Sort scripts based on frequency.

2007-10-28  Behdad Esfahbod  <behdad@gnome.org>

        * pango/pango-script-lang-table.h:
        * tools/gen-script-for-lang-new.c:
        Sort scripts based on frequency.

        * tools/Makefile.am: Don't build anything by default.


svn path=/trunk/; revision=2485

2 files changed, 106 insertions, 18 deletions

diff --git a/tools/Makefile.am b/tools/Makefile.am
index 8702a62e..fdc746e9 100644
--- a/tools/Makefile.am
+++ b/tools/Makefile.am
@@ -7,7 +7,7 @@ INCLUDES =					\
 	-DPANGO_DISABLE_DEPRECATED		\
 	$(PANGO_DEBUG_FLAGS)
 
-noinst_PROGRAMS = 		\
+EXTRA_PROGRAMS = 		\
 	gen-script-for-lang	\
 	gen-script-for-lang-new
 
diff --git a/tools/gen-script-for-lang-new.c b/tools/gen-script-for-lang-new.c
index fb5eebc8..2ef49859 100644
--- a/tools/gen-script-for-lang-new.c
+++ b/tools/gen-script-for-lang-new.c
@@ -24,6 +24,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <time.h>
 
 #include <pango/pango-enum-types.h>
 #include <pango/pango-script.h>
@@ -34,8 +35,13 @@
 #define MAX_SCRIPTS 3
 
 typedef struct {
+  PangoScript script;
+  int freq;
+} ScriptInfo;
+
+typedef struct {
   PangoLanguage *lang;
-  PangoScript scripts[MAX_SCRIPTS];
+  ScriptInfo scripts[MAX_SCRIPTS];
 } LangInfo;
 
 static const char *get_script_name (PangoScript script)
@@ -81,17 +87,19 @@ script_for_char (gunichar   ch,
 
       for (j = 0; j < MAX_SCRIPTS; j++)
 	{
-	  if (info->scripts[j] == script)
+	  if (info->scripts[j].script == script)
 	    break;
-	  if (info->scripts[j] == PANGO_SCRIPT_INVALID_CODE)
+	  if (info->scripts[j].script == PANGO_SCRIPT_COMMON)
 	    {
-	      info->scripts[j] = script;
+	      info->scripts[j].script = script;
 	      break;
 	    }
 	}
 
       if (j == MAX_SCRIPTS)
-	fail ("More than %d scripts found for %s\n", MAX_SCRIPTS, pango_language_to_string (info->lang));
+	fail ("More than %d scripts found for %s.  Increase MAX_SCRIPTS.\n", MAX_SCRIPTS, pango_language_to_string (info->lang));
+
+      info->scripts[j].freq++;
     }
 }
      
@@ -140,7 +148,10 @@ do_lang (GArray        *script_array,
   info.lang = pango_language_from_string ((const char *)lang);
 
   for (j = 0; j < MAX_SCRIPTS; j++)
-    info.scripts[j] = PANGO_SCRIPT_INVALID_CODE;
+    {
+      info.scripts[j].script = PANGO_SCRIPT_COMMON;
+      info.scripts[j].freq = 0;
+    }
   
   scripts_for_lang (&info);
 
