michael@0: /*
michael@0:  * Copyright © 2009  Red Hat, Inc.
michael@0:  * Copyright © 2011  Google, Inc.
michael@0:  *
michael@0:  *  This is part of HarfBuzz, a text shaping library.
michael@0:  *
michael@0:  * Permission is hereby granted, without written agreement and without
michael@0:  * license or royalty fees, to use, copy, modify, and distribute this
michael@0:  * software and its documentation for any purpose, provided that the
michael@0:  * above copyright notice and the following two paragraphs appear in
michael@0:  * all copies of this software.
michael@0:  *
michael@0:  * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
michael@0:  * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
michael@0:  * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
michael@0:  * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
michael@0:  * DAMAGE.
michael@0:  *
michael@0:  * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
michael@0:  * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
michael@0:  * FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
michael@0:  * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
michael@0:  * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
michael@0:  *
michael@0:  * Red Hat Author(s): Behdad Esfahbod
michael@0:  * Google Author(s): Behdad Esfahbod
michael@0:  */
michael@0: 
michael@0: #include "hb-private.hh"
michael@0: 
michael@0: #include "hb-glib.h"
michael@0: 
michael@0: #include "hb-unicode-private.hh"
michael@0: 
michael@0: 
michael@0: #if !GLIB_CHECK_VERSION(2,29,14)
michael@0: static const hb_script_t
michael@0: glib_script_to_script[] =
michael@0: {
michael@0:   HB_SCRIPT_COMMON,
michael@0:   HB_SCRIPT_INHERITED,
michael@0:   HB_SCRIPT_ARABIC,
michael@0:   HB_SCRIPT_ARMENIAN,
michael@0:   HB_SCRIPT_BENGALI,
michael@0:   HB_SCRIPT_BOPOMOFO,
michael@0:   HB_SCRIPT_CHEROKEE,
michael@0:   HB_SCRIPT_COPTIC,
michael@0:   HB_SCRIPT_CYRILLIC,
michael@0:   HB_SCRIPT_DESERET,
michael@0:   HB_SCRIPT_DEVANAGARI,
michael@0:   HB_SCRIPT_ETHIOPIC,
michael@0:   HB_SCRIPT_GEORGIAN,
michael@0:   HB_SCRIPT_GOTHIC,
michael@0:   HB_SCRIPT_GREEK,
michael@0:   HB_SCRIPT_GUJARATI,
michael@0:   HB_SCRIPT_GURMUKHI,
michael@0:   HB_SCRIPT_HAN,
michael@0:   HB_SCRIPT_HANGUL,
michael@0:   HB_SCRIPT_HEBREW,
michael@0:   HB_SCRIPT_HIRAGANA,
michael@0:   HB_SCRIPT_KANNADA,
michael@0:   HB_SCRIPT_KATAKANA,
michael@0:   HB_SCRIPT_KHMER,
michael@0:   HB_SCRIPT_LAO,
michael@0:   HB_SCRIPT_LATIN,
michael@0:   HB_SCRIPT_MALAYALAM,
michael@0:   HB_SCRIPT_MONGOLIAN,
michael@0:   HB_SCRIPT_MYANMAR,
michael@0:   HB_SCRIPT_OGHAM,
michael@0:   HB_SCRIPT_OLD_ITALIC,
michael@0:   HB_SCRIPT_ORIYA,
michael@0:   HB_SCRIPT_RUNIC,
michael@0:   HB_SCRIPT_SINHALA,
michael@0:   HB_SCRIPT_SYRIAC,
michael@0:   HB_SCRIPT_TAMIL,
michael@0:   HB_SCRIPT_TELUGU,
michael@0:   HB_SCRIPT_THAANA,
