Redo the handling of Emoji data

Use the data files from https://github.com/milesj/emojibase.git
as source for our Emoji data. Slightly change our data format by
adding a group to each item, in both the Emoji data and in the
setting for recent-emoji.

Install translated versions of the data as separate resource
bundles in $prefix/gtk-4.0/emoji, and load them when appropriate.

Currently, we have data for de, en, es, fr, zh, with data taken
from Unicode 13 and CLDR 13.

Fixes: #950 #1511

gtk/emoji/README.md

@@ -0,0 +1,23 @@
+Emoji data
+==========
+
+We use Emoji data from Unicode and the CLDR, stored in a GVariant.
+The immediate source for our data is the json files from
+
+  https://github.com/milesj/emojibase.git
+
+To convert the data from that repository to a GVariant that GTK
+can use, the convert-emoji tool can be used:
+
+  convert-emoji $emojibase/packages/data/de/data.raw.json de.data
+
+for example (for German).
+
+To make these usable by GTK, we wrap them in a resource bundle
+that has the GVariant as
+
+   /org/gtk/libgtk/emoji/de.data
+
+and install the resulting resource bundle at this location:
+
+  /usr/share/gtk-4.0/emoji/de.gresource

gtk/emoji/convert-emoji.c

@@ -18,26 +18,40 @@
 /* Build with gcc -o convert-emoji convert-emoji.c `pkg-config --cflags --libs json-glib-1.0`
  */
 
-/* The format of the generated data is: a(auss).
+/* Reads data from the json files in emojibase, expecting
+ * language-specific data.raw.json as input
+ */
+
+/* The format of the generated data is: a(ausasu).
  * Each member of the array has the following fields:
  * au - sequence of unicode codepoints. If the
  *      sequence contains a 0, it marks the point
  *      where skin tone modifiers should be inserted
  * s  - name, e.g. "man worker"
  * as - keywords, e.g. "man", "worker"
+ * u  - the group that this item belongs to:
+ *      0: smileys-emotion
+ *      1: people-body
+ *      2: component
+ *      3: animals-nature
+ *      4: food-drink
+ *      5: travel-places
+ *      6: activities
+ *      7: objects
+ *      8: symbols
+ *      9: flags
  */
 #include <json-glib/json-glib.h>
 #include <string.h>
 
 gboolean
 parse_code (GVariantBuilder *b,
-            const char      *code,
-            GString         *name_key)
+            const char      *code)
 {
   g_auto(GStrv) strv = NULL;
   int j;
 
-  strv = g_strsplit (code, " ", -1);
+  strv = g_strsplit (code, "-", -1);
   for (j = 0; strv[j]; j++)
     {
       guint32 u;
@@ -52,12 +66,7 @@ parse_code (GVariantBuilder *b,
       if (0x1f3fb <= u && u <= 0x1f3ff)
         g_variant_builder_add (b, "u", 0);
       else
-        {
-          g_variant_builder_add (b, "u", u);
-          if (j > 0)
-            g_string_append_c (name_key, '-');
-          g_string_append_printf (name_key, "%x", u);
-        }
+        g_variant_builder_add (b, "u", u);
     }
 
   return TRUE;
@@ -68,10 +77,10 @@ main (int argc, char *argv[])
 {
   JsonParser *parser;
   JsonNode *root;
-  JsonArray *array;
   JsonObject *ro;
+  JsonArray *array;
   JsonNode *node;
-  const char *name;
+  const char *unicode;
   JsonObjectIter iter;
   GError *error = NULL;
   guint length, i;
@@ -81,40 +90,14 @@ main (int argc, char *argv[])
   GHashTable *names;
   GString *name_key;
 
-  if (argc != 4)
+  if (argc != 3)
     {
-      g_print ("Usage: emoji-convert INPUT INPUT1 OUTPUT\n");
+      g_print ("Usage: emoji-convert INPUT OUTPUT\n");
       return 1;
     }
 
   parser = json_parser_new ();
 
-  if (!json_parser_load_from_file (parser, argv[2], &error))
-    {
-      g_error ("%s", error->message);
-      return 1;
-    }
-
-  root = json_parser_get_root (parser);
-  ro = json_node_get_object (root);
-  json_object_iter_init (&iter, ro);
-
-  names = g_hash_table_new_full (g_str_hash, g_str_equal, g_free, (GDestroyNotify)json_object_unref);
-  name_key = g_string_new ("");
-
-  while (json_object_iter_next (&iter, &name, &node))
-    {
-      JsonObject *obj = json_node_get_object (node);
-      const char *unicode;
-
-      unicode = json_object_get_string_member (obj, "unicode");
-      g_hash_table_insert (names, g_strdup (unicode), json_object_ref (obj));
-    }
-
-  g_object_unref (parser);
-
-  parser = json_parser_new ();
-
   if (!json_parser_load_from_file (parser, argv[1], &error))
     {
       g_error ("%s", error->message);
@@ -125,101 +108,59 @@ main (int argc, char *argv[])
   array = json_node_get_array (root);
   length = json_array_get_length (array);
 
