[Indic] Find syllables before any features are applied

With FreeSerif, it seems that the 'ccmp' feature does ligature
substituttions.  That was then causing syllable match failures.  We now
find syllables before any features have been applied.

Test sequence: U+0D9A,U+0DCA,U+200D,U+0DBB,U+0DCF

diff --git a/src/hb-ot-shape-complex-indic-machine.rl b/src/hb-ot-shape-complex-indic-machine.rl
index 03e3910..b745466 100644 (file)
--- a/src/hb-ot-shape-complex-indic-machine.rl
+++ b/src/hb-ot-shape-complex-indic-machine.rl
@@ -82,7 +82,7 @@ main := |*
        consonant_syllable      => { found_syllable (consonant_syllable); };
        vowel_syllable          => { found_syllable (vowel_syllable); };
        standalone_cluster      => { found_syllable (standalone_cluster); };
-       broken_cluster          => { found_syllable (broken_cluster); *had_broken_cluster = true; };
+       broken_cluster          => { found_syllable (broken_cluster); };
        other                   => { found_syllable (non_indic_cluster); };
 *|;
 
@@ -100,7 +100,7 @@ main := |*
   } HB_STMT_END
 
 static void
-find_syllables (const hb_ot_shape_plan_t *plan, hb_buffer_t *buffer, bool *had_broken_cluster)
+find_syllables (hb_buffer_t *buffer)
 {
   unsigned int p, pe, eof, ts, te, act;
   int cs;

diff --git a/src/hb-ot-shape-complex-indic.cc b/src/hb-ot-shape-complex-indic.cc
index 7f6b79a..6a87c1e 100644 (file)
--- a/src/hb-ot-shape-complex-indic.cc
+++ b/src/hb-ot-shape-complex-indic.cc
@@ -203,6 +203,10 @@ enum {
 };
 
 static void
+setup_syllables (const hb_ot_shape_plan_t *plan,
+                hb_font_t *font,
+                hb_buffer_t *buffer);
+static void
 initial_reordering (const hb_ot_shape_plan_t *plan,
                    hb_font_t *font,
                    hb_buffer_t *buffer);
@@ -216,6 +220,9 @@ collect_features_indic (hb_ot_shape_planner_t *plan)
 {
   hb_ot_map_builder_t *map = &plan->map;
 
+  /* Do this before any lookups have been applied. */
+  map->add_gsub_pause (setup_syllables);
+
   map->add_bool_feature (HB_TAG('l','o','c','l'));
   /* 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. */
@@ -349,6 +356,17 @@ consonant_position_from_face (const indic_shape_plan_t *indic_plan,
 }
 
 
+enum syllable_type_t {
+  consonant_syllable,
+  vowel_syllable,
+  standalone_cluster,
+  broken_cluster,
+  non_indic_cluster,
+};
+
+#include "hb-ot-shape-complex-indic-machine.hh"
+
+
 static void
 setup_masks_indic (const hb_ot_shape_plan_t *plan HB_UNUSED,
                   hb_buffer_t              *buffer,
@@ -365,6 +383,14 @@ setup_masks_indic (const hb_ot_shape_plan_t *plan HB_UNUSED,
     set_indic_properties (buffer->info[i]);
 }
 
+static void
+setup_syllables (const hb_ot_shape_plan_t *plan HB_UNUSED,
+                hb_font_t *font HB_UNUSED,
+                hb_buffer_t *buffer)
+{
+  find_syllables (buffer);
+}
+
 static int
 compare_indic_order (const hb_glyph_info_t *pa, const hb_glyph_info_t *pb)
 {
@@ -788,16 +814,6 @@ initial_reordering_non_indic_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
 }
 
 
-enum syllable_type_t {
-  consonant_syllable,
-  vowel_syllable,
-  standalone_cluster,
-  broken_cluster,
-  non_indic_cluster,
-};
-
-#include "hb-ot-shape-complex-indic-machine.hh"
-
 static void
 initial_reordering_syllable (const hb_ot_shape_plan_t *plan,
                             hb_buffer_t *buffer,
@@ -813,11 +829,23 @@ initial_reordering_syllable (const hb_ot_shape_plan_t *plan,
   }
 }
 
-static void
+static inline void
 insert_dotted_circles (const hb_ot_shape_plan_t *plan,
                       hb_font_t *font,
                       hb_buffer_t *buffer)
 {
+  /* Note: This loop is extra overhead, but should not be measurable. */
+  bool has_broken_syllables = false;
+  unsigned int count = buffer->len;
+  for (unsigned int i = 0; i < count; i++)
+    if ((buffer->info[i].syllable() & 0x0F) == broken_cluster) {
+      has_broken_syllables = true;
+      break;
+    }
+  if (likely (!has_broken_syllables))
+    return;
+
+
   hb_codepoint_t dottedcircle_glyph;
   if (!font->get_glyph (0x25CC, 0, &dottedcircle_glyph))
     return;
@@ -856,11 +884,7 @@ initial_reordering (const hb_ot_shape_plan_t *plan,
                    hb_buffer_t *buffer)
 {
   update_consonant_positions (plan, font, buffer);
-
-  bool had_broken_clusters = false;
-  find_syllables (plan, buffer, &had_broken_clusters);
-  if (unlikely (had_broken_clusters))
-    insert_dotted_circles (plan, font, buffer);
+  insert_dotted_circles (plan, font, buffer);
 
   hb_glyph_info_t *info = buffer->info;
   unsigned int count = buffer->len;

diff --git a/test/shaping/texts/in-tree/shaper-indic/indic/script-sinhala/misc/misc.txt b/test/shaping/texts/in-tree/shaper-indic/indic/script-sinhala/misc/misc.txt
index c8c939a..8715f72 100644 (file)
--- a/test/shaping/texts/in-tree/shaper-indic/indic/script-sinhala/misc/misc.txt
+++ b/test/shaping/texts/in-tree/shaper-indic/indic/script-sinhala/misc/misc.txt
@@ -36,3 +36,4 @@
 න්‍ගේ
 න‍්ගේ
 ර්‍
+ක්‍රා