[indic-like] Add per-lookup per-syllable flag

This allows mix-and-matching per-syllable and other lookups.
In fact, removes the clear-syllables call completely.

Fixes https://github.com/harfbuzz/harfbuzz/issues/3513

src/hb-ot-layout-gsubgpos.hh

@@ -408,9 +408,10 @@ struct hb_ot_apply_context_t :
   {
     matcher_t () :
 	     lookup_props (0),
+	     mask (-1),
 	     ignore_zwnj (false),
 	     ignore_zwj (false),
-	     mask (-1),
+	     per_syllable (false),
 	     syllable {0},
 	     match_func (nullptr),
 	     match_data (nullptr) {}
@@ -421,7 +422,8 @@ struct hb_ot_apply_context_t :
     void set_ignore_zwj (bool ignore_zwj_) { ignore_zwj = ignore_zwj_; }
     void set_lookup_props (unsigned int lookup_props_) { lookup_props = lookup_props_; }
     void set_mask (hb_mask_t mask_) { mask = mask_; }
-    void set_syllable (uint8_t syllable_)  { syllable = syllable_; }
+    void set_per_syllable (bool per_syllable_) { per_syllable = per_syllable_; }
+    void set_syllable (uint8_t syllable_)  { syllable = per_syllable ? syllable_ : 0; }
     void set_match_func (match_func_t match_func_,
 			 const void *match_data_)
     { match_func = match_func_; match_data = match_data_; }
@@ -467,9 +469,10 @@ struct hb_ot_apply_context_t :
 
     protected:
     unsigned int lookup_props;
+    hb_mask_t mask;
     bool ignore_zwnj;
     bool ignore_zwj;
-    hb_mask_t mask;
+    bool per_syllable;
     uint8_t syllable;
     match_func_t match_func;
     const void *match_data;
@@ -488,6 +491,7 @@ struct hb_ot_apply_context_t :
       /* Ignore ZWJ if we are matching context, or asked to. */
       matcher.set_ignore_zwj  (context_match || c->auto_zwj);
       matcher.set_mask (context_match ? -1 : c->lookup_mask);
+      matcher.set_per_syllable (c->per_syllable);
     }
     void set_lookup_props (unsigned int lookup_props)
     {
@@ -634,6 +638,7 @@ struct hb_ot_apply_context_t :
   bool has_glyph_classes;
   bool auto_zwnj;
   bool auto_zwj;
+  bool per_syllable;
   bool random;
 
   uint32_t random_state;
@@ -662,6 +667,7 @@ struct hb_ot_apply_context_t :
 			has_glyph_classes (gdef.has_glyph_classes ()),
 			auto_zwnj (true),
 			auto_zwj (true),
+			per_syllable (false),
 			random (false),
 			random_state (1) { init_iters (); }
 
@@ -674,6 +680,7 @@ struct hb_ot_apply_context_t :
   void set_lookup_mask (hb_mask_t mask) { lookup_mask = mask; init_iters (); }
   void set_auto_zwj (bool auto_zwj_) { auto_zwj = auto_zwj_; init_iters (); }
   void set_auto_zwnj (bool auto_zwnj_) { auto_zwnj = auto_zwnj_; init_iters (); }
+  void set_per_syllable (bool per_syllable_) { per_syllable = per_syllable_; init_iters (); }
   void set_random (bool random_) { random = random_; }
   void set_recurse_func (recurse_func_t func) { recurse_func = func; }
   void set_lookup_index (unsigned int lookup_index_) { lookup_index = lookup_index_; }

src/hb-ot-layout.cc

@@ -1931,6 +1931,7 @@ inline void hb_ot_map_t::apply (const Proxy &proxy,
       c.set_auto_zwj (lookups[table_index][i].auto_zwj);
       c.set_auto_zwnj (lookups[table_index][i].auto_zwnj);
       c.set_random (lookups[table_index][i].random);
+      c.set_per_syllable (lookups[table_index][i].per_syllable);
 
       apply_string<Proxy> (&c,
 			   proxy.table.get_lookup (lookup_index),

src/hb-ot-layout.hh

@@ -172,17 +172,6 @@ _hb_next_syllable (hb_buffer_t *buffer, unsigned int start)
   return start;
 }
 