michael@0:   HB_SCRIPT_THAI,
michael@0:   HB_SCRIPT_TIBETAN,
michael@0:   HB_SCRIPT_CANADIAN_SYLLABICS,
michael@0:   HB_SCRIPT_YI,
michael@0:   HB_SCRIPT_TAGALOG,
michael@0:   HB_SCRIPT_HANUNOO,
michael@0:   HB_SCRIPT_BUHID,
michael@0:   HB_SCRIPT_TAGBANWA,
michael@0: 
michael@0:   /* Unicode-4.0 additions */
michael@0:   HB_SCRIPT_BRAILLE,
michael@0:   HB_SCRIPT_CYPRIOT,
michael@0:   HB_SCRIPT_LIMBU,
michael@0:   HB_SCRIPT_OSMANYA,
michael@0:   HB_SCRIPT_SHAVIAN,
michael@0:   HB_SCRIPT_LINEAR_B,
michael@0:   HB_SCRIPT_TAI_LE,
michael@0:   HB_SCRIPT_UGARITIC,
michael@0: 
michael@0:   /* Unicode-4.1 additions */
michael@0:   HB_SCRIPT_NEW_TAI_LUE,
michael@0:   HB_SCRIPT_BUGINESE,
michael@0:   HB_SCRIPT_GLAGOLITIC,
michael@0:   HB_SCRIPT_TIFINAGH,
michael@0:   HB_SCRIPT_SYLOTI_NAGRI,
michael@0:   HB_SCRIPT_OLD_PERSIAN,
michael@0:   HB_SCRIPT_KHAROSHTHI,
michael@0: 
michael@0:   /* Unicode-5.0 additions */
michael@0:   HB_SCRIPT_UNKNOWN,
michael@0:   HB_SCRIPT_BALINESE,
michael@0:   HB_SCRIPT_CUNEIFORM,
michael@0:   HB_SCRIPT_PHOENICIAN,
michael@0:   HB_SCRIPT_PHAGS_PA,
michael@0:   HB_SCRIPT_NKO,
michael@0: 
michael@0:   /* Unicode-5.1 additions */
michael@0:   HB_SCRIPT_KAYAH_LI,
michael@0:   HB_SCRIPT_LEPCHA,
michael@0:   HB_SCRIPT_REJANG,
michael@0:   HB_SCRIPT_SUNDANESE,
michael@0:   HB_SCRIPT_SAURASHTRA,
michael@0:   HB_SCRIPT_CHAM,
michael@0:   HB_SCRIPT_OL_CHIKI,
michael@0:   HB_SCRIPT_VAI,
michael@0:   HB_SCRIPT_CARIAN,
michael@0:   HB_SCRIPT_LYCIAN,
michael@0:   HB_SCRIPT_LYDIAN,
michael@0: 
michael@0:   /* Unicode-5.2 additions */
michael@0:   HB_SCRIPT_AVESTAN,
michael@0:   HB_SCRIPT_BAMUM,
michael@0:   HB_SCRIPT_EGYPTIAN_HIEROGLYPHS,
michael@0:   HB_SCRIPT_IMPERIAL_ARAMAIC,
michael@0:   HB_SCRIPT_INSCRIPTIONAL_PAHLAVI,
michael@0:   HB_SCRIPT_INSCRIPTIONAL_PARTHIAN,
michael@0:   HB_SCRIPT_JAVANESE,
michael@0:   HB_SCRIPT_KAITHI,
michael@0:   HB_SCRIPT_TAI_THAM,
michael@0:   HB_SCRIPT_LISU,
michael@0:   HB_SCRIPT_MEETEI_MAYEK,
michael@0:   HB_SCRIPT_OLD_SOUTH_ARABIAN,
michael@0:   HB_SCRIPT_OLD_TURKIC,
michael@0:   HB_SCRIPT_SAMARITAN,
michael@0:   HB_SCRIPT_TAI_VIET,
michael@0: 
michael@0:   /* Unicode-6.0 additions */
michael@0:   HB_SCRIPT_BATAK,
michael@0:   HB_SCRIPT_BRAHMI,
michael@0:   HB_SCRIPT_MANDAIC,
michael@0: 
michael@0:   /* Unicode-6.1 additions */
michael@0:   HB_SCRIPT_CHAKMA,
michael@0:   HB_SCRIPT_MEROITIC_CURSIVE,
michael@0:   HB_SCRIPT_MEROITIC_HIEROGLYPHS,
michael@0:   HB_SCRIPT_MIAO,
michael@0:   HB_SCRIPT_SHARADA,
michael@0:   HB_SCRIPT_SORA_SOMPENG,
michael@0:   HB_SCRIPT_TAKRI
michael@0: };
michael@0: #endif
michael@0: 
michael@0: hb_script_t
michael@0: hb_glib_script_to_script (GUnicodeScript script)
michael@0: {
michael@0: #if GLIB_CHECK_VERSION(2,29,14)
michael@0:   return (hb_script_t) g_unicode_script_to_iso15924 (script);
michael@0: #else
michael@0:   if (likely ((unsigned int) script < ARRAY_LENGTH (glib_script_to_script)))
