Add API for setting invisible-codepoint

Fixes https://github.com/harfbuzz/harfbuzz/issues/1216

New API:
hb_buffer_set_invisible_codepoint()
hb_buffer_get_invisible_codepoint()

hb-shape / hb-view --invisible-codepoint

docs/harfbuzz-sections.txt

@@ -67,6 +67,8 @@ hb_buffer_set_user_data
 hb_buffer_get_user_data
 hb_buffer_get_glyph_infos
 hb_buffer_get_glyph_positions
+hb_buffer_get_invisible_codepoint
+hb_buffer_set_invisible_codepoint
 hb_buffer_set_replacement_codepoint
 hb_buffer_get_replacement_codepoint
 hb_buffer_normalize_glyphs

src/hb-buffer.cc

@@ -219,6 +219,7 @@ hb_buffer_t::reset (void)
   unicode = hb_unicode_funcs_reference (hb_unicode_funcs_get_default ());
   flags = HB_BUFFER_FLAG_DEFAULT;
   replacement = HB_BUFFER_REPLACEMENT_CODEPOINT_DEFAULT;
+  invisible = 0;
 
   clear ();
 }
@@ -665,6 +666,7 @@ DEFINE_NULL_INSTANCE (hb_buffer_t) =
   HB_BUFFER_FLAG_DEFAULT,
   HB_BUFFER_CLUSTER_LEVEL_DEFAULT,
   HB_BUFFER_REPLACEMENT_CODEPOINT_DEFAULT,
+  0, /* invisible */
   HB_BUFFER_SCRATCH_FLAG_DEFAULT,
   HB_BUFFER_MAX_LEN_DEFAULT,
   HB_BUFFER_MAX_OPS_DEFAULT,
@@ -1166,6 +1168,46 @@ hb_buffer_get_replacement_codepoint (hb_buffer_t    *buffer)
 }
 
 
+/**
+ * hb_buffer_set_invisible_codepoint:
+ * @buffer: an #hb_buffer_t.
+ * @invisible: the invisible #hb_codepoint_t
+ *
+ * Sets the #hb_codepoint_t that replaces invisible characters in
+ * the shaping result.  If set to zero (default), the glyph for the
+ * U+0020 SPACE character is used.  Otherwise, this value is used
+ * verbatim.
+ *
+ * Since: REPLACEME
+ **/
+void
+hb_buffer_set_invisible_codepoint (hb_buffer_t    *buffer,
+				   hb_codepoint_t  invisible)
+{
+  if (unlikely (hb_object_is_inert (buffer)))
+    return;
+
+  buffer->invisible = invisible;
+}
+
+/**
+ * hb_buffer_get_invisible_codepoint:
+ * @buffer: an #hb_buffer_t.
+ *
+ * See hb_buffer_set_invisible_codepoint().
+ *
+ * Return value: 
+ * The @buffer invisible #hb_codepoint_t.
+ *
+ * Since: REPLACEME
+ **/
+hb_codepoint_t
+hb_buffer_get_invisible_codepoint (hb_buffer_t    *buffer)
+{
+  return buffer->invisible;
+}
+
+
 /**
  * hb_buffer_reset:
  * @buffer: an #hb_buffer_t.

src/hb-buffer.h

@@ -341,6 +341,13 @@ hb_buffer_set_replacement_codepoint (hb_buffer_t    *buffer,
 HB_EXTERN hb_codepoint_t
 hb_buffer_get_replacement_codepoint (hb_buffer_t    *buffer);
 
+HB_EXTERN void
+hb_buffer_set_invisible_codepoint (hb_buffer_t    *buffer,
+				   hb_codepoint_t  invisible);
+
+HB_EXTERN hb_codepoint_t
+hb_buffer_get_invisible_codepoint (hb_buffer_t    *buffer);
+
 