-static inline void
-_hb_clear_syllables (const hb_ot_shape_plan_t *plan HB_UNUSED,
-		     hb_font_t *font HB_UNUSED,
-		     hb_buffer_t *buffer)
-{
-  hb_glyph_info_t *info = buffer->info;
-  unsigned int count = buffer->len;
-  for (unsigned int i = 0; i < count; i++)
-    info[i].syllable() = 0;
-}
-
 
 /* unicode_props */
 

src/hb-ot-map.cc

@@ -117,7 +117,8 @@ hb_ot_map_builder_t::add_lookups (hb_ot_map_t  &m,
 				  hb_mask_t     mask,
 				  bool          auto_zwnj,
 				  bool          auto_zwj,
-				  bool          random)
+				  bool          random,
+				  bool          per_syllable)
 {
   unsigned int lookup_indices[32];
   unsigned int offset, len;
@@ -145,6 +146,7 @@ hb_ot_map_builder_t::add_lookups (hb_ot_map_t  &m,
       lookup->auto_zwnj = auto_zwnj;
       lookup->auto_zwj = auto_zwj;
       lookup->random = random;
+      lookup->per_syllable = per_syllable;
     }
 
     offset += len;
@@ -277,6 +279,7 @@ hb_ot_map_builder_t::compile (hb_ot_map_t                  &m,
     map->auto_zwnj = !(info->flags & F_MANUAL_ZWNJ);
     map->auto_zwj = !(info->flags & F_MANUAL_ZWJ);
     map->random = !!(info->flags & F_RANDOM);
+    map->per_syllable = !!(info->flags & F_PER_SYLLABLE);
     if ((info->flags & F_GLOBAL) && info->max_value == 1) {
       /* Uses the global bit */
       map->shift = global_bit_shift;
@@ -319,7 +322,8 @@ hb_ot_map_builder_t::compile (hb_ot_map_t                  &m,
 		       m.features[i].mask,
 		       m.features[i].auto_zwnj,
 		       m.features[i].auto_zwj,
-		       m.features[i].random);
+		       m.features[i].random,
+		       m.features[i].per_syllable);
 
       /* Sort lookups and merge duplicates */
       if (last_num_lookups < m.lookups[table_index].length)

src/hb-ot-map.hh

@@ -56,6 +56,7 @@ struct hb_ot_map_t
     unsigned int auto_zwnj : 1;
     unsigned int auto_zwj : 1;
     unsigned int random : 1;
+    unsigned int per_syllable : 1;
 
     int cmp (const hb_tag_t tag_) const
     { return tag_ < tag ? -1 : tag_ > tag ? 1 : 0; }
@@ -66,6 +67,7 @@ struct hb_ot_map_t
     unsigned short auto_zwnj : 1;
     unsigned short auto_zwj : 1;
     unsigned short random : 1;
+    unsigned short per_syllable : 1;
     hb_mask_t mask;
 
     HB_INTERNAL static int cmp (const void *pa, const void *pb)
@@ -183,7 +185,8 @@ enum hb_ot_map_feature_flags_t
   F_GLOBAL_MANUAL_JOINERS= F_GLOBAL | F_MANUAL_JOINERS,
   F_GLOBAL_HAS_FALLBACK = F_GLOBAL | F_HAS_FALLBACK,
   F_GLOBAL_SEARCH	= 0x0010u, /* If feature not found in LangSys, look for it in global feature list and pick one. */
-  F_RANDOM		= 0x0020u  /* Randomly select a glyph from an AlternateSubstFormat1 subtable. */
+  F_RANDOM		= 0x0020u, /* Randomly select a glyph from an AlternateSubstFormat1 subtable. */
+  F_PER_SYLLABLE	= 0x0040u  /* Contain lookup application to within syllable. */
 };
 HB_MARK_AS_FLAG_T (hb_ot_map_feature_flags_t);
 