michael@0:     return glib_script_to_script[script];
michael@0: 
michael@0:   if (unlikely (script == G_UNICODE_SCRIPT_INVALID_CODE))
michael@0:     return HB_SCRIPT_INVALID;
michael@0: 
michael@0:   return HB_SCRIPT_UNKNOWN;
michael@0: #endif
michael@0: }
michael@0: 
michael@0: GUnicodeScript
michael@0: hb_glib_script_from_script (hb_script_t script)
michael@0: {
michael@0: #if GLIB_CHECK_VERSION(2,29,14)
michael@0:   return g_unicode_script_from_iso15924 (script);
michael@0: #else
michael@0:   unsigned int count = ARRAY_LENGTH (glib_script_to_script);
michael@0:   for (unsigned int i = 0; i < count; i++)
michael@0:     if (glib_script_to_script[i] == script)
michael@0:       return (GUnicodeScript) i;
michael@0: 
michael@0:   if (unlikely (script == HB_SCRIPT_INVALID))
michael@0:     return G_UNICODE_SCRIPT_INVALID_CODE;
michael@0: 
michael@0:   return G_UNICODE_SCRIPT_UNKNOWN;
michael@0: #endif
michael@0: }
michael@0: 
michael@0: 
michael@0: static hb_unicode_combining_class_t
michael@0: hb_glib_unicode_combining_class (hb_unicode_funcs_t *ufuncs HB_UNUSED,
michael@0: 				 hb_codepoint_t      unicode,
michael@0: 				 void               *user_data HB_UNUSED)
michael@0: 
michael@0: {
michael@0:   return (hb_unicode_combining_class_t) g_unichar_combining_class (unicode);
michael@0: }
michael@0: 
michael@0: static unsigned int
michael@0: hb_glib_unicode_eastasian_width (hb_unicode_funcs_t *ufuncs HB_UNUSED,
michael@0: 				 hb_codepoint_t      unicode,
michael@0: 				 void               *user_data HB_UNUSED)
michael@0: {
michael@0:   return g_unichar_iswide (unicode) ? 2 : 1;
michael@0: }
michael@0: 
michael@0: static hb_unicode_general_category_t
michael@0: hb_glib_unicode_general_category (hb_unicode_funcs_t *ufuncs HB_UNUSED,
michael@0: 				  hb_codepoint_t      unicode,
michael@0: 				  void               *user_data HB_UNUSED)
michael@0: 
michael@0: {
michael@0:   /* hb_unicode_general_category_t and GUnicodeType are identical */
michael@0:   return (hb_unicode_general_category_t) g_unichar_type (unicode);
michael@0: }
michael@0: 
michael@0: static hb_codepoint_t
michael@0: hb_glib_unicode_mirroring (hb_unicode_funcs_t *ufuncs HB_UNUSED,
michael@0: 			   hb_codepoint_t      unicode,
michael@0: 			   void               *user_data HB_UNUSED)
michael@0: {
michael@0:   g_unichar_get_mirror_char (unicode, &unicode);
michael@0:   return unicode;
michael@0: }
michael@0: 
michael@0: static hb_script_t
michael@0: hb_glib_unicode_script (hb_unicode_funcs_t *ufuncs HB_UNUSED,
michael@0: 			hb_codepoint_t      unicode,
michael@0: 			void               *user_data HB_UNUSED)
michael@0: {
michael@0:   return hb_glib_script_to_script (g_unichar_get_script (unicode));
michael@0: }
michael@0: 
michael@0: static hb_bool_t
michael@0: hb_glib_unicode_compose (hb_unicode_funcs_t *ufuncs HB_UNUSED,