-  g_variant_builder_init (&builder, G_VARIANT_TYPE ("a(ausas)"));
-  i = 0;
-  while (i < length)
+  g_variant_builder_init (&builder, G_VARIANT_TYPE ("a(ausasu)"));
+  for (i = 0; i < length; i++)
     {
-      JsonNode *node = json_array_get_element (array, i);
-      JsonObject *obj = json_node_get_object (node);
+      JsonObject *obj = json_array_get_object_element (array, i);
       GVariantBuilder b1;
       GVariantBuilder b2;
+      guint group;
       const char *name;
       char *code;
-      int j, k;
-      gboolean skip;
-      gboolean has_variations;
-      JsonObject *obj2;
-      JsonArray *kw;
-      char **name_tokens;
 
-      i++;
+      if (!json_object_has_member (obj, "group"))
+        continue;
+
+      group = json_object_get_int_member (obj, "group");
+      name = json_object_get_string_member (obj, "annotation");
+
+      if (json_object_has_member (obj, "skins"))
+        {
+          JsonArray *a2 = json_object_get_array_member (obj, "skins");
+          JsonNode *n2 = json_array_get_element (a2, 0);
+          JsonObject *o2 = json_node_get_object (n2);
+          code = g_strdup (json_object_get_string_member (o2, "hexcode"));
+        }
+      else
+        {
+          code = g_strdup (json_object_get_string_member (obj, "hexcode"));
+        }
 
       g_variant_builder_init (&b1, G_VARIANT_TYPE ("au"));
 
-      name = json_object_get_string_member (obj, "name");
-      code = g_strdup (json_object_get_string_member (obj, "code"));
-
-      has_variations = FALSE;
-      while (i < length)
-        {
-          JsonNode *node2 = json_array_get_element (array, i);
-          JsonObject *obj2 = json_node_get_object (node2);
-          const char *name2;
-          const char *code2;
-
-          name2 = json_object_get_string_member (obj2, "name");
-          code2 = json_object_get_string_member (obj2, "code");
-
-          if (!strstr (name2, "skin tone") || !g_str_has_prefix (name2, name))
-            break;
-
-          if (!has_variations)
-            {
-              has_variations = TRUE;
-              g_free (code);
-              code = g_strdup (code2);
-            }
-          i++;
-        }
-
-      g_string_set_size (name_key, 0);
-      if (!parse_code (&b1, code, name_key))
+      if (!parse_code (&b1, code))
         return 1;
 
       g_variant_builder_init (&b2, G_VARIANT_TYPE ("as"));
-      name_tokens = g_str_tokenize_and_fold (name, "en", NULL);
-      for (j = 0; j < g_strv_length (name_tokens); j++)
-        g_variant_builder_add (&b2, "s", name_tokens[j]);
-
-      obj2 = g_hash_table_lookup (names, name_key->str);
-      if (obj2)
+      if (json_object_has_member (obj, "tags"))
         {
-          kw = json_object_get_array_member (obj2, "keywords");
-          for (k = 0; k < json_array_get_length (kw); k++)
-            {
-              char **folded;
-              char **ascii;
-
-              folded = g_str_tokenize_and_fold (json_array_get_string_element (kw, k), "en", &ascii);
-              for (j = 0; j < g_strv_length (folded); j++)
-                {
-                  if (!g_strv_contains ((const char * const *)name_tokens, folded[j]))
-                    g_variant_builder_add (&b2, "s", folded[j]);
-                }
-              for (j = 0; j < g_strv_length (ascii); j++)
-                {
-                  if (!g_strv_contains ((const char * const *)name_tokens, ascii[j]))
-                    g_variant_builder_add (&b2, "s", ascii[j]);
-                }
-              g_strfreev (folded);
-              g_strfreev (ascii);
-            }
+          JsonArray *tags = json_object_get_array_member (obj, "tags");
+          for (int j = 0; j < json_array_get_length (tags); j++)
+            g_variant_builder_add (&b2, "s", json_array_get_string_element (tags, j));
         }
 
