michael@0: /*
michael@0:  * Copyright © 2009,2010  Red Hat, Inc.
michael@0:  * Copyright © 2011,2012  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-mutex-private.hh"
michael@0: #include "hb-object-private.hh"
michael@0: 
michael@0: #include <locale.h>
michael@0: 
michael@0: 
michael@0: /* hb_options_t */
michael@0: 
michael@0: hb_options_union_t _hb_options;
michael@0: 
michael@0: void
michael@0: _hb_options_init (void)
michael@0: {
michael@0:   hb_options_union_t u;
michael@0:   u.i = 0;
michael@0:   u.opts.initialized = 1;
michael@0: 
michael@0:   char *c = getenv ("HB_OPTIONS");
michael@0:   u.opts.uniscribe_bug_compatible = c && strstr (c, "uniscribe-bug-compatible");
michael@0: 
michael@0:   /* This is idempotent and threadsafe. */
michael@0:   _hb_options = u;
michael@0: }
michael@0: 
michael@0: 
michael@0: /* hb_tag_t */
michael@0: 
michael@0: /**
michael@0:  * hb_tag_from_string:
michael@0:  * @str: (array length=len): 
michael@0:  * @len: 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: hb_tag_t
michael@0: hb_tag_from_string (const char *str, int len)
michael@0: {
michael@0:   char tag[4];
michael@0:   unsigned int i;
michael@0: 
michael@0:   if (!str || !len || !*str)
michael@0:     return HB_TAG_NONE;
michael@0: 
michael@0:   if (len < 0 || len > 4)
michael@0:     len = 4;
michael@0:   for (i = 0; i < (unsigned) len && str[i]; i++)
michael@0:     tag[i] = str[i];
michael@0:   for (; i < 4; i++)
michael@0:     tag[i] = ' ';
michael@0: 
michael@0:   return HB_TAG_CHAR4 (tag);
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * hb_tag_to_string:
michael@0:  * @tag: 
michael@0:  * @buf: (array fixed-size=4): 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: void
michael@0: hb_tag_to_string (hb_tag_t tag, char *buf)
michael@0: {
michael@0:   buf[0] = (char) (uint8_t) (tag >> 24);
michael@0:   buf[1] = (char) (uint8_t) (tag >> 16);
michael@0:   buf[2] = (char) (uint8_t) (tag >>  8);
michael@0:   buf[3] = (char) (uint8_t) (tag >>  0);
michael@0: }
michael@0: 
michael@0: 
michael@0: /* hb_direction_t */
michael@0: 
michael@0: const char direction_strings[][4] = {
michael@0:   "ltr",
michael@0:   "rtl",
michael@0:   "ttb",
michael@0:   "btt"
michael@0: };
michael@0: 
michael@0: /**
michael@0:  * hb_direction_from_string:
michael@0:  * @str: (array length=len): 
michael@0:  * @len: 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: hb_direction_t
michael@0: hb_direction_from_string (const char *str, int len)
michael@0: {
michael@0:   if (unlikely (!str || !len || !*str))
michael@0:     return HB_DIRECTION_INVALID;
michael@0: 
michael@0:   /* Lets match loosely: just match the first letter, such that
michael@0:    * all of "ltr", "left-to-right", etc work!
michael@0:    */
michael@0:   char c = TOLOWER (str[0]);
michael@0:   for (unsigned int i = 0; i < ARRAY_LENGTH (direction_strings); i++)
michael@0:     if (c == direction_strings[i][0])
michael@0:       return (hb_direction_t) (HB_DIRECTION_LTR + i);
michael@0: 
michael@0:   return HB_DIRECTION_INVALID;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * hb_direction_to_string:
michael@0:  * @direction: 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: (transfer none): 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: const char *
michael@0: hb_direction_to_string (hb_direction_t direction)
michael@0: {
michael@0:   if (likely ((unsigned int) (direction - HB_DIRECTION_LTR)
michael@0: 	      < ARRAY_LENGTH (direction_strings)))
michael@0:     return direction_strings[direction - HB_DIRECTION_LTR];
michael@0: 
michael@0:   return "invalid";
michael@0: }
michael@0: 
michael@0: 
michael@0: /* hb_language_t */
michael@0: 
michael@0: struct hb_language_impl_t {
michael@0:   const char s[1];
michael@0: };
michael@0: 
michael@0: static const char canon_map[256] = {
michael@0:    0,   0,   0,   0,   0,   0,   0,   0,    0,   0,   0,   0,   0,   0,   0,   0,
michael@0:    0,   0,   0,   0,   0,   0,   0,   0,    0,   0,   0,   0,   0,   0,   0,   0,
