdout.write('\tcategories_%s, // %s\n' % (table, codepoint))
+					sys.stdout.write('\tcategories_%s, /* %s */\n' % (table, codepoint))
 				else:
 					sys.stdout.write('\tcategories_%s,\n' % codepoint)
 			sys.stdout.write('};\n')
@@ -197,14 +198,14 @@ if __name__ == '__main__':
 	sys.stdout.write('{\n')
 	for codepoints, category, comment in category_sets:
 		if category:
-			sys.stdout.write('\tif (c <= 0x%s) return %s; // %s : %s\n' % (codepoints.last, category, codepoints, comment))
+			sys.stdout.write('\tif (c <= 0x%s) return %s; /* %s : %s */\n' % (codepoints.last, category, codepoints, comment))
 		else:
-			sys.stdout.write('\tif (c <= 0x%s) // %s\n' % (codepoints.last, codepoints))
+			sys.stdout.write('\tif (c <= 0x%s) /* %s */\n' % (codepoints.last, codepoints))
 			sys.stdout.write('\t{\n')
 			sys.stdout.write('\t\tconst uint8_t *table = categories_%s_%s[(c - 0x%s) / 256];\n' % (codepoints.first, codepoints.last, codepoints.first))
 			sys.stdout.write('\t\treturn (ucd_category)table[c % 256];\n')
 			sys.stdout.write('\t}\n')
-	sys.stdout.write('\treturn Ii; // Invalid Unicode Codepoint\n')
+	sys.stdout.write('\treturn Ii; /* Invalid Unicode Codepoint */\n')
 	sys.stdout.write('}\n')
 
 	sys.stdout.write("""

src/ucd-tools/tools/printdata.py

@@ -166,6 +166,8 @@ def properties(data):
 	props += (2 ** 34) * data.get('Emoji_Presentation', 0) # emoji-data
 	props += (2 ** 35) * data.get('Emoji_Modifier', 0) # emoji-data
 	props += (2 ** 36) * data.get('Emoji_Modifier_Base', 0) # emoji-data
+	props += (2 ** 37) * data.get('Regional_Indicator', 0) # PropList 10.0.0
+	props += (2 ** 38) * data.get('Emoji_Component', 0) # emoji-data 5.0
         # eSpeak NG extended properties:
 	props += (2 ** 52) * data.get('Inverted_Terminal_Punctuation', 0)
 	props += (2 ** 53) * data.get('Punctuation_In_Word', 0)

src/ucd-tools/tools/scripts.py

@@ -104,8 +104,9 @@ if __name__ == '__main__':
  * along with ucd-tools.  If not, see <http://www.gnu.org/licenses/>.
  */
 
-// NOTE: This file is automatically generated from the Scripts.txt file in
-// the Unicode Character database by the ucd-tools/tools/scripts.py script.
+/* NOTE: This file is automatically generated from the Scripts.txt file in
+ * the Unicode Character database by the ucd-tools/tools/scripts.py script.
+ */
 
 #include "ucd/ucd.h"
 
@@ -152,6 +153,7 @@ if __name__ == '__main__':
 #define Geok UCD_SCRIPT_Geok
 #define Geor UCD_SCRIPT_Geor
 #define Glag UCD_SCRIPT_Glag
+#define Gonm UCD_SCRIPT_Gonm
 #define Goth UCD_SCRIPT_Goth
 #define Gran UCD_SCRIPT_Gran
 #define Grek UCD_SCRIPT_Grek
@@ -249,6 +251,7 @@ if __name__ == '__main__':
 #define Sind UCD_SCRIPT_Sind
 #define Sinh UCD_SCRIPT_Sinh
 #define Sora UCD_SCRIPT_Sora
+#define Soyo UCD_SCRIPT_Soyo
 #define Sund UCD_SCRIPT_Sund
 #define Sylo UCD_SCRIPT_Sylo
 #define Syrc UCD_SCRIPT_Syrc
@@ -278,6 +281,7 @@ if __name__ == '__main__':
 #define Xpeo UCD_SCRIPT_Xpeo
 #define Xsux UCD_SCRIPT_Xsux
 #define Yiii UCD_SCRIPT_Yiii
+#define Zanb UCD_SCRIPT_Zanb
 #define Zinh UCD_SCRIPT_Zinh
 #define Zmth UCD_SCRIPT_Zmth
 #define Zsym UCD_SCRIPT_Zsym
@@ -285,7 +289,7 @@ if __name__ == '__main__':
 #define Zyyy UCD_SCRIPT_Zyyy
 #define Zzzz UCD_SCRIPT_Zzzz
 
-// Unicode Character Data %s
+/* Unicode Character Data %s */
 """ % ucd_version)
 
 	for script in special_scripts:
@@ -323,7 +327,7 @@ if __name__ == '__main__':
 			sys.stdout.write('{\n')
 			for codepoint, table in sorted(script_tables[table_index].items()):
 				if isinstance(table, str):
-					sys.stdout.write('\tscripts_%s, // %s\n' % (table, codepoint))
+					sys.stdout.write('\tscripts_%s, /* %s */\n' % (table, codepoint))
 				else:
 					sys.stdout.write('\tscripts_%s,\n' % codepoint)
 			sys.stdout.write('};\n')
@@ -333,12 +337,12 @@ if __name__ == '__main__':
 	sys.stdout.write('{\n')
 	for codepoints, script, comment in script_sets:
 		if script:
-			sys.stdout.write('\tif (c <= 0x%s) return %s; // %s : %s\n' % (codepoints.last, script, codepoints, comment))
+			sys.stdout.write('\tif (c <= 0x%s) return %s; /* %s : %s */\n' % (codepoints.last, script, codepoints, comment))
 		else:
-			sys.stdout.write('\tif (c <= 0x%s) // %s\n' % (codepoints.last, codepoints))
+			sys.stdout.write('\tif (c <= 0x%s) /* %s */\n' % (codepoints.last, codepoints))
 			sys.stdout.write('\t{\n')
 			sys.stdout.write('\t\tconst uint8_t *table = scripts_%s_%s[(c - 0x%s) / 256];\n' % (codepoints.first, codepoints.last, codepoints.first))
 			sys.stdout.write('\t\treturn (ucd_script)table[c % 256];\n')
 			sys.stdout.write('\t}\n')
-	sys.stdout.write('\treturn Zzzz; // Invalid Unicode Codepoint\n')
+	sys.stdout.write('\treturn Zzzz; /* Invalid Unicode Codepoint */\n')
 	sys.stdout.write('}\n')