-      g_strfreev (name_tokens);
-
-      g_variant_builder_add (&builder, "(ausas)", &b1, name, &b2);
+      g_variant_builder_add (&builder, "(ausasu)", &b1, name, &b2, group);
     }
 
   v = g_variant_builder_end (&builder);
-  if (g_str_has_suffix (argv[3], ".json"))
+  if (g_str_has_suffix (argv[2], ".json"))
     {
       JsonNode *node;
       char *out;
 
       node = json_gvariant_serialize (v);
       out = json_to_string (node, TRUE);
-      if (!g_file_set_contents (argv[3], out, -1, &error))
+      if (!g_file_set_contents (argv[2], out, -1, &error))
         {
           g_error ("%s", error->message);
           return 1;
@@ -230,7 +171,7 @@ main (int argc, char *argv[])
       GBytes *bytes;
 
       bytes = g_variant_get_data_as_bytes (v);
-      if (!g_file_set_contents (argv[3], g_bytes_get_data (bytes, NULL), g_bytes_get_size (bytes), &error))
+      if (!g_file_set_contents (argv[2], g_bytes_get_data (bytes, NULL), g_bytes_get_size (bytes), &error))
         {
           g_error ("%s", error->message);
           return 1;

gtk/emoji/gresource.xml.in

@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<gresources>
+  <gresource prefix="/org/gtk/libgtk/emoji/">
+    <file>@lang@.data</file>
+  </gresource>
+</gresources>

gtk/gen-gtk-gresources-xml.py

@@ -86,7 +86,7 @@ for f in get_files('inspector', '.ui'):
 xml += '''
     <file>inspector/logo.png</file>
     <file>inspector/inspector.css</file>
-    <file>emoji/emoji.data</file>
+    <file>emoji/en.data</file>
   </gresource>
 </gresources>'''
 

gtk/gtkemojichooser.c

@@ -38,6 +38,8 @@
 #include "gtknative.h"
 #include "gtkwidgetprivate.h"
 #include "gdk/gdkprofilerprivate.h"
+#include "gtkmain.h"
+#include "gtkprivate.h"
 
 /**
  * SECTION:gtkemojichooser
@@ -184,7 +186,7 @@ typedef struct {
   GtkWidget *box;
   GtkWidget *heading;
   GtkWidget *button;
-  const char *first;
+  int group;
   gunichar label;
   gboolean empty;
 } EmojiSection;
@@ -337,8 +339,8 @@ add_recent_item (GtkEmojiChooser *chooser,
 
   g_variant_ref (item);
 
-  g_variant_builder_init (&builder, G_VARIANT_TYPE ("a((ausas)u)"));
-  g_variant_builder_add (&builder, "(@(ausas)u)", item, modifier);
+  g_variant_builder_init (&builder, G_VARIANT_TYPE ("a((ausasu)u)"));
+  g_variant_builder_add (&builder, "(@(ausasu)u)", item, modifier);
 
   children = NULL;
   for (child = gtk_widget_get_last_child (chooser->recent.box);
@@ -363,7 +365,7 @@ add_recent_item (GtkEmojiChooser *chooser,
           continue;
         }
 
-      g_variant_builder_add (&builder, "(@(ausas)u)", item2, modifier2);
+      g_variant_builder_add (&builder, "(@(ausasu)u)", item2, modifier2);
     }
   g_list_free (children);
 
@@ -591,6 +593,79 @@ add_emoji (GtkWidget    *box,
   gtk_flow_box_insert (GTK_FLOW_BOX (box), child, prepend ? 0 : -1);
 }
 
