Merge branch 'wip/smcv/compose-files' into 'master'

Fix Compose handling on big-endian host architectures

Closes #4217

See merge request GNOME/gtk!3902

gtk/compose/compose-parse.c

@@ -7,7 +7,8 @@
  * character data, and definitions for the builtin compose table of GTK.
  * Run it like this:
  *
- *   compose-parse Compose sequences chars gtkcomposedata.h
+ *   cpp -DXCOMM='#' Compose.pre | sed -e 's/^ *#/#/' > Compose
+ *   compose-parse Compose sequences-little-endian sequences-big-endian chars gtkcomposedata.h
  *
  * The GTK build expects the output files to be in the source tree, in
  * the gtk/compose directory.
@@ -15,15 +16,19 @@
 int
 main (int argc, char *argv[])
 {
+  const guint16 *sequences_le;
+  const guint16 *sequences_be;
+  guint16 *other_sequences;
   GtkComposeTable *table;
   GError *error = NULL;
   GString *str;
+  gsize i;
 
   setlocale (LC_ALL, "");
 
   if (argc < 5)
     {
-      g_print ("Usage: compose-parse INPUT OUTPUT1 OUTPUT2 OUTPUT3\n");
+      g_print ("Usage: compose-parse INPUT SEQUENCES-LE SEQUENCES-BE CHARS HEADER\n");
       exit (1);
     }
 
@@ -31,11 +36,26 @@ main (int argc, char *argv[])
   if (!table)
     g_error ("Failed to parse %s", argv[1]);
 
+#if G_BYTE_ORDER == G_BIG_ENDIAN
+  sequences_le = other_sequences = g_new0 (guint16, table->data_size);
+  sequences_be = (const guint16 *) table->data;
+#else
+  sequences_le = (const guint16 *) table->data;
+  sequences_be = other_sequences = g_new0 (guint16, table->data_size);
+#endif
+
+  for (i = 0; i < table->data_size; i++)
+    other_sequences[i] = GUINT16_SWAP_LE_BE (table->data[i]);
+
   /* data_size is the size in guint16 */
-  if (!g_file_set_contents (argv[2], (char *)table->data, 2 * table->data_size, &error))
+  if (!g_file_set_contents (argv[2], (char *) sequences_le, 2 * table->data_size, &error))
     g_error ("%s", error->message);
 
-  if (!g_file_set_contents (argv[3], table->char_data, table->n_chars + 1, &error))
+  /* data_size is the size in guint16 */
+  if (!g_file_set_contents (argv[3], (char *) sequences_be, 2 * table->data_size, &error))
+    g_error ("%s", error->message);
+
+  if (!g_file_set_contents (argv[4], table->char_data, table->n_chars + 1, &error))
     g_error ("%s", error->message);
 
   str = g_string_new ("");
@@ -55,10 +75,11 @@ main (int argc, char *argv[])
                    "\n"
                    "#endif\n");
 
-  if (!g_file_set_contents (argv[4], str->str, str->len, &error))
+  if (!g_file_set_contents (argv[5], str->str, str->len, &error))
     g_error ("%s", error->message);
 
   g_string_free (str, TRUE);
+  g_free (other_sequences);
 
   return 0;
 }

gtk/compose/gtkcomposedata.h

@@ -3,7 +3,7 @@
 
 #define MAX_SEQ_LEN 5
 #define N_INDEX_SIZE 30
-#define DATA_SIZE 16447
+#define DATA_SIZE 16477
 #define N_CHARS 1241
 
 #endif

gtk/gen-gtk-gresources-xml.py

@@ -21,6 +21,7 @@ def replace_if_changed(new, old):
     os.remove(new)
 
 srcdir = sys.argv[1]
+endian = sys.argv[2]
 
 xml = '''<?xml version='1.0' encoding='UTF-8'?>
 <gresources>
@@ -84,13 +85,13 @@ for f in get_files('inspector', '.ui'):
 xml += '''
     <file>inspector/inspector.css</file>
     <file>emoji/en.data</file>
-    <file>compose/sequences</file>
+    <file alias="compose/sequences">compose/sequences-{0}-endian</file>
     <file>compose/chars</file>
   </gresource>
-</gresources>'''
+</gresources>'''.format(endian)
 
-if len(sys.argv) > 2:
-  outfile = sys.argv[2]
+if len(sys.argv) > 3:
+  outfile = sys.argv[3]
   tmpfile = outfile + '~'
   with open(tmpfile, 'w') as f:
     f.write(xml)

gtk/meson.build

@@ -838,6 +838,7 @@ gtk_gresources_xml = configure_file(output: 'gtk.gresources.xml',
   command: [
     gen_gtk_gresources_xml,
     meson.current_source_dir(),
+    host_machine.endian(),
     '@OUTPUT@'
   ],
 )

testsuite/gtk/compose/system.expected

@@ -1,7 +1,7 @@
-# n_sequences: 4874
+# n_sequences: 4884
 # max_seq_len: 5
 # n_index_size: 30
-# data_size: 16447
+# data_size: 16477
 # n_chars: 1241
 <U7ae> <U7e9> : "ΐ" # U390
 <U7ae> <U7f5> : "ΰ" # U3b0
@@ -1959,6 +1959,7 @@
 <Uff20> <U47> <U2e> : "Ġ" # U120
 <Uff20> <U47> <U54> : ">" # U3e
 <Uff20> <U47> <U55> : "Ğ" # U11e
+<Uff20> <U47> <U75> : "Ğ" # U11e
 <Uff20> <U47> <U1a2> : "Ğ" # U11e
 <Uff20> <U48> <U2c> : "Ḩ" # U1e28
 <Uff20> <U49> <U22> : "Ï" # Ucf
@@ -2512,6 +2513,7 @@
 <Uff20> <U67> <U2e> : "ġ" # U121
 <Uff20> <U67> <U55> : "ğ" # U11f
 <Uff20> <U67> <U74> : ">" # U3e
+<Uff20> <U67> <U75> : "ğ" # U11f
 <Uff20> <U67> <U1a2> : "ğ" # U11f
 <Uff20> <U68> <U2c> : "ḩ" # U1e29
 <Uff20> <U69> <U22> : "ï" # Uef
@@ -2611,11 +2613,19 @@
 <Uff20> <U75> <U3b> : "ų" # U173
 <Uff20> <U75> <U3e> : "û" # Ufb
 <Uff20> <U75> <U41> : "Ă" # U102
+<Uff20> <U75> <U45> : "Ĕ" # U114
+<Uff20> <U75> <U47> : "Ğ" # U11e
+<Uff20> <U75> <U49> : "Ĭ" # U12c
+<Uff20> <U75> <U4f> : "Ŏ" # U14e
 <Uff20> <U75> <U55> : "Ŭ" # U16c
 <Uff20> <U75> <U5e> : "û" # Ufb
 <Uff20> <U75> <U5f> : "ū" # U16b
 <Uff20> <U75> <U60> : "ù" # Uf9
 <Uff20> <U75> <U61> : "ă" # U103
+<Uff20> <U75> <U65> : "ĕ" # U115
+<Uff20> <U75> <U67> : "ğ" # U11f
+<Uff20> <U75> <U69> : "ĭ" # U12d
+<Uff20> <U75> <U6f> : "ŏ" # U14f
 <Uff20> <U75> <U75> : "ŭ" # U16d
 <Uff20> <U75> <U7e> : "ũ" # U169
 <Uff20> <U75> <Ua8> : "ü" # Ufc