Merge pull request #3699 from googlefonts/filter_scripts

[subset] Add support for --layout-scripts

src/hb-ot-layout-common.hh

@@ -764,6 +764,8 @@ struct Script
   {
     TRACE_SUBSET (this);
     if (!l->visitScript ()) return_trace (false);
+    if (tag && !c->plan->layout_scripts->has (*tag))
+      return false;
 
     auto *out = c->serializer->start_embed (*this);
     if (unlikely (!out || !c->serializer->extend_min (out))) return_trace (false);

src/hb-ot-layout-gdef-table.hh

@@ -645,21 +645,25 @@ struct GDEF
     bool subset_ligcaretlist = out->ligCaretList.serialize_subset (c, ligCaretList, this);
     bool subset_markattachclassdef = out->markAttachClassDef.serialize_subset (c, markAttachClassDef, this, nullptr, false, true);
 
-    bool subset_markglyphsetsdef = true;
+    bool subset_markglyphsetsdef = false;
     if (version.to_int () >= 0x00010002u)
     {
       subset_markglyphsetsdef = out->markGlyphSetsDef.serialize_subset (c, markGlyphSetsDef, this);
-      if (!subset_markglyphsetsdef &&
-	  version.to_int () == 0x00010002u)
-	out->version.minor = 0;
     }
 
-    bool subset_varstore = true;
+    bool subset_varstore = false;
     if (version.to_int () >= 0x00010003u)
     {
       subset_varstore = out->varStore.serialize_subset (c, varStore, this);
-      if (!subset_varstore && version.to_int () == 0x00010003u)
+    }
+
+    if (subset_varstore)
+    {
+      out->version.minor = 3;
+    } else if (subset_markglyphsetsdef) {
       out->version.minor = 2;
+    } else  {
+      out->version.minor = 0;
     }
 
     return_trace (subset_glyphclassdef || subset_attachlist ||

src/hb-subset-input.cc

@@ -203,6 +203,8 @@ hb_subset_input_create_or_fail (void)
 
   input->sets.layout_features->add_array (default_layout_features, ARRAY_LENGTH (default_layout_features));
 
+  input->sets.layout_scripts->invert (); // Default to all scripts.
+
   if (input->in_error ())
   {
     hb_subset_input_destroy (input);

src/hb-subset-input.hh

@@ -50,6 +50,7 @@ struct hb_subset_input_t
     hb_set_t *name_ids;
     hb_set_t *name_languages;
     hb_set_t *layout_features;
+    hb_set_t *layout_scripts;
   };
 
   union {

src/hb-subset-plan.cc

@@ -91,94 +91,100 @@ _remap_indexes (const hb_set_t *indexes,
 typedef void (*layout_collect_func_t) (hb_face_t *face, hb_tag_t table_tag, const hb_tag_t *scripts, const hb_tag_t *languages, const hb_tag_t *features, hb_set_t *lookup_indexes /* OUT */);
 
 
+/*
+ * Removes all tags from 'tags' that are not in filter. Additionally eliminates any duplicates.
+ * Returns true if anything was removed (not including duplicates).
+ */
+static bool _filter_tag_list(hb_vector_t<hb_tag_t>* tags, /* IN/OUT */
+                             const hb_set_t* filter)
+{
+  hb_vector_t<hb_tag_t> out;
+  out.alloc (tags->get_size() + 1); // +1 is to allocate room for the null terminator.
+
+  bool removed = false;
+  hb_set_t visited;
+
+  for (hb_tag_t tag : *tags)
+  {
+    if (!tag) continue;
+    if (visited.has (tag)) continue;
+
+    if (!filter->has (tag))
+    {
+      removed = true;
+      continue;
+    }
+
+    visited.add (tag);
+    out.push (tag);
+  }
+
+  // The collect function needs a null element to signal end of the array.
+  out.push (HB_TAG_NONE);
+
+  hb_swap (out, *tags);
+  return removed;
+}
+
 template <typename T>
-static void _collect_layout_indices (hb_face_t		  *face,
+static void _collect_layout_indices (hb_subset_plan_t     *plan,
                                      const T&              table,
-                                     const hb_set_t	  *layout_features_to_retain,
                                      layout_collect_func_t layout_collect_func,
                                      hb_set_t		  *indices /* OUT */)
 {
+  unsigned num_features = table.get_feature_count ();
   hb_vector_t<hb_tag_t> features;
-  if (!features.alloc (table.get_feature_count () + 1))
+  if (!plan->check_success (features.resize (num_features))) return;
+  table.get_feature_tags (0, &num_features, features.arrayZ);
+  bool retain_all_features = !_filter_tag_list (&features, plan->layout_features);
+
+  unsigned num_scripts = table.get_script_count ();
+  hb_vector_t<hb_tag_t> scripts;
+  if (!plan->check_success (scripts.resize (num_scripts))) return;
+  table.get_script_tags (0, &num_scripts, scripts.arrayZ);
+  bool retain_all_scripts = !_filter_tag_list (&scripts, plan->layout_scripts);
+
+  if (!plan->check_success (!features.in_error ()) || !features
+      || !plan->check_success (!scripts.in_error ()) || !scripts)
     return;
 
