add basic support for arabic.

Makefile.common

@@ -284,7 +284,8 @@ OBJ += intl/msg_hash_de.o \
        intl/msg_hash_ru.o \
        intl/msg_hash_vn.o \
        intl/msg_hash_chs.o \
-       intl/msg_hash_cht.o
+       intl/msg_hash_cht.o \
+       intl/msg_hash_ar.o
 
 endif
 

gfx/font_driver.c

@@ -470,14 +470,255 @@ static bool font_init_first(
    return false;
 }
 
+#ifdef HAVE_LANGEXTRA
+
+/* ACII:        0xxxxxxx  (c & 0x80) == 0x00
+ * other start: 11xxxxxx  (c & 0xC0) == 0xC0
+ * other cont:  10xxxxxx  (c & 0xC0) == 0x80
+ * Neutral :
+ * 0020 - 002F : 001xxxxx (c & 0xE0) == 0x20
+ * Arabic:
+ * 0600 - 07FF : 11011xxx (c & 0xF8) == 0xD8 (2 bytes)
+ * 0800 - 08FF : 11100000 101000xx  c == 0xE0 && (c1 & 0xAC) == 0xA0 (3 bytes) */
+
+/* clang-format off */
+#define IS_ASCII(p)        ((*(p)&0x80) == 0x00)
+#define IS_MBSTART(p)      ((*(p)&0xC0) == 0xC0)
+#define IS_MBCONT(p)       ((*(p)&0xC0) == 0x80)
+#define IS_DIR_NEUTRAL(p)  ((*(p)&0xE0) == 0x20)
+#define IS_ARABIC0(p)      ((*(p)&0xF8) == 0xD8)
+#define IS_ARABIC1(p)      ((*(p) == 0xE0) && ((*((p) + 1) & 0xAC) == 0xA0))
+#define IS_ARABIC(p)       (IS_ARABIC0(p) || IS_ARABIC1(p))
+#define IS_RTL(p)          IS_ARABIC(p)
+
+/* 0x0620 to 0x064F */
+static const unsigned arabic_shape_map[0x50 - 0x20][0x4] = {
+   { 0 },                              /* 0x0620 */
+   { 0xFE80 },
+   { 0xFE81, 0xFE82 },
+   { 0xFE83, 0xFE84 },
+   { 0xFE85, 0xFE86 },
+   { 0xFE87, 0xFE88 },
+   { 0xFE89, 0xFE8A, 0xFE8B, 0xFE8C },
+   { 0xFE8D, 0xFE8E },
+
+   { 0xFE8F, 0xFE90, 0xFE91, 0xFE92 },
+   { 0xFE93, 0xFE94 },
+   { 0xFE95, 0xFE96, 0xFE97, 0xFE98 },
+   { 0xFE99, 0xFE9A, 0xFE9B, 0xFE9C },
+   { 0xFE9D, 0xFE9E, 0xFE9F, 0xFEA0 },
+   { 0xFEA1, 0xFEA2, 0xFEA3, 0xFEA4 },
+   { 0xFEA5, 0xFEA6, 0xFEA7, 0xFEA8 },
+   { 0xFEA9, 0xFEAA },
+
+   { 0xFEAB, 0xFEAC },                  /* 0x0630 */
+   { 0xFEAD, 0xFEAE },
+   { 0xFEAF, 0xFEB0 },
+   { 0xFEB1, 0xFEB2, 0xFEB3, 0xFEB4 },
+   { 0xFEB5, 0xFEB6, 0xFEB7, 0xFEB8 },
+   { 0xFEB9, 0xFEBA, 0xFEBB, 0xFEBC },
+   { 0xFEBD, 0xFEBE, 0xFEBF, 0xFEC0 },
+   { 0xFEC1, 0xFEC2, 0xFEC3, 0xFEC4 },
+   { 0xFEC5, 0xFEC6, 0xFEC7, 0xFEC8 },
+
+   { 0xFEC9, 0xFECA, 0xFECB, 0xFECC },
+   { 0xFECD, 0xFECE, 0xFECF, 0xFED0 },
+   { 0 },
+   { 0 },
+   { 0 },
+   { 0 },
+   { 0 },
+   { 0 },
+
+   { 0xFED1, 0xFED2, 0xFED3, 0xFED4 }, /* 0x0640 */
+   { 0xFED5, 0xFED6, 0xFED7, 0xFED8 },
+   { 0xFED9, 0xFEDA, 0xFEDB, 0xFEDC },
+   { 0xFEDD, 0xFEDE, 0xFEDF, 0xFEE0 },
+   { 0xFEE1, 0xFEE2, 0xFEE3, 0xFEE4 },
+   { 0xFEE5, 0xFEE6, 0xFEE7, 0xFEE8 },
+   { 0xFEE9, 0xFEEA, 0xFEEB, 0xFEEC },
+   { 0xFEED, 0xFEEE },
+
+   { 0xFEEF, 0xFEF0, 0xFBE8, 0xFBE9 },
+   { 0xFEF1, 0xFEF2, 0xFEF3, 0xFEF4 },
+};
+/* clang-format on */
+
+static INLINE unsigned font_get_replacement(const char* src, const char* start)
+{
+   if ((*src & 0xFC) == 0xD8) /* 0x0600 to 0x06FF */
+   {
+      int           lookup;
+      unsigned      result;
+      bool          prev_connected = false;
+      bool          next_connected = false;
+      unsigned char id             = (src[0] << 6) | (src[1] & 0x3F);
+      const char*   prev1          = src - 2;
+      const char*   prev2          = src - 4;
+
+      if (id < 0x21 || id > 0x4A)
+         return 0;
+
+      if(prev2 < start)
+      {
+         prev2 = NULL;
+         if(prev1 < start)
+            prev1 = NULL;
+      }
+
+      if (prev1 && (*prev1 & 0xFC) == 0xD8)
+      {
+         unsigned char prev1_id = 0;
+
+         if (prev1)
+            prev1_id = (prev1[0] << 6) | (prev1[1] & 0x3F);
+
+         if (prev1_id == 0x44)
+         {
+            unsigned char prev2_id = 0;
+
+            if (prev2)
+               prev2_id = (prev2[0] << 6) | (prev2[1] & 0x3F);
+
+            if (prev2_id > 0x20 || prev2_id < 0x50)
+               prev_connected = !!arabic_shape_map[prev2_id - 0x20][2];
+
+            switch (id)
+            {
+               case 0x22:
+                  return 0xFEF5 + prev_connected;
+               case 0x23:
