Roll HarfBuzz to 1.1.3

third_party/harfbuzz-ng/src/hb-ot-shape-complex-indic.cc

 /*
  * Copyright © 2011,2012  Google, Inc.
  *
  *  This is part of HarfBuzz, a text shaping library.
  *
  * Permission is hereby granted, without written agreement and without
  * license or royalty fees, to use, copy, modify, and distribute this
  * software and its documentation for any purpose, provided that the
  * above copyright notice and the following two paragraphs appear in
  * all copies of this software.
@@ skipping 158 matching lines @@
   hb_codepoint_t u = info.codepoint;
   unsigned int type = hb_indic_get_categories (u);
   indic_category_t cat = (indic_category_t) (type & 0x7Fu);
   indic_position_t pos = (indic_position_t) (type >> 8);
 
 
   /*
    * Re-assign category
    */
 
-
-  /* The spec says U+0952 is OT_A.  However, testing shows that Uniscribe
-   * treats a whole bunch of characters similarly.
-   * TESTS: For example, for U+0951:
-   * U+092E,U+0947,U+0952
-   * U+092E,U+0952,U+0947
-   * U+092E,U+0947,U+0951
-   * U+092E,U+0951,U+0947
-   * U+092E,U+0951,U+0952
-   * U+092E,U+0952,U+0951
-   */
-  if (unlikely (hb_in_ranges (u, 0x0951u, 0x0952u,
-                                 0x1CD0u, 0x1CD2u,
-                                 0x1CD4u, 0x1CE1u) ||
-                            u == 0x1CF4u))
-    cat = OT_A;
   /* The following act more like the Bindus. */
-  else if (unlikely (hb_in_range (u, 0x0953u, 0x0954u)))
+  if (unlikely (hb_in_range (u, 0x0953u, 0x0954u)))
     cat = OT_SM;
   /* The following act like consonants. */
   else if (unlikely (hb_in_ranges (u, 0x0A72u, 0x0A73u,
                                       0x1CF5u, 0x1CF6u)))
     cat = OT_C;
   /* TODO: The following should only be allowed after a Visarga.
    * For now, just treat them like regular tone marks. */
   else if (unlikely (hb_in_range (u, 0x1CE2u, 0x1CE8u)))
     cat = OT_A;
   /* TODO: The following should only be allowed after some of
    * the nasalization marks, maybe only for U+1CE9..U+1CF1.
    * For now, just treat them like tone marks. */
   else if (unlikely (u == 0x1CEDu))
     cat = OT_A;
   /* The following take marks in standalone clusters, similar to Avagraha. */
   else if (unlikely (hb_in_ranges (u, 0xA8F2u, 0xA8F7u,
                                       0x1CE9u, 0x1CECu,
                                       0x1CEEu, 0x1CF1u)))
   {
     cat = OT_Symbol;
     ASSERT_STATIC ((int) INDIC_SYLLABIC_CATEGORY_AVAGRAHA == OT_Symbol);
   }
-  else if (unlikely (hb_in_range (u, 0x17CDu, 0x17D1u) ||
-                     u == 0x17CBu || u == 0x17D3u || u == 0x17DDu)) /* Khmer Various signs */
+  else if (unlikely (u == 0x17DDu)) /* https://github.com/roozbehp/unicode-data/issues/2 */
   {
-    /* These are like Top Matras. */
     cat = OT_M;
     pos = POS_ABOVE_C;
   }
   else if (unlikely (u == 0x17C6u)) cat = OT_N; /* Khmer Bindu doesn't like to be repositioned. */
-  else if (unlikely (u == 0x17D2u)) cat = OT_Coeng; /* Khmer coeng */
   else if (unlikely (hb_in_range (u, 0x2010u, 0x2011u)))
                                     cat = OT_PLACEHOLDER;
   else if (unlikely (u == 0x25CCu)) cat = OT_DOTTEDCIRCLE;
   else if (unlikely (u == 0xA982u)) cat = OT_SM; /* Javanese repha. */
   else if (unlikely (u == 0xA9BEu)) cat = OT_CM2; /* Javanese medial ya. */
   else if (unlikely (u == 0xA9BDu)) { cat = OT_M; pos = POS_POST_C; } /* Javanese vocalic r. */
 
 
   /*
    * Re-assign position.
@@ skipping 312 matching lines @@
   for (unsigned int i = 1; i < ARRAY_LENGTH (indic_configs); i++)
     if (plan->props.script == indic_configs[i].script) {
       indic_plan->config = &indic_configs[i];
       break;
     }
 
   indic_plan->is_old_spec = indic_plan->config->has_old_spec && ((plan->map.chosen_script[0] & 0x000000FFu) != '2');
   indic_plan->virama_glyph = (hb_codepoint_t) -1;
 
