Make PangoEngine{,Lang,Shape} GObjects, and use a GTypeModule-based

Sat Aug  2 23:19:16 2003  Owen Taylor  <otaylor@redhat.com>

        * pango/pango-engine.[ch] modules/*/*-{fc,win32,x}.c
        pango/modules.c pango/break.c pango/pango-context.c
        pango/pango-layout.c pango/pango-modules.h
        pango/querymodules.c pango/shape.c: Make
        PangoEngine{,Lang,Shape} GObjects, and use a
        GTypeModule-based module-loading system closely based
        on the one used for GtkIMContext and GtkThemeEngine.

        * pango/pango-impl-utils.h: OK, I'm tired of typing
        in get_type() functions.

        * pango/pango-script.[ch] pango/pango-script-table.h
        tests/testscript.c tools/gen-script-table.pl: Add port
        of script-range code from ICU in preparation for future
        use. (#91542)

        * tools/gen-script-for-lang.c: Utility program to determine
        the script for each fontconfig .orth file.

        * docs/tmpl/{scripts.sgml,pango-engine-lang.sgml,
        pango-engine-shape.sgml} docs/pango-sections.txt docs/pango-docs.sgml:
        Redo to go along with the above changes.

        * configure.in: chmod +x tests/runtests.sh

4 files changed, 315 insertions, 0 deletions

diff --git a/tools/.cvsignore b/tools/.cvsignore
index c038ed78..24f67f13 100644
--- a/tools/.cvsignore
+++ b/tools/.cvsignore
@@ -1,2 +1,3 @@
+gen-script-for-lang
 Makefile
 Makefile.in
\ No newline at end of file
diff --git a/tools/Makefile.am b/tools/Makefile.am
index d43cf4a6..2206e601 100644
--- a/tools/Makefile.am
+++ b/tools/Makefile.am
@@ -1,8 +1,20 @@
 ## Process this file with automake to create Makefile.in.
 