+GBytes *
+get_emoji_data (void)
+{
+  GBytes *bytes;
+  const char *lang;
+  char q[10];
+  char *path;
+  GError *error = NULL;
+
+  lang = pango_language_to_string (gtk_get_default_language ());
+  if (strchr (lang, '-'))
+    {
+      int i;
+      for (i = 0; lang[i] != '-' && i < 9; i++)
+        q[i] = lang[i];
+      q[i] = '\0';
+      lang = q;
+    }
+
+  path = g_strconcat ("/org/gtk/libgtk/emoji/", lang, ".data", NULL);
+  bytes = g_resources_lookup_data (path, 0, &error);
+  if (bytes)
+    {
+      g_debug ("Found emoji data for %s in resource %s", lang, path);
+      g_free (path);
+      return bytes;
+    }
+
+  if (g_error_matches (error, G_RESOURCE_ERROR, G_RESOURCE_ERROR_NOT_FOUND))
+    {
+      char *filename;
+      GMappedFile *file;
+
+      g_clear_error (&error);
+
+      filename = g_strconcat ("/usr/share/gtk-4.0/emoji/", lang, ".gresource", NULL);
+      file = g_mapped_file_new (filename, FALSE, NULL);
+
+      if (file)
+        {
+          GBytes *data;
+          GResource *resource;
+
+          data = g_mapped_file_get_bytes (file);
+          g_mapped_file_unref (file);
+
+          resource = g_resource_new_from_data (data, NULL);
+          g_bytes_unref (data);
+
+          g_debug ("Registering resource for Emoji data for %s from file %s", lang, filename);
+          g_resources_register (resource);
+          g_resource_unref (resource);
+
+          bytes = g_resources_lookup_data (path, 0, NULL);
+          if (bytes)
+            {
+              g_debug ("Found emoji data for %s in resource %s", lang, path);
+              g_free (path);
+              g_free (filename);
+              return bytes;
+            }
+        }
+
+      g_free (filename);
+    }
+
+  g_clear_error (&error);
+
+  g_free (path);
+
+  return g_resources_lookup_data ("/org/gtk/libgtk/emoji/en.data", 0, NULL);
+}
+
 static gboolean
 populate_emoji_chooser (gpointer data)
 {
@@ -602,8 +677,11 @@ populate_emoji_chooser (gpointer data)
 
   if (!chooser->data)
     {
-      GBytes *bytes = g_resources_lookup_data ("/org/gtk/libgtk/emoji/emoji.data", 0, NULL);
-      chooser->data = g_variant_ref_sink (g_variant_new_from_bytes (G_VARIANT_TYPE ("a(ausas)"), bytes, TRUE));
+      GBytes *bytes;
+
+      bytes = get_emoji_data ();
+
+      chooser->data = g_variant_ref_sink (g_variant_new_from_bytes (G_VARIANT_TYPE ("a(ausasu)"), bytes, TRUE));
       g_bytes_unref (bytes);
     }
 
@@ -615,25 +693,27 @@ populate_emoji_chooser (gpointer data)
 
   while ((item = g_variant_iter_next_value (chooser->iter)))
     {
-      const char *name;
+      guint group;
 
-      g_variant_get_child (item, 1, "&s", &name);
+      g_variant_get_child (item, 3, "u", &group);
 
-      if (strcmp (name, chooser->body.first) == 0)
+      if (group == chooser->people.group)
+        chooser->box = chooser->people.box;
+      else if (group == chooser->body.group)
         chooser->box = chooser->body.box;
-      else if (strcmp (name, chooser->nature.first) == 0)
+      else if (group == chooser->nature.group)
         chooser->box = chooser->nature.box;
-      else if (strcmp (name, chooser->food.first) == 0)
+      else if (group == chooser->food.group)
         chooser->box = chooser->food.box;
-      else if (strcmp (name, chooser->travel.first) == 0)
+      else if (group == chooser->travel.group)
         chooser->box = chooser->travel.box;
-      else if (strcmp (name, chooser->activities.first) == 0)
+      else if (group == chooser->activities.group)
         chooser->box = chooser->activities.box;
-      else if (strcmp (name, chooser->objects.first) == 0)
+      else if (group == chooser->objects.group)
         chooser->box = chooser->objects.box;
-      else if (strcmp (name, chooser->symbols.first) == 0)
+      else if (group == chooser->symbols.group)
         chooser->box = chooser->symbols.box;
-      else if (strcmp (name, chooser->flags.first) == 0)
+      else if (group == chooser->flags.group)
         chooser->box = chooser->flags.box;
 
       add_emoji (chooser->box, FALSE, item, 0, chooser);
@@ -848,11 +928,11 @@ stop_search (GtkEntry *entry,
 
 static void
 setup_section (GtkEmojiChooser *chooser,
-               EmojiSection   *section,
-               const char     *first,
-               const char     *icon)
+               EmojiSection    *section,
+               int              group,
+               const char      *icon)
 {
-  section->first = first;
+  section->group = group;
 
   gtk_button_set_icon_name (GTK_BUTTON (section->button), icon);
 
@@ -898,16 +978,16 @@ gtk_emoji_chooser_init (GtkEmojiChooser *chooser)
   adj = gtk_scrolled_window_get_vadjustment (GTK_SCROLLED_WINDOW (chooser->scrolled_window));
   g_signal_connect (adj, "value-changed", G_CALLBACK (adj_value_changed), chooser);
 