   /* Use zero-context would_substitute() matching for new-spec of the main
-   * Indic scripts, and scripts with one spec only, but not for old-specs. */
-  bool zero_context = !indic_plan->is_old_spec;
+   * Indic scripts, and scripts with one spec only, but not for old-specs.
+   * The new-spec for all dual-spec scripts says zero-context matching happens.
+   *
+   * However, testing with Malayalam shows that old and new spec both allow
+   * context.  Testing with Bengali new-spec however shows that it doesn't.
+   * So, the heuristic here is the way it is.  It should *only* be changed,
+   * as we discover more cases of what Windows does.  DON'T TOUCH OTHERWISE.
+   */
+  bool zero_context = !indic_plan->is_old_spec && plan->props.script != HB_SCRIPT_MALAYALAM;
   indic_plan->rphf.init (&plan->map, HB_TAG('r','p','h','f'), zero_context);
   indic_plan->pref.init (&plan->map, HB_TAG('p','r','e','f'), zero_context);
   indic_plan->blwf.init (&plan->map, HB_TAG('b','l','w','f'), zero_context);
   indic_plan->pstf.init (&plan->map, HB_TAG('p','s','t','f'), zero_context);
 
   for (unsigned int i = 0; i < ARRAY_LENGTH (indic_plan->mask_array); i++)
     indic_plan->mask_array[i] = (indic_features[i].flags & F_GLOBAL) ?
                                  0 : plan->map.get_1_mask (indic_features[i].tag);
 
   return indic_plan;
@@ skipping 769 matching lines @@
               base = i;
               while (base < end && is_halant_or_coeng (info[base]))
                 base++;
               info[base].indic_position() = POS_BASE_C;
 
               try_pref = false;
             }
             break;
           }
       }
+      /* For Malayalam, skip over unformed below- (but NOT post-) forms. */
+      if (buffer->props.script == HB_SCRIPT_MALAYALAM)
+      {
+        for (unsigned int i = base + 1; i < end; i++)
+        {
+          while (i < end && is_joiner (info[i]))
+            i++;
+          if (i == end || !is_halant_or_coeng (info[i]))
+            break;
+          i++; /* Skip halant. */
+          while (i < end && is_joiner (info[i]))
+            i++;
+          if (i < end && is_consonant (info[i]) && info[i].indic_position() == POS_BELOW_C)
+          {
+            base = i;
+            info[base].indic_position() = POS_BASE_C;
+          }
+        }
+      }
 
       if (start < base && info[base].indic_position() > POS_BASE_C)
         base--;
       break;
     }
   if (base == end && start < base &&
       is_one_of (info[base - 1], FLAG (OT_ZWJ)))
     base--;
   if (base < end)
     while (start < base &&
@@ skipping 438 matching lines @@
         (c->font->get_glyph (ab, 0, &glyph) &&
          indic_plan->pstf.would_substitute (&glyph, 1, c->font->face)))
     {
       /* Ok, safe to use Uniscribe-style decomposition. */
       *a = 0x0DD9u;
       *b = ab;
       return true;
     }
   }
 
-  return c->unicode->decompose (ab, a, b);
+  return (bool) c->unicode->decompose (ab, a, b);
 }
 
 static bool
 compose_indic (const hb_ot_shape_normalize_context_t *c,
                hb_codepoint_t  a,
                hb_codepoint_t  b,
                hb_codepoint_t *ab)
 {
   /* Avoid recomposing split matras. */
   if (HB_UNICODE_GENERAL_CATEGORY_IS_MARK (c->unicode->general_category (a)))
     return false;
 
   /* Composition-exclusion exceptions that we want to recompose. */
   if (a == 0x09AFu && b == 0x09BCu) { *ab = 0x09DFu; return true; }
 
-  return c->unicode->compose (a, b, ab);
+  return (bool) c->unicode->compose (a, b, ab);
 }
 
 
 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_indic =
 {
   "indic",
   collect_features_indic,
   override_features_indic,
   data_create_indic,
   data_destroy_indic,
   NULL, /* preprocess_text */
   NULL, /* postprocess_glyphs */
   HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS_NO_SHORT_CIRCUIT,
   decompose_indic,
   compose_indic,
   setup_masks_indic,
   HB_OT_SHAPE_ZERO_WIDTH_MARKS_NONE,
   false, /* fallback_position */
 };