@@ -237,7 +240,8 @@ struct hb_ot_map_builder_t
 				hb_mask_t     mask,
 				bool          auto_zwnj = true,
 				bool          auto_zwj = true,
-				bool          random = false);
+				bool          random = false,
+				bool          per_syllable = false);
 
   struct feature_info_t {
     hb_tag_t tag;

src/hb-ot-shape-complex-indic.cc

@@ -109,17 +109,17 @@ indic_features[] =
    * These features are applied in order, one at a time, after initial_reordering,
    * constrained to the syllable.
    */
-  {HB_TAG('n','u','k','t'), F_GLOBAL_MANUAL_JOINERS},
+  {HB_TAG('n','u','k','t'), F_GLOBAL_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('a','k','h','n'), F_GLOBAL_MANUAL_JOINERS},
+  {HB_TAG('a','k','h','n'), F_GLOBAL_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('r','p','h','f'),        F_MANUAL_JOINERS},
+  {HB_TAG('r','p','h','f'),        F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('r','k','r','f'), F_GLOBAL_MANUAL_JOINERS},
+  {HB_TAG('r','k','r','f'), F_GLOBAL_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('p','r','e','f'),        F_MANUAL_JOINERS},
+  {HB_TAG('p','r','e','f'),        F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('b','l','w','f'),        F_MANUAL_JOINERS},
+  {HB_TAG('b','l','w','f'),        F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('a','b','v','f'),        F_MANUAL_JOINERS},
+  {HB_TAG('a','b','v','f'),        F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('h','a','l','f'),        F_MANUAL_JOINERS},
+  {HB_TAG('h','a','l','f'),        F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('p','s','t','f'),        F_MANUAL_JOINERS},
+  {HB_TAG('p','s','t','f'),        F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('v','a','t','u'), F_GLOBAL_MANUAL_JOINERS},
+  {HB_TAG('v','a','t','u'), F_GLOBAL_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('c','j','c','t'), F_GLOBAL_MANUAL_JOINERS},
+  {HB_TAG('c','j','c','t'), F_GLOBAL_MANUAL_JOINERS | F_PER_SYLLABLE},
   /*
    * Other features.
    * These features are applied all at once, after final_reordering, constrained
@@ -127,12 +127,12 @@ indic_features[] =
    * Default Bengali font in Windows for example has intermixed
    * lookups for init,pres,abvs,blws features.
    */
-  {HB_TAG('i','n','i','t'),        F_MANUAL_JOINERS},
+  {HB_TAG('i','n','i','t'),        F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('p','r','e','s'), F_GLOBAL_MANUAL_JOINERS},
+  {HB_TAG('p','r','e','s'), F_GLOBAL_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('a','b','v','s'), F_GLOBAL_MANUAL_JOINERS},
+  {HB_TAG('a','b','v','s'), F_GLOBAL_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('b','l','w','s'), F_GLOBAL_MANUAL_JOINERS},
+  {HB_TAG('b','l','w','s'), F_GLOBAL_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('p','s','t','s'), F_GLOBAL_MANUAL_JOINERS},
+  {HB_TAG('p','s','t','s'), F_GLOBAL_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('h','a','l','n'), F_GLOBAL_MANUAL_JOINERS},
+  {HB_TAG('h','a','l','n'), F_GLOBAL_MANUAL_JOINERS | F_PER_SYLLABLE},
 };
 
 /*
@@ -183,10 +183,10 @@ collect_features_indic (hb_ot_shape_planner_t *plan)
   /* Do this before any lookups have been applied. */
   map->add_gsub_pause (setup_syllables_indic);
 