 HB_EXTERN void
 hb_buffer_reset (hb_buffer_t *buffer);

src/hb-buffer.hh

@@ -93,6 +93,7 @@ struct hb_buffer_t
   hb_buffer_flags_t flags; /* BOT / EOT / etc. */
   hb_buffer_cluster_level_t cluster_level;
   hb_codepoint_t replacement; /* U+FFFD or something else. */
+  hb_codepoint_t invisible; /* 0 or something else. */
   hb_buffer_scratch_flags_t scratch_flags; /* Have space-fallback, etc. */
   unsigned int max_len; /* Maximum allowed len. */
   int max_ops; /* Maximum allowed operations. */

src/hb-ot-shape.cc

@@ -548,15 +548,15 @@ hb_ot_hide_default_ignorables (hb_ot_shape_context_t *c)
   if (i == count)
     return;
 
-  hb_codepoint_t space;
+  hb_codepoint_t invisible = c->buffer->invisible;
   if (!(buffer->flags & HB_BUFFER_FLAG_REMOVE_DEFAULT_IGNORABLES) &&
-      c->font->get_nominal_glyph (' ', &space))
+      (invisible || c->font->get_nominal_glyph (' ', &invisible)))
   {
-    /* Replace default-ignorables with a zero-advance space glyph. */
+    /* Replace default-ignorables with a zero-advance invisible glyph. */
     for (/*continue*/; i < count; i++)
     {
       if (_hb_glyph_info_is_default_ignorable (&info[i]))
-	info[i].codepoint = space;
+	info[i].codepoint = invisible;
     }
   }
   else

util/options.cc

@@ -415,6 +415,7 @@ shape_options_t::add_options (option_parser_t *parser)
     {"eot",		0, 0, G_OPTION_ARG_NONE,	&this->eot,			"Treat text as end-of-paragraph",	nullptr},
     {"preserve-default-ignorables",0, 0, G_OPTION_ARG_NONE,	&this->preserve_default_ignorables,	"Preserve Default-Ignorable characters",	nullptr},
     {"remove-default-ignorables",0, 0, G_OPTION_ARG_NONE,	&this->remove_default_ignorables,	"Remove Default-Ignorable characters",	nullptr},
+    {"invisible-codepoint",0, 0, G_OPTION_ARG_INT,	&this->invisible_codepoint,	"Value to replace Default-Ignorables with",	nullptr},
     {"utf8-clusters",	0, 0, G_OPTION_ARG_NONE,	&this->utf8_clusters,		"Use UTF8 byte indices, not char indices",	nullptr},
     {"cluster-level",	0, 0, G_OPTION_ARG_INT,		&this->cluster_level,		"Cluster merging level (default: 0)",	"0/1/2"},
     {"normalize-glyphs",0, 0, G_OPTION_ARG_NONE,	&this->normalize_glyphs,	"Rearrange glyph clusters in nominal order",	nullptr},

util/options.hh

@@ -155,6 +155,7 @@ struct shape_options_t : option_group_t
     num_features = 0;
     shapers = nullptr;
     utf8_clusters = false;
+    invisible_codepoint = 0;
     cluster_level = HB_BUFFER_CLUSTER_LEVEL_DEFAULT;
     normalize_glyphs = false;
     verify = false;
@@ -185,6 +186,7 @@ struct shape_options_t : option_group_t
 				  (preserve_default_ignorables ? HB_BUFFER_FLAG_PRESERVE_DEFAULT_IGNORABLES : 0) |
 				  (remove_default_ignorables ? HB_BUFFER_FLAG_REMOVE_DEFAULT_IGNORABLES : 0) |
 				  0));
+    hb_buffer_set_invisible_codepoint (buffer, invisible_codepoint);
     hb_buffer_set_cluster_level (buffer, cluster_level);
     hb_buffer_guess_segment_properties (buffer);
   }
@@ -435,6 +437,7 @@ struct shape_options_t : option_group_t
   unsigned int num_features;
   char **shapers;
   hb_bool_t utf8_clusters;
+  hb_codepoint_t invisible_codepoint;
   hb_buffer_cluster_level_t cluster_level;
   hb_bool_t normalize_glyphs;
   hb_bool_t verify;