michael@0:    0,   0,   0,   0,   0,   0,   0,   0,    0,   0,   0,   0,   0,  '-',  0,   0,
michael@0:   '0', '1', '2', '3', '4', '5', '6', '7',  '8', '9',  0,   0,   0,   0,   0,   0,
michael@0:   '-', 'a', 'b', 'c', 'd', 'e', 'f', 'g',  'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o',
michael@0:   'p', 'q', 'r', 's', 't', 'u', 'v', 'w',  'x', 'y', 'z',  0,   0,   0,   0,  '-',
michael@0:    0,  'a', 'b', 'c', 'd', 'e', 'f', 'g',  'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o',
michael@0:   'p', 'q', 'r', 's', 't', 'u', 'v', 'w',  'x', 'y', 'z',  0,   0,   0,   0,   0
michael@0: };
michael@0: 
michael@0: static hb_bool_t
michael@0: lang_equal (hb_language_t  v1,
michael@0: 	    const void    *v2)
michael@0: {
michael@0:   const unsigned char *p1 = (const unsigned char *) v1;
michael@0:   const unsigned char *p2 = (const unsigned char *) v2;
michael@0: 
michael@0:   while (*p1 && *p1 == canon_map[*p2])
michael@0:     p1++, p2++;
michael@0: 
michael@0:   return *p1 == canon_map[*p2];
michael@0: }
michael@0: 
michael@0: #if 0
michael@0: static unsigned int
michael@0: lang_hash (const void *key)
michael@0: {
michael@0:   const unsigned char *p = key;
michael@0:   unsigned int h = 0;
michael@0:   while (canon_map[*p])
michael@0:     {
michael@0:       h = (h << 5) - h + canon_map[*p];
michael@0:       p++;
michael@0:     }
michael@0: 
michael@0:   return h;
michael@0: }
michael@0: #endif
michael@0: 
michael@0: 
michael@0: struct hb_language_item_t {
michael@0: 
michael@0:   struct hb_language_item_t *next;
michael@0:   hb_language_t lang;
michael@0: 
michael@0:   inline bool operator == (const char *s) const {
michael@0:     return lang_equal (lang, s);
michael@0:   }
michael@0: 
michael@0:   inline hb_language_item_t & operator = (const char *s) {
michael@0:     lang = (hb_language_t) strdup (s);
michael@0:     for (unsigned char *p = (unsigned char *) lang; *p; p++)
michael@0:       *p = canon_map[*p];
michael@0: 
michael@0:     return *this;
michael@0:   }
michael@0: 
michael@0:   void finish (void) { free ((void *) lang); }
michael@0: };
michael@0: 
michael@0: 
michael@0: /* Thread-safe lock-free language list */
michael@0: 
michael@0: static hb_language_item_t *langs;
michael@0: 
michael@0: static inline
michael@0: void free_langs (void)
michael@0: {
michael@0:   while (langs) {
michael@0:     hb_language_item_t *next = langs->next;
michael@0:     langs->finish ();
michael@0:     free (langs);
michael@0:     langs = next;
michael@0:   }
michael@0: }
michael@0: 
michael@0: static hb_language_item_t *
michael@0: lang_find_or_insert (const char *key)
michael@0: {
michael@0: retry:
michael@0:   hb_language_item_t *first_lang = (hb_language_item_t *) hb_atomic_ptr_get (&langs);
michael@0: 
michael@0:   for (hb_language_item_t *lang = first_lang; lang; lang = lang->next)
michael@0:     if (*lang == key)
michael@0:       return lang;
michael@0: 
michael@0:   /* Not found; allocate one. */
michael@0:   hb_language_item_t *lang = (hb_language_item_t *) calloc (1, sizeof (hb_language_item_t));
michael@0:   if (unlikely (!lang))
michael@0:     return NULL;
michael@0:   lang->next = first_lang;
michael@0:   *lang = key;
michael@0: 
michael@0:   if (!hb_atomic_ptr_cmpexch (&langs, first_lang, lang)) {
michael@0:     free (lang);
michael@0:     goto retry;
michael@0:   }
michael@0: 
michael@0: #ifdef HAVE_ATEXIT
michael@0:   if (!first_lang)
michael@0:     atexit (free_langs); /* First person registers atexit() callback. */
michael@0: #endif
michael@0: 
michael@0:   return lang;
michael@0: }
michael@0: 
michael@0: 
michael@0: /**
michael@0:  * hb_language_from_string:
michael@0:  * @str: (array length=len): 
michael@0:  * @len: 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: hb_language_t
michael@0: hb_language_from_string (const char *str, int len)
michael@0: {
michael@0:   char strbuf[64];
michael@0: 
michael@0:   if (!str || !len || !*str)
michael@0:     return HB_LANGUAGE_INVALID;
michael@0: 
michael@0:   if (len >= 0)
michael@0:   {