+INCLUDES =					\
+	-I$(top_srcdir)				\
+	$(GLIB_CFLAGS)				\
+	-DG_DISABLE_DEPRECATED			\
+	-DPANGO_DISABLE_DEPRECATED		\
+	$(PANGO_DEBUG_FLAGS)
+
+noinst_PROGRAMS = gen-script-for-lang
+
 EXTRA_DIST= 			\
 	add-copyright 		\
 	compress-table.pl 	\
 	make-table.sh 		\
 	maps/README		\
 	maps/tis-620
+
+gen_script_for_lang_SOURCES = gen-script-for-lang.c
+gen_script_for_lang_LDADD = ../pango/libpango-$(PANGO_API_VERSION).la
diff --git a/tools/gen-script-for-lang.c b/tools/gen-script-for-lang.c
new file mode 100644
index 00000000..ee1580e2
--- /dev/null
+++ b/tools/gen-script-for-lang.c
@@ -0,0 +1,232 @@
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include <pango/pango-enum-types.h>
+#include <pango/pango-script.h>
+#include <pango/pango-types.h>
+#include <pango/pango-utils.h>
+
+PangoScript script_for_file (const char *base_dir,
+			     const char *file_part);
+
+const char *get_script_name (PangoScript script)
+{
+  static GEnumClass *class = NULL;
+  if (!class)
+    class = g_type_class_ref (PANGO_TYPE_SCRIPT);
+  
+  GEnumValue *value = g_enum_get_value (class, script);
+  g_assert (value);
+
+  return value->value_name;
+}
+
+int fail (char *format, ...)
+{
+  va_list vap;
+  
+  va_start (vap, format);
+  vfprintf (stderr, format, vap);
+  va_end (vap);
+  
+  exit (1);
+}
+
+gboolean scan_hex (const char **str, gunichar *result)
+{
+  const char *end;
+
+  *result = strtol (*str, (char **)&end, 16);
+  if (end == *str)
+    return FALSE;
+
+  *str = end;
+  return TRUE;
+}
+
+void warn_mismatch (const char *file_part,
+		    PangoScript script1,
+		    PangoScript script2)
+{
+  g_printerr ("%s has characters for both %s and %s\n",
+	      file_part, get_script_name (script1), get_script_name (script2));
+}
+
+PangoScript script_for_line (const char *base_dir,
+			     const char *file_part,
+			     const char *str)
+{
+  gunichar start_char;
+  gunichar end_char;
+  gunichar ch;
+  PangoScript result = PANGO_SCRIPT_COMMON;
+  const char *p = str;
+
+  if (g_str_has_prefix (str, "include"))
+    {
+      GString *file_part = g_string_new (NULL);
+      
+      str += strlen ("include");
+      if (!pango_skip_space (&str))
+	goto err;
+
+      if (!pango_scan_string (&str, file_part) ||
+	  pango_skip_space (&str))
+	goto err;
+
+      result = script_for_file (base_dir, file_part->str);
+      g_string_free (file_part, TRUE);
+
+      return result;
+    }
+  
+  /* Format is HEX_DIGITS or HEX_DIGITS-HEX_DIGITS */
+  if (!scan_hex (&p, &start_char))
+    goto err;
+  
+  end_char = start_char;
+
+  pango_skip_space (&p);
+  if (*p == '-')
+    {
+      p++;
+      if (!scan_hex (&p, &end_char))
+	goto err;
+
+      pango_skip_space (&p);
+    }
+  if (*p != '\0')
+    goto err;
+
+  for (ch = start_char; ch <= end_char; ch++)
+    {
+      PangoScript script = pango_script_for_unichar (ch);
+      if (script != PANGO_SCRIPT_COMMON &&
+	  script != PANGO_SCRIPT_INHERITED)
+	{
+	  if (result != PANGO_SCRIPT_COMMON && script != result)
+	    {
+	      warn_mismatch (file_part, script, result);
+	      return PANGO_SCRIPT_INVALID_CODE;
+	    }
+	  result = script;
+	}
+    }
+
+  return result;
+  
+ err:
+  fail ("While processing '%s', cannot parse line: '%s'\n", file_part, str);
+  return PANGO_SCRIPT_INVALID_CODE; /* Not reached */
+}
+     
+PangoScript script_for_file (const char *base_dir,
+			     const char *file_part)
+{
+  GError *error = NULL;
+  char *filename = g_build_filename (base_dir, file_part, NULL);
+  GIOChannel *channel = g_io_channel_new_file (filename, "r", &error);
+  GIOStatus status = G_IO_STATUS_NORMAL;
+  PangoScript result = PANGO_SCRIPT_COMMON;
+
+  if (!channel)
+    fail ("Error opening '%s': %s\n", filename, error);
+
+  /* The files have ISO-8859-1 copyright signs in them */
+  if (!g_io_channel_set_encoding (channel, "ISO-8859-1", &error))