michael@0: 			 hb_codepoint_t      a,
michael@0: 			 hb_codepoint_t      b,
michael@0: 			 hb_codepoint_t     *ab,
michael@0: 			 void               *user_data HB_UNUSED)
michael@0: {
michael@0: #if GLIB_CHECK_VERSION(2,29,12)
michael@0:   return g_unichar_compose (a, b, ab);
michael@0: #endif
michael@0: 
michael@0:   /* We don't ifdef-out the fallback code such that compiler always
michael@0:    * sees it and makes sure it's compilable. */
michael@0: 
michael@0:   gchar utf8[12];
michael@0:   gchar *normalized;
michael@0:   int len;
michael@0:   hb_bool_t ret;
michael@0: 
michael@0:   len = g_unichar_to_utf8 (a, utf8);
michael@0:   len += g_unichar_to_utf8 (b, utf8 + len);
michael@0:   normalized = g_utf8_normalize (utf8, len, G_NORMALIZE_NFC);
michael@0:   len = g_utf8_strlen (normalized, -1);
michael@0:   if (unlikely (!len))
michael@0:     return false;
michael@0: 
michael@0:   if (len == 1) {
michael@0:     *ab = g_utf8_get_char (normalized);
michael@0:     ret = true;
michael@0:   } else {
michael@0:     ret = false;
michael@0:   }
michael@0: 
michael@0:   g_free (normalized);
michael@0:   return ret;
michael@0: }
michael@0: 
michael@0: static hb_bool_t
michael@0: hb_glib_unicode_decompose (hb_unicode_funcs_t *ufuncs HB_UNUSED,
michael@0: 			   hb_codepoint_t      ab,
michael@0: 			   hb_codepoint_t     *a,
michael@0: 			   hb_codepoint_t     *b,
michael@0: 			   void               *user_data HB_UNUSED)
michael@0: {
michael@0: #if GLIB_CHECK_VERSION(2,29,12)
michael@0:   return g_unichar_decompose (ab, a, b);
michael@0: #endif
michael@0: 
michael@0:   /* We don't ifdef-out the fallback code such that compiler always
michael@0:    * sees it and makes sure it's compilable. */
michael@0: 
michael@0:   gchar utf8[6];
michael@0:   gchar *normalized;
michael@0:   int len;
michael@0:   hb_bool_t ret;
michael@0: 
michael@0:   len = g_unichar_to_utf8 (ab, utf8);
michael@0:   normalized = g_utf8_normalize (utf8, len, G_NORMALIZE_NFD);
michael@0:   len = g_utf8_strlen (normalized, -1);
michael@0:   if (unlikely (!len))
michael@0:     return false;
michael@0: 
michael@0:   if (len == 1) {
michael@0:     *a = g_utf8_get_char (normalized);
michael@0:     *b = 0;
michael@0:     ret = *a != ab;
michael@0:   } else if (len == 2) {
michael@0:     *a = g_utf8_get_char (normalized);
michael@0:     *b = g_utf8_get_char (g_utf8_next_char (normalized));
michael@0:     /* Here's the ugly part: if ab decomposes to a single character and
michael@0:      * that character decomposes again, we have to detect that and undo
michael@0:      * the second part :-(. */
michael@0:     gchar *recomposed = g_utf8_normalize (normalized, -1, G_NORMALIZE_NFC);
michael@0:     hb_codepoint_t c = g_utf8_get_char (recomposed);
michael@0:     if (c != ab && c != *a) {
michael@0:       *a = c;
michael@0:       *b = 0;
michael@0:     }
michael@0:     g_free (recomposed);
