svn commit: r196140 - in user/edwin/locale: . etc tools

[Edwin Groothuis]

Author: edwin
Date: Wed Aug 12 12:49:52 2009
New Revision: 196140
URL: http://svn.freebsd.org/changeset/base/196140

Log:
  - Follow up of move from tools/*.xml and tools/charmaps to etc/.
  - Add etc/unicode.conf to specify the unicode.org data.
  - Improve argument handling in cldr2def.

Added:
  user/edwin/locale/etc/unicode.conf
Modified:
  user/edwin/locale/Makefile
  user/edwin/locale/tools/charmaps.pm
  user/edwin/locale/tools/cldr2def.pl
  user/edwin/locale/tools/whatis.pl

Modified: user/edwin/locale/Makefile
==============================================================================
--- user/edwin/locale/Makefile	Wed Aug 12 12:47:47 2009	(r196139)
+++ user/edwin/locale/Makefile	Wed Aug 12 12:49:52 2009	(r196140)
@@ -2,17 +2,17 @@
 # $FreeBSD$
 #
 
-UNICODEDIR?=	/home/edwin/unicode/
-CLDRVERSION?=	1.7.1
-CLDRDIR?=	${UNICODEDIR}/cldr/${CLDRVERSION}/
-UNIDATAVERSION?=5.2.0
-UNIDATADIR?=	${UNICODEDIR}/UNIDATA/${UNIDATAVERSION}/
+CLDRDIR!=	grep ^cldr etc/unicode.conf | cut -f 2 -d " "
+UNIDATADIR!=	grep ^unidata etc/unicode.conf | cut -f 2 -d " "
 
-XMLDIR?=	/home/edwin/svn/edwin/locale/tools/
-XMLFILE?=	charmaps.xml
+ETCDIR=		${.CURDIR}/etc
 
 TYPES?=		monetdef numericdef msgdef timedef
 
+.if defined(LC)
+LC:=	--lc=${LC}
+.endif
+
 all:
 .for t in ${TYPES}
 	test -d ${t} || mkdir ${t}
@@ -30,7 +30,11 @@ install:
 .for t in ${TYPES}
 build-${t}:
 	test -d ${t} || mkdir ${t}
-	perl -I tools tools/cldr2def.pl ${CLDRDIR} ${UNIDATADIR} ${XMLDIR} ${XMLDIR}/${XMLFILE} ${t} ${LC}
+	perl -I tools tools/cldr2def.pl \
+		--cldr=$$(realpath ${CLDRDIR}) \
+		--unidata=$$(realpath ${UNIDATADIR}) \
+		--etc=$$(realpath ${ETCDIR}) \
+		--type=${t} ${LC}
 .endfor
 
 clean:

Added: user/edwin/locale/etc/unicode.conf
==============================================================================
--- /dev/null	00:00:00 1970	(empty, because file is newly added)
+++ user/edwin/locale/etc/unicode.conf	Wed Aug 12 12:49:52 2009	(r196140)
@@ -0,0 +1,2 @@
+cldr ~/unicode/cldr/1.7.1
+unidata ~/unicode/UNIDATA/5.2.0

Modified: user/edwin/locale/tools/charmaps.pm
==============================================================================
--- user/edwin/locale/tools/charmaps.pm	Wed Aug 12 12:47:47 2009	(r196139)
+++ user/edwin/locale/tools/charmaps.pm	Wed Aug 12 12:49:52 2009	(r196140)
@@ -13,9 +13,9 @@ my %d = ();
 my $index = -1;
 
 sub get_xmldata {
-	my $xmlfile = shift;
+	my $etcdir = shift;
 
-	open(FIN, $xmlfile);
+	open(FIN, "$etcdir/charmaps.xml");
 	my @xml = <FIN>;
 	chomp(@xml);
 	close(FIN);

Modified: user/edwin/locale/tools/cldr2def.pl
==============================================================================
--- user/edwin/locale/tools/cldr2def.pl	Wed Aug 12 12:47:47 2009	(r196139)
+++ user/edwin/locale/tools/cldr2def.pl	Wed Aug 12 12:49:52 2009	(r196140)
@@ -8,23 +8,33 @@ use strict;
 use XML::Parser;
 use Tie::IxHash;
 use Data::Dumper;
+use Getopt::Long;
 use Digest::SHA qw(sha1_hex);
 require "charmaps.pm";
 