+      fail ("Cannot set encoding when reading '%s': %s\n", filename, error);
+  
+  while (status == G_IO_STATUS_NORMAL)
+    {
+      char *str;
+      size_t term;
+      char *comment;
+      
+      status = g_io_channel_read_line  (channel, &str, NULL, &term, &error);
+      switch (status)
+	{
+	case G_IO_STATUS_NORMAL:
+	  str[term] = '\0';
+	  comment = strchr (str, '#');
+	  if (comment)
+	    *comment = '\0';
+	  g_strstrip (str);
+	  if (str[0] != '\0') 	/* Empty */
+	    {
+	      PangoScript script = script_for_line (base_dir, file_part, str);
+	      if (script != PANGO_SCRIPT_COMMON &&
+		  script != PANGO_SCRIPT_INHERITED)
+		{
+		  if (result != PANGO_SCRIPT_COMMON && script != result)
+		    {
+		      warn_mismatch (file_part, script, result);
+		      result = PANGO_SCRIPT_INVALID_CODE;
+		      status = G_IO_STATUS_EOF;
+		    }
+		  else
+		    result = script;
+		}
+	    }
+	  g_free (str);
+	  break;
+	case G_IO_STATUS_EOF:
+	  break;
+	case G_IO_STATUS_ERROR:
+	  fail ("Error reading '%s': %s\n", filename, error->message);
+	  break;
+	case G_IO_STATUS_AGAIN:
+	  g_assert_not_reached ();
+	  break;
+	}
+    }
+
+  if (!g_io_channel_shutdown (channel, FALSE, &error))
+    fail ("Error closing '%s': %s\n", channel, error);
+
+  g_free (filename);
+
+  return result;
+}
+
+int main (int argc, char **argv)
+{
+  GDir *dir;
+  GError *error = NULL;
+
+  g_type_init ();
+
+  if (argc != 2)
+    fail ("Usage: gen-script-for-lang DIR > script-for-lang.h\n");
+
+  dir = g_dir_open (argv[1], 0, &error);
+  if (!dir)
+    fail ("%s\n", error->message);
+
+  while (TRUE)
+    {
+      const char *name = g_dir_read_name (dir);
+      if (!name)
+	break;
+
+      if (g_str_has_suffix (name, ".orth"))
+	{
+	  char *langpart = g_strndup (name,
+				      strlen (name) - strlen (".orth"));
+	  PangoLanguage *lang = pango_language_from_string (langpart);
+	  PangoScript script;
+
+	  script = script_for_file (argv[1], name);
+	  g_print ("%s - %s\n",
+		   pango_language_to_string (lang),
+		   get_script_name (script));
+	  g_free (langpart);
+	}
+    }
+
+  g_dir_close (dir);
+  
+  return 0;
+}
+
diff --git a/tools/gen-script-table.pl b/tools/gen-script-table.pl
new file mode 100755
index 00000000..6b524ef7
--- /dev/null
+++ b/tools/gen-script-table.pl
@@ -0,0 +1,70 @@
+#!/usr/bin/perl -w 
+#
+# Script to convert http://www.unicode.org/Public/UNIDATA/Scripts.txt
+# into a machine-readable table.
+#
+######################################################################
+
+if (@ARGV != 1) {
+    die "Usage: gen-script-table.pl Scripts.txt > pango-script-table.h\n";
+}
+
+open IN, $ARGV[0] || die "Cannot open $ARGV[0]: $!\n";
+
+my @ranges;
+my $file;
+
+while (<IN>) {
+    if (/^\#\s+(Scripts-.*.txt)/) {
+	$file = $1;
+    }
+    
+    s/#.*//;
+    next if /^\s*$/;
+    if (!/^([0-9A-F]+)(?:\.\.([0-9A-F]+))?\s+;\s+([A-Z_]+)\s+$/) {
+	die "Cannot parse line: $_\n";
+    }
+
+    if (defined $2) {
+	push @ranges, [ hex $1, hex $2, $3 ];
+    } else {
+	push @ranges, [ hex $1, hex $1, $3 ];
+    }
+}
+
+$date = gmtime;
+
+print <<"EOT";
+/* pango-script-table.h: Generated by gen-script-table.pl
+ *
+ *  Date: $date
+ *  Source: $file
+ *
+ * Do not edit.   
+ */
+static const struct {
+    gunichar    start;
+    guint16     chars;
+    guint16     script;		/* PangoScript */
+} pango_script_table[] = { 
+EOT
+
+@ranges = sort { $a->[0] <=> $b->[0] } @ranges;
+
+for (my $i = 0; $i <= $#ranges; $i++) {
+    my $start = $ranges[$i]->[0];
+    my $end = $ranges[$i]->[1];
+    my $script = $ranges[$i]->[2];
+
+    while ($i <= $#ranges - 1 &&
+	   $ranges[$i + 1]->[0] == $end + 1 &&
+	   $ranges[$i + 1]->[2] eq $script) {
+	$i++;
+	$end = $ranges[$i]->[1];
+    }
+
+    printf " { %#06x, %5d, PANGO_SCRIPT_%s },\n", $start, $end - $start + 1, $script;
+}
+
+printf "};\n";
+