-  hb_set_t visited_features;
+  layout_collect_func (plan->source,
-  bool retain_all_features = true;
-  for (unsigned i = 0; i < table.get_feature_count (); i++)
-  {
-    hb_tag_t tag = table.get_feature_tag (i);
-    if (!tag) continue;
-    if (!layout_features_to_retain->has (tag))
-    {
-      retain_all_features = false;
-      continue;
-    }
-
-    if (visited_features.has (tag))
-      continue;
-
-    features.push (tag);
-    visited_features.add (tag);
-  }
-
-  if (!features)
-    return;
-
-  // The collect function needs a null element to signal end of the array.
-  features.push (0);
-
-  if (retain_all_features)
-  {
-    // Looking for all features, trigger the faster collection method.
-    layout_collect_func (face,
                        T::tableTag,
+                       retain_all_scripts ? nullptr : scripts.arrayZ,
 		       nullptr,
-                         nullptr,
+		       retain_all_features ? nullptr : features.arrayZ,
-                         nullptr,
-                         indices);
-    return;
-  }
-
-  layout_collect_func (face,
-                       T::tableTag,
-		       nullptr,
-		       nullptr,
-		       features.arrayZ,
 		       indices);
 }
 
 template <typename T>
 static inline void
-_closure_glyphs_lookups_features (hb_face_t	     *face,
+_closure_glyphs_lookups_features (hb_subset_plan_t   *plan,
 				  hb_set_t	     *gids_to_retain,
-				  const hb_set_t     *layout_features_to_retain,
 				  hb_map_t	     *lookups,
 				  hb_map_t	     *features,
 				  script_langsys_map *langsys_map)
 {
-  hb_blob_ptr_t<T> table = hb_sanitize_context_t ().reference_table<T> (face);
+  hb_blob_ptr_t<T> table = hb_sanitize_context_t ().reference_table<T> (plan->source);
   hb_tag_t table_tag = table->tableTag;
   hb_set_t lookup_indices;
-  _collect_layout_indices<T> (face,
+  _collect_layout_indices<T> (plan,
                               *table,
-                              layout_features_to_retain,
                               hb_ot_layout_collect_lookups,
                               &lookup_indices);
 
   if (table_tag == HB_OT_TAG_GSUB)
-    hb_ot_layout_lookups_substitute_closure (face,
+    hb_ot_layout_lookups_substitute_closure (plan->source,
                                              &lookup_indices,
 					     gids_to_retain);
-  table->closure_lookups (face,
+  table->closure_lookups (plan->source,
 			  gids_to_retain,
                           &lookup_indices);
   _remap_indexes (&lookup_indices, lookups);
 
   // Collect and prune features
   hb_set_t feature_indices;
-  _collect_layout_indices<T> (face,
+  _collect_layout_indices<T> (plan,
                               *table,
-                              layout_features_to_retain,
                               hb_ot_layout_collect_features,
                               &feature_indices);
 
@@ -395,18 +401,16 @@ _populate_gids_to_retain (hb_subset_plan_t* plan,
   if (close_over_gsub)
     // closure all glyphs/lookups/features needed for GSUB substitutions.
     _closure_glyphs_lookups_features<GSUB> (
-        plan->source,
+        plan,
         plan->_glyphset_gsub,
-        plan->layout_features,
         plan->gsub_lookups,
         plan->gsub_features,
         plan->gsub_langsys);
 