-  map->enable_feature (HB_TAG('l','o','c','l'));
+  map->enable_feature (HB_TAG('l','o','c','l'), F_PER_SYLLABLE);
   /* The Indic specs do not require ccmp, but we apply it here since if
    * there is a use of it, it's typically at the beginning. */
-  map->enable_feature (HB_TAG('c','c','m','p'));
+  map->enable_feature (HB_TAG('c','c','m','p'), F_PER_SYLLABLE);
 
 
   unsigned int i = 0;
@@ -201,8 +201,6 @@ collect_features_indic (hb_ot_shape_planner_t *plan)
 
   for (; i < INDIC_NUM_FEATURES; i++)
     map->add_feature (indic_features[i]);
-
-  map->add_gsub_pause (_hb_clear_syllables);
 }
 
 static void

src/hb-ot-shape-complex-khmer.cc

@@ -45,11 +45,11 @@ khmer_features[] =
    * These features are applied all at once, before reordering, constrained
    * to the syllable.
    */
-  {HB_TAG('p','r','e','f'), F_MANUAL_JOINERS},
+  {HB_TAG('p','r','e','f'), F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('b','l','w','f'), F_MANUAL_JOINERS},
+  {HB_TAG('b','l','w','f'), F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('a','b','v','f'), F_MANUAL_JOINERS},
+  {HB_TAG('a','b','v','f'), F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('p','s','t','f'), F_MANUAL_JOINERS},
+  {HB_TAG('p','s','t','f'), F_MANUAL_JOINERS | F_PER_SYLLABLE},
-  {HB_TAG('c','f','a','r'), F_MANUAL_JOINERS},
+  {HB_TAG('c','f','a','r'), F_MANUAL_JOINERS | F_PER_SYLLABLE},
   /*
    * Other features.
    * These features are applied all at once after clearing syllables.
@@ -107,16 +107,10 @@ collect_features_khmer (hb_ot_shape_planner_t *plan)
    *
    * https://github.com/harfbuzz/harfbuzz/issues/974
    */
-  map->enable_feature (HB_TAG('l','o','c','l'));
+  map->enable_feature (HB_TAG('l','o','c','l'), F_PER_SYLLABLE);
-  map->enable_feature (HB_TAG('c','c','m','p'));
+  map->enable_feature (HB_TAG('c','c','m','p'), F_PER_SYLLABLE);
 
-  unsigned int i = 0;
+  for (unsigned i = 0; i < KHMER_NUM_FEATURES; i++)
-  for (; i < KHMER_BASIC_FEATURES; i++)
-    map->add_feature (khmer_features[i]);
-
-  map->add_gsub_pause (_hb_clear_syllables);
-
-  for (; i < KHMER_NUM_FEATURES; i++)
     map->add_feature (khmer_features[i]);
 }
 

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

@@ -79,22 +79,20 @@ collect_features_myanmar (hb_ot_shape_planner_t *plan)
   /* Do this before any lookups have been applied. */
   map->add_gsub_pause (setup_syllables_myanmar);
 
-  map->enable_feature (HB_TAG('l','o','c','l'));
+  map->enable_feature (HB_TAG('l','o','c','l'), F_PER_SYLLABLE);
   /* The Indic specs do not require ccmp, but we apply it here since if
    * there is a use of it, it's typically at the beginning. */
-  map->enable_feature (HB_TAG('c','c','m','p'));
+  map->enable_feature (HB_TAG('c','c','m','p'), F_PER_SYLLABLE);
 
 
   map->add_gsub_pause (reorder_myanmar);
 
   for (unsigned int i = 0; i < ARRAY_LENGTH (myanmar_basic_features); i++)
   {
-    map->enable_feature (myanmar_basic_features[i], F_MANUAL_ZWJ);
+    map->enable_feature (myanmar_basic_features[i], F_MANUAL_ZWJ | F_PER_SYLLABLE);
     map->add_gsub_pause (nullptr);
   }
 
