Add hb_unicode_combining_class_t

src/hb-common.h

@@ -141,47 +141,11 @@ hb_language_t
 hb_language_get_default (void);
 
 
-/* hb_unicode_general_category_t */
-
-typedef enum
-{
-  HB_UNICODE_GENERAL_CATEGORY_CONTROL,			/* Cc */
-  HB_UNICODE_GENERAL_CATEGORY_FORMAT,			/* Cf */
-  HB_UNICODE_GENERAL_CATEGORY_UNASSIGNED,		/* Cn */
-  HB_UNICODE_GENERAL_CATEGORY_PRIVATE_USE,		/* Co */
-  HB_UNICODE_GENERAL_CATEGORY_SURROGATE,		/* Cs */
-  HB_UNICODE_GENERAL_CATEGORY_LOWERCASE_LETTER,		/* Ll */
-  HB_UNICODE_GENERAL_CATEGORY_MODIFIER_LETTER,		/* Lm */
-  HB_UNICODE_GENERAL_CATEGORY_OTHER_LETTER,		/* Lo */
-  HB_UNICODE_GENERAL_CATEGORY_TITLECASE_LETTER,		/* Lt */
-  HB_UNICODE_GENERAL_CATEGORY_UPPERCASE_LETTER,		/* Lu */
-  HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK,		/* Mc */
-  HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK,		/* Me */
-  HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK,		/* Mn */
-  HB_UNICODE_GENERAL_CATEGORY_DECIMAL_NUMBER,		/* Nd */
-  HB_UNICODE_GENERAL_CATEGORY_LETTER_NUMBER,		/* Nl */
-  HB_UNICODE_GENERAL_CATEGORY_OTHER_NUMBER,		/* No */
-  HB_UNICODE_GENERAL_CATEGORY_CONNECT_PUNCTUATION,	/* Pc */
-  HB_UNICODE_GENERAL_CATEGORY_DASH_PUNCTUATION,		/* Pd */
-  HB_UNICODE_GENERAL_CATEGORY_CLOSE_PUNCTUATION,	/* Pe */
-  HB_UNICODE_GENERAL_CATEGORY_FINAL_PUNCTUATION,	/* Pf */
-  HB_UNICODE_GENERAL_CATEGORY_INITIAL_PUNCTUATION,	/* Pi */
-  HB_UNICODE_GENERAL_CATEGORY_OTHER_PUNCTUATION,	/* Po */
-  HB_UNICODE_GENERAL_CATEGORY_OPEN_PUNCTUATION,		/* Ps */
-  HB_UNICODE_GENERAL_CATEGORY_CURRENCY_SYMBOL,		/* Sc */
-  HB_UNICODE_GENERAL_CATEGORY_MODIFIER_SYMBOL,		/* Sk */
-  HB_UNICODE_GENERAL_CATEGORY_MATH_SYMBOL,		/* Sm */
-  HB_UNICODE_GENERAL_CATEGORY_OTHER_SYMBOL,		/* So */
-  HB_UNICODE_GENERAL_CATEGORY_LINE_SEPARATOR,		/* Zl */
-  HB_UNICODE_GENERAL_CATEGORY_PARAGRAPH_SEPARATOR,	/* Zp */
-  HB_UNICODE_GENERAL_CATEGORY_SPACE_SEPARATOR		/* Zs */
-} hb_unicode_general_category_t;
-
-
 /* hb_script_t */
 
 /* http://unicode.org/iso15924/ */
 /* http://goo.gl/x9ilM */
+/* Unicode Character Database property: Script (sc) */
 typedef enum
 {
   /* Unicode-1.1 additions */

src/hb-glib.cc

@@ -192,13 +192,13 @@ hb_glib_script_from_script (hb_script_t script)
 }
 
 
-static unsigned int
+static hb_unicode_combining_class_t
 hb_glib_unicode_combining_class (hb_unicode_funcs_t *ufuncs HB_UNUSED,
 				 hb_codepoint_t      unicode,
 				 void               *user_data HB_UNUSED)
 