+                  return 0xFEF7 + prev_connected;
+               case 0x25:
+                  return 0xFEF9 + prev_connected;
+               case 0x27:
+                  return 0xFEFB + prev_connected;
+            }
+         }
+         if (prev1_id > 0x20 || prev1_id < 0x50)
+            prev_connected = !!arabic_shape_map[prev1_id - 0x20][2];
+      }
+
+      if ((src[2] & 0xFC) == 0xD8)
+      {
+         unsigned char next_id = (src[2] << 6) | (src[3] & 0x3F);
+
+         if (next_id > 0x20 || next_id < 0x50)
+            next_connected = true;
+      }
+
+      result = arabic_shape_map[id - 0x20][prev_connected | (next_connected << 1)];
+
+      if (result)
+         return result;
+
+      return arabic_shape_map[id - 0x20][prev_connected];
+   }
+
+   return 0;
+}
+
+static char* font_driver_reshape_msg(const char* msg)
+{
+   /* worst case transformations are 2 bytes to 4 bytes */
+   char*       buffer  = (char*)malloc((strlen(msg) * 2) + 1);
+   const char* src     = msg;
+   char*       dst     = buffer;
+   bool        reverse = false;
+
+   while (*src || reverse)
+   {
+      if (reverse)
+      {
+         src--;
+         while (IS_MBCONT(src))
+            src--;
+
+         if (IS_RTL(src) || IS_DIR_NEUTRAL(src))
+         {
+            unsigned replacement = font_get_replacement(src, msg);
+            if (replacement)
+            {
+               if (replacement < 0x80)
+                  *dst++ = replacement;
+               else if (replacement < 0x8000)
+               {
+                  *dst++ = 0xC0 | (replacement >> 6);
+                  *dst++ = 0x80 | (replacement & 0x3F);
+               }
+               else if (replacement < 0x10000)
+               {
+                  /* merged glyphs */
+                  if ((replacement >= 0xFEF5) && (replacement <= 0xFEFC))
+                     src -= 2;
+
+                  *dst++ = 0xE0 | (replacement >> 12);
+                  *dst++ = 0x80 | ((replacement >> 6) & 0x3F);
+                  *dst++ = 0x80 | (replacement & 0x3F);
+               }
+               else
+               {
+                  *dst++ = 0xF0 | (replacement >> 18);
+                  *dst++ = 0x80 | ((replacement >> 12) & 0x3F);
+                  *dst++ = 0x80 | ((replacement >> 6) & 0x3F);
+                  *dst++ = 0x80 | (replacement & 0x3F);
+               }
+
+               continue;
+            }
+
+            *dst++ = *src++;
+            while (IS_MBCONT(src))
+               *dst++ = *src++;
+            src--;
+
+            while (IS_MBCONT(src))
+               src--;
+         }
+         else
+         {
+            reverse = false;
+            src++;
+            while (IS_MBCONT(src) || IS_RTL(src) || IS_DIR_NEUTRAL(src))
+               src++;
+         }
+      }
+      else
+      {
+         if (IS_RTL(src))
+         {
+            reverse = true;
+            while (IS_MBCONT(src) || IS_RTL(src) || IS_DIR_NEUTRAL(src))
+               src++;
+         }
+         else
+            *dst++ = *src++;
+      }
+   }
+
+   *dst = '\0';
+
+   return buffer;
+}
+#endif
+
 void font_driver_render_msg(
       video_frame_info_t *video_info,
       void *font_data,
       const char *msg, const void *params)
 {
    font_data_t *font = (font_data_t*)(font_data ? font_data : video_font_driver);
+
    if (font && font->renderer && font->renderer->render_msg)
+   {
+#ifdef HAVE_LANGEXTRA
+      char* new_msg = font_driver_reshape_msg(msg);
+      font->renderer->render_msg(video_info, font->renderer_data, new_msg, params);
+      free(new_msg);
+#else
       font->renderer->render_msg(video_info, font->renderer_data, msg, params);
+#endif
+   }
 }
 
 void font_driver_bind_block(void *font_data, void *block)

griffin/griffin.c

@@ -995,6 +995,7 @@ RETROARCH
 #include "../intl/msg_hash_vn.c"
 #include "../intl/msg_hash_chs.c"
 #include "../intl/msg_hash_cht.c"
+#include "../intl/msg_hash_ar.c"
 #endif
 
 #include "../intl/msg_hash_us.c"