Add script to/from ISO 15924 tag support

Also adds --script support to hb-view.

If a script tag is not known to us, we pass the ISO 15924 tag around.
Right now, the OT layer ignores that, but we can fix it to blindly
convert that to an OT script tag.

src/hb-ot-tag.c

@@ -152,6 +152,8 @@ hb_ot_tags_from_script (hb_script_t script)
 {
   static const hb_tag_t def_tag[] = {HB_OT_TAG_DEFAULT_SCRIPT, HB_TAG_NONE};
 
+  /* XXX Handle non-enum scripts */
+
   if (unlikely ((unsigned int) script >= ARRAY_LENGTH (ot_scripts)))
     return def_tag;
 
@@ -170,6 +172,8 @@ hb_ot_tag_to_script (hb_tag_t tag)
         return i;
   }
 
+  /* XXX Convert to non-enum scripts */
+
   return HB_SCRIPT_UNKNOWN;
 }
 

src/hb-unicode.c

@@ -234,6 +234,279 @@ hb_unicode_get_eastasian_width (hb_unicode_funcs_t *ufuncs,
 }
 
 
+/* hb_script_t */
+
+static const hb_tag_t script_to_iso15924_tag[] =
+{
+  HB_TAG('Z','y','y','y'),	/* HB_SCRIPT_COMMON */
+  HB_TAG('Q','a','a','i'),	/* HB_SCRIPT_INHERITED */
+  HB_TAG('A','r','a','b'),	/* HB_SCRIPT_ARABIC */
+  HB_TAG('A','r','m','n'),	/* HB_SCRIPT_ARMENIAN */
+  HB_TAG('B','e','n','g'),	/* HB_SCRIPT_BENGALI */
+  HB_TAG('B','o','p','o'),	/* HB_SCRIPT_BOPOMOFO */
+  HB_TAG('C','h','e','r'),	/* HB_SCRIPT_CHEROKEE */
+  HB_TAG('Q','a','a','c'),	/* HB_SCRIPT_COPTIC */
+  HB_TAG('C','y','r','l'),	/* HB_SCRIPT_CYRILLIC */
+  HB_TAG('D','s','r','t'),	/* HB_SCRIPT_DESERET */
+  HB_TAG('D','e','v','a'),	/* HB_SCRIPT_DEVANAGARI */
+  HB_TAG('E','t','h','i'),	/* HB_SCRIPT_ETHIOPIC */
+  HB_TAG('G','e','o','r'),	/* HB_SCRIPT_GEORGIAN */
+  HB_TAG('G','o','t','h'),	/* HB_SCRIPT_GOTHIC */
+  HB_TAG('G','r','e','k'),	/* HB_SCRIPT_GREEK */
+  HB_TAG('G','u','j','r'),	/* HB_SCRIPT_GUJARATI */
+  HB_TAG('G','u','r','u'),	/* HB_SCRIPT_GURMUKHI */
+  HB_TAG('H','a','n','i'),	/* HB_SCRIPT_HAN */
+  HB_TAG('H','a','n','g'),	/* HB_SCRIPT_HANGUL */
+  HB_TAG('H','e','b','r'),	/* HB_SCRIPT_HEBREW */
+  HB_TAG('H','i','r','a'),	/* HB_SCRIPT_HIRAGANA */
+  HB_TAG('K','n','d','a'),	/* HB_SCRIPT_KANNADA */
+  HB_TAG('K','a','n','a'),	/* HB_SCRIPT_KATAKANA */
+  HB_TAG('K','h','m','r'),	/* HB_SCRIPT_KHMER */
+  HB_TAG('L','a','o','o'),	/* HB_SCRIPT_LAO */
+  HB_TAG('L','a','t','n'),	/* HB_SCRIPT_LATIN */
+  HB_TAG('M','l','y','m'),	/* HB_SCRIPT_MALAYALAM */
+  HB_TAG('M','o','n','g'),	/* HB_SCRIPT_MONGOLIAN */
+  HB_TAG('M','y','m','r'),	/* HB_SCRIPT_MYANMAR */
+  HB_TAG('O','g','a','m'),	/* HB_SCRIPT_OGHAM */
+  HB_TAG('I','t','a','l'),	/* HB_SCRIPT_OLD_ITALIC */
+  HB_TAG('O','r','y','a'),	/* HB_SCRIPT_ORIYA */