 {
-  return g_unichar_combining_class (unicode);
+  return (hb_unicode_combining_class_t) g_unichar_combining_class (unicode);
 }
 
 static unsigned int

src/hb-icu.cc

@@ -63,13 +63,13 @@ hb_icu_script_from_script (hb_script_t script)
 }
 
 
-static unsigned int
+static hb_unicode_combining_class_t
 hb_icu_unicode_combining_class (hb_unicode_funcs_t *ufuncs HB_UNUSED,
 				hb_codepoint_t      unicode,
 				void               *user_data HB_UNUSED)
 
 {
-  return u_getCombiningClass (unicode);
+  return (hb_unicode_combining_class_t) u_getCombiningClass (unicode);
 }
 
 static unsigned int

src/hb-unicode-private.hh

@@ -55,7 +55,7 @@
 
 /* Simple callbacks are those taking a hb_codepoint_t and returning a hb_codepoint_t */
 #define HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS_SIMPLE \
-  HB_UNICODE_FUNC_IMPLEMENT (unsigned int, combining_class) \
+  HB_UNICODE_FUNC_IMPLEMENT (hb_unicode_combining_class_t, combining_class) \
   HB_UNICODE_FUNC_IMPLEMENT (unsigned int, eastasian_width) \
   HB_UNICODE_FUNC_IMPLEMENT (hb_unicode_general_category_t, general_category) \
   HB_UNICODE_FUNC_IMPLEMENT (hb_codepoint_t, mirroring) \

src/hb-unicode.cc

@@ -38,12 +38,12 @@
  * hb_unicode_funcs_t
  */
 
-static unsigned int
+static hb_unicode_combining_class_t
 hb_unicode_combining_class_nil (hb_unicode_funcs_t *ufuncs    HB_UNUSED,
 				hb_codepoint_t      unicode   HB_UNUSED,
 				void               *user_data HB_UNUSED)
 {
-  return 0;
+  return HB_UNICODE_COMBINING_CLASS_NOT_REORDERED;
 }
 
 static unsigned int
@@ -354,11 +354,11 @@ _hb_unicode_modified_combining_class (hb_unicode_funcs_t *ufuncs,
      */
     c = c == 33 ? 27 : c + 1;
   }
-  else if (unlikely (hb_in_range<int> (c, 10, 25)))
+  else if (unlikely (hb_in_range<int> (c, 10, 26)))
   {
     /* The equivalent fix for Hebrew is more complex.
      *
-     * We permute the "fixed-position" classes 10-25 into the order
+     * We permute the "fixed-position" classes 10-26 into the order
      * described in the SBL Hebrew manual:
      *
      * http://www.sbl-site.org/Fonts/SBLHebrewUserManual1.5x.pdf
@@ -369,7 +369,7 @@ _hb_unicode_modified_combining_class (hb_unicode_funcs_t *ufuncs,
      * More details here:
      * https://bugzilla.mozilla.org/show_bug.cgi?id=662055
      */
-    static const int permuted_hebrew_classes[25 - 10 + 1] = {
+    static const int permuted_hebrew_classes[26 - 10 + 1] = {
       /* 10 sheva */        22,
       /* 11 hataf segol */  15,
       /* 12 hataf patah */  16,
@@ -386,6 +386,7 @@ _hb_unicode_modified_combining_class (hb_unicode_funcs_t *ufuncs,
       /* 23 rafe */         13,
       /* 24 shin dot */     10,
       /* 25 sin dot */      11,
+      /* 26 point varika */ 26,
     };
     c = permuted_hebrew_classes[c - 10];
   }