-  setup_section (chooser, &chooser->recent, NULL, "emoji-recent-symbolic");
-  setup_section (chooser, &chooser->people, "grinning face", "emoji-people-symbolic");
-  setup_section (chooser, &chooser->body, "selfie", "emoji-body-symbolic");
-  setup_section (chooser, &chooser->nature, "monkey face", "emoji-nature-symbolic");
-  setup_section (chooser, &chooser->food, "grapes", "emoji-food-symbolic");
-  setup_section (chooser, &chooser->travel, "globe showing Europe-Africa", "emoji-travel-symbolic");
-  setup_section (chooser, &chooser->activities, "jack-o-lantern", "emoji-activities-symbolic");
-  setup_section (chooser, &chooser->objects, "muted speaker", "emoji-objects-symbolic");
-  setup_section (chooser, &chooser->symbols, "ATM sign", "emoji-symbols-symbolic");
-  setup_section (chooser, &chooser->flags, "chequered flag", "emoji-flags-symbolic");
+  setup_section (chooser, &chooser->recent, -1, "emoji-recent-symbolic");
+  setup_section (chooser, &chooser->people, 0, "emoji-people-symbolic");
+  setup_section (chooser, &chooser->body, 1, "emoji-body-symbolic");
+  setup_section (chooser, &chooser->nature, 3, "emoji-nature-symbolic");
+  setup_section (chooser, &chooser->food, 4, "emoji-food-symbolic");
+  setup_section (chooser, &chooser->travel, 5, "emoji-travel-symbolic");
+  setup_section (chooser, &chooser->activities, 6, "emoji-activities-symbolic");
+  setup_section (chooser, &chooser->objects, 7, "emoji-objects-symbolic");
+  setup_section (chooser, &chooser->symbols, 8, "emoji-symbols-symbolic");
+  setup_section (chooser, &chooser->flags, 9, "emoji-flags-symbolic");
 
   populate_recent_section (chooser);
 

gtk/gtkemojicompletion.c

@@ -623,8 +623,8 @@ gtk_emoji_completion_init (GtkEmojiCompletion *completion)
 
   gtk_widget_init_template (GTK_WIDGET (completion));
 
-  bytes = g_resources_lookup_data ("/org/gtk/libgtk/emoji/emoji.data", 0, NULL);
-  completion->data = g_variant_ref_sink (g_variant_new_from_bytes (G_VARIANT_TYPE ("a(ausas)"), bytes, TRUE));
+  bytes = get_emoji_data ();
+  completion->data = g_variant_ref_sink (g_variant_new_from_bytes (G_VARIANT_TYPE ("a(ausasu)"), bytes, TRUE));
 
   g_bytes_unref (bytes);
 

gtk/gtkprivate.h

@@ -126,6 +126,8 @@ void  gtk_set_display_debug_flags (GdkDisplay *display,
 guint gtk_get_display_debug_flags (GdkDisplay *display);
 gboolean gtk_get_any_display_debug_flag_set (void);
 
+GBytes *get_emoji_data (void);
+
 #ifdef G_ENABLE_DEBUG
 
 #define GTK_DISPLAY_DEBUG_CHECK(display,type) (gtk_get_any_display_debug_flag_set () && G_UNLIKELY (gtk_get_display_debug_flags (display) & GTK_DEBUG_##type))

gtk/meson.build

@@ -861,6 +861,23 @@ gtkresources = gnome.compile_resources('gtkresources',
   extra_args: '--manual-register',
 )
 
+foreach lang : [ 'de', 'fr', 'es', 'zh' ]
+  conf = configuration_data()
+  conf.set('lang', lang)
+  resxml = configure_file(input: 'emoji/gresource.xml.in',
+                          output: lang + '.gresource.xml',
+                          configuration: conf
+  )
+
+  gnome.compile_resources(lang,
+    resxml,
+    source_dir: 'emoji',
+    gresource_bundle: true,
+    install: true,
+    install_dir: join_paths(gtk_datadir, 'gtk-4.0', 'emoji')
+  )
+endforeach
+
 gtk_x11_sources = files([
   'gtkapplication-x11.c',
   'gtkmountoperation-x11.c',

gtk/org.gtk.gtk4.Settings.EmojiChooser.gschema.xml

@@ -2,7 +2,7 @@
 <schemalist>
 
   <schema id='org.gtk.gtk4.Settings.EmojiChooser' path='/org/gtk/gtk4/settings/emoji-chooser/'>
-    <key name='recent-emoji' type='a((ausas)u)'>
+    <key name='recent-emoji' type='a((ausasu)u)'>
       <default>[]</default>
       <summary>Recently used Emoji</summary>
       <description>