@@ -148,6 +159,36 @@ do_lang (GArray        *script_array,
 }
 
 static int
+compare_script (gconstpointer a,
+		gconstpointer b,
+		gpointer      data)
+{
+  const ScriptInfo *info_a = a;
+  const ScriptInfo *info_b = b;
+  G_GNUC_UNUSED LangInfo *lang_info = data;
+
+  /* first compare frequencies, higher first */
+  if (info_a->freq > info_b->freq)
+    return -1;
+  if (info_a->freq < info_b->freq)
+    return +1;
+
+  /* next compare script indices, higher first (it's more specific) */
+  if (info_a->script > info_b->script)
+    return -1;
+  if (info_a->script < info_b->script)
+    return +1;
+
+  /* for stability, next compare pointers themselves, smaller first */
+  if (info_a < info_b)
+    return -1;
+  if (info_a > info_b)
+    return +1;
+
+  return 0;
+}
+
+static int
 compare_lang (gconstpointer a,
 	      gconstpointer b)
 {
@@ -161,12 +202,21 @@ compare_lang (gconstpointer a,
 int main (void)
 {
   GArray *script_array;
+
   unsigned int i;
   int j;
   int max_lang_len = 0;
+  int max_script_len = 0;
+
   FcStrList *langs;
   FcChar8* lang;
 
+  char date_buf[200];
+  const char *date_str = "unknown";
+  time_t t;
+  struct tm *tmp;
+  int fc_version;
+
   g_type_init ();
 
   script_array = g_array_new (FALSE, FALSE, sizeof (LangInfo));
@@ -185,36 +235,74 @@ int main (void)
   for (i = 0; i < script_array->len; i++)
     {
       LangInfo *info = &g_array_index (script_array, LangInfo, i);
-      
+
       max_lang_len = MAX (max_lang_len,
-			  1 + (int)strlen (pango_language_to_string (info->lang)));
-    }
+			  (int)strlen (pango_language_to_string (info->lang)));
+
+      g_qsort_with_data (info->scripts,
+			 G_N_ELEMENTS (info->scripts),
+			 sizeof (info->scripts[0]), 
+			 compare_script,
+			 info);
+
+      for (j = 0; j < MAX_SCRIPTS; j++)
+        if (!info->scripts[j].freq)
+	  break;
       
-  g_print ("typedef struct {\n"
+      max_script_len = MAX (max_script_len, j);
+    }
+
+  if ((t = time(NULL), tmp = localtime (&t)) && strftime(date_buf, sizeof(date_buf), "%F", tmp))
+    date_str = date_buf;
+
+  fc_version = FcGetVersion ();
+
+  g_print ("/* pango-script-lang-table.h:\n"
+	   " * \n"
+	   " * Generated by %s\n"
+	   " * Date: %s\n"
+	   " * Source: fontconfig-%d.%d.%d\n"
+	   " * \n"
+	   " * Do not edit.\n"
+	   " */\n",
+	   __FILE__,
+	   date_str,
+	   fc_version / 10000, (fc_version / 100) % 100, fc_version % 100);
+
+  g_print ("typedef struct _PangoScriptForLang {\n"
 	   "  const char lang[%d];\n"
 	   "  PangoScript scripts[%d];\n"
 	   "} PangoScriptForLang;\n"
 	   "\n"
 	   "static const PangoScriptForLang pango_script_for_lang[] = {\n",
-	   max_lang_len,
-	   MAX_SCRIPTS);
+	   max_lang_len + 1,
+	   max_script_len);
   
   for (i = 0; i < script_array->len; i++)
     {
       LangInfo *info = &g_array_index (script_array, LangInfo, i);
       
-      g_print ("  { \"%s\", { ", pango_language_to_string (info->lang));
+      g_print ("  { \"%s\", %*s{ ",
+	       pango_language_to_string (info->lang),
+	       max_lang_len - strlen (pango_language_to_string (info->lang)), "");
       for (j = 0; j < MAX_SCRIPTS; j++)
 	{
+	  if (!info->scripts[j].freq)
+	    break;
+
 	  if (j != 0)
 	    g_print (", ");
-	  g_print ("%s", get_script_name (info->scripts[j]));
+	  g_print ("%s/*%d*/",
+		   get_script_name (info->scripts[j].script),
+		   info->scripts[j].freq);
 	}
-      g_print (" } },\n");
+      g_print (" } }");
+      if (i + 1 != script_array->len)
+	g_print (",");
+      g_print ("\n");
     }
 
   g_print ("};\n");
-  
-  
+
   return 0;
 }