michael@0:     len = MIN (len, (int) sizeof (strbuf) - 1);
michael@0:     str = (char *) memcpy (strbuf, str, len);
michael@0:     strbuf[len] = '\0';
michael@0:   }
michael@0: 
michael@0:   hb_language_item_t *item = lang_find_or_insert (str);
michael@0: 
michael@0:   return likely (item) ? item->lang : HB_LANGUAGE_INVALID;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * hb_language_to_string:
michael@0:  * @language: 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: (transfer none): 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: const char *
michael@0: hb_language_to_string (hb_language_t language)
michael@0: {
michael@0:   /* This is actually NULL-safe! */
michael@0:   return language->s;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * hb_language_get_default:
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: hb_language_t
michael@0: hb_language_get_default (void)
michael@0: {
michael@0:   static hb_language_t default_language = HB_LANGUAGE_INVALID;
michael@0: 
michael@0:   hb_language_t language = (hb_language_t) hb_atomic_ptr_get (&default_language);
michael@0:   if (unlikely (language == HB_LANGUAGE_INVALID)) {
michael@0:     language = hb_language_from_string (setlocale (LC_CTYPE, NULL), -1);
michael@0:     hb_atomic_ptr_cmpexch (&default_language, HB_LANGUAGE_INVALID, language);
michael@0:   }
michael@0: 
michael@0:   return default_language;
michael@0: }
michael@0: 
michael@0: 
michael@0: /* hb_script_t */
michael@0: 
michael@0: /**
michael@0:  * hb_script_from_iso15924_tag:
michael@0:  * @tag: 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: hb_script_t
michael@0: hb_script_from_iso15924_tag (hb_tag_t tag)
michael@0: {
michael@0:   if (unlikely (tag == HB_TAG_NONE))
michael@0:     return HB_SCRIPT_INVALID;
michael@0: 
michael@0:   /* Be lenient, adjust case (one capital letter followed by three small letters) */
michael@0:   tag = (tag & 0xDFDFDFDF) | 0x00202020;
michael@0: 
michael@0:   switch (tag) {
michael@0: 
michael@0:     /* These graduated from the 'Q' private-area codes, but
michael@0:      * the old code is still aliased by Unicode, and the Qaai
michael@0:      * one in use by ICU. */
michael@0:     case HB_TAG('Q','a','a','i'): return HB_SCRIPT_INHERITED;
michael@0:     case HB_TAG('Q','a','a','c'): return HB_SCRIPT_COPTIC;
michael@0: 
michael@0:     /* Script variants from http://unicode.org/iso15924/ */
michael@0:     case HB_TAG('C','y','r','s'): return HB_SCRIPT_CYRILLIC;
michael@0:     case HB_TAG('L','a','t','f'): return HB_SCRIPT_LATIN;
michael@0:     case HB_TAG('L','a','t','g'): return HB_SCRIPT_LATIN;
michael@0:     case HB_TAG('S','y','r','e'): return HB_SCRIPT_SYRIAC;
michael@0:     case HB_TAG('S','y','r','j'): return HB_SCRIPT_SYRIAC;
michael@0:     case HB_TAG('S','y','r','n'): return HB_SCRIPT_SYRIAC;
michael@0:   }
michael@0: 
michael@0:   /* If it looks right, just use the tag as a script */
michael@0:   if (((uint32_t) tag & 0xE0E0E0E0) == 0x40606060)
michael@0:     return (hb_script_t) tag;
michael@0: 
michael@0:   /* Otherwise, return unknown */
michael@0:   return HB_SCRIPT_UNKNOWN;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * hb_script_from_string:
michael@0:  * @s: (array length=len): 
michael@0:  * @len: 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: hb_script_t
michael@0: hb_script_from_string (const char *s, int len)
michael@0: {
michael@0:   return hb_script_from_iso15924_tag (hb_tag_from_string (s, len));
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * hb_script_to_iso15924_tag:
michael@0:  * @script: 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: hb_tag_t
michael@0: hb_script_to_iso15924_tag (hb_script_t script)
michael@0: {
michael@0:   return (hb_tag_t) script;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * hb_script_get_horizontal_direction:
michael@0:  * @script: 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: hb_direction_t
michael@0: hb_script_get_horizontal_direction (hb_script_t script)
michael@0: {
michael@0:   /* http://goo.gl/x9ilM */