+  HB_TAG('R','u','n','r'),	/* HB_SCRIPT_RUNIC */
+  HB_TAG('S','i','n','h'),	/* HB_SCRIPT_SINHALA */
+  HB_TAG('S','y','r','c'),	/* HB_SCRIPT_SYRIAC */
+  HB_TAG('T','a','m','l'),	/* HB_SCRIPT_TAMIL */
+  HB_TAG('T','e','l','u'),	/* HB_SCRIPT_TELUGU */
+  HB_TAG('T','h','a','a'),	/* HB_SCRIPT_THAANA */
+  HB_TAG('T','h','a','i'),	/* HB_SCRIPT_THAI */
+  HB_TAG('T','i','b','t'),	/* HB_SCRIPT_TIBETAN */
+  HB_TAG('C','a','n','s'),	/* HB_SCRIPT_CANADIAN_ABORIGINAL */
+  HB_TAG('Y','i','i','i'),	/* HB_SCRIPT_YI */
+  HB_TAG('T','g','l','g'),	/* HB_SCRIPT_TAGALOG */
+  HB_TAG('H','a','n','o'),	/* HB_SCRIPT_HANUNOO */
+  HB_TAG('B','u','h','d'),	/* HB_SCRIPT_BUHID */
+  HB_TAG('T','a','g','b'),	/* HB_SCRIPT_TAGBANWA */
+
+  /* Unicode-4.0 additions */
+  HB_TAG('B','r','a','i'),	/* HB_SCRIPT_BRAILLE */
+  HB_TAG('C','p','r','t'),	/* HB_SCRIPT_CYPRIOT */
+  HB_TAG('L','i','m','b'),	/* HB_SCRIPT_LIMBU */
+  HB_TAG('O','s','m','a'),	/* HB_SCRIPT_OSMANYA */
+  HB_TAG('S','h','a','w'),	/* HB_SCRIPT_SHAVIAN */
+  HB_TAG('L','i','n','b'),	/* HB_SCRIPT_LINEAR_B */
+  HB_TAG('T','a','l','e'),	/* HB_SCRIPT_TAI_LE */
+  HB_TAG('U','g','a','r'),	/* HB_SCRIPT_UGARITIC */
+
+  /* Unicode-4.1 additions */
+  HB_TAG('T','a','l','u'),	/* HB_SCRIPT_NEW_TAI_LUE */
+  HB_TAG('B','u','g','i'),	/* HB_SCRIPT_BUGINESE */
+  HB_TAG('G','l','a','g'),	/* HB_SCRIPT_GLAGOLITIC */
+  HB_TAG('T','f','n','g'),	/* HB_SCRIPT_TIFINAGH */
+  HB_TAG('S','y','l','o'),	/* HB_SCRIPT_SYLOTI_NAGRI */
+  HB_TAG('X','p','e','o'),	/* HB_SCRIPT_OLD_PERSIAN */
+  HB_TAG('K','h','a','r'),	/* HB_SCRIPT_KHAROSHTHI */
+
+  /* Unicode-5.0 additions */
+  HB_TAG('Z','z','z','z'),	/* HB_SCRIPT_UNKNOWN */
+  HB_TAG('B','a','l','i'),	/* HB_SCRIPT_BALINESE */
+  HB_TAG('X','s','u','x'),	/* HB_SCRIPT_CUNEIFORM */
+  HB_TAG('P','h','n','x'),	/* HB_SCRIPT_PHOENICIAN */
+  HB_TAG('P','h','a','g'),	/* HB_SCRIPT_PHAGS_PA */
+  HB_TAG('N','k','o','o'),	/* HB_SCRIPT_NKO */
+
+  /* Unicode-5.1 additions */
+  HB_TAG('K','a','l','i'),	/* HB_SCRIPT_KAYAH_LI */
+  HB_TAG('L','e','p','c'),	/* HB_SCRIPT_LEPCHA */
+  HB_TAG('R','j','n','g'),	/* HB_SCRIPT_REJANG */
+  HB_TAG('S','u','n','d'),	/* HB_SCRIPT_SUNDANESE */
+  HB_TAG('S','a','u','r'),	/* HB_SCRIPT_SAURASHTRA */
+  HB_TAG('C','h','a','m'),	/* HB_SCRIPT_CHAM */
+  HB_TAG('O','l','c','k'),	/* HB_SCRIPT_OL_CHIKI */
+  HB_TAG('V','a','i','i'),	/* HB_SCRIPT_VAI */
+  HB_TAG('C','a','r','i'),	/* HB_SCRIPT_CARIAN */
+  HB_TAG('L','y','c','i'),	/* HB_SCRIPT_LYCIAN */
+  HB_TAG('L','y','d','i'),	/* HB_SCRIPT_LYDIAN */
+
+  /* Unicode-5.2 additions */