   if (close_over_gpos)
     _closure_glyphs_lookups_features<GPOS> (
-        plan->source,
+        plan,
         plan->_glyphset_gsub,
-        plan->layout_features,
         plan->gpos_lookups,
         plan->gpos_features,
         plan->gpos_langsys);
@@ -549,6 +553,7 @@ hb_subset_plan_create_or_fail (hb_face_t	 *face,
   _nameid_closure (face, plan->name_ids);
   plan->name_languages = hb_set_copy (input->sets.name_languages);
   plan->layout_features = hb_set_copy (input->sets.layout_features);
+  plan->layout_scripts = hb_set_copy (input->sets.layout_scripts);
   plan->glyphs_requested = hb_set_copy (input->sets.glyphs);
   plan->drop_tables = hb_set_copy (input->sets.drop_tables);
   plan->no_subset_tables = hb_set_copy (input->sets.no_subset_tables);
@@ -636,6 +641,7 @@ hb_subset_plan_destroy (hb_subset_plan_t *plan)
   hb_set_destroy (plan->name_ids);
   hb_set_destroy (plan->name_languages);
   hb_set_destroy (plan->layout_features);
+  hb_set_destroy (plan->layout_scripts);
   hb_set_destroy (plan->glyphs_requested);
   hb_set_destroy (plan->drop_tables);
   hb_set_destroy (plan->no_subset_tables);

src/hb-subset-plan.hh

@@ -55,6 +55,9 @@ struct hb_subset_plan_t
   //layout features which will be preserved
   hb_set_t *layout_features;
 
+  // layout scripts which will be preserved.
+  hb_set_t *layout_scripts;
+
   //glyph ids requested to retain
   hb_set_t *glyphs_requested;
 

src/hb-subset.h

@@ -100,6 +100,8 @@ typedef enum { /*< flags >*/
  * @HB_SUBSET_SETS_NAME_LANG_ID: the set of name lang ids that will be retained.
  * @HB_SUBSET_SETS_LAYOUT_FEATURE_TAG: the set of layout feature tags that will be retained
  * in the subset.
+ * @HB_SUBSET_SETS_LAYOUT_SCRIPT_TAG: the set of layout script tags that will be retained
+ * in the subset. Defaults to all tags. Since: REPLACEME
  *
  * List of sets that can be configured on the subset input.
  *
@@ -113,6 +115,7 @@ typedef enum {
   HB_SUBSET_SETS_NAME_ID,
   HB_SUBSET_SETS_NAME_LANG_ID,
   HB_SUBSET_SETS_LAYOUT_FEATURE_TAG,
+  HB_SUBSET_SETS_LAYOUT_SCRIPT_TAG,
 } hb_subset_sets_t;
 
 HB_EXTERN hb_subset_input_t *

test/subset/data/profiles/filter-scripts-features.2.txt

@@ -0,0 +1,2 @@
+--layout-scripts=grek,latn
+--layout-features=liga

test/subset/data/profiles/filter-scripts-features.txt

@@ -0,0 +1,2 @@
+--layout-scripts=grek,cyrl
+--layout-features=numr,onum

test/subset/data/profiles/filter-scripts.txt

@@ -0,0 +1 @@
+--layout-scripts=grek,cyrl

test/subset/data/profiles/no-scripts.txt

@@ -0,0 +1 @@
+--layout-scripts-=*

test/subset/data/tests/full-font.tests

@@ -5,6 +5,10 @@ SourceSerifVariable-Roman.ttf
 PROFILES:
 default.txt
 drop-hints.txt
+no-scripts.txt
+filter-scripts.txt
+filter-scripts-features.txt
+filter-scripts-features.2.txt
 
 SUBSETS:
 abc

util/hb-subset.cc

@@ -532,7 +532,8 @@ set_flag (const char *name,
 }
 
 static gboolean
-parse_layout_features (const char *name,
+parse_layout_tag_list (hb_subset_sets_t set_type,
+                       const char *name,
                        const char *arg,
                        gpointer    data,
                        GError    **error G_GNUC_UNUSED)
@@ -540,22 +541,22 @@ parse_layout_features (const char *name,
   subset_main_t *subset_main = (subset_main_t *) data;
   hb_bool_t is_remove = (name[strlen (name) - 1] == '-');
   hb_bool_t is_add = (name[strlen (name) - 1] == '+');
-  hb_set_t *layout_features = hb_subset_input_set (subset_main->input, HB_SUBSET_SETS_LAYOUT_FEATURE_TAG);
+  hb_set_t *layout_tags = hb_subset_input_set (subset_main->input, set_type);
 
