bug 941470 - update harfbuzz to latest upstream commit (205bf834...), to pick up plan-caching for user features and misc fixes. r=jdaggett

gfx/harfbuzz/src/hb-atomic-private.hh

@@ -78,7 +78,7 @@ typedef int32_t hb_atomic_int_t;
 #if (MAC_OS_X_VERSION_MIN_REQUIRED > MAC_OS_X_VERSION_10_4 || __IPHONE_VERSION_MIN_REQUIRED >= 20100)
 #define hb_atomic_ptr_cmpexch(P,O,N)	OSAtomicCompareAndSwapPtrBarrier ((void *) (O), (void *) (N), (void **) (P))
 #else
-#if __ppc64__ || __x86_64__
+#if __ppc64__ || __x86_64__ || __arm64__
 #define hb_atomic_ptr_cmpexch(P,O,N)    OSAtomicCompareAndSwap64Barrier ((int64_t) (O), (int64_t) (N), (int64_t*) (P))
 #else
 #define hb_atomic_ptr_cmpexch(P,O,N)    OSAtomicCompareAndSwap32Barrier ((int32_t) (O), (int32_t) (N), (int32_t*) (P))

gfx/harfbuzz/src/hb-buffer-private.hh

@@ -181,12 +181,13 @@ struct hb_buffer_t {
   HB_INTERNAL bool enlarge (unsigned int size);
 
   inline bool ensure (unsigned int size)
-  { return likely (size < allocated) ? true : enlarge (size); }
+  { return likely (!size || size < allocated) ? true : enlarge (size); }
 
   HB_INTERNAL bool make_room_for (unsigned int num_in, unsigned int num_out);
   HB_INTERNAL bool shift_forward (unsigned int count);
 
-  HB_INTERNAL void *get_scratch_buffer (unsigned int *size);
+  typedef long scratch_buffer_t;
+  HB_INTERNAL scratch_buffer_t *get_scratch_buffer (unsigned int *size);
 
   inline void clear_context (unsigned int side) { context_len[side] = 0; }
 };

gfx/harfbuzz/src/hb-buffer.cc

@@ -152,7 +152,7 @@ hb_buffer_t::shift_forward (unsigned int count)
   return true;
 }
 
-void *
+hb_buffer_t::scratch_buffer_t *
 hb_buffer_t::get_scratch_buffer (unsigned int *size)
 {
   have_output = false;
@@ -161,8 +161,9 @@ hb_buffer_t::get_scratch_buffer (unsigned int *size)
   out_len = 0;
   out_info = info;
 
-  *size = allocated * sizeof (pos[0]);
+  assert ((uintptr_t) pos % sizeof (scratch_buffer_t) == 0);
-  return pos;
+  *size = allocated * sizeof (pos[0]) / sizeof (scratch_buffer_t);
+  return (scratch_buffer_t *) (void *) pos;
 }
 
 
@@ -1149,7 +1150,10 @@ hb_buffer_set_length (hb_buffer_t  *buffer,
   buffer->len = length;
 
   if (!length)
+  {
+    buffer->content_type = HB_BUFFER_CONTENT_TYPE_INVALID;
     buffer->clear_context (0);
+  }
   buffer->clear_context (1);
 
   return true;

gfx/harfbuzz/src/hb-coretext.cc

@@ -394,6 +394,7 @@ _hb_coretext_shape (hb_shape_plan_t    *shape_plan,
                     unsigned int        num_features)
 {
   hb_face_t *face = font->face;
+  hb_coretext_shaper_face_data_t *face_data = HB_SHAPER_DATA_GET (face);
   hb_coretext_shaper_font_data_t *font_data = HB_SHAPER_DATA_GET (font);
 
   /*
@@ -553,11 +554,21 @@ _hb_coretext_shape (hb_shape_plan_t    *shape_plan,
   } HB_STMT_END;
 
   unsigned int scratch_size;
-  char *scratch = (char *) buffer->get_scratch_buffer (&scratch_size);
+  hb_buffer_t::scratch_buffer_t *scratch = buffer->get_scratch_buffer (&scratch_size);
+
+#define ALLOCATE_ARRAY(Type, name, len) \
+  Type *name = (Type *) scratch; \
+  { \
+    unsigned int _consumed = DIV_CEIL ((len) * sizeof (Type), sizeof (*scratch)); \
+    assert (_consumed <= scratch_size); \
+    scratch += _consumed; \
+    scratch_size -= _consumed; \
+  }
 
 #define utf16_index() var1.u32
 
-  UniChar *pchars = (UniChar *) scratch;
+  ALLOCATE_ARRAY (UniChar, pchars, buffer->len * 2);
+
   unsigned int chars_len = 0;
   for (unsigned int i = 0; i < buffer->len; i++) {
     hb_codepoint_t c = buffer->info[i].codepoint;
@@ -586,7 +597,7 @@ _hb_coretext_shape (hb_shape_plan_t    *shape_plan,
 
   if (num_features)
   {
-    unsigned int *log_clusters = (unsigned int *) (pchars + chars_len);
+    ALLOCATE_ARRAY (unsigned int, log_clusters, chars_len);
 
     /* Need log_clusters to assign features. */
     chars_len = 0;
@@ -638,26 +649,60 @@ _hb_coretext_shape (hb_shape_plan_t    *shape_plan,
 
   const CFRange range_all = CFRangeMake (0, 0);
 
-  for (unsigned int i = 0; i < num_runs; i++) {
+  for (unsigned int i = 0; i < num_runs; i++)
+  {
     CTRunRef run = (CTRunRef) CFArrayGetValueAtIndex (glyph_runs, i);
 
+    /* CoreText does automatic font fallback (AKA "cascading") for  characters
+     * not supported by the requested font, and provides no way to turn it off,
+     * so we detect if the returned run uses a font other than the requested
+     * one and fill in the buffer with .notdef glyphs instead of random glyph
+     * indices from a different font.
+     */
+    CFDictionaryRef attributes = CTRunGetAttributes (run);
+    CTFontRef run_ct_font = static_cast<CTFontRef>(CFDictionaryGetValue (attributes, kCTFontAttributeName));
+    CGFontRef run_cg_font = CTFontCopyGraphicsFont (run_ct_font, 0);
+    if (!CFEqual (run_cg_font, face_data->cg_font))
+    {
+        CFRelease (run_cg_font);
+
+	CFRange range = CTRunGetStringRange (run);
+	buffer->ensure (buffer->len + range.length);
+	if (buffer->in_error)
+	  FAIL ("Buffer resize failed");
+	hb_glyph_info_t *info = buffer->info + buffer->len;
+	buffer->len += range.length;
+
+        for (CFIndex j = 0; j < range.length; j++)
+	{
+            CGGlyph notdef = 0;
+            double advance = CTFontGetAdvancesForGlyphs (font_data->ct_font, kCTFontHorizontalOrientation, &notdef, NULL, 1);
+
+            info->codepoint = notdef;
+	    /* TODO We have to fixup clusters later.  See vis_clusters in
+	     * hb-uniscribe.cc for example. */
+            info->cluster = range.location + j;
+
+            info->mask = advance;
+            info->var1.u32 = 0;
+            info->var2.u32 = 0;
+
+	    info++;
+        }
+        continue;
+    }
+    CFRelease (run_cg_font);
+
     unsigned int num_glyphs = CTRunGetGlyphCount (run);
     if (num_glyphs == 0)
       continue;
 
     buffer->ensure (buffer->len + num_glyphs);
 
-    /* Testing indicates that CTRunGetGlyphsPtr (almost?) always succeeds,
+    scratch = buffer->get_scratch_buffer (&scratch_size);
-     * and so copying data to our own buffer with CTRunGetGlyphs will be
-     * extremely rare. */
 
-    unsigned int scratch_size;
+    /* Testing indicates that CTRunGetGlyphsPtr, etc (almost?) always
-    char *scratch = (char *) buffer->get_scratch_buffer (&scratch_size);
+     * succeed, and so copying data to our own buffer will be rare. */
-
-#define ALLOCATE_ARRAY(Type, name, len) \
-  Type *name = (Type *) scratch; \
-  scratch += (len) * sizeof ((name)[0]); \
-  scratch_size -= (len) * sizeof ((name)[0]);
 
     const CGGlyph* glyphs = CTRunGetGlyphsPtr (run);
     if (!glyphs) {

gfx/harfbuzz/src/hb-fallback-shape.cc

@@ -95,6 +95,16 @@ _hb_fallback_shape (hb_shape_plan_t    *shape_plan HB_UNUSED,
 		    const hb_feature_t *features HB_UNUSED,
 		    unsigned int        num_features HB_UNUSED)
 {
+  /* TODO
+   *
+   * - Apply fallback kern.
+   * - Handle Variation Selectors?
+   * - Apply normalization?
+   *
+   * This will make the fallback shaper into a dumb "TrueType"
+   * shaper which many people unfortunately still request.
+   */
+
   hb_codepoint_t space;
   font->get_glyph (' ', 0, &space);
 

gfx/harfbuzz/src/hb-font-private.hh

@@ -118,12 +118,12 @@ struct hb_font_t {
   /* Convert from parent-font user-space to our user-space */
   inline hb_position_t parent_scale_x_distance (hb_position_t v) {
     if (unlikely (parent && parent->x_scale != x_scale))
-      return v * (int64_t) this->x_scale / this->parent->x_scale;
+      return (hb_position_t) (v * (int64_t) this->x_scale / this->parent->x_scale);
     return v;
   }
   inline hb_position_t parent_scale_y_distance (hb_position_t v) {
     if (unlikely (parent && parent->y_scale != y_scale))
-      return v * (int64_t) this->y_scale / this->parent->y_scale;
+      return (hb_position_t) (v * (int64_t) this->y_scale / this->parent->y_scale);
     return v;
   }
   inline hb_position_t parent_scale_x_position (hb_position_t v) {
@@ -397,7 +397,7 @@ struct hb_font_t {
   }
 
   private:
-  inline hb_position_t em_scale (int16_t v, int scale) { return v * (int64_t) scale / face->get_upem (); }
+  inline hb_position_t em_scale (int16_t v, int scale) { return (hb_position_t) (v * (int64_t) scale / face->get_upem ()); }
 };
 
 #define HB_SHAPER_DATA_CREATE_FUNC_EXTRA_ARGS

gfx/harfbuzz/src/hb-ft.cc

@@ -418,8 +418,8 @@ hb_ft_font_create (FT_Face           ft_face,
 		     _hb_ft_get_font_funcs (),
 		     ft_face, (hb_destroy_func_t) _do_nothing);
   hb_font_set_scale (font,
-		     ((uint64_t) ft_face->size->metrics.x_scale * (uint64_t) ft_face->units_per_EM + (1<<15)) >> 16,
+		     (int) (((uint64_t) ft_face->size->metrics.x_scale * (uint64_t) ft_face->units_per_EM + (1<<15)) >> 16),
-		     ((uint64_t) ft_face->size->metrics.y_scale * (uint64_t) ft_face->units_per_EM + (1<<15)) >> 16);
+		     (int) (((uint64_t) ft_face->size->metrics.y_scale * (uint64_t) ft_face->units_per_EM + (1<<15)) >> 16));
   hb_font_set_ppem (font,
 		    ft_face->size->metrics.x_ppem,
 		    ft_face->size->metrics.y_ppem);

gfx/harfbuzz/src/hb-graphite2.cc

@@ -243,14 +243,9 @@ _hb_graphite2_shape (hb_shape_plan_t    *shape_plan,
   float curradvx = 0., curradvy = 0.;
 
   unsigned int scratch_size;
-  char *scratch = (char *) buffer->get_scratch_buffer (&scratch_size);
+  hb_buffer_t::scratch_buffer_t *scratch = buffer->get_scratch_buffer (&scratch_size);
 
-#define ALLOCATE_ARRAY(Type, name, len) \
+  uint32_t *chars = (uint32_t *) scratch;
-  Type *name = (Type *) scratch; \
-  scratch += (len) * sizeof ((name)[0]); \
-  scratch_size -= (len) * sizeof ((name)[0]);
-
-  ALLOCATE_ARRAY (uint32_t, chars, buffer->len);
 
   for (unsigned int i = 0; i < buffer->len; ++i)
     chars[i] = buffer->info[i].codepoint;
@@ -276,9 +271,9 @@ _hb_graphite2_shape (hb_shape_plan_t    *shape_plan,
     return false;
   }
 
-  scratch = (char *) buffer->get_scratch_buffer (&scratch_size);
+  scratch = buffer->get_scratch_buffer (&scratch_size);
-  while ((sizeof (hb_graphite2_cluster_t) * buffer->len +
+  while ((DIV_CEIL (sizeof (hb_graphite2_cluster_t) * buffer->len, sizeof (*scratch)) +
-	  sizeof (hb_codepoint_t) * glyph_count) > scratch_size)
+	  DIV_CEIL (sizeof (hb_codepoint_t) * glyph_count, sizeof (*scratch))) > scratch_size)
   {
     buffer->ensure (buffer->allocated * 2);
     if (unlikely (buffer->in_error)) {
@@ -286,12 +281,23 @@ _hb_graphite2_shape (hb_shape_plan_t    *shape_plan,
       gr_seg_destroy (seg);
       return false;
     }
-    scratch = (char *) buffer->get_scratch_buffer (&scratch_size);
+    scratch = buffer->get_scratch_buffer (&scratch_size);
+  }
+
+#define ALLOCATE_ARRAY(Type, name, len) \
+  Type *name = (Type *) scratch; \
+  { \
+    unsigned int _consumed = DIV_CEIL ((len) * sizeof (Type), sizeof (*scratch)); \
+    assert (_consumed <= scratch_size); \
+    scratch += _consumed; \
+    scratch_size -= _consumed; \
   }
 
   ALLOCATE_ARRAY (hb_graphite2_cluster_t, clusters, buffer->len);
   ALLOCATE_ARRAY (hb_codepoint_t, gids, glyph_count);
 
+#undef ALLOCATE_ARRAY
+
   memset (clusters, 0, sizeof (clusters[0]) * buffer->len);
 
   hb_codepoint_t *pg = gids;

gfx/harfbuzz/src/hb-ot-layout-common-private.hh

@@ -377,7 +377,7 @@ struct FeatureParamsStylisticSet
     return TRACE_RETURN (c->check_struct (this));
   }
 