+  HB_TAG('A','v','s','t'),	/* HB_SCRIPT_AVESTAN */
+  HB_TAG('B','a','m','u'),	/* HB_SCRIPT_BAMUM */
+  HB_TAG('E','g','y','p'),	/* HB_SCRIPT_EGYPTIAN_HIEROGLYPHS */
+  HB_TAG('A','r','m','i'),	/* HB_SCRIPT_IMPERIAL_ARAMAIC */
+  HB_TAG('P','h','l','i'),	/* HB_SCRIPT_INSCRIPTIONAL_PAHLAVI */
+  HB_TAG('P','r','t','i'),	/* HB_SCRIPT_INSCRIPTIONAL_PARTHIAN */
+  HB_TAG('J','a','v','a'),	/* HB_SCRIPT_JAVANESE */
+  HB_TAG('K','t','h','i'),	/* HB_SCRIPT_KAITHI */
+  HB_TAG('L','i','s','u'),	/* HB_SCRIPT_LISU */
+  HB_TAG('M','t','e','i'),	/* HB_SCRIPT_MEETEI_MAYEK */
+  HB_TAG('S','a','r','b'),	/* HB_SCRIPT_OLD_SOUTH_ARABIAN */
+  HB_TAG('O','r','k','h'),	/* HB_SCRIPT_OLD_TURKIC */
+  HB_TAG('S','a','m','r'),	/* HB_SCRIPT_SAMARITAN */
+  HB_TAG('L','a','n','a'),	/* HB_SCRIPT_TAI_THAM */
+  HB_TAG('T','a','v','t'),	/* HB_SCRIPT_TAI_VIET */
+
+  /* Unicode-6.0 additions */
+  HB_TAG('B','a','t','k'),	/* HB_SCRIPT_BATAK */
+  HB_TAG('B','r','a','h'),	/* HB_SCRIPT_BRAHMI */
+  HB_TAG('M','a','n','d') 	/* HB_SCRIPT_MANDAIC */
+};
+
+struct tag_script_pair {
+  hb_tag_t tag;
+  hb_script_t script;
+};
+static const struct tag_script_pair script_from_iso15924_tag[] =
+{
+  {HB_TAG('A','r','a','b'), HB_SCRIPT_ARABIC},
+  {HB_TAG('A','r','m','i'), HB_SCRIPT_IMPERIAL_ARAMAIC},
+  {HB_TAG('A','r','m','n'), HB_SCRIPT_ARMENIAN},
+  {HB_TAG('A','v','s','t'), HB_SCRIPT_AVESTAN},
+  {HB_TAG('B','a','l','i'), HB_SCRIPT_BALINESE},
+  {HB_TAG('B','a','m','u'), HB_SCRIPT_BAMUM},
+  {HB_TAG('B','a','t','k'), HB_SCRIPT_BATAK},
+  {HB_TAG('B','e','n','g'), HB_SCRIPT_BENGALI},
+  {HB_TAG('B','o','p','o'), HB_SCRIPT_BOPOMOFO},
+  {HB_TAG('B','r','a','h'), HB_SCRIPT_BRAHMI},
+  {HB_TAG('B','r','a','i'), HB_SCRIPT_BRAILLE},
+  {HB_TAG('B','u','g','i'), HB_SCRIPT_BUGINESE},
+  {HB_TAG('B','u','h','d'), HB_SCRIPT_BUHID},
+  {HB_TAG('C','a','n','s'), HB_SCRIPT_CANADIAN_ABORIGINAL},
+  {HB_TAG('C','a','r','i'), HB_SCRIPT_CARIAN},
+  {HB_TAG('C','h','a','m'), HB_SCRIPT_CHAM},
+  {HB_TAG('C','h','e','r'), HB_SCRIPT_CHEROKEE},
+  {HB_TAG('C','p','r','t'), HB_SCRIPT_CYPRIOT},
+  {HB_TAG('C','y','r','l'), HB_SCRIPT_CYRILLIC},
+  {HB_TAG('C','y','r','s'), HB_SCRIPT_CYRILLIC},
+  {HB_TAG('D','e','v','a'), HB_SCRIPT_DEVANAGARI},
+  {HB_TAG('D','s','r','t'), HB_SCRIPT_DESERET},
+  {HB_TAG('E','g','y','p'), HB_SCRIPT_EGYPTIAN_HIEROGLYPHS},
+  {HB_TAG('E','t','h','i'), HB_SCRIPT_ETHIOPIC},
+  {HB_TAG('G','e','o','a'), HB_SCRIPT_GEORGIAN},
+  {HB_TAG('G','e','o','n'), HB_SCRIPT_GEORGIAN},
+  {HB_TAG('G','e','o','r'), HB_SCRIPT_GEORGIAN},