-  map->add_gsub_pause (_hb_clear_syllables);
-
   for (unsigned int i = 0; i < ARRAY_LENGTH (myanmar_other_features); i++)
     map->enable_feature (myanmar_other_features[i], F_MANUAL_ZWJ);
 }

src/hb-ot-shape-complex-use.cc

@@ -115,25 +115,24 @@ collect_features_use (hb_ot_shape_planner_t *plan)
   map->add_gsub_pause (setup_syllables_use);
 
   /* "Default glyph pre-processing group" */
-  map->enable_feature (HB_TAG('l','o','c','l'));
+  map->enable_feature (HB_TAG('l','o','c','l'), F_PER_SYLLABLE);
-  map->enable_feature (HB_TAG('c','c','m','p'));
+  map->enable_feature (HB_TAG('c','c','m','p'), F_PER_SYLLABLE);
-  map->enable_feature (HB_TAG('n','u','k','t'));
+  map->enable_feature (HB_TAG('n','u','k','t'), F_PER_SYLLABLE);
-  map->enable_feature (HB_TAG('a','k','h','n'), F_MANUAL_ZWJ);
+  map->enable_feature (HB_TAG('a','k','h','n'), F_MANUAL_ZWJ | F_PER_SYLLABLE);
 
   /* "Reordering group" */
   map->add_gsub_pause (_hb_clear_substitution_flags);
-  map->add_feature (HB_TAG('r','p','h','f'), F_MANUAL_ZWJ);
+  map->add_feature (HB_TAG('r','p','h','f'), F_MANUAL_ZWJ | F_PER_SYLLABLE);
   map->add_gsub_pause (record_rphf_use);
   map->add_gsub_pause (_hb_clear_substitution_flags);
-  map->enable_feature (HB_TAG('p','r','e','f'), F_MANUAL_ZWJ);
+  map->enable_feature (HB_TAG('p','r','e','f'), F_MANUAL_ZWJ | F_PER_SYLLABLE);
   map->add_gsub_pause (record_pref_use);
 
   /* "Orthographic unit shaping group" */
   for (unsigned int i = 0; i < ARRAY_LENGTH (use_basic_features); i++)
-    map->enable_feature (use_basic_features[i], F_MANUAL_ZWJ);
+    map->enable_feature (use_basic_features[i], F_MANUAL_ZWJ | F_PER_SYLLABLE);
 
   map->add_gsub_pause (reorder_use);
-  map->add_gsub_pause (_hb_clear_syllables);
 
   /* "Topographical features" */
   for (unsigned int i = 0; i < ARRAY_LENGTH (use_topographical_features); i++)

test/shape/data/in-house/tests/indic-syllable.tests

@@ -11,3 +11,4 @@
 ../fonts/b3075ca42b27dde7341c2d0ae16703c5b6640df0.ttf;;U+0B2C,U+0B55,U+0B3E;[uni0B2C=0+641|uni0B55=0+0|uni0B3E=0+253]
 ../fonts/b3075ca42b27dde7341c2d0ae16703c5b6640df0.ttf;;U+0B2C,U+0B3E,U+0B55;[uni0B2C=0+641|uni0B3E=0+253|uni0B55=0+0]
 ../fonts/e2b17207c4b7ad78d843e1b0c4d00b09398a1137.ttf;;U+0BAA,U+0BAA,U+0BCD;[pa-tamil=0+778|pa-tamil.001=1+778|pulli-tamil=1@-385,0+0]
+../fonts/41071178fbce4956d151f50967af458dbf555f7b.ttf;;U+0926,U+093F,U+0938,U+0902,U+092C,U+0930;[isigndeva=0+266|dadeva=0+541|sadeva=2+709|anusvaradeva=2@0,-1+0|badeva=4+537|radeva=5+436]