-  USHORT	minorVersion;	/* (set to 0): This corresponds to a “minor”
+  USHORT	version;	/* (set to 0): This corresponds to a “minor”
 				 * version number. Additional data may be
 				 * added to the end of this Feature Parameters
 				 * table in the future. */
@@ -400,6 +400,7 @@ struct FeatureParamsStylisticSet
   DEFINE_SIZE_STATIC (4);
 };
 
+/* http://www.microsoft.com/typography/otspec/features_ae.htm#cv01-cv99 */
 struct FeatureParamsCharacterVariants
 {
   inline bool sanitize (hb_sanitize_context_t *c) {
@@ -1112,7 +1113,7 @@ struct Device
 
     if (!pixels) return 0;
 
-    return pixels * (int64_t) scale / ppem;
+    return (int) (pixels * (int64_t) scale / ppem);
   }
 
 

gfx/harfbuzz/src/hb-ot-shape-complex-myanmar-machine.hh

@@ -34,196 +34,197 @@
 
 #line 36 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
 static const unsigned char _myanmar_syllable_machine_trans_keys[] = {
-	1u, 30u, 3u, 30u, 5u, 29u, 5u, 8u, 5u, 29u, 3u, 25u, 5u, 25u, 5u, 25u, 
+	1u, 31u, 3u, 30u, 5u, 29u, 5u, 8u, 5u, 29u, 3u, 25u, 5u, 25u, 5u, 25u, 
 	3u, 29u, 3u, 29u, 3u, 29u, 3u, 29u, 1u, 16u, 3u, 29u, 3u, 29u, 3u, 29u, 
 	3u, 29u, 3u, 29u, 3u, 29u, 3u, 29u, 3u, 29u, 3u, 29u, 5u, 29u, 5u, 8u, 
 	5u, 29u, 3u, 25u, 5u, 25u, 5u, 25u, 3u, 29u, 3u, 29u, 3u, 29u, 3u, 29u, 
 	3u, 30u, 3u, 29u, 1u, 30u, 3u, 29u, 3u, 29u, 3u, 29u, 3u, 29u, 3u, 29u, 
-	3u, 29u, 3u, 29u, 3u, 29u, 3u, 29u, 0
+	3u, 29u, 3u, 29u, 3u, 29u, 3u, 29u, 8u, 8u, 0
 };
 
 static const char _myanmar_syllable_machine_key_spans[] = {
-	30, 28, 25, 4, 25, 23, 21, 21, 
+	31, 28, 25, 4, 25, 23, 21, 21, 
 	27, 27, 27, 27, 16, 27, 27, 27, 
 	27, 27, 27, 27, 27, 27, 25, 4, 
 	25, 23, 21, 21, 27, 27, 27, 27, 
 	28, 27, 30, 27, 27, 27, 27, 27, 
-	27, 27, 27, 27
+	27, 27, 27, 27, 1
 };
 
 static const short _myanmar_syllable_machine_index_offsets[] = {
-	0, 31, 60, 86, 91, 117, 141, 163, 
+	0, 32, 61, 87, 92, 118, 142, 164, 
-	185, 213, 241, 269, 297, 314, 342, 370, 
+	186, 214, 242, 270, 298, 315, 343, 371, 
-	398, 426, 454, 482, 510, 538, 566, 592, 
+	399, 427, 455, 483, 511, 539, 567, 593, 
-	597, 623, 647, 669, 691, 719, 747, 775, 
+	598, 624, 648, 670, 692, 720, 748, 776, 
-	803, 832, 860, 891, 919, 947, 975, 1003, 
+	804, 833, 861, 892, 920, 948, 976, 1004, 
-	1031, 1059, 1087, 1115
+	1032, 1060, 1088, 1116, 1144
 };
 
 static const char _myanmar_syllable_machine_indicies[] = {
 	1, 1, 2, 3, 4, 4, 0, 5, 
 	0, 6, 0, 1, 0, 0, 0, 7, 
 	0, 8, 1, 0, 9, 10, 11, 12, 
-	13, 14, 15, 16, 17, 18, 0, 20, 
+	13, 14, 15, 16, 17, 18, 19, 0, 
-	21, 22, 22, 19, 23, 19, 24, 19, 
+	21, 22, 23, 23, 20, 24, 20, 25, 
-	19, 19, 19, 19, 19, 19, 25, 19, 
+	20, 20, 20, 20, 20, 20, 20, 26, 
-	19, 26, 27, 28, 29, 30, 31, 32, 
+	20, 20, 27, 28, 29, 30, 31, 32, 
-	33, 34, 35, 19, 22, 22, 19, 23, 
+	33, 34, 35, 36, 20, 23, 23, 20, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	24, 20, 20, 20, 20, 20, 20, 20, 
-	19, 36, 19, 19, 19, 19, 19, 19, 
+	20, 20, 37, 20, 20, 20, 20, 20, 
-	30, 19, 19, 19, 34, 19, 22, 22, 
+	20, 31, 20, 20, 20, 35, 20, 23, 
-	19, 23, 19, 22, 22, 19, 23, 19, 
+	23, 20, 24, 20, 23, 23, 20, 24, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	19, 19, 19, 19, 19, 19, 19, 30, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	19, 19, 19, 34, 19, 37, 19, 22, 
+	31, 20, 20, 20, 35, 20, 38, 20, 
-	22, 19, 23, 19, 30, 19, 19, 19, 
+	23, 23, 20, 24, 20, 31, 20, 20, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 39, 20, 20, 
-	19, 19, 19, 30, 19, 22, 22, 19, 
+	20, 20, 20, 20, 31, 20, 23, 23, 
-	23, 19, 19, 19, 19, 19, 19, 19, 
+	20, 24, 20, 20, 20, 20, 20, 20, 
-	19, 19, 38, 19, 19, 19, 19, 19, 
+	20, 20, 20, 39, 20, 20, 20, 20, 
-	19, 30, 19, 22, 22, 19, 23, 19, 
+	20, 20, 31, 20, 23, 23, 20, 24, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	19, 19, 19, 19, 19, 19, 19, 30, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	19, 20, 19, 22, 22, 19, 23, 19, 
+	31, 20, 21, 20, 23, 23, 20, 24, 
-	24, 19, 19, 19, 19, 19, 19, 19, 
+	20, 25, 20, 20, 20, 20, 20, 20, 
-	39, 19, 19, 39, 19, 19, 19, 30, 
+	20, 40, 20, 20, 40, 20, 20, 20, 
-	40, 19, 19, 34, 19, 20, 19, 22, 
+	31, 41, 20, 20, 35, 20, 21, 20, 
-	22, 19, 23, 19, 24, 19, 19, 19, 
+	23, 23, 20, 24, 20, 25, 20, 20, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	19, 19, 19, 30, 19, 19, 19, 34, 
+	20, 20, 20, 20, 31, 20, 20, 20, 
-	19, 20, 19, 22, 22, 19, 23, 19, 
+	35, 20, 21, 20, 23, 23, 20, 24, 
-	24, 19, 19, 19, 19, 19, 19, 19, 
+	20, 25, 20, 20, 20, 20, 20, 20, 
-	39, 19, 19, 19, 19, 19, 19, 30, 
+	20, 40, 20, 20, 20, 20, 20, 20, 
-	40, 19, 19, 34, 19, 20, 19, 22, 
+	31, 41, 20, 20, 35, 20, 21, 20, 
-	22, 19, 23, 19, 24, 19, 19, 19, 
+	23, 23, 20, 24, 20, 25, 20, 20, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	19, 19, 19, 30, 40, 19, 19, 34, 
+	20, 20, 20, 20, 31, 41, 20, 20, 
-	19, 1, 1, 19, 19, 19, 19, 19, 
+	35, 20, 1, 1, 20, 20, 20, 20, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	1, 19, 20, 19, 22, 22, 19, 23, 
+	20, 1, 20, 21, 20, 23, 23, 20, 
-	19, 24, 19, 19, 19, 19, 19, 19, 
+	24, 20, 25, 20, 20, 20, 20, 20, 
-	19, 25, 19, 19, 26, 27, 28, 29, 
+	20, 20, 26, 20, 20, 27, 28, 29, 
-	30, 31, 32, 33, 34, 19, 20, 19, 
+	30, 31, 32, 33, 34, 35, 20, 21, 
-	22, 22, 19, 23, 19, 24, 19, 19, 
+	20, 23, 23, 20, 24, 20, 25, 20, 
-	19, 19, 19, 19, 19, 33, 19, 19, 
+	20, 20, 20, 20, 20, 20, 34, 20, 
-	19, 19, 19, 19, 30, 31, 32, 33, 
+	20, 20, 20, 20, 20, 31, 32, 33, 
-	34, 19, 20, 19, 22, 22, 19, 23, 
+	34, 35, 20, 21, 20, 23, 23, 20, 
-	19, 24, 19, 19, 19, 19, 19, 19, 
+	24, 20, 25, 20, 20, 20, 20, 20, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	30, 31, 32, 33, 34, 19, 20, 19, 
+	20, 31, 32, 33, 34, 35, 20, 21, 
-	22, 22, 19, 23, 19, 24, 19, 19, 
+	20, 23, 23, 20, 24, 20, 25, 20, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	19, 19, 19, 19, 30, 31, 32, 19, 
+	20, 20, 20, 20, 20, 31, 32, 33, 
-	34, 19, 20, 19, 22, 22, 19, 23, 
+	20, 35, 20, 21, 20, 23, 23, 20, 
-	19, 24, 19, 19, 19, 19, 19, 19, 
+	24, 20, 25, 20, 20, 20, 20, 20, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	30, 19, 32, 19, 34, 19, 20, 19, 
+	20, 31, 20, 33, 20, 35, 20, 21, 
-	22, 22, 19, 23, 19, 24, 19, 19, 
+	20, 23, 23, 20, 24, 20, 25, 20, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	26, 19, 28, 19, 30, 31, 32, 33, 
+	20, 27, 20, 29, 20, 31, 32, 33, 
-	34, 19, 20, 19, 22, 22, 19, 23, 
+	34, 35, 20, 21, 20, 23, 23, 20, 
-	19, 24, 19, 19, 19, 19, 19, 19, 
+	24, 20, 25, 20, 20, 20, 20, 20, 
-	19, 33, 19, 19, 26, 19, 19, 19, 
+	20, 20, 34, 20, 20, 27, 20, 20, 
-	30, 31, 32, 33, 34, 19, 20, 19, 
+	20, 31, 32, 33, 34, 35, 20, 21, 
-	22, 22, 19, 23, 19, 24, 19, 19, 
+	20, 23, 23, 20, 24, 20, 25, 20, 
-	19, 19, 19, 19, 19, 19, 19, 19, 
+	20, 20, 20, 20, 20, 20, 20, 20, 
-	26, 27, 28, 19, 30, 31, 32, 33, 
+	20, 27, 28, 29, 20, 31, 32, 33, 
-	34, 19, 20, 21, 22, 22, 19, 23, 
+	34, 35, 20, 21, 22, 23, 23, 20, 
-	19, 24, 19, 19, 19, 19, 19, 19, 
+	24, 20, 25, 20, 20, 20, 20, 20, 
-	19, 25, 19, 19, 26, 27, 28, 29, 
+	20, 20, 26, 20, 20, 27, 28, 29, 
-	30, 31, 32, 33, 34, 19, 3, 3, 
+	30, 31, 32, 33, 34, 35, 20, 3, 
-	41, 5, 41, 41, 41, 41, 41, 41, 
+	3, 42, 5, 42, 42, 42, 42, 42, 
-	41, 41, 41, 42, 41, 41, 41, 41, 
+	42, 42, 42, 42, 43, 42, 42, 42, 
-	41, 41, 13, 41, 41, 41, 17, 41, 
+	42, 42, 42, 13, 42, 42, 42, 17, 
-	3, 3, 41, 5, 41, 3, 3, 41, 
+	42, 3, 3, 42, 5, 42, 3, 3, 
-	5, 41, 41, 41, 41, 41, 41, 41, 
+	42, 5, 42, 42, 42, 42, 42, 42, 
-	41, 41, 41, 41, 41, 41, 41, 41, 
+	42, 42, 42, 42, 42, 42, 42, 42, 
-	41, 13, 41, 41, 41, 17, 41, 43, 
+	42, 42, 13, 42, 42, 42, 17, 42, 
-	41, 3, 3, 41, 5, 41, 13, 41, 
+	44, 42, 3, 3, 42, 5, 42, 13, 
-	41, 41, 41, 41, 41, 41, 41, 41, 
+	42, 42, 42, 42, 42, 42, 42, 45, 
-	41, 41, 41, 41, 41, 13, 41, 3, 
+	42, 42, 42, 42, 42, 42, 13, 42, 
-	3, 41, 5, 41, 41, 41, 41, 41, 
+	3, 3, 42, 5, 42, 42, 42, 42, 
-	41, 41, 41, 41, 44, 41, 41, 41, 
+	42, 42, 42, 42, 42, 45, 42, 42, 
-	41, 41, 41, 13, 41, 3, 3, 41, 
+	42, 42, 42, 42, 13, 42, 3, 3, 
-	5, 41, 41, 41, 41, 41, 41, 41, 
+	42, 5, 42, 42, 42, 42, 42, 42, 
-	41, 41, 41, 41, 41, 41, 41, 41, 
+	42, 42, 42, 42, 42, 42, 42, 42, 
-	41, 13, 41, 2, 41, 3, 3, 41, 
+	42, 42, 13, 42, 2, 42, 3, 3, 
-	5, 41, 6, 41, 41, 41, 41, 41, 
+	42, 5, 42, 6, 42, 42, 42, 42, 
-	41, 41, 45, 41, 41, 45, 41, 41, 
+	42, 42, 42, 46, 42, 42, 46, 42, 
-	41, 13, 46, 41, 41, 17, 41, 2, 
+	42, 42, 13, 47, 42, 42, 17, 42, 
-	41, 3, 3, 41, 5, 41, 6, 41, 
+	2, 42, 3, 3, 42, 5, 42, 6, 
-	41, 41, 41, 41, 41, 41, 41, 41, 
+	42, 42, 42, 42, 42, 42, 42, 42, 
-	41, 41, 41, 41, 41, 13, 41, 41, 
+	42, 42, 42, 42, 42, 42, 13, 42, 
-	41, 17, 41, 2, 41, 3, 3, 41, 
+	42, 42, 17, 42, 2, 42, 3, 3, 
-	5, 41, 6, 41, 41, 41, 41, 41, 
+	42, 5, 42, 6, 42, 42, 42, 42, 
-	41, 41, 45, 41, 41, 41, 41, 41, 
+	42, 42, 42, 46, 42, 42, 42, 42, 
-	41, 13, 46, 41, 41, 17, 41, 2, 
+	42, 42, 13, 47, 42, 42, 17, 42, 
-	41, 3, 3, 41, 5, 41, 6, 41, 
+	2, 42, 3, 3, 42, 5, 42, 6, 
-	41, 41, 41, 41, 41, 41, 41, 41, 
+	42, 42, 42, 42, 42, 42, 42, 42, 
-	41, 41, 41, 41, 41, 13, 46, 41, 
+	42, 42, 42, 42, 42, 42, 13, 47, 
-	41, 17, 41, 20, 21, 22, 22, 19, 
+	42, 42, 17, 42, 21, 22, 23, 23, 
-	23, 19, 24, 19, 19, 19, 19, 19, 
+	20, 24, 20, 25, 20, 20, 20, 20, 
-	19, 19, 47, 19, 19, 26, 27, 28, 
+	20, 20, 20, 48, 20, 20, 27, 28, 
-	29, 30, 31, 32, 33, 34, 35, 19, 
+	29, 30, 31, 32, 33, 34, 35, 36, 
-	20, 48, 22, 22, 19, 23, 19, 24, 
+	20, 21, 49, 23, 23, 20, 24, 20, 
-	19, 19, 19, 19, 19, 19, 19, 25, 
+	25, 20, 20, 20, 20, 20, 20, 20, 
-	19, 19, 26, 27, 28, 29, 30, 31, 
+	26, 20, 20, 27, 28, 29, 30, 31, 
-	32, 33, 34, 19, 1, 1, 2, 3, 
+	32, 33, 34, 35, 20, 1, 1, 2, 
-	3, 3, 41, 5, 41, 6, 41, 1, 
+	3, 3, 3, 42, 5, 42, 6, 42, 
-	41, 41, 41, 1, 41, 8, 1, 41, 
+	1, 42, 42, 42, 1, 42, 8, 1, 
-	9, 10, 11, 12, 13, 14, 15, 16, 
+	42, 9, 10, 11, 12, 13, 14, 15, 
-	17, 18, 41, 2, 41, 3, 3, 41, 
+	16, 17, 18, 42, 2, 42, 3, 3, 
-	5, 41, 6, 41, 41, 41, 41, 41, 
+	42, 5, 42, 6, 42, 42, 42, 42, 
-	41, 41, 8, 41, 41, 9, 10, 11, 
+	42, 42, 42, 8, 42, 42, 9, 10, 
-	12, 13, 14, 15, 16, 17, 41, 2, 
+	11, 12, 13, 14, 15, 16, 17, 42, 
-	41, 3, 3, 41, 5, 41, 6, 41, 
+	2, 42, 3, 3, 42, 5, 42, 6, 
-	41, 41, 41, 41, 41, 41, 16, 41, 
+	42, 42, 42, 42, 42, 42, 42, 16, 
-	41, 41, 41, 41, 41, 13, 14, 15, 
+	42, 42, 42, 42, 42, 42, 13, 14, 
-	16, 17, 41, 2, 41, 3, 3, 41, 
+	15, 16, 17, 42, 2, 42, 3, 3, 
-	5, 41, 6, 41, 41, 41, 41, 41, 
+	42, 5, 42, 6, 42, 42, 42, 42, 
-	41, 41, 41, 41, 41, 41, 41, 41, 
+	42, 42, 42, 42, 42, 42, 42, 42, 
-	41, 13, 14, 15, 16, 17, 41, 2, 
+	42, 42, 13, 14, 15, 16, 17, 42, 
-	41, 3, 3, 41, 5, 41, 6, 41, 
+	2, 42, 3, 3, 42, 5, 42, 6, 
-	41, 41, 41, 41, 41, 41, 41, 41, 
+	42, 42, 42, 42, 42, 42, 42, 42, 
-	41, 41, 41, 41, 41, 13, 14, 15, 
+	42, 42, 42, 42, 42, 42, 13, 14, 
-	41, 17, 41, 2, 41, 3, 3, 41, 
+	15, 42, 17, 42, 2, 42, 3, 3, 
-	5, 41, 6, 41, 41, 41, 41, 41, 
+	42, 5, 42, 6, 42, 42, 42, 42, 
-	41, 41, 41, 41, 41, 41, 41, 41, 
+	42, 42, 42, 42, 42, 42, 42, 42, 
-	41, 13, 41, 15, 41, 17, 41, 2, 
+	42, 42, 13, 42, 15, 42, 17, 42, 
-	41, 3, 3, 41, 5, 41, 6, 41, 
+	2, 42, 3, 3, 42, 5, 42, 6, 
-	41, 41, 41, 41, 41, 41, 41, 41, 
+	42, 42, 42, 42, 42, 42, 42, 42, 
-	41, 9, 41, 11, 41, 13, 14, 15, 
+	42, 42, 9, 42, 11, 42, 13, 14, 
-	16, 17, 41, 2, 41, 3, 3, 41, 
+	15, 16, 17, 42, 2, 42, 3, 3, 
-	5, 41, 6, 41, 41, 41, 41, 41, 
+	42, 5, 42, 6, 42, 42, 42, 42, 
-	41, 41, 16, 41, 41, 9, 41, 41, 
+	42, 42, 42, 16, 42, 42, 9, 42, 
-	41, 13, 14, 15, 16, 17, 41, 2, 
+	42, 42, 13, 14, 15, 16, 17, 42, 
-	41, 3, 3, 41, 5, 41, 6, 41, 
+	2, 42, 3, 3, 42, 5, 42, 6, 
-	41, 41, 41, 41, 41, 41, 41, 41, 
+	42, 42, 42, 42, 42, 42, 42, 42, 
-	41, 9, 10, 11, 41, 13, 14, 15, 
+	42, 42, 9, 10, 11, 42, 13, 14, 
-	16, 17, 41, 2, 3, 3, 3, 41, 
+	15, 16, 17, 42, 2, 3, 3, 3, 
-	5, 41, 6, 41, 41, 41, 41, 41, 
+	42, 5, 42, 6, 42, 42, 42, 42, 
-	41, 41, 8, 41, 41, 9, 10, 11, 
+	42, 42, 42, 8, 42, 42, 9, 10, 
-	12, 13, 14, 15, 16, 17, 41, 0
+	11, 12, 13, 14, 15, 16, 17, 42, 
+	51, 50, 0
 };
 