+  {HB_TAG('G','l','a','g'), HB_SCRIPT_GLAGOLITIC},
+  {HB_TAG('G','o','t','h'), HB_SCRIPT_GOTHIC},
+  {HB_TAG('G','r','e','k'), HB_SCRIPT_GREEK},
+  {HB_TAG('G','u','j','r'), HB_SCRIPT_GUJARATI},
+  {HB_TAG('G','u','r','u'), HB_SCRIPT_GURMUKHI},
+  {HB_TAG('H','a','n','g'), HB_SCRIPT_HANGUL},
+  {HB_TAG('H','a','n','i'), HB_SCRIPT_HAN},
+  {HB_TAG('H','a','n','o'), HB_SCRIPT_HANUNOO},
+  {HB_TAG('H','e','b','r'), HB_SCRIPT_HEBREW},
+  {HB_TAG('H','i','r','a'), HB_SCRIPT_HIRAGANA},
+  {HB_TAG('I','t','a','l'), HB_SCRIPT_OLD_ITALIC},
+  {HB_TAG('J','a','v','a'), HB_SCRIPT_JAVANESE},
+  {HB_TAG('K','a','l','i'), HB_SCRIPT_KAYAH_LI},
+  {HB_TAG('K','a','n','a'), HB_SCRIPT_KATAKANA},
+  {HB_TAG('K','h','a','r'), HB_SCRIPT_KHAROSHTHI},
+  {HB_TAG('K','h','m','r'), HB_SCRIPT_KHMER},
+  {HB_TAG('K','n','d','a'), HB_SCRIPT_KANNADA},
+  {HB_TAG('K','t','h','i'), HB_SCRIPT_KAITHI},
+  {HB_TAG('L','a','n','a'), HB_SCRIPT_TAI_THAM},
+  {HB_TAG('L','a','o','o'), HB_SCRIPT_LAO},
+  {HB_TAG('L','a','t','f'), HB_SCRIPT_LATIN},
+  {HB_TAG('L','a','t','g'), HB_SCRIPT_LATIN},
+  {HB_TAG('L','a','t','n'), HB_SCRIPT_LATIN},
+  {HB_TAG('L','e','p','c'), HB_SCRIPT_LEPCHA},
+  {HB_TAG('L','i','m','b'), HB_SCRIPT_LIMBU},
+  {HB_TAG('L','i','n','b'), HB_SCRIPT_LINEAR_B},
+  {HB_TAG('L','i','s','u'), HB_SCRIPT_LISU},
+  {HB_TAG('L','y','c','i'), HB_SCRIPT_LYCIAN},
+  {HB_TAG('L','y','d','i'), HB_SCRIPT_LYDIAN},
+  {HB_TAG('M','a','n','d'), HB_SCRIPT_MANDAIC},
+  {HB_TAG('M','l','y','m'), HB_SCRIPT_MALAYALAM},
+  {HB_TAG('M','o','n','g'), HB_SCRIPT_MONGOLIAN},
+  {HB_TAG('M','t','e','i'), HB_SCRIPT_MEETEI_MAYEK},
+  {HB_TAG('M','y','m','r'), HB_SCRIPT_MYANMAR},
+  {HB_TAG('N','k','o','o'), HB_SCRIPT_NKO},
+  {HB_TAG('O','g','a','m'), HB_SCRIPT_OGHAM},
+  {HB_TAG('O','l','c','k'), HB_SCRIPT_OL_CHIKI},
+  {HB_TAG('O','r','k','h'), HB_SCRIPT_OLD_TURKIC},
+  {HB_TAG('O','r','y','a'), HB_SCRIPT_ORIYA},
+  {HB_TAG('O','s','m','a'), HB_SCRIPT_OSMANYA},
+  {HB_TAG('P','h','a','g'), HB_SCRIPT_PHAGS_PA},
+  {HB_TAG('P','h','l','i'), HB_SCRIPT_INSCRIPTIONAL_PAHLAVI},
+  {HB_TAG('P','h','n','x'), HB_SCRIPT_PHOENICIAN},
+  {HB_TAG('P','r','t','i'), HB_SCRIPT_INSCRIPTIONAL_PARTHIAN},
+  {HB_TAG('Q','a','a','c'), HB_SCRIPT_COPTIC},
+  {HB_TAG('Q','a','a','i'), HB_SCRIPT_INHERITED},
+  {HB_TAG('R','j','n','g'), HB_SCRIPT_REJANG},
+  {HB_TAG('R','u','n','r'), HB_SCRIPT_RUNIC},
+  {HB_TAG('S','a','m','r'), HB_SCRIPT_SAMARITAN},
+  {HB_TAG('S','a','r','b'), HB_SCRIPT_OLD_SOUTH_ARABIAN},
+  {HB_TAG('S','a','u','r'), HB_SCRIPT_SAURASHTRA},