michael@0:   switch ((hb_tag_t) script)
michael@0:   {
michael@0:     /* Unicode-1.1 additions */
michael@0:     case HB_SCRIPT_ARABIC:
michael@0:     case HB_SCRIPT_HEBREW:
michael@0: 
michael@0:     /* Unicode-3.0 additions */
michael@0:     case HB_SCRIPT_SYRIAC:
michael@0:     case HB_SCRIPT_THAANA:
michael@0: 
michael@0:     /* Unicode-4.0 additions */
michael@0:     case HB_SCRIPT_CYPRIOT:
michael@0: 
michael@0:     /* Unicode-4.1 additions */
michael@0:     case HB_SCRIPT_KHAROSHTHI:
michael@0: 
michael@0:     /* Unicode-5.0 additions */
michael@0:     case HB_SCRIPT_PHOENICIAN:
michael@0:     case HB_SCRIPT_NKO:
michael@0: 
michael@0:     /* Unicode-5.1 additions */
michael@0:     case HB_SCRIPT_LYDIAN:
michael@0: 
michael@0:     /* Unicode-5.2 additions */
michael@0:     case HB_SCRIPT_AVESTAN:
michael@0:     case HB_SCRIPT_IMPERIAL_ARAMAIC:
michael@0:     case HB_SCRIPT_INSCRIPTIONAL_PAHLAVI:
michael@0:     case HB_SCRIPT_INSCRIPTIONAL_PARTHIAN:
michael@0:     case HB_SCRIPT_OLD_SOUTH_ARABIAN:
michael@0:     case HB_SCRIPT_OLD_TURKIC:
michael@0:     case HB_SCRIPT_SAMARITAN:
michael@0: 
michael@0:     /* Unicode-6.0 additions */
michael@0:     case HB_SCRIPT_MANDAIC:
michael@0: 
michael@0:     /* Unicode-6.1 additions */
michael@0:     case HB_SCRIPT_MEROITIC_CURSIVE:
michael@0:     case HB_SCRIPT_MEROITIC_HIEROGLYPHS:
michael@0: 
michael@0:       return HB_DIRECTION_RTL;
michael@0:   }
michael@0: 
michael@0:   return HB_DIRECTION_LTR;
michael@0: }
michael@0: 
michael@0: 
michael@0: /* hb_user_data_array_t */
michael@0: 
michael@0: bool
michael@0: hb_user_data_array_t::set (hb_user_data_key_t *key,
michael@0: 			   void *              data,
michael@0: 			   hb_destroy_func_t   destroy,
michael@0: 			   hb_bool_t           replace)
michael@0: {
michael@0:   if (!key)
michael@0:     return false;
michael@0: 
michael@0:   if (replace) {
michael@0:     if (!data && !destroy) {
michael@0:       items.remove (key, lock);
michael@0:       return true;
michael@0:     }
michael@0:   }
michael@0:   hb_user_data_item_t item = {key, data, destroy};
michael@0:   bool ret = !!items.replace_or_insert (item, lock, replace);
michael@0: 
michael@0:   return ret;
michael@0: }
michael@0: 
michael@0: void *
michael@0: hb_user_data_array_t::get (hb_user_data_key_t *key)
michael@0: {
michael@0:   hb_user_data_item_t item = {NULL };
michael@0: 
michael@0:   return items.find (key, &item, lock) ? item.data : NULL;
michael@0: }
michael@0: 
michael@0: 
michael@0: /* hb_version */
michael@0: 
michael@0: /**
michael@0:  * hb_version:
michael@0:  * @major: (out): Library major version component.
michael@0:  * @minor: (out): Library minor version component.
michael@0:  * @micro: (out): Library micro version component.
michael@0:  *
michael@0:  * Returns library version as three integer components.
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: void
michael@0: hb_version (unsigned int *major,
michael@0: 	    unsigned int *minor,
michael@0: 	    unsigned int *micro)
michael@0: {
michael@0:   *major = HB_VERSION_MAJOR;
michael@0:   *minor = HB_VERSION_MINOR;
michael@0:   *micro = HB_VERSION_MICRO;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * hb_version_string:
michael@0:  *
michael@0:  * Returns library version as a string with three components.
michael@0:  *
michael@0:  * Return value: library version string.
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: const char *
michael@0: hb_version_string (void)
michael@0: {
michael@0:   return HB_VERSION_STRING;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * hb_version_check:
michael@0:  * @major: 
michael@0:  * @minor: 
michael@0:  * @micro: 
michael@0:  *
michael@0:  * 
michael@0:  *
michael@0:  * Return value: 
michael@0:  *
michael@0:  * Since: 1.0
michael@0:  **/
michael@0: hb_bool_t
michael@0: hb_version_check (unsigned int major,
michael@0: 		  unsigned int minor,
michael@0: 		  unsigned int micro)
michael@0: {
michael@0:   return HB_VERSION_CHECK (major, minor, micro);
michael@0: }