 static const char _myanmar_syllable_machine_trans_targs[] = {
 	0, 1, 22, 0, 0, 23, 29, 32, 
 	35, 36, 40, 41, 42, 25, 38, 39, 
-	37, 28, 43, 0, 2, 12, 0, 3, 
+	37, 28, 43, 44, 0, 2, 12, 0, 
-	9, 13, 14, 18, 19, 20, 5, 16, 
+	3, 9, 13, 14, 18, 19, 20, 5, 
-	17, 15, 8, 21, 4, 6, 7, 10, 
+	16, 17, 15, 8, 21, 4, 6, 7, 
-	11, 0, 24, 26, 27, 30, 31, 33, 
+	10, 11, 0, 24, 26, 27, 30, 31, 
-	34
+	33, 34, 0, 0
 };
 
 static const char _myanmar_syllable_machine_trans_actions[] = {
 	3, 0, 0, 4, 5, 0, 0, 0, 
 	0, 0, 0, 0, 0, 0, 0, 0, 
-	0, 0, 0, 6, 0, 0, 7, 0, 
+	0, 0, 0, 0, 6, 0, 0, 7, 
 	0, 0, 0, 0, 0, 0, 0, 0, 
 	0, 0, 0, 0, 0, 0, 0, 0, 
-	0, 8, 0, 0, 0, 0, 0, 0, 
+	0, 0, 8, 0, 0, 0, 0, 0, 
-	0
+	0, 0, 9, 10
 };
 
 static const char _myanmar_syllable_machine_to_state_actions[] = {
@@ -232,7 +233,7 @@ static const char _myanmar_syllable_machine_to_state_actions[] = {
 	0, 0, 0, 0, 0, 0, 0, 0, 
 	0, 0, 0, 0, 0, 0, 0, 0, 
 	0, 0, 0, 0, 0, 0, 0, 0, 
-	0, 0, 0, 0
+	0, 0, 0, 0, 0
 };
 
 static const char _myanmar_syllable_machine_from_state_actions[] = {
@@ -241,16 +242,16 @@ static const char _myanmar_syllable_machine_from_state_actions[] = {
 	0, 0, 0, 0, 0, 0, 0, 0, 
 	0, 0, 0, 0, 0, 0, 0, 0, 
 	0, 0, 0, 0, 0, 0, 0, 0, 
-	0, 0, 0, 0
+	0, 0, 0, 0, 0
 };
 
 static const short _myanmar_syllable_machine_eof_trans[] = {
-	0, 20, 20, 20, 20, 20, 20, 20, 
+	0, 21, 21, 21, 21, 21, 21, 21, 
-	20, 20, 20, 20, 20, 20, 20, 20, 
+	21, 21, 21, 21, 21, 21, 21, 21, 
-	20, 20, 20, 20, 20, 20, 42, 42, 
+	21, 21, 21, 21, 21, 21, 43, 43, 
-	42, 42, 42, 42, 42, 42, 42, 42, 
+	43, 43, 43, 43, 43, 43, 43, 43, 
-	20, 20, 42, 42, 42, 42, 42, 42, 
+	21, 21, 43, 43, 43, 43, 43, 43, 
-	42, 42, 42, 42
+	43, 43, 43, 43, 51
 };
 
 static const int myanmar_syllable_machine_start = 0;
@@ -264,7 +265,7 @@ static const int myanmar_syllable_machine_en_main = 0;
 
 
 
-#line 90 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
+#line 93 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
 
 
 #define found_syllable(syllable_type) \
@@ -284,7 +285,7 @@ find_syllables (hb_buffer_t *buffer)
   int cs;
   hb_glyph_info_t *info = buffer->info;
   
-#line 288 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
+#line 289 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
 	{
 	cs = myanmar_syllable_machine_start;
 	ts = 0;
@@ -292,7 +293,7 @@ find_syllables (hb_buffer_t *buffer)
 	act = 0;
 	}
 
-#line 111 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
+#line 114 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
 
 
   p = 0;
@@ -301,7 +302,7 @@ find_syllables (hb_buffer_t *buffer)
   unsigned int last = 0;
   unsigned int syllable_serial = 1;
   
-#line 305 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
+#line 306 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
 	{
 	int _slen;
 	int _trans;
@@ -315,7 +316,7 @@ _resume:
 #line 1 "NONE"
 	{ts = p;}
 	break;
-#line 319 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
+#line 320 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
 	}
 
 	_keys = _myanmar_syllable_machine_trans_keys + (cs<<1);
@@ -334,30 +335,38 @@ _eof_trans:
 
 	switch ( _myanmar_syllable_machine_trans_actions[_trans] ) {
 	case 7:
-#line 83 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
+#line 85 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
 	{te = p+1;{ found_syllable (consonant_syllable); }}
 	break;
 	case 5:
-#line 84 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
-	{te = p+1;{ found_syllable (non_myanmar_cluster); }}
-	break;
-	case 4:
-#line 85 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
-	{te = p+1;{ found_syllable (broken_cluster); }}
-	break;
-	case 3:
 #line 86 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
+	{te = p+1;{ found_syllable (non_myanmar_cluster); }}
+	break;
+	case 10:
+#line 87 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
+	{te = p+1;{ found_syllable (punctuation_cluster); }}
+	break;
+	case 4:
+#line 88 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
+	{te = p+1;{ found_syllable (broken_cluster); }}
+	break;
+	case 3:
+#line 89 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
 	{te = p+1;{ found_syllable (non_myanmar_cluster); }}
 	break;
 	case 6:
-#line 83 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
+#line 85 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
 	{te = p;p--;{ found_syllable (consonant_syllable); }}
 	break;
 	case 8:
-#line 85 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
+#line 88 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
 	{te = p;p--;{ found_syllable (broken_cluster); }}
 	break;
-#line 361 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
+	case 9:
+#line 89 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
+	{te = p;p--;{ found_syllable (non_myanmar_cluster); }}
+	break;
+#line 370 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
 	}
 
 _again:
@@ -366,7 +375,7 @@ _again:
 #line 1 "NONE"
 	{ts = 0;}
 	break;
-#line 370 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
+#line 379 "hb-ot-shape-complex-myanmar-machine.hh.tmp"
 	}
 
 	if ( ++p != pe )
@@ -382,7 +391,7 @@ _again:
 
 	}
 
-#line 120 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
+#line 123 "../../src/hb-ot-shape-complex-myanmar-machine.rl"
 
 }
 

gfx/harfbuzz/src/hb-ot-shape-complex-myanmar-machine.rl

@@ -61,6 +61,7 @@ VS   = 30;
 ZWJ  = 6;
 ZWNJ = 5;
 Ra   = 16;
+P    = 31;
 
 j = ZWJ|ZWNJ;			# Joiners
 k = (Ra As H);			# Kinzi
@@ -70,18 +71,20 @@ c = C|Ra;			# is_consonant
 medial_group = MY? MR? ((MW MH? | MH) As?)?;
 main_vowel_group = VPre* VAbv* VBlw* A* (DB As?)?;
 post_vowel_group = VPst MH? As* VAbv* A* (DB As?)?;
-pwo_tone_group = PT A* (DB As?)?;
+pwo_tone_group = PT A* DB? As?;
 
 complex_syllable_tail = As* medial_group main_vowel_group post_vowel_group* pwo_tone_group* V* j?;
 syllable_tail = (H | complex_syllable_tail);
 
 consonant_syllable =	k? (c|IV|D|GB).VS? (H (c|IV).VS?)* syllable_tail;
+punctuation_cluster = 	P V;
 broken_cluster =	k? VS? syllable_tail;
 other =			any;
 
 main := |*
 	consonant_syllable	=> { found_syllable (consonant_syllable); };
 	j			=> { found_syllable (non_myanmar_cluster); };
+	punctuation_cluster	=> { found_syllable (punctuation_cluster); };
 	broken_cluster		=> { found_syllable (broken_cluster); };
 	other			=> { found_syllable (non_myanmar_cluster); };
 *|;

gfx/harfbuzz/src/hb-ot-shape-complex-myanmar.cc

@@ -119,6 +119,7 @@ override_features_myanmar (hb_ot_shape_planner_t *plan)
 
 enum syllable_type_t {
   consonant_syllable,
+  punctuation_cluster,
   broken_cluster,
   non_myanmar_cluster,
 };
@@ -143,7 +144,8 @@ enum myanmar_category_t {
   OT_VBlw = 27,
   OT_VPre = 28,
   OT_VPst = 29,
-  OT_VS   = 30 /* Variation selectors */
+  OT_VS   = 30, /* Variation selectors */
+  OT_P    = 31  /* Punctuation */
 };
 
 
@@ -186,6 +188,10 @@ set_myanmar_properties (hb_glyph_info_t &info)
 
   switch (u)
   {
+    case 0x104E:
+      cat = (indic_category_t) OT_C; /* The spec says C, IndicSyllableCategory doesn't have. */
+      break;
+
     case 0x002D: case 0x00A0: case 0x00D7: case 0x2012:
     case 0x2013: case 0x2014: case 0x2015: case 0x2022:
     case 0x25CC: case 0x25FB: case 0x25FC: case 0x25FD:
@@ -243,6 +249,10 @@ set_myanmar_properties (hb_glyph_info_t &info)
     case 0x108F: case 0x109A: case 0x109B: case 0x109C:
       cat = (indic_category_t) OT_SM;
       break;
+
+    case 0x104A: case 0x104B:
+      cat = (indic_category_t) OT_P;
+      break;
   }
 
   if (cat == OT_M)
@@ -405,6 +415,16 @@ initial_reordering_broken_cluster (const hb_ot_shape_plan_t *plan,
   initial_reordering_consonant_syllable (plan, face, buffer, start, end);
 }
 
+static void
+initial_reordering_punctuation_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
+					hb_face_t *face HB_UNUSED,
+					hb_buffer_t *buffer HB_UNUSED,
+					unsigned int start HB_UNUSED, unsigned int end HB_UNUSED)
+{
+  /* Nothing to do right now.  If we ever switch to using the output
+   * buffer in the reordering process, we'd need to next_glyph() here. */
+}
+
 static void
 initial_reordering_non_myanmar_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
 					hb_face_t *face HB_UNUSED,
@@ -425,6 +445,7 @@ initial_reordering_syllable (const hb_ot_shape_plan_t *plan,
   syllable_type_t syllable_type = (syllable_type_t) (buffer->info[start].syllable() & 0x0F);
   switch (syllable_type) {
   case consonant_syllable:	initial_reordering_consonant_syllable  (plan, face, buffer, start, end); return;
+  case punctuation_cluster:	initial_reordering_punctuation_cluster (plan, face, buffer, start, end); return;
   case broken_cluster:		initial_reordering_broken_cluster      (plan, face, buffer, start, end); return;
   case non_myanmar_cluster:	initial_reordering_non_myanmar_cluster (plan, face, buffer, start, end); return;
   }

gfx/harfbuzz/src/hb-ot-shape-private.hh

@@ -66,7 +66,7 @@ struct hb_ot_shape_planner_t
   hb_ot_map_builder_t map;
 
   hb_ot_shape_planner_t (const hb_shape_plan_t *master_plan) :
-			 face (master_plan->face),
+			 face (master_plan->face_unsafe),
 			 props (master_plan->props),
 			 shaper (NULL),
 			 map (face, &props) {}

gfx/harfbuzz/src/hb-ot-tag.cc

@@ -167,9 +167,12 @@ typedef struct {
  *
  * Generated by intersecting the OpenType language tag list from
  * Draft OpenType 1.5 spec, with with the ISO 639-3 codes from
- * 2008/08/04, matching on name, and finally adjusted manually.
+ * 2008-08-04, matching on name, and finally adjusted manually.
  *
- * Updated on 2012/12/07 with more research into remaining codes.
+ * Updated on 2012-12-07 with more research into remaining codes.
+ *
+ * Updated on 2013-11-23 based on usage in SIL and Microsoft fonts,
+ * the new proposal from Microsoft, and latest ISO 639-3 names.
  *
  * Some items still missing.  Those are commented out at the end.
  * Keep sorted for bsearch.
@@ -179,57 +182,90 @@ static const LangTag ot_languages[] = {
   {"aa",	HB_TAG('A','F','R',' ')},	/* Afar */
   {"ab",	HB_TAG('A','B','K',' ')},	/* Abkhazian */
   {"abq",	HB_TAG('A','B','A',' ')},	/* Abaza */
+  {"ach",	HB_TAG('A','C','H',' ')},	/* Acoli */
   {"ada",	HB_TAG('D','N','G',' ')},	/* Dangme */
   {"ady",	HB_TAG('A','D','Y',' ')},	/* Adyghe */
   {"af",	HB_TAG('A','F','K',' ')},	/* Afrikaans */
   {"aii",	HB_TAG('S','W','A',' ')},	/* Swadaya Aramaic */
+  {"aio",	HB_TAG('A','I','O',' ')},	/* Aiton */
   {"aiw",	HB_TAG('A','R','I',' ')},	/* Aari */
+  {"ak",	HB_TAG('T','W','I',' ')},	/* Akan [macrolanguage] */
   {"alt",	HB_TAG('A','L','T',' ')},	/* [Southern] Altai */
   {"am",	HB_TAG('A','M','H',' ')},	/* Amharic */
   {"amf",	HB_TAG('H','B','N',' ')},	/* Hammer-Banna */
-  {"ar",	HB_TAG('A','R','A',' ')},	/* Arabic */
+  {"an",	HB_TAG('A','R','G',' ')},	/* Aragonese */
+  {"ang",	HB_TAG('A','N','G',' ')},	/* Old English (ca. 450-1100) */
+  {"ar",	HB_TAG('A','R','A',' ')},	/* Arabic [macrolanguage] */
+  {"arb",	HB_TAG('A','R','A',' ')},	/* Standard Arabic */
   {"arn",	HB_TAG('M','A','P',' ')},	/* Mapudungun */
+  {"ary",	HB_TAG('M','O','R',' ')},	/* Moroccan Arabic */
   {"as",	HB_TAG('A','S','M',' ')},	/* Assamese */
+  {"ast",	HB_TAG('A','S','T',' ')},	/* Asturian/Asturleonese/Bable/Leonese */
   {"ath",	HB_TAG('A','T','H',' ')},	/* Athapaskan [family] */
   {"atv",	HB_TAG('A','L','T',' ')},	/* [Northern] Altai */
   {"av",	HB_TAG('A','V','R',' ')},	/* Avaric */
   {"awa",	HB_TAG('A','W','A',' ')},	/* Awadhi */
-  {"ay",	HB_TAG('A','Y','M',' ')},	/* Aymara */
+  {"ay",	HB_TAG('A','Y','M',' ')},	/* Aymara [macrolanguage] */
-  {"az",	HB_TAG('A','Z','E',' ')},	/* Azerbaijani */
+  {"az",	HB_TAG('A','Z','E',' ')},	/* Azerbaijani [macrolanguage] */
+  {"azb",	HB_TAG('A','Z','B',' ')},	/* South Azerbaijani */
+  {"azj",	HB_TAG('A','Z','E',' ')},	/* North Azerbaijani */
   {"ba",	HB_TAG('B','S','H',' ')},	/* Bashkir */
   {"bai",	HB_TAG('B','M','L',' ')},	/* Bamileke [family] */
-  {"bal",	HB_TAG('B','L','I',' ')},	/* Baluchi */
+  {"bal",	HB_TAG('B','L','I',' ')},	/* Baluchi [macrolangauge] */
-  {"bci",	HB_TAG('B','A','U',' ')},	/* Baule */
+  {"ban",	HB_TAG('B','A','N',' ')},	/* Balinese */
+  {"bar",	HB_TAG('B','A','R',' ')},	/* Bavarian */
+  {"bbc",	HB_TAG('B','B','C',' ')},	/* Batak Toba */
+  {"bci",	HB_TAG('B','A','U',' ')},	/* Baoulé */
+  {"bcl",	HB_TAG('B','I','K',' ')},	/* Central Bikol */
   {"bcq",	HB_TAG('B','C','H',' ')},	/* Bench */
-  {"be",	HB_TAG('B','E','L',' ')},  	/* Belarussian */
+  {"be",	HB_TAG('B','E','L',' ')},  	/* Belarusian */
   {"bem",	HB_TAG('B','E','M',' ')},	/* Bemba (Zambia) */
   {"ber",	HB_TAG('B','E','R',' ')},  	/* Berber [family] */
   {"bfq",	HB_TAG('B','A','D',' ')},	/* Badaga */
   {"bft",	HB_TAG('B','L','T',' ')},	/* Balti */
   {"bfy",	HB_TAG('B','A','G',' ')},	/* Baghelkhandi */
   {"bg",	HB_TAG('B','G','R',' ')},	/* Bulgarian */
+  {"bgc",	HB_TAG('B','G','C',' ')},	/* Haryanvi */
+  {"bgq",	HB_TAG('B','G','Q',' ')},	/* Bagri */
   {"bhb",	HB_TAG('B','H','I',' ')},	/* Bhili */
+  {"bhk",	HB_TAG('B','I','K',' ')},	/* Albay Bicolano (retired code) */
   {"bho",	HB_TAG('B','H','O',' ')},	/* Bhojpuri */
-  {"bik",	HB_TAG('B','I','K',' ')},	/* Bikol */
+  {"bi",	HB_TAG('B','I','S',' ')},	/* Bislama */
+  {"bik",	HB_TAG('B','I','K',' ')},	/* Bikol [macrolanguage] */
   {"bin",	HB_TAG('E','D','O',' ')},	/* Bini */
+  {"bjj",	HB_TAG('B','J','J',' ')},	/* Kanauji */
   {"bjt",	HB_TAG('B','L','N',' ')},	/* Balanta-Ganja */
   {"bla",	HB_TAG('B','K','F',' ')},	/* Blackfoot */
   {"ble",	HB_TAG('B','L','N',' ')},	/* Balanta-Kentohe */
+  {"blk",	HB_TAG('B','L','K',' ')},	/* Pa'O/Pa'o Karen */
+  {"bln",	HB_TAG('B','I','K',' ')},	/* Southern Catanduanes Bikol */
   {"bm",	HB_TAG('B','M','B',' ')},	/* Bambara */
   {"bn",	HB_TAG('B','E','N',' ')},	/* Bengali */
   {"bo",	HB_TAG('T','I','B',' ')},	/* Tibetan */
+  {"bpy",	HB_TAG('B','P','Y',' ')},	/* Bishnupriya */
+  {"bqi",	HB_TAG('L','R','C',' ')},	/* Bakhtiari */
   {"br",	HB_TAG('B','R','E',' ')},	/* Breton */
   {"bra",	HB_TAG('B','R','I',' ')},	/* Braj Bhasha */
   {"brh",	HB_TAG('B','R','H',' ')},	/* Brahui */
+  {"brx",	HB_TAG('B','R','X',' ')},	/* Bodo (India) */
   {"bs",	HB_TAG('B','O','S',' ')},	/* Bosnian */
   {"btb",	HB_TAG('B','T','I',' ')},	/* Beti (Cameroon) */
+  {"bto",	HB_TAG('B','I','K',' ')},	/* Rinconada Bikol */
+  {"bts",	HB_TAG('B','T','S',' ')},	/* Batak Simalungun */
+  {"bug",	HB_TAG('B','U','G',' ')},	/* Buginese */
   {"bxr",	HB_TAG('R','B','U',' ')},	/* Russian Buriat */
   {"byn",	HB_TAG('B','I','L',' ')},	/* Bilen */
   {"ca",	HB_TAG('C','A','T',' ')},	/* Catalan */
+  {"cbk",	HB_TAG('C','B','K',' ')},	/* Chavacano */
   {"ce",	HB_TAG('C','H','E',' ')},	/* Chechen */
   {"ceb",	HB_TAG('C','E','B',' ')},	/* Cebuano */
+  {"cgg",	HB_TAG('C','G','G',' ')},	/* Chiga */
+  {"ch",	HB_TAG('C','H','A',' ')},	/* Chamorro */
+  {"cho",	HB_TAG('C','H','O',' ')},	/* Choctaw */
   {"chp",	HB_TAG('C','H','P',' ')},	/* Chipewyan */
   {"chr",	HB_TAG('C','H','R',' ')},	/* Cherokee */
+  {"chy",	HB_TAG('C','H','Y',' ')},	/* Cheyenne */
+  {"ckb",	HB_TAG('K','U','R',' ')},	/* Central Kurdish (Sorani) */
   {"ckt",	HB_TAG('C','H','K',' ')},	/* Chukchi */
   {"cop",	HB_TAG('C','O','P',' ')},	/* Coptic */
   {"cr",	HB_TAG('C','R','E',' ')},	/* Cree */
@@ -239,6 +275,9 @@ static const LangTag ot_languages[] = {
   {"crm",	HB_TAG('M','C','R',' ')},	/* Moose Cree */
   {"crx",	HB_TAG('C','R','R',' ')},	/* Carrier */
   {"cs",	HB_TAG('C','S','Y',' ')},	/* Czech */
+  {"csb",	HB_TAG('C','S','B',' ')},	/* Kashubian */
+  {"ctg",	HB_TAG('C','T','G',' ')},	/* Chittagonian */
+  {"cts",	HB_TAG('B','I','K',' ')},	/* Northern Catanduanes Bikol */
   {"cu",	HB_TAG('C','S','L',' ')},	/* Church Slavic */
   {"cv",	HB_TAG('C','H','U',' ')},	/* Chuvash */
   {"cwd",	HB_TAG('D','C','R',' ')},	/* Woods Cree */
@@ -247,34 +286,42 @@ static const LangTag ot_languages[] = {
   {"dap",	HB_TAG('N','I','S',' ')},	/* Nisi (India) */
   {"dar",	HB_TAG('D','A','R',' ')},	/* Dargwa */
   {"de",	HB_TAG('D','E','U',' ')},	/* German */
-  {"din",	HB_TAG('D','N','K',' ')},	/* Dinka */
+  {"dgo",	HB_TAG('D','G','O',' ')},	/* Dogri */
-  {"dje",	HB_TAG('D','J','R',' ')},	/* Djerma */
+  {"dhd",	HB_TAG('M','A','W',' ')},	/* Dhundari */
+  {"din",	HB_TAG('D','N','K',' ')},	/* Dinka [macrolanguage] */
+  {"diq",	HB_TAG('D','I','Q',' ')},	/* Dimli */
+  {"dje",	HB_TAG('D','J','R',' ')},	/* Zarma */
   {"dng",	HB_TAG('D','U','N',' ')},	/* Dungan */
-  {"doi",	HB_TAG('D','G','R',' ')},	/* Dogri */
+  {"doi",	HB_TAG('D','G','R',' ')},	/* Dogri [macrolanguage] */
   {"dsb",	HB_TAG('L','S','B',' ')},	/* Lower Sorbian */
-  {"dv",	HB_TAG('D','I','V',' ')},	/* Dhivehi */
+  {"dv",	HB_TAG('D','I','V',' ')},	/* Dhivehi/Divehi/Maldivian */
   {"dyu",	HB_TAG('J','U','L',' ')},	/* Jula */
   {"dz",	HB_TAG('D','Z','N',' ')},	/* Dzongkha */
   {"ee",	HB_TAG('E','W','E',' ')},	/* Ewe */
   {"efi",	HB_TAG('E','F','I',' ')},	/* Efik */
+  {"ekk",	HB_TAG('E','T','I',' ')},	/* Standard Estonian */
   {"el",	HB_TAG('E','L','L',' ')},	/* Modern Greek (1453-) */
+  {"emk",	HB_TAG('M','N','K',' ')},	/* Eastern Maninkakan */
   {"en",	HB_TAG('E','N','G',' ')},	/* English */
   {"eo",	HB_TAG('N','T','O',' ')},	/* Esperanto */
   {"eot",	HB_TAG('B','T','I',' ')},	/* Beti (Côte d'Ivoire) */
   {"es",	HB_TAG('E','S','P',' ')},	/* Spanish */
-  {"et",	HB_TAG('E','T','I',' ')},	/* Estonian */
+  {"et",	HB_TAG('E','T','I',' ')},	/* Estonian [macrolanguage] */
   {"eu",	HB_TAG('E','U','Q',' ')},	/* Basque */
   {"eve",	HB_TAG('E','V','N',' ')},	/* Even */
   {"evn",	HB_TAG('E','V','K',' ')},	/* Evenki */
-  {"fa",	HB_TAG('F','A','R',' ')},	/* Persian */
+  {"fa",	HB_TAG('F','A','R',' ')},	/* Persian [macrolanguage] */
-  {"ff",	HB_TAG('F','U','L',' ')},	/* Fulah */
+  {"ff",	HB_TAG('F','U','L',' ')},	/* Fulah [macrolanguage] */
   {"fi",	HB_TAG('F','I','N',' ')},	/* Finnish */
   {"fil",	HB_TAG('P','I','L',' ')},	/* Filipino */
   {"fj",	HB_TAG('F','J','I',' ')},	/* Fijian */
   {"fo",	HB_TAG('F','O','S',' ')},	/* Faroese */
   {"fon",	HB_TAG('F','O','N',' ')},	/* Fon */
   {"fr",	HB_TAG('F','R','A',' ')},	/* French */
+  {"frc",	HB_TAG('F','R','C',' ')},	/* Cajun French */
+  {"frp",	HB_TAG('F','R','P',' ')},	/* Arpitan/Francoprovençal */
   {"fur",	HB_TAG('F','R','L',' ')},	/* Friulian */
+  {"fuv",	HB_TAG('F','U','V',' ')},	/* Nigerian Fulfulde */
   {"fy",	HB_TAG('F','R','I',' ')},	/* Western Frisian */
   {"ga",	HB_TAG('I','R','I',' ')},	/* Irish */
   {"gaa",	HB_TAG('G','A','D',' ')},	/* Ga */
@@ -282,113 +329,167 @@ static const LangTag ot_languages[] = {
   {"gbm",	HB_TAG('G','A','W',' ')},	/* Garhwali */
   {"gd",	HB_TAG('G','A','E',' ')},	/* Scottish Gaelic */
   {"gez",	HB_TAG('G','E','Z',' ')},	/* Ge'ez */
+  {"ggo",	HB_TAG('G','O','N',' ')},	/* Southern Gondi */
   {"gl",	HB_TAG('G','A','L',' ')},	/* Galician */
   {"gld",	HB_TAG('N','A','N',' ')},	/* Nanai */
-  {"gn",	HB_TAG('G','U','A',' ')},	/* Guarani */
+  {"glk",	HB_TAG('G','L','K',' ')},	/* Gilaki */
-  {"gon",	HB_TAG('G','O','N',' ')},	/* Gondi */
+  {"gn",	HB_TAG('G','U','A',' ')},	/* Guarani [macrolanguage] */
+  {"gno",	HB_TAG('G','O','N',' ')},	/* Northern Gondi */
+  {"gog",	HB_TAG('G','O','G',' ')},	/* Gogo */
+  {"gon",	HB_TAG('G','O','N',' ')},	/* Gondi [macrolanguage] */
   {"grt",	HB_TAG('G','R','O',' ')},	/* Garo */
   {"gru",	HB_TAG('S','O','G',' ')},	/* Sodo Gurage */
   {"gu",	HB_TAG('G','U','J',' ')},	/* Gujarati */
+  {"guc",	HB_TAG('G','U','C',' ')},	/* Wayuu */
   {"guk",	HB_TAG('G','M','Z',' ')},	/* Gumuz */
-  {"gv",	HB_TAG('M','N','X',' ')},	/* Manx Gaelic */
+/*{"guk",	HB_TAG('G','U','K',' ')},*/	/* Gumuz (in SIL fonts) */
+  {"guz",	HB_TAG('G','U','Z',' ')},	/* Ekegusii/Gusii */
+  {"gv",	HB_TAG('M','N','X',' ')},	/* Manx */
   {"ha",	HB_TAG('H','A','U',' ')},	/* Hausa */
   {"har",	HB_TAG('H','R','I',' ')},	/* Harari */
-  {"haw",	HB_TAG('H','A','W',' ')},  	/* Hawaiin */
+  {"haw",	HB_TAG('H','A','W',' ')},  	/* Hawaiian */
+  {"hay",	HB_TAG('H','A','Y',' ')},  	/* Haya */
+  {"haz",	HB_TAG('H','A','Z',' ')},  	/* Hazaragi */
   {"he",	HB_TAG('I','W','R',' ')},	/* Hebrew */
+  {"hz",	HB_TAG('H','E','R',' ')},	/* Herero */
   {"hi",	HB_TAG('H','I','N',' ')},	/* Hindi */
   {"hil",	HB_TAG('H','I','L',' ')},	/* Hiligaynon */
   {"hnd",	HB_TAG('H','N','D',' ')},	/* [Southern] Hindko */
   {"hne",	HB_TAG('C','H','H',' ')},	/* Chattisgarhi */
   {"hno",	HB_TAG('H','N','D',' ')},	/* [Northern] Hindko */
+  {"ho",	HB_TAG('H','M','O',' ')},	/* Hiri Motu */
   {"hoc",	HB_TAG('H','O',' ',' ')},	/* Ho */
   {"hoj",	HB_TAG('H','A','R',' ')},	/* Harauti */
   {"hr",	HB_TAG('H','R','V',' ')},	/* Croatian */
   {"hsb",	HB_TAG('U','S','B',' ')},	/* Upper Sorbian */
-  {"ht",	HB_TAG('H','A','I',' ')},	/* Haitian */
+  {"ht",	HB_TAG('H','A','I',' ')},	/* Haitian/Haitian Creole */
   {"hu",	HB_TAG('H','U','N',' ')},	/* Hungarian */
   {"hy",	HB_TAG('H','Y','E',' ')},	/* Armenian */
+  {"hz",	HB_TAG('H','E','R',' ')},	/* Herero */
+  {"ia",	HB_TAG('I','N','A',' ')},	/* Interlingua (International Auxiliary Language Association) */
+  {"ibb",	HB_TAG('I','B','B',' ')},	/* Ibibio */
   {"id",	HB_TAG('I','N','D',' ')},	/* Indonesian */
+  {"ie",	HB_TAG('I','L','E',' ')},	/* Interlingue/Occidental */
   {"ig",	HB_TAG('I','B','O',' ')},	/* Igbo */
   {"igb",	HB_TAG('E','B','I',' ')},	/* Ebira */
+  {"ijc",	HB_TAG('I','J','O',' ')},	/* Izon */
   {"ijo",	HB_TAG('I','J','O',' ')},	/* Ijo [family] */
+  {"ik",	HB_TAG('I','P','K',' ')},	/* Inupiaq [macrolanguage] */
   {"ilo",	HB_TAG('I','L','O',' ')},	/* Ilokano */
   {"inh",	HB_TAG('I','N','G',' ')},	/* Ingush */
+  {"io",	HB_TAG('I','D','O',' ')},	/* Ido */
   {"is",	HB_TAG('I','S','L',' ')},	/* Icelandic */
   {"it",	HB_TAG('I','T','A',' ')},	/* Italian */
-  {"iu",	HB_TAG('I','N','U',' ')},	/* Inuktitut */
+  {"iu",	HB_TAG('I','N','U',' ')},	/* Inuktitut [macrolanguage] */
   {"ja",	HB_TAG('J','A','N',' ')},	/* Japanese */
+  {"jam",	HB_TAG('J','A','M',' ')},	/* Jamaican Creole English */
+  {"jbo",	HB_TAG('J','B','O',' ')},	/* Lojban */
   {"jv",	HB_TAG('J','A','V',' ')},	/* Javanese */
   {"ka",	HB_TAG('K','A','T',' ')},	/* Georgian */
   {"kaa",	HB_TAG('K','R','K',' ')},	/* Karakalpak */
+  {"kab",	HB_TAG('K','A','B',' ')},	/* Kabyle */
   {"kam",	HB_TAG('K','M','B',' ')},	/* Kamba (Kenya) */
   {"kar",	HB_TAG('K','R','N',' ')},	/* Karen [family] */
   {"kbd",	HB_TAG('K','A','B',' ')},	/* Kabardian */
+  {"kde",	HB_TAG('K','D','E',' ')},	/* Makonde */
   {"kdr",	HB_TAG('K','R','M',' ')},	/* Karaim */
   {"kdt",	HB_TAG('K','U','Y',' ')},	/* Kuy */
   {"kex",	HB_TAG('K','K','N',' ')},	/* Kokni */
   {"kfr",	HB_TAG('K','A','C',' ')},	/* Kachchi */
   {"kfy",	HB_TAG('K','M','N',' ')},	/* Kumaoni */
+  {"kg",	HB_TAG('K','O','N',' ')},	/* Kongo [macrolanguage] */
   {"kha",	HB_TAG('K','S','I',' ')},	/* Khasi */
-  {"khb",	HB_TAG('X','B','D',' ')},	/* Tai Lue */
+  {"khb",	HB_TAG('X','B','D',' ')},	/* Lü */
+  {"kht",	HB_TAG('K','H','N',' ')},	/* Khamti (Microsoft fonts) */
+/*{"kht",	HB_TAG('K','H','T',' ')},*/	/* Khamti (OpenType spec and SIL fonts) */
   {"khw",	HB_TAG('K','H','W',' ')},	/* Khowar */
-  {"ki",	HB_TAG('K','I','K',' ')},	/* Kikuyu */
+  {"ki",	HB_TAG('K','I','K',' ')},	/* Gikuyu/Kikuyu */
+  {"kj",	HB_TAG('K','U','A',' ')},	/* Kuanyama/Kwanyama */
   {"kjh",	HB_TAG('K','H','A',' ')},	/* Khakass */
+  {"kjp",	HB_TAG('K','J','P',' ')},	/* Pwo Eastern Karen */
   {"kk",	HB_TAG('K','A','Z',' ')},	/* Kazakh */
   {"kl",	HB_TAG('G','R','N',' ')},	/* Kalaallisut */
   {"kln",	HB_TAG('K','A','L',' ')},	/* Kalenjin */
   {"km",	HB_TAG('K','H','M',' ')},	/* Central Khmer */
-  {"kmb",	HB_TAG('M','B','N',' ')},	/* [North] Mbundu */
+  {"kmb",	HB_TAG('M','B','N',' ')},	/* Kimbundu */
   {"kmw",	HB_TAG('K','M','O',' ')},	/* Komo (Democratic Republic of Congo) */
   {"kn",	HB_TAG('K','A','N',' ')},	/* Kannada */
+  {"knn",	HB_TAG('K','O','K',' ')},	/* Konkani */
   {"ko",	HB_TAG('K','O','R',' ')},	/* Korean */
   {"koi",	HB_TAG('K','O','P',' ')},	/* Komi-Permyak */
-  {"kok",	HB_TAG('K','O','K',' ')},	/* Konkani */
+  {"kok",	HB_TAG('K','O','K',' ')},	/* Konkani [macrolanguage] */
-  {"kpe",	HB_TAG('K','P','L',' ')},	/* Kpelle */
+  {"kpe",	HB_TAG('K','P','L',' ')},	/* Kpelle [macrolanguage] */
   {"kpv",	HB_TAG('K','O','Z',' ')},	/* Komi-Zyrian */
   {"kpy",	HB_TAG('K','Y','K',' ')},	/* Koryak */
   {"kqy",	HB_TAG('K','R','T',' ')},	/* Koorete */
-  {"kr",	HB_TAG('K','N','R',' ')},	/* Kanuri */
+  {"kr",	HB_TAG('K','N','R',' ')},	/* Kanuri [macrolanguage] */
   {"kri",	HB_TAG('K','R','I',' ')},	/* Krio */
   {"krl",	HB_TAG('K','R','L',' ')},	/* Karelian */
   {"kru",	HB_TAG('K','U','U',' ')},	/* Kurukh */
   {"ks",	HB_TAG('K','S','H',' ')},	/* Kashmiri */
-  {"ku",	HB_TAG('K','U','R',' ')},	/* Kurdish */
+  {"ksh",	HB_TAG('K','S','H',' ')},	/* Kölsch */
+/*{"ksw",	HB_TAG('K','R','N',' ')},*/	/* S'gaw Karen (Microsoft fonts?) */
+  {"ksw",	HB_TAG('K','S','W',' ')},	/* S'gaw Karen (OpenType spec and SIL fonts) */
+  {"ku",	HB_TAG('K','U','R',' ')},	/* Kurdish [macrolanguage] */
   {"kum",	HB_TAG('K','U','M',' ')},	/* Kumyk */
+  {"kv",	HB_TAG('K','O','M',' ')},	/* Komi [macrolanguage] */
   {"kvd",	HB_TAG('K','U','I',' ')},	/* Kui (Indonesia) */
+  {"kw",	HB_TAG('C','O','R',' ')},	/* Cornish */
   {"kxc",	HB_TAG('K','M','S',' ')},	/* Komso */
   {"kxu",	HB_TAG('K','U','I',' ')},	/* Kui (India) */
-  {"ky",	HB_TAG('K','I','R',' ')},	/* Kirghiz */
+  {"ky",	HB_TAG('K','I','R',' ')},	/* Kirghiz/Kyrgyz */
+  {"kyu",	HB_TAG('K','Y','U',' ')},	/* Western Kayah */
   {"la",	HB_TAG('L','A','T',' ')},	/* Latin */
   {"lad",	HB_TAG('J','U','D',' ')},	/* Ladino */
   {"lb",	HB_TAG('L','T','Z',' ')},	/* Luxembourgish */
   {"lbe",	HB_TAG('L','A','K',' ')},	/* Lak */
   {"lbj",	HB_TAG('L','D','K',' ')},	/* Ladakhi */
   {"lez",	HB_TAG('L','E','Z',' ')},	/* Lezgi */
-  {"lg",	HB_TAG('L','U','G',' ')},	/* Luganda */
+  {"lg",	HB_TAG('L','U','G',' ')},	/* Ganda */
+  {"li",	HB_TAG('L','I','M',' ')},	/* Limburgan/Limburger/Limburgish */
   {"lif",	HB_TAG('L','M','B',' ')},	/* Limbu */
+  {"lij",	HB_TAG('L','I','J',' ')},	/* Ligurian */
+  {"lis",	HB_TAG('L','I','S',' ')},	/* Lisu */
+  {"ljp",	HB_TAG('L','J','P',' ')},	/* Lampung Api */
+  {"lki",	HB_TAG('L','K','I',' ')},	/* Laki */
   {"lld",	HB_TAG('L','A','D',' ')},	/* Ladin */
   {"lmn",	HB_TAG('L','A','M',' ')},	/* Lambani */
+  {"lmo",	HB_TAG('L','M','O',' ')},	/* Lombard */
   {"ln",	HB_TAG('L','I','N',' ')},	/* Lingala */
   {"lo",	HB_TAG('L','A','O',' ')},	/* Lao */
+  {"lrc",	HB_TAG('L','R','C',' ')},	/* Northern Luri */
   {"lt",	HB_TAG('L','T','H',' ')},	/* Lithuanian */
   {"lu",	HB_TAG('L','U','B',' ')},	/* Luba-Katanga */
   {"lua",	HB_TAG('L','U','B',' ')},	/* Luba-Kasai */
   {"luo",	HB_TAG('L','U','O',' ')},	/* Luo (Kenya and Tanzania) */
   {"lus",	HB_TAG('M','I','Z',' ')},	/* Mizo */
-  {"luy",	HB_TAG('L','U','H',' ')},	/* Luhya [macrolanguage] */
+  {"luy",	HB_TAG('L','U','H',' ')},	/* Luyia/Oluluyia [macrolanguage] */
+  {"luz",	HB_TAG('L','R','C',' ')},	/* Southern Luri */
   {"lv",	HB_TAG('L','V','I',' ')},	/* Latvian */
   {"lzz",	HB_TAG('L','A','Z',' ')},	/* Laz */
+  {"mad",	HB_TAG('M','A','D',' ')},	/* Madurese */
+  {"mag",	HB_TAG('M','A','G',' ')},	/* Magahi */
   {"mai",	HB_TAG('M','T','H',' ')},	/* Maithili */
+  {"mak",	HB_TAG('M','K','R',' ')},	/* Makasar */
+  {"man",	HB_TAG('M','N','K',' ')},	/* Manding/Mandingo [macrolanguage] */
   {"mdc",	HB_TAG('M','L','E',' ')},	/* Male (Papua New Guinea) */
   {"mdf",	HB_TAG('M','O','K',' ')},	/* Moksha */
+  {"mdr",	HB_TAG('M','D','R',' ')},	/* Mandar */
   {"mdy",	HB_TAG('M','L','E',' ')},	/* Male (Ethiopia) */
   {"men",	HB_TAG('M','D','E',' ')},	/* Mende (Sierra Leone) */
-  {"mg",	HB_TAG('M','L','G',' ')},	/* Malagasy */
+  {"mer",	HB_TAG('M','E','R',' ')},	/* Meru */
+  {"mfe",	HB_TAG('M','F','E',' ')},	/* Morisyen */
+  {"mg",	HB_TAG('M','L','G',' ')},	/* Malagasy [macrolanguage] */
+  {"mh",	HB_TAG('M','A','H',' ')},	/* Marshallese */
   {"mhr",	HB_TAG('L','M','A',' ')},	/* Low Mari */
   {"mi",	HB_TAG('M','R','I',' ')},	/* Maori */
+  {"min",	HB_TAG('M','I','N',' ')},	/* Minangkabau */
   {"mk",	HB_TAG('M','K','D',' ')},	/* Macedonian */
+  {"mku",	HB_TAG('M','N','K',' ')},	/* Konyanka Maninka */
+  {"mkw",	HB_TAG('M','K','W',' ')},	/* Kituba (Congo) */
   {"ml",	HB_TAG('M','L','R',' ')},	/* Malayalam */
-  {"mn",	HB_TAG('M','N','G',' ')},	/* Mongolian */
+  {"mlq",	HB_TAG('M','N','K',' ')},	/* Western Maninkakan */
+  {"mn",	HB_TAG('M','N','G',' ')},	/* Mongolian [macrolanguage] */
   {"mnc",	HB_TAG('M','C','H',' ')},	/* Manchu */
   {"mni",	HB_TAG('M','N','I',' ')},	/* Manipuri */
   {"mnk",	HB_TAG('M','N','D',' ')},	/* Mandinka */
@@ -396,72 +497,119 @@ static const LangTag ot_languages[] = {
   {"mnw",	HB_TAG('M','O','N',' ')},	/* Mon */
   {"mo",	HB_TAG('M','O','L',' ')},	/* Moldavian */
   {"moh",	HB_TAG('M','O','H',' ')},	/* Mohawk */
+  {"mos",	HB_TAG('M','O','S',' ')},	/* Mossi */
   {"mpe",	HB_TAG('M','A','J',' ')},	/* Majang */
   {"mr",	HB_TAG('M','A','R',' ')},	/* Marathi */
   {"mrj",	HB_TAG('H','M','A',' ')},	/* High Mari */
-  {"ms",	HB_TAG('M','L','Y',' ')},	/* Malay */
+  {"ms",	HB_TAG('M','L','Y',' ')},	/* Malay [macrolanguage] */
+  {"msc",	HB_TAG('M','N','K',' ')},	/* Sankaran Maninka */
   {"mt",	HB_TAG('M','T','S',' ')},	/* Maltese */
-  {"mwr",	HB_TAG('M','A','W',' ')},	/* Marwari */
+  {"mtr",	HB_TAG('M','A','W',' ')},	/* Mewari */
+  {"mus",	HB_TAG('M','U','S',' ')},	/* Creek */
+  {"mve",	HB_TAG('M','A','W',' ')},	/* Marwari (Pakistan) */
+  {"mwk",	HB_TAG('M','N','K',' ')},	/* Kita Maninkakan */
+  {"mwl",	HB_TAG('M','W','L',' ')},	/* Mirandese */
+  {"mwr",	HB_TAG('M','A','W',' ')},	/* Marwari [macrolanguage] */
+  {"mww",	HB_TAG('M','W','W',' ')},	/* Hmong Daw */
   {"my",	HB_TAG('B','R','M',' ')},	/* Burmese */
   {"mym",	HB_TAG('M','E','N',' ')},	/* Me'en */
+  {"myq",	HB_TAG('M','N','K',' ')},	/* Forest Maninka (retired code) */
   {"myv",	HB_TAG('E','R','Z',' ')},	/* Erzya */
+  {"mzn",	HB_TAG('M','Z','N',' ')},	/* Mazanderani */
+  {"na",	HB_TAG('N','A','U',' ')},	/* Nauru */
   {"nag",	HB_TAG('N','A','G',' ')},	/* Naga-Assamese */
+  {"nah",	HB_TAG('N','A','H',' ')},	/* Nahuatl [family] */
+  {"nap",	HB_TAG('N','A','P',' ')},	/* Neapolitan */
   {"nb",	HB_TAG('N','O','R',' ')},	/* Norwegian Bokmål */
   {"nco",	HB_TAG('S','I','B',' ')},	/* Sibe */
   {"nd",	HB_TAG('N','D','B',' ')},	/* [North] Ndebele */
+  {"ndc",	HB_TAG('N','D','C',' ')},	/* Ndau */
+  {"nds",	HB_TAG('N','D','S',' ')},	/* Low German/Low Saxon */
   {"ne",	HB_TAG('N','E','P',' ')},	/* Nepali */
   {"new",	HB_TAG('N','E','W',' ')},	/* Newari */
   {"ng",	HB_TAG('N','D','G',' ')},	/* Ndonga */
+  {"nga",	HB_TAG('N','G','A',' ')},	/* Ngabaka */
   {"ngl",	HB_TAG('L','M','W',' ')},	/* Lomwe */
   {"niu",	HB_TAG('N','I','U',' ')},	/* Niuean */
   {"niv",	HB_TAG('G','I','L',' ')},	/* Gilyak */
   {"nl",	HB_TAG('N','L','D',' ')},	/* Dutch */
   {"nn",	HB_TAG('N','Y','N',' ')},	/* Norwegian Nynorsk */
-  {"no",	HB_TAG('N','O','R',' ')},	/* Norwegian (deprecated) */
+  {"no",	HB_TAG('N','O','R',' ')},	/* Norwegian [macrolanguage] */
-  {"nod",	HB_TAG('N','T','A',' ')},	/* Northern Tai */
+  {"nod",	HB_TAG('N','T','A',' ')},	/* Northern Thai */
+  {"noe",	HB_TAG('N','O','E',' ')},	/* Nimadi */
   {"nog",	HB_TAG('N','O','G',' ')},	/* Nogai */
+  {"nov",	HB_TAG('N','O','V',' ')},	/* Novial */
   {"nqo",	HB_TAG('N','K','O',' ')},	/* N'Ko */
   {"nr",	HB_TAG('N','D','B',' ')},	/* [South] Ndebele */
   {"nsk",	HB_TAG('N','A','S',' ')},	/* Naskapi */
   {"nso",	HB_TAG('S','O','T',' ')},	/* [Northern] Sotho */
-  {"ny",	HB_TAG('C','H','I',' ')},	/* Nyanja */
+  {"ny",	HB_TAG('C','H','I',' ')},	/* Chewa/Chichwa/Nyanja */
-  {"nyn",	HB_TAG('N','K','L',' ')},	/* Nkole */
+  {"nym",	HB_TAG('N','Y','M',' ')},	/* Nyamwezi */
+  {"nyn",	HB_TAG('N','K','L',' ')},	/* Nyankole */
   {"oc",	HB_TAG('O','C','I',' ')},	/* Occitan (post 1500) */
-  {"oj",	HB_TAG('O','J','B',' ')},	/* Ojibwa */
+  {"oj",	HB_TAG('O','J','B',' ')},	/* Ojibwa [macrolanguage] */
   {"ojs",	HB_TAG('O','C','R',' ')},	/* Oji-Cree */
-  {"om",	HB_TAG('O','R','O',' ')},	/* Oromo */
+  {"om",	HB_TAG('O','R','O',' ')},	/* Oromo [macrolanguage] */
   {"or",	HB_TAG('O','R','I',' ')},	/* Oriya */
   {"os",	HB_TAG('O','S','S',' ')},	/* Ossetian */
   {"pa",	HB_TAG('P','A','N',' ')},	/* Panjabi */
+  {"pag",	HB_TAG('P','A','G',' ')},	/* Pangasinan */
+  {"pam",	HB_TAG('P','A','M',' ')},	/* Kapampangan/Pampanga */
+  {"pap",	HB_TAG('P','A','P',' ')},	/* Papiamento */
+  {"pcc",	HB_TAG('P','C','C',' ')},	/* Bouyei */
+  {"pcd",	HB_TAG('P','C','D',' ')},	/* Picard */
   {"pce",	HB_TAG('P','L','G',' ')},	/* [Ruching] Palaung */
+  {"pdc",	HB_TAG('P','D','C',' ')},	/* Pennsylvania German */
+  {"pes",	HB_TAG('F','A','R',' ')},	/* Iranian Persian */
+  {"phk",	HB_TAG('P','H','K',' ')},	/* Phake */
   {"pi",	HB_TAG('P','A','L',' ')},	/* Pali */
+  {"pih",	HB_TAG('P','I','H',' ')},	/* Pitcairn-Norfolk */
   {"pl",	HB_TAG('P','L','K',' ')},	/* Polish */
   {"pll",	HB_TAG('P','L','G',' ')},	/* [Shwe] Palaung */
   {"plp",	HB_TAG('P','A','P',' ')},	/* Palpa */
-  {"prs",	HB_TAG('D','R','I',' ')},	/* Dari */
+  {"pms",	HB_TAG('P','M','S',' ')},	/* Piemontese */
-  {"ps",	HB_TAG('P','A','S',' ')},	/* Pushto */
+  {"pnb",	HB_TAG('P','N','B',' ')},	/* Western Panjabi */
+  {"prs",	HB_TAG('D','R','I',' ')},	/* Afghan Persian/Dari */
+  {"ps",	HB_TAG('P','A','S',' ')},	/* Pashto/Pushto [macrolanguage] */
   {"pt",	HB_TAG('P','T','G',' ')},	/* Portuguese */
-  {"raj",	HB_TAG('R','A','J',' ')},	/* Rajasthani */
+  {"pwo",	HB_TAG('P','W','O',' ')},	/* Pwo Western Karen */
-  {"rbb",	HB_TAG('P','L','G',' ')},	/* [Rumai] Palaung */
+  {"qu",	HB_TAG('Q','U','Z',' ')},	/* Quechua [macrolanguage] */
+  {"quc",	HB_TAG('Q','U','C',' ')},	/* K'iche'/Quiché */
+  {"quz",	HB_TAG('Q','U','Z',' ')},	/* Cusco Quechua */
+  {"raj",	HB_TAG('R','A','J',' ')},	/* Rajasthani [macrolanguage] */
+  {"rbb",	HB_TAG('P','L','G',' ')},	/* Rumai Palaung */
+  {"rej",	HB_TAG('R','E','J',' ')},	/* Rejang */
   {"ria",	HB_TAG('R','I','A',' ')},	/* Riang (India) */
   {"ril",	HB_TAG('R','I','A',' ')},	/* Riang (Myanmar) */
-  {"rki",	HB_TAG('A','R','K',' ')},	/* Arakanese */
+  {"rki",	HB_TAG('A','R','K',' ')},	/* Rakhine */
-  {"rm",	HB_TAG('R','M','S',' ')},	/* Rhaeto-Romanic */
+  {"rm",	HB_TAG('R','M','S',' ')},	/* Romansh */
+  {"rmy",	HB_TAG('R','M','Y',' ')},	/* Vlax Romani */
+  {"rn",	HB_TAG('R','U','N',' ')},	/* Rundi */
   {"ro",	HB_TAG('R','O','M',' ')},	/* Romanian */
-  {"rom",	HB_TAG('R','O','Y',' ')},	/* Romany */
+  {"rom",	HB_TAG('R','O','Y',' ')},	/* Romany [macrolanguage] */
   {"ru",	HB_TAG('R','U','S',' ')},	/* Russian */
   {"rue",	HB_TAG('R','S','Y',' ')},	/* Rusyn */
-  {"rw",	HB_TAG('R','U','A',' ')},	/* Ruanda */
+  {"rup",	HB_TAG('R','U','P',' ')},	/* Aromanian/Arumanian/Macedo-Romanian */
+  {"rw",	HB_TAG('R','U','A',' ')},	/* Kinyarwanda */
+  {"rwr",	HB_TAG('M','A','W',' ')},	/* Marwari (India) */
   {"sa",	HB_TAG('S','A','N',' ')},	/* Sanskrit */
   {"sah",	HB_TAG('Y','A','K',' ')},	/* Yakut */
+  {"sas",	HB_TAG('S','A','S',' ')},	/* Sasak */
   {"sat",	HB_TAG('S','A','T',' ')},	/* Santali */
   {"sck",	HB_TAG('S','A','D',' ')},	/* Sadri */
+  {"sc",	HB_TAG('S','R','D',' ')},	/* Sardinian [macrolanguage] */
+  {"scn",	HB_TAG('S','C','N',' ')},	/* Sicilian */
+  {"sco",	HB_TAG('S','C','O',' ')},	/* Scots */
   {"scs",	HB_TAG('S','L','A',' ')},	/* [North] Slavey */
   {"sd",	HB_TAG('S','N','D',' ')},	/* Sindhi */
   {"se",	HB_TAG('N','S','M',' ')},	/* Northern Sami */
   {"seh",	HB_TAG('S','N','A',' ')},	/* Sena */
   {"sel",	HB_TAG('S','E','L',' ')},	/* Selkup */
   {"sg",	HB_TAG('S','G','O',' ')},	/* Sango */
+  {"sga",	HB_TAG('S','G','A',' ')},	/* Old Irish (to 900) */
+  {"sgs",	HB_TAG('S','G','S',' ')},	/* Samogitian */
+  {"sgw",	HB_TAG('C','H','G',' ')},	/* Sebat Bet Gurage */
+/*{"sgw",	HB_TAG('S','G','W',' ')},*/	/* Sebat Bet Gurage (in SIL fonts) */
   {"shn",	HB_TAG('S','H','N',' ')},	/* Shan */
   {"si",	HB_TAG('S','N','H',' ')},	/* Sinhala */
   {"sid",	HB_TAG('S','I','D',' ')},	/* Sidamo */
@@ -474,60 +622,98 @@ static const LangTag ot_languages[] = {
   {"smj",	HB_TAG('L','S','M',' ')},	/* Lule Sami */
   {"smn",	HB_TAG('I','S','M',' ')},	/* Inari Sami */
   {"sms",	HB_TAG('S','K','S',' ')},	/* Skolt Sami */
+  {"sn",	HB_TAG('S','N','A',' ')},	/* Shona */
   {"snk",	HB_TAG('S','N','K',' ')},	/* Soninke */
   {"so",	HB_TAG('S','M','L',' ')},	/* Somali */
-  {"sq",	HB_TAG('S','Q','I',' ')},	/* Albanian */
+  {"sop",	HB_TAG('S','O','P',' ')},	/* Songe */
+  {"sq",	HB_TAG('S','Q','I',' ')},	/* Albanian [macrolanguage] */
   {"sr",	HB_TAG('S','R','B',' ')},	/* Serbian */
   {"srr",	HB_TAG('S','R','R',' ')},	/* Serer */
-  {"ss",	HB_TAG('S','W','Z',' ')},	/* Swazi */
+  {"ss",	HB_TAG('S','W','Z',' ')},	/* Swati */
   {"st",	HB_TAG('S','O','T',' ')},	/* [Southern] Sotho */
+  {"stq",	HB_TAG('S','T','Q',' ')},	/* Saterfriesisch */
+  {"stv",	HB_TAG('S','I','G',' ')},	/* Silt'e */
+  {"su",	HB_TAG('S','U','N',' ')},	/* Sundanese */
+  {"suk",	HB_TAG('S','U','K',' ')},	/* Sukama */
   {"suq",	HB_TAG('S','U','R',' ')},	/* Suri */
   {"sv",	HB_TAG('S','V','E',' ')},	/* Swedish */
   {"sva",	HB_TAG('S','V','A',' ')},	/* Svan */
-  {"sw",	HB_TAG('S','W','K',' ')},	/* Swahili */
+  {"sw",	HB_TAG('S','W','K',' ')},	/* Swahili [macrolanguage] */
   {"swb",	HB_TAG('C','M','R',' ')},	/* Comorian */
-  {"syr",	HB_TAG('S','Y','R',' ')},	/* Syriac */
+  {"swh",	HB_TAG('S','W','K',' ')},	/* Kiswahili/Swahili */
+  {"swv",	HB_TAG('M','A','W',' ')},	/* Shekhawati */
+  {"sxu",	HB_TAG('S','X','U',' ')},	/* Upper Saxon */
+  {"syl",	HB_TAG('S','Y','L',' ')},	/* Sylheti */
+  {"syr",	HB_TAG('S','Y','R',' ')},	/* Syriac [macrolanguage] */
+  {"szl",	HB_TAG('S','Z','L',' ')},	/* Silesian */
   {"ta",	HB_TAG('T','A','M',' ')},	/* Tamil */
   {"tab",	HB_TAG('T','A','B',' ')},	/* Tabasaran */
   {"tcy",	HB_TAG('T','U','L',' ')},	/* Tulu */
+  {"tdd",	HB_TAG('T','D','D',' ')},	/* Tai Nüa */
   {"te",	HB_TAG('T','E','L',' ')},	/* Telugu */
   {"tem",	HB_TAG('T','M','N',' ')},	/* Temne */
+  {"tet",	HB_TAG('T','E','T',' ')},	/* Tetum */
   {"tg",	HB_TAG('T','A','J',' ')},	/* Tajik */
   {"th",	HB_TAG('T','H','A',' ')},	/* Thai */
   {"ti",	HB_TAG('T','G','Y',' ')},	/* Tigrinya */
   {"tig",	HB_TAG('T','G','R',' ')},	/* Tigre */
+  {"tiv",	HB_TAG('T','I','V',' ')},	/* Tiv */
   {"tk",	HB_TAG('T','K','M',' ')},	/* Turkmen */
+  {"tl",	HB_TAG('T','G','L',' ')},	/* Tagalog */
+  {"tmh",	HB_TAG('t','m','h',' ')},	/* Tamashek [macrolanguage] */
   {"tn",	HB_TAG('T','N','A',' ')},	/* Tswana */
   {"to",	HB_TAG('T','G','N',' ')},	/* Tonga (Tonga Islands) */
+  {"tpi",	HB_TAG('T','P','I',' ')},	/* Tok Pisin */
   {"tr",	HB_TAG('T','R','K',' ')},	/* Turkish */
   {"tru",	HB_TAG('T','U','A',' ')},	/* Turoyo Aramaic */
   {"ts",	HB_TAG('T','S','G',' ')},	/* Tsonga */
   {"tt",	HB_TAG('T','A','T',' ')},	/* Tatar */
+  {"tum",	HB_TAG('T','U','M',' ')},	/* Tumbuka */
   {"tw",	HB_TAG('T','W','I',' ')},	/* Twi */
   {"ty",	HB_TAG('T','H','T',' ')},	/* Tahitian */
   {"tyv",	HB_TAG('T','U','V',' ')},	/* Tuvin */
+  {"tyz",	HB_TAG('T','Y','Z',' ')},	/* Tày */
+  {"tzm",	HB_TAG('T','Z','M',' ')},	/* Central Atlas Tamazight */
   {"udm",	HB_TAG('U','D','M',' ')},	/* Udmurt */
   {"ug",	HB_TAG('U','Y','G',' ')},	/* Uighur */
   {"uk",	HB_TAG('U','K','R',' ')},	/* Ukrainian */
-  {"umb",	HB_TAG('M','B','N',' ')},	/* [South] Mbundu */
+  {"umb",	HB_TAG('U','M','B',' ')},	/* Umbundu */
   {"unr",	HB_TAG('M','U','N',' ')},	/* Mundari */
   {"ur",	HB_TAG('U','R','D',' ')},	/* Urdu */
-  {"uz",	HB_TAG('U','Z','B',' ')},	/* Uzbek */
+  {"uz",	HB_TAG('U','Z','B',' ')},	/* Uzbek [macrolanguage] */
+  {"uzn",	HB_TAG('U','Z','B',' ')},	/* Northern Uzbek */
+  {"uzs",	HB_TAG('U','Z','B',' ')},	/* Southern Uzbek */
   {"ve",	HB_TAG('V','E','N',' ')},	/* Venda */
+  {"vec",	HB_TAG('V','E','C',' ')},	/* Venetian */
+  {"vls",	HB_TAG('F','L','E',' ')},	/* Vlaams */
   {"vi",	HB_TAG('V','I','T',' ')},	/* Vietnamese */
-  {"vmw",	HB_TAG('M','A','K',' ')},	/* Makua */
+  {"vmw",	HB_TAG('M','A','K',' ')},	/* Makhuwa */
+  {"vo",	HB_TAG('V','O','L',' ')},	/* Volapük */
+  {"vro",	HB_TAG('V','R','O',' ')},	/* Võro */
+  {"wa",	HB_TAG('W','L','N',' ')},	/* Walloon */
+  {"war",	HB_TAG('W','A','R',' ')},	/* Waray (Philippines) */
   {"wbm",	HB_TAG('W','A',' ',' ')},	/* Wa */
   {"wbr",	HB_TAG('W','A','G',' ')},	/* Wagdi */
+  {"wle",	HB_TAG('S','I','G',' ')},	/* Wolane */
+  {"wry",	HB_TAG('M','A','W',' ')},	/* Merwari */
+  {"wtm",	HB_TAG('W','T','M',' ')},	/* Mewati */
   {"wo",	HB_TAG('W','L','F',' ')},	/* Wolof */
   {"xal",	HB_TAG('K','L','M',' ')},	/* Kalmyk */
   {"xh",	HB_TAG('X','H','S',' ')},	/* Xhosa */
+  {"xog",	HB_TAG('X','O','G',' ')},	/* Soga */
   {"xom",	HB_TAG('K','M','O',' ')},	/* Komo (Sudan) */
   {"xsl",	HB_TAG('S','S','L',' ')},	/* South Slavey */
-  {"yi",	HB_TAG('J','I','I',' ')},	/* Yiddish */
+  {"xst",	HB_TAG('S','I','G',' ')},	/* Silt'e (retired code) */
+  {"xwo",	HB_TAG('T','O','D',' ')},	/* Written Oirat (Todo) */
+  {"yao",	HB_TAG('Y','A','O',' ')},	/* Yao */
+  {"yi",	HB_TAG('J','I','I',' ')},	/* Yiddish [macrolanguage] */
   {"yo",	HB_TAG('Y','B','A',' ')},	/* Yoruba */
   {"yso",	HB_TAG('N','I','S',' ')},	/* Nisi (China) */
+  {"za",	HB_TAG('Z','H','A',' ')},	/* Chuang/Zhuang [macrolanguage] */
+  {"zea",	HB_TAG('Z','E','A',' ')},	/* Zeeuws */
   {"zne",	HB_TAG('Z','N','D',' ')},	/* Zande */
-  {"zu",	HB_TAG('Z','U','L',' ')} 	/* Zulu */
+  {"zu",	HB_TAG('Z','U','L',' ')}, 	/* Zulu */
+  {"zum",	HB_TAG('L','R','C',' ')}	/* Kumzari */
 
   /* The corresponding languages IDs for the following IDs are unclear,
    * overlap, or are architecturally weird. Needs more research. */
@@ -536,13 +722,13 @@ static const LangTag ot_languages[] = {
 /*{"gsw?/gsw-FR?",	HB_TAG('A','L','S',' ')},*/	/* Alsatian */
 /*{"krc",	HB_TAG('B','A','L',' ')},*/	/* Balkar */
 /*{"??",	HB_TAG('B','C','R',' ')},*/	/* Bible Cree */
-/*{"sgw?",	HB_TAG('C','H','G',' ')},*/	/* Chaha Gurage */
+/*{"zh?",	HB_TAG('C','H','N',' ')},*/	/* Chinese (seen in Microsoft fonts) */
 /*{"acf/gcf?",	HB_TAG('F','A','N',' ')},*/	/* French Antillean */
-/*{"vls/nl-be",	HB_TAG('F','L','E',' ')},*/	/* Flemish */
 /*{"enf?/yrk?",	HB_TAG('F','N','E',' ')},*/	/* Forest Nenets */
 /*{"fuf?",	HB_TAG('F','T','A',' ')},*/	/* Futa */
 /*{"ar-Syrc?",	HB_TAG('G','A','R',' ')},*/	/* Garshuni */
 /*{"cfm/rnl?",	HB_TAG('H','A','L',' ')},*/	/* Halam */
+/*{"fonipa",	HB_TAG('I','P','P','H')},*/	/* Phonetic transcription—IPA conventions */
 /*{"ga-Latg?/Latg?",	HB_TAG('I','R','T',' ')},*/	/* Irish Traditional */
 /*{"krc",	HB_TAG('K','A','R',' ')},*/	/* Karachay */
 /*{"alw?/ktb?",	HB_TAG('K','E','B',' ')},*/	/* Kebena */
@@ -559,8 +745,6 @@ static const LangTag ot_languages[] = {
 /*{"??",	HB_TAG('L','C','R',' ')},*/	/* L-Cree */
 /*{"??",	HB_TAG('M','A','L',' ')},*/	/* Malayalam Traditional */
 /*{"mnk?/mlq?/...",	HB_TAG('M','L','N',' ')},*/	/* Malinke */
-/*{"man?/myq?/mku?/msc?/...",	HB_TAG('M','N','K',' ')},*/	/* Maninka */
-/*{"??",	HB_TAG('M','O','R',' ')},*/	/* Moroccan */
 /*{"??",	HB_TAG('N','C','R',' ')},*/	/* N-Cree */
 /*{"??",	HB_TAG('N','H','C',' ')},*/	/* Norway House Cree */
 /*{"jpa?/sam?",	HB_TAG('P','A','A',' ')},*/	/* Palestinian Aramaic */
@@ -569,14 +753,12 @@ static const LangTag ot_languages[] = {
 /*{"??",	HB_TAG('R','C','R',' ')},*/	/* R-Cree */
 /*{"chp?",	HB_TAG('S','A','Y',' ')},*/	/* Sayisi */
 /*{"xan?",	HB_TAG('S','E','K',' ')},*/	/* Sekota */
-/*{"stv/wle?/xst?",	HB_TAG('S','I','G',' ')},*/	/* Silte Gurage */
 /*{"ngo?",	HB_TAG('S','X','T',' ')},*/	/* Sutu */
 /*{"??",	HB_TAG('T','C','R',' ')},*/	/* TH-Cree */
 /*{"tnz?/tog?/toi?",	HB_TAG('T','N','G',' ')},*/	/* Tonga */
 /*{"enh?/yrk?",	HB_TAG('T','N','E',' ')},*/	/* Tundra Nenets */
-/*{"??",	HB_TAG('T','O','D',' ')},*/	/* Todo */
 /*{"??",	HB_TAG('W','C','R',' ')},*/	/* West-Cree */
-/*{"??",	HB_TAG('Y','C','R',' ')},*/	/* Y-Cree */
+/*{"cre?",	HB_TAG('Y','C','R',' ')},*/	/* Y-Cree */
 /*{"??",	HB_TAG('Y','I','C',' ')},*/	/* Yi Classic */
 /*{"ii?/Yiii?",	HB_TAG('Y','I','M',' ')},*/	/* Yi Modern */
 /*{"??",	HB_TAG('Z','H','P',' ')},*/	/* Chinese Phonetic */

gfx/harfbuzz/src/hb-private.hh

@@ -79,6 +79,9 @@ static inline Type MIN (const Type &a, const Type &b) { return a < b ? a : b; }
 template <typename Type>
 static inline Type MAX (const Type &a, const Type &b) { return a > b ? a : b; }
 
+static inline unsigned int DIV_CEIL (const unsigned int a, unsigned int b)
+{ return (a + (b - 1)) / b; }
+
 
 #undef  ARRAY_LENGTH
 template <typename Type, unsigned int n>

gfx/harfbuzz/src/hb-shape-plan-private.hh

@@ -39,12 +39,15 @@ struct hb_shape_plan_t
   ASSERT_POD ();
 
   hb_bool_t default_shaper_list;
-  hb_face_t *face;
+  hb_face_t *face_unsafe; /* We don't carry a reference to face. */
   hb_segment_properties_t props;
 
   hb_shape_func_t *shaper_func;
   const char *shaper_name;
 
+  hb_feature_t *user_features;
+  unsigned int num_user_features;
+
   struct hb_shaper_data_t shaper_data;
 };
 

gfx/harfbuzz/src/hb-shape-plan.cc

@@ -46,7 +46,7 @@ hb_shape_plan_plan (hb_shape_plan_t    *shape_plan,
 
 #define HB_SHAPER_PLAN(shaper) \
 	HB_STMT_START { \
-	  if (hb_##shaper##_shaper_face_data_ensure (shape_plan->face)) { \
+	  if (hb_##shaper##_shaper_face_data_ensure (shape_plan->face_unsafe)) { \
 	    HB_SHAPER_DATA (shaper, shape_plan) = \
 	      HB_SHAPER_DATA_CREATE_FUNC (shaper, shape_plan) (shape_plan, user_features, num_user_features); \
 	    shape_plan->shaper_func = _hb_##shaper##_shape; \
@@ -107,18 +107,27 @@ hb_shape_plan_create (hb_face_t                     *face,
   assert (props->direction != HB_DIRECTION_INVALID);
 
   hb_shape_plan_t *shape_plan;
+  hb_feature_t *features = NULL;
 
   if (unlikely (!face))
     face = hb_face_get_empty ();
   if (unlikely (!props || hb_object_is_inert (face)))
     return hb_shape_plan_get_empty ();
-  if (!(shape_plan = hb_object_create<hb_shape_plan_t> ()))
+  if (num_user_features && !(features = (hb_feature_t *) malloc (num_user_features * sizeof (hb_feature_t))))
     return hb_shape_plan_get_empty ();
+  if (!(shape_plan = hb_object_create<hb_shape_plan_t> ())) {
+    free (features);
+    return hb_shape_plan_get_empty ();
+  }
 
   hb_face_make_immutable (face);
   shape_plan->default_shaper_list = shaper_list == NULL;
-  shape_plan->face = hb_face_reference (face);
+  shape_plan->face_unsafe = face;
   shape_plan->props = *props;
+  shape_plan->num_user_features = num_user_features;
+  shape_plan->user_features = features;
+  if (num_user_features)
+    memcpy (features, user_features, num_user_features * sizeof (hb_feature_t));
 
   hb_shape_plan_plan (shape_plan, user_features, num_user_features, shaper_list);
 
@@ -147,6 +156,9 @@ hb_shape_plan_get_empty (void)
     NULL, /* shaper_func */
     NULL, /* shaper_name */
 
+    NULL, /* user_features */
+    0,    /* num_user_featurs */
+
     {
 #define HB_SHAPER_IMPLEMENT(shaper) HB_SHAPER_DATA_INVALID,
 #include "hb-shaper-list.hh"
@@ -190,7 +202,7 @@ hb_shape_plan_destroy (hb_shape_plan_t *shape_plan)
 #include "hb-shaper-list.hh"
 #undef HB_SHAPER_IMPLEMENT
 
-  hb_face_destroy (shape_plan->face);
+  free (shape_plan->user_features);
 
   free (shape_plan);
 }
@@ -264,7 +276,7 @@ hb_shape_plan_execute (hb_shape_plan_t    *shape_plan,
 		hb_object_is_inert (buffer)))
     return false;
 
-  assert (shape_plan->face == font->face);
+  assert (shape_plan->face_unsafe == font->face);
   assert (hb_segment_properties_equal (&shape_plan->props, &buffer->props));
 
 #define HB_SHAPER_EXECUTE(shaper) \
@@ -301,23 +313,55 @@ hb_shape_plan_hash (const hb_shape_plan_t *shape_plan)
 }
 #endif
 
-/* TODO no user-feature caching for now. */
+/* User-feature caching is currently somewhat dumb:
+ * it only finds matches where the feature array is identical,
+ * not cases where the feature lists would be compatible for plan purposes
+ * but have different ranges, for example.
+ */
 struct hb_shape_plan_proposal_t
 {
   const hb_segment_properties_t  props;
   const char * const            *shaper_list;
+  const hb_feature_t            *user_features;
+  unsigned int                   num_user_features;
   hb_shape_func_t               *shaper_func;
 };
 
+static inline hb_bool_t
+hb_shape_plan_user_features_match (const hb_shape_plan_t          *shape_plan,
+				   const hb_shape_plan_proposal_t *proposal)
+{
+  if (proposal->num_user_features != shape_plan->num_user_features) return false;
+  for (unsigned int i = 0, n = proposal->num_user_features; i < n; i++)
+    if (proposal->user_features[i].tag   != shape_plan->user_features[i].tag   ||
+        proposal->user_features[i].value != shape_plan->user_features[i].value ||
+        proposal->user_features[i].start != shape_plan->user_features[i].start ||
+        proposal->user_features[i].end   != shape_plan->user_features[i].end) return false;
+  return true;
+}
+
 static hb_bool_t
 hb_shape_plan_matches (const hb_shape_plan_t          *shape_plan,
 		       const hb_shape_plan_proposal_t *proposal)
 {
   return hb_segment_properties_equal (&shape_plan->props, &proposal->props) &&
+	 hb_shape_plan_user_features_match (shape_plan, proposal) &&
 	 ((shape_plan->default_shaper_list && proposal->shaper_list == NULL) ||
 	  (shape_plan->shaper_func == proposal->shaper_func));
 }
 
+static inline hb_bool_t
+hb_non_global_user_features_present (const hb_feature_t *user_features,
+				     unsigned int        num_user_features)
+{
+  while (num_user_features)
+    if (user_features->start != 0 || user_features->end != (unsigned int) -1)
+      return true;
+    else
+      num_user_features--, user_features++;
+  return false;
+}
+
 /**
  * hb_shape_plan_create_cached:
  * @face: 
@@ -339,12 +383,11 @@ hb_shape_plan_create_cached (hb_face_t                     *face,
 			     unsigned int                   num_user_features,
 			     const char * const            *shaper_list)
 {
-  if (num_user_features)
-    return hb_shape_plan_create (face, props, user_features, num_user_features, shaper_list);
-
   hb_shape_plan_proposal_t proposal = {
     *props,
     shaper_list,
+    user_features,
+    num_user_features,
     NULL
   };
 
@@ -382,6 +425,11 @@ retry:
 
   hb_shape_plan_t *shape_plan = hb_shape_plan_create (face, props, user_features, num_user_features, shaper_list);
 
+  /* Don't add the plan to the cache if there were user features with non-global ranges */
+
+  if (hb_non_global_user_features_present (user_features, num_user_features))
+    return shape_plan;
+
   hb_face_t::plan_node_t *node = (hb_face_t::plan_node_t *) calloc (1, sizeof (hb_face_t::plan_node_t));
   if (unlikely (!node))
     return shape_plan;
@@ -395,9 +443,6 @@ retry:
     goto retry;
   }
 
-  /* Release our reference on face. */
-  hb_face_destroy (face);
-
   return hb_shape_plan_reference (shape_plan);
 }
 

gfx/harfbuzz/src/hb-uniscribe.cc

@@ -729,18 +729,21 @@ _hb_uniscribe_shape (hb_shape_plan_t    *shape_plan,
 retry:
 
   unsigned int scratch_size;
-  char *scratch = (char *) buffer->get_scratch_buffer (&scratch_size);
+  hb_buffer_t::scratch_buffer_t *scratch = buffer->get_scratch_buffer (&scratch_size);
-
-  /* Allocate char buffers; they all fit */
 
 #define ALLOCATE_ARRAY(Type, name, len) \
   Type *name = (Type *) scratch; \
-  scratch += (len) * sizeof ((name)[0]); \
+  { \
-  scratch_size -= (len) * sizeof ((name)[0]);
+    unsigned int _consumed = DIV_CEIL ((len) * sizeof (Type), sizeof (*scratch)); \
+    assert (_consumed <= scratch_size); \
+    scratch += _consumed; \
+    scratch_size -= _consumed; \
+  }
 
 #define utf16_index() var1.u32
 
-  WCHAR *pchars = (WCHAR *) scratch;
+  ALLOCATE_ARRAY (WCHAR, pchars, buffer->len * 2);
+
   unsigned int chars_len = 0;
   for (unsigned int i = 0; i < buffer->len; i++)
   {
@@ -756,7 +759,6 @@ retry:
     }
   }
 
-  ALLOCATE_ARRAY (WCHAR, wchars, chars_len);
   ALLOCATE_ARRAY (WORD, log_clusters, chars_len);
   ALLOCATE_ARRAY (SCRIPT_CHARPROP, char_props, chars_len);
 
@@ -774,12 +776,13 @@ retry:
     }
   }
 
-  /* On Windows, we don't care about alignment...*/
+  /* All the following types are sized in multiples of sizeof(int). */
-  unsigned int glyphs_size = scratch_size / (sizeof (WORD) +
+  unsigned int glyphs_size = scratch_size / ((sizeof (WORD) +
-					     sizeof (SCRIPT_GLYPHPROP) +
+					      sizeof (SCRIPT_GLYPHPROP) +
-					     sizeof (int) +
+					      sizeof (int) +
-					     sizeof (GOFFSET) +
+					      sizeof (GOFFSET) +
-					     sizeof (uint32_t));
+					      sizeof (uint32_t))
+					     / sizeof (int));
 
   ALLOCATE_ARRAY (WORD, glyphs, glyphs_size);
   ALLOCATE_ARRAY (SCRIPT_GLYPHPROP, glyph_props, glyphs_size);
@@ -812,7 +815,7 @@ retry:
   bidi_state.uBidiLevel = HB_DIRECTION_IS_FORWARD (buffer->props.direction) ? 0 : 1;
   bidi_state.fOverrideDirection = 1;
 
-  hr = funcs->ScriptItemizeOpenType (wchars,
+  hr = funcs->ScriptItemizeOpenType (pchars,
 				     chars_len,
 				     MAX_ITEMS,
 				     &bidi_control,
@@ -887,7 +890,7 @@ retry:
 				     range_char_counts.array,
 				     range_properties.array,
 				     range_properties.len,
-				     wchars + chars_offset,
+				     pchars + chars_offset,
 				     item_chars_len,
 				     glyphs_size - glyphs_offset,
 				     /* out */
@@ -929,7 +932,7 @@ retry:
 				     range_char_counts.array,
 				     range_properties.array,
 				     range_properties.len,
-				     wchars + chars_offset,
+				     pchars + chars_offset,
 				     log_clusters + chars_offset,
 				     char_props + chars_offset,
 				     item_chars_len,

gfx/harfbuzz/src/hb-version.h

@@ -38,9 +38,9 @@ HB_BEGIN_DECLS
 
 #define HB_VERSION_MAJOR 0
 #define HB_VERSION_MINOR 9
-#define HB_VERSION_MICRO 23
+#define HB_VERSION_MICRO 24
 
-#define HB_VERSION_STRING "0.9.23"
+#define HB_VERSION_STRING "0.9.24"
 
 #define HB_VERSION_CHECK(major,minor,micro) \
 	((major)*10000+(minor)*100+(micro) >= \