+  {HB_TAG('S','h','a','w'), HB_SCRIPT_SHAVIAN},
+  {HB_TAG('S','i','n','h'), HB_SCRIPT_SINHALA},
+  {HB_TAG('S','u','n','d'), HB_SCRIPT_SUNDANESE},
+  {HB_TAG('S','y','l','o'), HB_SCRIPT_SYLOTI_NAGRI},
+  {HB_TAG('S','y','r','c'), HB_SCRIPT_SYRIAC},
+  {HB_TAG('S','y','r','e'), HB_SCRIPT_SYRIAC},
+  {HB_TAG('S','y','r','n'), HB_SCRIPT_SYRIAC},
+  {HB_TAG('T','a','g','b'), HB_SCRIPT_TAGBANWA},
+  {HB_TAG('T','a','l','e'), HB_SCRIPT_TAI_LE},
+  {HB_TAG('T','a','l','u'), HB_SCRIPT_NEW_TAI_LUE},
+  {HB_TAG('T','a','m','l'), HB_SCRIPT_TAMIL},
+  {HB_TAG('T','a','v','t'), HB_SCRIPT_TAI_VIET},
+  {HB_TAG('T','e','l','u'), HB_SCRIPT_TELUGU},
+  {HB_TAG('T','f','n','g'), HB_SCRIPT_TIFINAGH},
+  {HB_TAG('T','g','l','g'), HB_SCRIPT_TAGALOG},
+  {HB_TAG('T','h','a','a'), HB_SCRIPT_THAANA},
+  {HB_TAG('T','h','a','i'), HB_SCRIPT_THAI},
+  {HB_TAG('T','i','b','t'), HB_SCRIPT_TIBETAN},
+  {HB_TAG('U','g','a','r'), HB_SCRIPT_UGARITIC},
+  {HB_TAG('V','a','i','i'), HB_SCRIPT_VAI},
+  {HB_TAG('X','p','e','o'), HB_SCRIPT_OLD_PERSIAN},
+  {HB_TAG('X','s','u','x'), HB_SCRIPT_CUNEIFORM},
+  {HB_TAG('Y','i','i','i'), HB_SCRIPT_YI},
+  {HB_TAG('Z','y','y','y'), HB_SCRIPT_COMMON},
+  {HB_TAG('Z','z','z','z'), HB_SCRIPT_UNKNOWN}
+};
+
+static int
+_tag_cmp (hb_tag_t *pa, hb_tag_t *pb)
+{
+  hb_tag_t a = *pa, b = *pb;
+  return a < b ? -1 : a == b ? 0 : +1;
+}
+
+
+hb_script_t
+hb_script_from_iso15924_tag (hb_tag_t tag)
+{
+  const struct tag_script_pair *pair;
+
+  /* Be lenient, adjust case (one capital letter followed by three small letters) */
+  tag = (tag & 0xDFDFDFDF) | 0x00202020;
+
+  pair = (const struct tag_script_pair *) bsearch (&tag,
+						   script_from_iso15924_tag,
+						   ARRAY_LENGTH (script_from_iso15924_tag),
+						   sizeof (script_from_iso15924_tag[0]),
+						   (hb_compare_func_t) _tag_cmp);
+
+  if (pair)
+    return pair->script;
+
+  /* If it looks right, just use the tag as a script */
+  if (((uint32_t) tag & 0xE0E0E0E0) == 0x40606060)
+    return (hb_script_t) tag;
+
+  /* Otherwise, return unknown */
+  return HB_SCRIPT_UNKNOWN;
+}
+
+hb_tag_t
+hb_script_to_iso15924_tag (hb_script_t script)
+{
+  if (likely ((unsigned int) script < ARRAY_LENGTH (script_to_iso15924_tag)))
+    return script_to_iso15924_tag[script];
+
+  /* if script is of the right shape (one capital letter followed by three small letters),
+   * return as is. */
+  if (((uint32_t) script & 0xE0E0E0E0) == 0x40606060)
+    return (hb_tag_t) script;
+
+  /* Otherwise, we don't know what that is */
+  return script_to_iso15924_tag[HB_SCRIPT_UNKNOWN];
+}
+
 