src/hb-unicode.h

@@ -40,6 +40,126 @@
 HB_BEGIN_DECLS
 
 
+/* hb_unicode_general_category_t */
+
+/* Unicode Character Database property: General_Category (gc) */
+typedef enum
+{
+  HB_UNICODE_GENERAL_CATEGORY_CONTROL,			/* Cc */
+  HB_UNICODE_GENERAL_CATEGORY_FORMAT,			/* Cf */
+  HB_UNICODE_GENERAL_CATEGORY_UNASSIGNED,		/* Cn */
+  HB_UNICODE_GENERAL_CATEGORY_PRIVATE_USE,		/* Co */
+  HB_UNICODE_GENERAL_CATEGORY_SURROGATE,		/* Cs */
+  HB_UNICODE_GENERAL_CATEGORY_LOWERCASE_LETTER,		/* Ll */
+  HB_UNICODE_GENERAL_CATEGORY_MODIFIER_LETTER,		/* Lm */
+  HB_UNICODE_GENERAL_CATEGORY_OTHER_LETTER,		/* Lo */
+  HB_UNICODE_GENERAL_CATEGORY_TITLECASE_LETTER,		/* Lt */
+  HB_UNICODE_GENERAL_CATEGORY_UPPERCASE_LETTER,		/* Lu */
+  HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK,		/* Mc */
+  HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK,		/* Me */
+  HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK,		/* Mn */
+  HB_UNICODE_GENERAL_CATEGORY_DECIMAL_NUMBER,		/* Nd */
+  HB_UNICODE_GENERAL_CATEGORY_LETTER_NUMBER,		/* Nl */
+  HB_UNICODE_GENERAL_CATEGORY_OTHER_NUMBER,		/* No */
+  HB_UNICODE_GENERAL_CATEGORY_CONNECT_PUNCTUATION,	/* Pc */
+  HB_UNICODE_GENERAL_CATEGORY_DASH_PUNCTUATION,		/* Pd */
+  HB_UNICODE_GENERAL_CATEGORY_CLOSE_PUNCTUATION,	/* Pe */
+  HB_UNICODE_GENERAL_CATEGORY_FINAL_PUNCTUATION,	/* Pf */
+  HB_UNICODE_GENERAL_CATEGORY_INITIAL_PUNCTUATION,	/* Pi */
+  HB_UNICODE_GENERAL_CATEGORY_OTHER_PUNCTUATION,	/* Po */
+  HB_UNICODE_GENERAL_CATEGORY_OPEN_PUNCTUATION,		/* Ps */
+  HB_UNICODE_GENERAL_CATEGORY_CURRENCY_SYMBOL,		/* Sc */
+  HB_UNICODE_GENERAL_CATEGORY_MODIFIER_SYMBOL,		/* Sk */
+  HB_UNICODE_GENERAL_CATEGORY_MATH_SYMBOL,		/* Sm */
+  HB_UNICODE_GENERAL_CATEGORY_OTHER_SYMBOL,		/* So */
+  HB_UNICODE_GENERAL_CATEGORY_LINE_SEPARATOR,		/* Zl */
+  HB_UNICODE_GENERAL_CATEGORY_PARAGRAPH_SEPARATOR,	/* Zp */
+  HB_UNICODE_GENERAL_CATEGORY_SPACE_SEPARATOR		/* Zs */
+} hb_unicode_general_category_t;
+
+/* hb_unicode_combining_class_t */
+
+/* Unicode Character Database property: Canonical_Combining_Class (ccc) */
+typedef enum
+{
+  HB_UNICODE_COMBINING_CLASS_NOT_REORDERED	= 0,
+  HB_UNICODE_COMBINING_CLASS_OVERLAY		= 1,
+  HB_UNICODE_COMBINING_CLASS_NUKTA		= 7,
+  HB_UNICODE_COMBINING_CLASS_KANA_VOICING	= 8,
+  HB_UNICODE_COMBINING_CLASS_VIRAMA		= 9,
+
+  /* Hebrew */
+  HB_UNICODE_COMBINING_CLASS_CCC10	=  10,
+  HB_UNICODE_COMBINING_CLASS_CCC11	=  11,
+  HB_UNICODE_COMBINING_CLASS_CCC12	=  12,
+  HB_UNICODE_COMBINING_CLASS_CCC13	=  13,
+  HB_UNICODE_COMBINING_CLASS_CCC14	=  14,
+  HB_UNICODE_COMBINING_CLASS_CCC15	=  15,
+  HB_UNICODE_COMBINING_CLASS_CCC16	=  16,
+  HB_UNICODE_COMBINING_CLASS_CCC17	=  17,
+  HB_UNICODE_COMBINING_CLASS_CCC18	=  18,
+  HB_UNICODE_COMBINING_CLASS_CCC19	=  19,
+  HB_UNICODE_COMBINING_CLASS_CCC20	=  20,