-  if (!is_remove && !is_add) hb_set_clear (layout_features);
+  if (!is_remove && !is_add) hb_set_clear (layout_tags);
 
   if (0 == strcmp (arg, "*"))
   {
-    hb_set_clear (layout_features);
+    hb_set_clear (layout_tags);
     if (!is_remove)
-      hb_set_invert (layout_features);
+      hb_set_invert (layout_tags);
     return true;
   }
 
   char *s = strtok((char *) arg, ", ");
   while (s)
   {
-    if (strlen (s) > 4) // table tags are at most 4 bytes
+    if (strlen (s) > 4) // tags are at most 4 bytes
     {
       g_set_error (error, G_OPTION_ERROR, G_OPTION_ERROR_BAD_VALUE,
                    "Failed parsing table tag at: '%s'", s);
@@ -565,9 +566,9 @@ parse_layout_features (const char *name,
     hb_tag_t tag = hb_tag_from_string (s, strlen (s));
 
     if (!is_remove)
-      hb_set_add (layout_features, tag);
+      hb_set_add (layout_tags, tag);
     else
-      hb_set_del (layout_features, tag);
+      hb_set_del (layout_tags, tag);
 
     s = strtok(nullptr, ", ");
   }
@@ -575,6 +576,34 @@ parse_layout_features (const char *name,
   return true;
 }
 
+static gboolean
+parse_layout_features (const char *name,
+		       const char *arg,
+		       gpointer    data,
+		       GError    **error)
+
+{
+  return parse_layout_tag_list (HB_SUBSET_SETS_LAYOUT_FEATURE_TAG,
+                                name,
+                                arg,
+                                data,
+                                error);
+}
+
+static gboolean
+parse_layout_scripts (const char *name,
+		       const char *arg,
+		       gpointer    data,
+		       GError    **error)
+
+{
+  return parse_layout_tag_list (HB_SUBSET_SETS_LAYOUT_SCRIPT_TAG,
+                                name,
+                                arg,
+                                data,
+                                error);
+}
+
 static gboolean
 parse_drop_tables (const char *name,
 		   const char *arg,
@@ -777,9 +806,15 @@ subset_main_t::add_options ()
     {"name-languages",	0, 0, G_OPTION_ARG_CALLBACK, (gpointer) &parse_name_languages,	"Subset nameRecords with specified language IDs. Use --name-languages-=... to substract from the current set.", "list of int numbers or *"},
     {"name-languages-",	0, G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, (gpointer) &parse_name_languages,	"Subset nameRecords with specified language IDs", "list of int numbers or *"},
     {"name-languages+",	0, G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, (gpointer) &parse_name_languages,	"Subset nameRecords with specified language IDs", "list of int numbers or *"},
+
     {"layout-features",	0, 0, G_OPTION_ARG_CALLBACK, (gpointer) &parse_layout_features,	"Specify set of layout feature tags that will be preserved. Use --layout-features-=... to substract from the current set.", "list of string table tags or *"},
-    {"layout-features+",0, G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, (gpointer) &parse_layout_features,	"Specify set of layout feature tags that will be preserved", "list of string table tags or *"},
+    {"layout-features+",0, G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, (gpointer) &parse_layout_features,	"Specify set of layout feature tags that will be preserved", "list of string tags or *"},
-    {"layout-features-",0, G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, (gpointer) &parse_layout_features,	"Specify set of layout feature tags that will be preserved", "list of string table tags or *"},
+    {"layout-features-",0, G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, (gpointer) &parse_layout_features,	"Specify set of layout feature tags that will be preserved", "list of string tags or *"},
+
+    {"layout-scripts",	0, 0, G_OPTION_ARG_CALLBACK, (gpointer) &parse_layout_scripts,	"Specify set of layout script tags that will be preserved. Use --layout-scripts-=... to substract from the current set.", "list of string table tags or *"},
+    {"layout-scripts+",0, G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, (gpointer) &parse_layout_scripts,	"Specify set of layout script tags that will be preserved", "list of string tags or *"},
+    {"layout-scripts-",0, G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, (gpointer) &parse_layout_scripts,	"Specify set of layout script tags that will be preserved", "list of string tags or *"},
+
     {"drop-tables",	0, 0, G_OPTION_ARG_CALLBACK, (gpointer) &parse_drop_tables,	"Drop the specified tables. Use --drop-tables-=... to substract from the current set.", "list of string table tags or *"},
     {"drop-tables+",	0, G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, (gpointer) &parse_drop_tables,	"Drop the specified tables.", "list of string table tags or *"},
     {"drop-tables-",	0, G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, (gpointer) &parse_drop_tables,	"Drop the specified tables.", "list of string table tags or *"},