From 919adb4b57977d5e375dab0fa943b6e81fa145ab Mon Sep 17 00:00:00 2001
From: Ito Hiroyuki <ZXB01226@nifty.com>
Date: Tue, 24 Aug 2010 10:11:51 +0000
Subject: [w3m-dev 04393] [patch] locale-related character management

---
 libwc/map/mk_ucs_isdigit_map.pl | 56 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 56 insertions(+)
 create mode 100644 libwc/map/mk_ucs_isdigit_map.pl

(limited to 'libwc/map/mk_ucs_isdigit_map.pl')

diff --git a/libwc/map/mk_ucs_isdigit_map.pl b/libwc/map/mk_ucs_isdigit_map.pl
new file mode 100644
index 0000000..c67f645
--- /dev/null
+++ b/libwc/map/mk_ucs_isdigit_map.pl
@@ -0,0 +1,56 @@
+
+open(MAP, "> ucs_isdigit.map");
+print MAP <<EOF;
+/*
+   File generated from UnicodeData-4.1.0.txt.
+*/
+
+EOF
+
+for (<DATA>) {
+  chop;
+  ($name, $class) = split;
+  
+  @cp = ();
+  
+  open(UCD, "< private/UnicodeData-4.1.0.txt");
+  while(<UCD>) {
+    chop;
+    @entry = split(';');
+    last if $entry[0] =~ m/.{5,}/;
+    if ($entry[2] eq $class) {
+      push (@cp, $entry[0]);
+    }
+  }
+  close UCD;
+
+  @bs = ();
+  $last = -1;
+  $seq = -1;
+  for my $e (@cp) {
+    if (++$last != hex $e) {
+      $seq = $e;
+      $last = hex $e;
+      push (@bs, $seq);
+    }
+    $end{$seq} = $e;
+  }
+  $nobs = @bs;
+  
+  print MAP <<EOF;
+
+#define N_ucs_${name}_map ${nobs}
+
+static wc_map ucs_${name}_map[ N_ucs_${name}_map ] = {
+EOF
+  
+  for (@bs) {
+    print MAP "  { 0x$_, 0x$end{$_} },\n";
+  }
+  print MAP <<EOF
+};
+EOF
+}
+
+__END__
+isdigit Nd
-- 
cgit v1.2.3