michael@0:     ret = true;
michael@0:   } else {
michael@0:     /* If decomposed to more than two characters, take the last one,
michael@0:      * and recompose the rest to get the first component. */
michael@0:     gchar *end = g_utf8_offset_to_pointer (normalized, len - 1);
michael@0:     gchar *recomposed;
michael@0:     *b = g_utf8_get_char (end);
michael@0:     recomposed = g_utf8_normalize (normalized, end - normalized, G_NORMALIZE_NFC);
michael@0:     /* We expect that recomposed has exactly one character now. */
michael@0:     *a = g_utf8_get_char (recomposed);
michael@0:     g_free (recomposed);
michael@0:     ret = true;
michael@0:   }
michael@0: 
michael@0:   g_free (normalized);
michael@0:   return ret;
michael@0: }
michael@0: 
michael@0: static unsigned int
michael@0: hb_glib_unicode_decompose_compatibility (hb_unicode_funcs_t *ufuncs HB_UNUSED,
michael@0: 					 hb_codepoint_t      u,
michael@0: 					 hb_codepoint_t     *decomposed,
michael@0: 					 void               *user_data HB_UNUSED)
michael@0: {
michael@0: #if GLIB_CHECK_VERSION(2,29,12)
michael@0:   return g_unichar_fully_decompose (u, TRUE, decomposed, HB_UNICODE_MAX_DECOMPOSITION_LEN);
michael@0: #endif
michael@0: 
michael@0:   /* If the user doesn't have GLib >= 2.29.12 we have to perform
michael@0:    * a round trip to UTF-8 and the associated memory management dance. */
michael@0:   gchar utf8[6];
michael@0:   gchar *utf8_decomposed, *c;
michael@0:   gsize utf8_len, utf8_decomposed_len, i;
michael@0: 
michael@0:   /* Convert @u to UTF-8 and normalise it in NFKD mode. This performs the compatibility decomposition. */
michael@0:   utf8_len = g_unichar_to_utf8 (u, utf8);
michael@0:   utf8_decomposed = g_utf8_normalize (utf8, utf8_len, G_NORMALIZE_NFKD);
michael@0:   utf8_decomposed_len = g_utf8_strlen (utf8_decomposed, -1);
michael@0: 
michael@0:   assert (utf8_decomposed_len <= HB_UNICODE_MAX_DECOMPOSITION_LEN);
michael@0: 
michael@0:   for (i = 0, c = utf8_decomposed; i < utf8_decomposed_len; i++, c = g_utf8_next_char (c))
michael@0:     *decomposed++ = g_utf8_get_char (c);
michael@0: 
michael@0:   g_free (utf8_decomposed);
michael@0: 
michael@0:   return utf8_decomposed_len;
michael@0: }
michael@0: 
michael@0: hb_unicode_funcs_t *
michael@0: hb_glib_get_unicode_funcs (void)
michael@0: {
michael@0:   static const hb_unicode_funcs_t _hb_glib_unicode_funcs = {
michael@0:     HB_OBJECT_HEADER_STATIC,
michael@0: 
michael@0:     NULL, /* parent */
michael@0:     true, /* immutable */
michael@0:     {
michael@0: #define HB_UNICODE_FUNC_IMPLEMENT(name) hb_glib_unicode_##name,
michael@0:       HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS
michael@0: #undef HB_UNICODE_FUNC_IMPLEMENT
michael@0:     }
michael@0:   };
michael@0: 
michael@0:   return const_cast<hb_unicode_funcs_t *> (&_hb_glib_unicode_funcs);
michael@0: }
michael@0: