Merge branch 'main' into indic-cleanup

src/hb-buffer.cc

@@ -552,8 +552,6 @@ hb_buffer_t::delete_glyph ()
     goto done;
   }
 
-  /* TODO We need to just merge one direction, with whichever is smaller. */
-
   if (out_len)
   {
     /* Merge cluster backward. */

src/hb-buffer.hh

@@ -70,6 +70,7 @@ enum hb_buffer_scratch_flags_t {
   HB_BUFFER_SCRATCH_FLAG_HAS_GPOS_ATTACHMENT		= 0x00000008u,
   HB_BUFFER_SCRATCH_FLAG_HAS_CGJ			= 0x00000010u,
   HB_BUFFER_SCRATCH_FLAG_HAS_GLYPH_FLAGS		= 0x00000020u,
+  HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE		= 0x00000040u,
 
   /* Reserved for shapers' internal use. */
   HB_BUFFER_SCRATCH_FLAG_SHAPER0			= 0x01000000u,

src/hb-font.hh

@@ -446,7 +446,6 @@ struct hb_font_t
   {
     *x = get_glyph_h_advance (glyph) / 2;
 
-    /* TODO cache this somehow?! */
     hb_font_extents_t extents;
     get_h_extents_with_fallback (&extents);
     *y = extents.ascender;

src/hb-iter.hh

@@ -43,17 +43,12 @@
  * is writable, then the iterator returns lvalues, otherwise it
  * returns rvalues.
  *
- * TODO Document more.
- *
- * If iterator implementation implements operator!=, then can be
+ * If iterator implementation implements operator!=, then it can be
  * used in range-based for loop.  That already happens if the iterator
  * is random-access.  Otherwise, the range-based for loop incurs
  * one traversal to find end(), which can be avoided if written
  * as a while-style for loop, or if iterator implements a faster
- * __end__() method.
- * TODO When opting in for C++17, address this by changing return
- * type of .end()?
- */
+ * __end__() method. */
 
 /*
  * Base classes for iterators.
@@ -75,10 +70,6 @@ struct hb_iter_t
 	iter_t* thiz ()       { return static_cast<      iter_t *> (this); }
   public:
 
-  /* TODO:
-   * Port operators below to use hb_enable_if to sniff which method implements
-   * an operator and use it, and remove hb_iter_fallback_mixin_t completely. */
-
   /* Operators. */
   iter_t iter () const { return *thiz(); }
   iter_t operator + () const { return *thiz(); }
@@ -87,8 +78,7 @@ struct hb_iter_t
   explicit operator bool () const { return thiz()->__more__ (); }
   unsigned len () const { return thiz()->__len__ (); }
   /* The following can only be enabled if item_t is reference type.  Otherwise
-   * it will be returning pointer to temporary rvalue.
-   * TODO Use a wrapper return type to fix for non-reference type. */
+   * it will be returning pointer to temporary rvalue. */
   template <typename T = item_t,
 	    hb_enable_if (std::is_reference<T>::value)>
   hb_remove_reference<item_t>* operator -> () const { return std::addressof (**thiz()); }

src/hb-ot-shape-normalize.cc

@@ -223,7 +223,7 @@ handle_variation_selector_cluster (const hb_ot_shape_normalize_context_t *c,
 				   unsigned int end,
 				   bool short_circuit HB_UNUSED)
 {
-  /* TODO Currently if there's a variation-selector we give-up, it's just too hard. */
+  /* Currently if there's a variation-selector we give-up on normalization, it's just too hard. */
   hb_buffer_t * const buffer = c->buffer;
   hb_font_t * const font = c->font;
   for (; buffer->idx < end - 1 && buffer->successful;) {

src/hb-ot-shaper-arabic-fallback.hh

@@ -168,7 +168,6 @@ arabic_fallback_synthesize_lookup_ligature (const hb_ot_shape_plan_t *plan HB_UN
 					 hb_array (component_count_list, num_ligatures),
 					 hb_array (component_list, num_ligatures));
   c.end_serialize ();
-  /* TODO sanitize the results? */
 
   return ret && !c.in_error () ? c.copy<OT::SubstLookup> () : nullptr;
 }
@@ -232,7 +231,6 @@ arabic_fallback_plan_init_win1256 (arabic_fallback_plan_t *fallback_plan HB_UNUS
   const Manifest &manifest = reinterpret_cast<const Manifest&> (arabic_win1256_gsub_lookups.manifest);
   static_assert (sizeof (arabic_win1256_gsub_lookups.manifestData) ==
 		 ARABIC_FALLBACK_MAX_LOOKUPS * sizeof (ManifestLookup), "");
-  /* TODO sanitize the table? */
 
   unsigned j = 0;
   unsigned int count = manifest.len;

src/hb-ot-shaper-indic-machine.hh

@@ -506,7 +506,7 @@ _eof_trans:
 	break;
 	case 15:
 #line 110 "hb-ot-shaper-indic-machine.rl"
-	{te = p;p--;{ found_syllable (indic_broken_cluster); }}
+	{te = p;p--;{ found_syllable (indic_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; }}
 	break;
 	case 16:
 #line 111 "hb-ot-shaper-indic-machine.rl"
@@ -530,7 +530,7 @@ _eof_trans:
 	break;
 	case 4:
 #line 110 "hb-ot-shaper-indic-machine.rl"
-	{{p = ((te))-1;}{ found_syllable (indic_broken_cluster); }}
+	{{p = ((te))-1;}{ found_syllable (indic_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; }}
 	break;
 	case 6:
 #line 1 "NONE"
@@ -539,7 +539,7 @@ _eof_trans:
 	{{p = ((te))-1;} found_syllable (indic_consonant_syllable); }
 	break;
 	case 5:
-	{{p = ((te))-1;} found_syllable (indic_broken_cluster); }
+	{{p = ((te))-1;} found_syllable (indic_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; }
 	break;
 	case 6:
 	{{p = ((te))-1;} found_syllable (indic_non_indic_cluster); }

src/hb-ot-shaper-indic-machine.rl

@@ -107,7 +107,7 @@ main := |*
 	vowel_syllable		=> { found_syllable (indic_vowel_syllable); };
 	standalone_cluster	=> { found_syllable (indic_standalone_cluster); };
 	symbol_cluster		=> { found_syllable (indic_symbol_cluster); };
-	broken_cluster		=> { found_syllable (indic_broken_cluster); };
+	broken_cluster		=> { found_syllable (indic_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; };
 	other			=> { found_syllable (indic_non_indic_cluster); };
 *|;
 

src/hb-ot-shaper-khmer-machine.hh

@@ -338,7 +338,7 @@ _eof_trans:
 	break;
 	case 12:
 #line 97 "hb-ot-shaper-khmer-machine.rl"
-	{te = p;p--;{ found_syllable (khmer_broken_cluster); }}
+	{te = p;p--;{ found_syllable (khmer_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; }}
 	break;
 	case 11:
 #line 98 "hb-ot-shaper-khmer-machine.rl"
@@ -350,13 +350,13 @@ _eof_trans:
 	break;
 	case 5:
 #line 97 "hb-ot-shaper-khmer-machine.rl"
-	{{p = ((te))-1;}{ found_syllable (khmer_broken_cluster); }}
+	{{p = ((te))-1;}{ found_syllable (khmer_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; }}
 	break;
 	case 3:
 #line 1 "NONE"
 	{	switch( act ) {
 	case 2:
-	{{p = ((te))-1;} found_syllable (khmer_broken_cluster); }
+	{{p = ((te))-1;} found_syllable (khmer_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; }
 	break;
 	case 3:
 	{{p = ((te))-1;} found_syllable (khmer_non_khmer_cluster); }

src/hb-ot-shaper-khmer-machine.rl

@@ -94,7 +94,7 @@ other =			any;
 
 main := |*
 	consonant_syllable	=> { found_syllable (khmer_consonant_syllable); };
-	broken_cluster		=> { found_syllable (khmer_broken_cluster); };
+	broken_cluster		=> { found_syllable (khmer_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; };
 	other			=> { found_syllable (khmer_non_khmer_cluster); };
 *|;
 

src/hb-ot-shaper-myanmar-machine.hh

@@ -516,7 +516,7 @@ _eof_trans:
 	break;
 	case 8:
 #line 116 "hb-ot-shaper-myanmar-machine.rl"
-	{te = p+1;{ found_syllable (myanmar_broken_cluster); }}
+	{te = p+1;{ found_syllable (myanmar_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; }}
 	break;
 	case 3:
 #line 117 "hb-ot-shaper-myanmar-machine.rl"
@@ -528,7 +528,7 @@ _eof_trans:
 	break;
 	case 7:
 #line 116 "hb-ot-shaper-myanmar-machine.rl"
-	{te = p;p--;{ found_syllable (myanmar_broken_cluster); }}
+	{te = p;p--;{ found_syllable (myanmar_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; }}
 	break;
 	case 9:
 #line 117 "hb-ot-shaper-myanmar-machine.rl"

src/hb-ot-shaper-myanmar-machine.rl

@@ -113,7 +113,7 @@ main := |*
 	consonant_syllable	=> { found_syllable (myanmar_consonant_syllable); };
 	j			=> { found_syllable (myanmar_non_myanmar_cluster); };
 	punctuation_cluster	=> { found_syllable (myanmar_punctuation_cluster); };
-	broken_cluster		=> { found_syllable (myanmar_broken_cluster); };
+	broken_cluster		=> { found_syllable (myanmar_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; };
 	other			=> { found_syllable (myanmar_non_myanmar_cluster); };
 *|;
 

src/hb-ot-shaper-syllabic.cc

@@ -39,22 +39,9 @@ hb_syllabic_insert_dotted_circles (hb_font_t *font,
 {
   if (unlikely (buffer->flags & HB_BUFFER_FLAG_DO_NOT_INSERT_DOTTED_CIRCLE))
     return;
-
-  /* Note: This loop is extra overhead, but should not be measurable.
-   * TODO Use a buffer scratch flag to remove the loop. */
-  bool has_broken_syllables = false;
-  unsigned int count = buffer->len;
-  hb_glyph_info_t *info = buffer->info;
-  for (unsigned int i = 0; i < count; i++)
-    if ((info[i].syllable() & 0x0F) == broken_syllable_type)
-    {
-      has_broken_syllables = true;
-      break;
-    }
-  if (likely (!has_broken_syllables))
+  if (likely (!(buffer->scratch_flags & HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE)))
     return;
 
-
   hb_codepoint_t dottedcircle_glyph;
   if (!font->get_nominal_glyph (0x25CCu, &dottedcircle_glyph))
     return;

src/hb-ot-shaper-use-machine.hh

@@ -753,7 +753,7 @@ _eof_trans:
 	break;
 	case 4:
 #line 176 "hb-ot-shaper-use-machine.rl"
-	{te = p+1;{ found_syllable (use_broken_cluster); }}
+	{te = p+1;{ found_syllable (use_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; }}
 	break;
 	case 3:
 #line 177 "hb-ot-shaper-use-machine.rl"
@@ -785,7 +785,7 @@ _eof_trans:
 	break;
 	case 15:
 #line 176 "hb-ot-shaper-use-machine.rl"
-	{te = p;p--;{ found_syllable (use_broken_cluster); }}
+	{te = p;p--;{ found_syllable (use_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; }}
 	break;
 	case 16:
 #line 177 "hb-ot-shaper-use-machine.rl"

src/hb-ot-shaper-use-machine.rl

@@ -173,7 +173,7 @@ main := |*
 	numeral_cluster				=> { found_syllable (use_numeral_cluster); };
 	symbol_cluster				=> { found_syllable (use_symbol_cluster); };
 	hieroglyph_cluster			=> { found_syllable (use_hieroglyph_cluster); };
-	broken_cluster				=> { found_syllable (use_broken_cluster); };
+	broken_cluster				=> { found_syllable (use_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; };
 	other					=> { found_syllable (use_non_cluster); };
 *|;
 

src/hb-serialize.hh

@@ -695,7 +695,7 @@ struct hb_serialize_context_t
     check_assign (off, offset, HB_SERIALIZE_ERROR_OFFSET_OVERFLOW);
   }
 
-  public: /* TODO Make private. */
+  public:
   char *start, *head, *tail, *end;
   unsigned int debug_depth;
   hb_serialize_error_t errors;

src/hb-unicode.hh

@@ -105,12 +105,9 @@ HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS_SIMPLE
   unsigned int
   modified_combining_class (hb_codepoint_t u)
   {
-    /* XXX This hack belongs to the USE shaper (for Tai Tham):
-     * Reorder SAKOT to ensure it comes after any tone marks. */
+    /* Reorder SAKOT to ensure it comes after any tone marks. */
     if (unlikely (u == 0x1A60u)) return 254;
-
-    /* XXX This hack belongs to the Tibetan shaper:
-     * Reorder PADMA to ensure it comes after any vowel marks. */
+    /* Reorder PADMA to ensure it comes after any vowel marks. */
     if (unlikely (u == 0x0FC6u)) return 254;
     /* Reorder TSA -PHRU to reorder before U+0F74 */
     if (unlikely (u == 0x0F39u)) return 127;