+  HB_UNICODE_COMBINING_CLASS_CCC21	=  21,
+  HB_UNICODE_COMBINING_CLASS_CCC22	=  22,
+  HB_UNICODE_COMBINING_CLASS_CCC23	=  23,
+  HB_UNICODE_COMBINING_CLASS_CCC24	=  24,
+  HB_UNICODE_COMBINING_CLASS_CCC25	=  25,
+  HB_UNICODE_COMBINING_CLASS_CCC26	=  26,
+
+  /* Arabic */
+  HB_UNICODE_COMBINING_CLASS_CCC27	=  27,
+  HB_UNICODE_COMBINING_CLASS_CCC28	=  28,
+  HB_UNICODE_COMBINING_CLASS_CCC29	=  29,
+  HB_UNICODE_COMBINING_CLASS_CCC30	=  30,
+  HB_UNICODE_COMBINING_CLASS_CCC31	=  31,
+  HB_UNICODE_COMBINING_CLASS_CCC32	=  32,
+  HB_UNICODE_COMBINING_CLASS_CCC33	=  33,
+  HB_UNICODE_COMBINING_CLASS_CCC34	=  34,
+  HB_UNICODE_COMBINING_CLASS_CCC35	=  35,
+
+  /* Syriac */
+  HB_UNICODE_COMBINING_CLASS_CCC36	=  36,
+
+  /* Telugu */
+  HB_UNICODE_COMBINING_CLASS_CCC84	=  84,
+  HB_UNICODE_COMBINING_CLASS_CCC91	=  91,
+
+  /* Thai */
+  HB_UNICODE_COMBINING_CLASS_CCC103	= 103,
+  HB_UNICODE_COMBINING_CLASS_CCC107	= 107,
+
+  /* Lao */
+  HB_UNICODE_COMBINING_CLASS_CCC118	= 118,
+  HB_UNICODE_COMBINING_CLASS_CCC122	= 122,
+
+  /* Tibetan */
+  HB_UNICODE_COMBINING_CLASS_CCC129	= 129,
+  HB_UNICODE_COMBINING_CLASS_CCC130	= 130,
+  HB_UNICODE_COMBINING_CLASS_CCC133	= 132,
+
+
+  HB_UNICODE_COMBINING_CLASS_ATTACHED_BELOW_LEFT	= 200,
+  HB_UNICODE_COMBINING_CLASS_ATTACHED_BELOW		= 202,
+  HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE		= 214,
+  HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE_RIGHT	= 216,
+  HB_UNICODE_COMBINING_CLASS_BELOW_LEFT			= 218,
+  HB_UNICODE_COMBINING_CLASS_BELOW			= 220,
+  HB_UNICODE_COMBINING_CLASS_BELOW_RIGHT		= 222,
+  HB_UNICODE_COMBINING_CLASS_LEFT			= 224,
+  HB_UNICODE_COMBINING_CLASS_RIGHT			= 226,
+  HB_UNICODE_COMBINING_CLASS_ABOVE_LEFT			= 228,
+  HB_UNICODE_COMBINING_CLASS_ABOVE			= 230,
+  HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT		= 232,
+  HB_UNICODE_COMBINING_CLASS_DOUBLE_BELOW		= 233,
+  HB_UNICODE_COMBINING_CLASS_DOUBLE_ABOVE		= 234,
+
+  HB_UNICODE_COMBINING_CLASS_IOTA_SUBSCRIPT		= 240,
+
+  HB_UNICODE_COMBINING_CLASS_INVALID	= 255
+} hb_unicode_combining_class_t;
+
+
 /*
  * hb_unicode_funcs_t
  */
@@ -95,7 +215,7 @@ hb_unicode_funcs_get_parent (hb_unicode_funcs_t *ufuncs);
 
 /* typedefs */
 
-typedef unsigned int			(*hb_unicode_combining_class_func_t)	(hb_unicode_funcs_t *ufuncs,
+typedef hb_unicode_combining_class_t	(*hb_unicode_combining_class_func_t)	(hb_unicode_funcs_t *ufuncs,
 										 hb_codepoint_t      unicode,
 										 void               *user_data);
 typedef unsigned int			(*hb_unicode_eastasian_width_func_t)	(hb_unicode_funcs_t *ufuncs,
@@ -192,7 +312,7 @@ hb_unicode_funcs_set_decompose_compatibility_func (hb_unicode_funcs_t *ufuncs,
 
 /* accessors */
 
-unsigned int
+hb_unicode_combining_class_t
 hb_unicode_combining_class (hb_unicode_funcs_t *ufuncs,
 			    hb_codepoint_t unicode);