 #define LTR HB_DIRECTION_LTR
 #define RTL HB_DIRECTION_RTL

src/hb-unicode.h

@@ -290,7 +290,13 @@ hb_unicode_get_eastasian_width (hb_unicode_funcs_t *ufuncs,
 				hb_codepoint_t unicode);
 
 
-/* Misc functions */
+/* Script functions */
+
+hb_script_t
+hb_script_from_iso15924_tag (hb_tag_t tag);
+
+hb_tag_t
+hb_script_to_iso15924_tag (hb_script_t script);
 
 hb_direction_t
 hb_script_get_horizontal_direction (hb_script_t script);

src/hb-view.c

@@ -55,6 +55,7 @@ static const char *text = NULL;
 static const char *font_file = NULL;
 static const char *out_file = "/dev/stdout";
 static const char *language = NULL;
+static const char *script = NULL;
 
 /* Ugh, global vars.  Ugly, but does the job */
 static int width = 0;
@@ -97,6 +98,7 @@ parse_opts (int argc, char **argv)
 	{"foreground", 1, 0, 'F'},
 	{"background", 1, 0, 'B'},
 	{"language", 1, 0, 'L'},
+	{"script", 1, 0, 'S'},
 	{"output", 1, 0, 'o'},
 	{0, 0, 0, 0}
       };
@@ -143,6 +145,9 @@ parse_opts (int argc, char **argv)
 	case 'L':
 	  language = optarg;
 	  break;
+	case 'S':
+	  script = optarg;
+	  break;
 	case 'o':
 	  out_file = optarg;
 	  break;
@@ -184,7 +189,10 @@ _hb_cr_text_glyphs (cairo_t *cr,
   hb_buffer_set_unicode_funcs (hb_buffer, hb_glib_get_unicode_funcs ());
 
   hb_buffer_add_utf8 (hb_buffer, text, len, 0, len);
-  hb_buffer_set_script (hb_buffer, HB_SCRIPT_INVALID);
+  if (script)
+    hb_buffer_set_script (hb_buffer, hb_script_from_iso15924_tag (hb_tag_from_string (script)));
+  else
+    hb_buffer_set_script (hb_buffer, HB_SCRIPT_INVALID);
   hb_buffer_set_direction (hb_buffer, HB_DIRECTION_INVALID);
   hb_buffer_set_language (hb_buffer, hb_language_from_string (language));