+
 if ($#ARGV < 2) {
-	print "Usage: $0 <cldrdir> <unidatadir> <xmldirs> <charmaps> <type> [la_CC]\n";
+	print "Usage: $0 --cldr=<cldrdir> --unidata=<unidatadir> --etc=<etcdir> --type=<type> [--lc=<la_CC>]\n";
 	exit(1);
 }
 
 my $DEFENCODING = "UTF-8";
-my $CLDRDIR = shift(@ARGV);
-my $UNIDATADIR = shift(@ARGV);
-my $XMLDIR = shift(@ARGV);
-my $CHARMAPS = shift(@ARGV);
-my $TYPE = shift(@ARGV);
-my $doonly = shift(@ARGV);
 my @filter = ();
 
+my $CLDRDIR = undef;
+my $UNIDATADIR = undef;
+my $ETCDIR = undef;
+my $TYPE = undef;
+my $doonly = undef;
+
+my $result = GetOptions (
+		"cldr=s"	=> \$CLDRDIR,
+		"unidata=s"	=> \$UNIDATADIR,
+		"etc=s"		=> \$ETCDIR,
+		"type=s"	=> \$TYPE,
+		"lc=s"		=> \$doonly
+	    );
+
 my %convertors = ();
 
 my %ucd = ();
@@ -40,7 +50,7 @@ my %utf8map = ();
 my %utf8aliases = ();
 get_unidata($UNIDATADIR);
 get_utf8map("$CLDRDIR/posix/$DEFENCODING.cm");
-get_encodings("$XMLDIR/charmaps");
+get_encodings("$ETCDIR/charmaps");
 
 my %keys = ();
 tie(%keys, "Tie::IxHash");
@@ -198,7 +208,8 @@ sub callback_altmon {
 sub get_unidata {
 	my $directory = shift;
 
-	open(FIN, "$directory/UnicodeData.txt");
+	open(FIN, "$directory/UnicodeData.txt")
+	    or die("Cannot open $directory/UnicodeData.txt");;
 	my @lines = <FIN>;
 	chomp(@lines);
 	close(FIN);
@@ -276,7 +287,7 @@ sub get_encodings {
 }
 
 sub get_languages {
-	my %data = get_xmldata($CHARMAPS);
+	my %data = get_xmldata($ETCDIR);
 	%languages = %{$data{L}}; 
 	%translations = %{$data{T}}; 
 	%alternativemonths = %{$data{AM}}; 
@@ -315,7 +326,7 @@ sub get_fields {
 		$file .= $c;
 
 		my $filename = "$CLDRDIR/posix/$file.$DEFENCODING.src";
-		$filename = "$XMLDIR/$file.$DEFENCODING.src"
+		$filename = "$ETCDIR/$file.$DEFENCODING.src"
 		    if (! -f $filename);
 		if (! -f $filename
 		 && defined $languages{$l}{$f}{fallback}) {

Modified: user/edwin/locale/tools/whatis.pl
==============================================================================
--- user/edwin/locale/tools/whatis.pl	Wed Aug 12 12:47:47 2009	(r196139)
+++ user/edwin/locale/tools/whatis.pl	Wed Aug 12 12:49:52 2009	(r196140)
@@ -1,8 +1,8 @@
 #!/bin/sh
 
-UNIDATA=/home/edwin/unicode/UNIDATA/5.2.0/UnicodeData.txt
-CHARMAPS=/home/edwin/svn/edwin/locale/tools/charmaps
-UTF8=~/unicode/cldr/1.7.1/posix/UTF-8.cm
+UNIDATA=$(grep ^unidata etc/unicode.conf | cut -f 2 -d " ")
+UTF8=$(grep ^cldr etc/unicode.conf | cut -f 2 -d " ")/UTF-8.cm
+CHARMAPS=etc/charmaps
 
 if [ -z "$1" ]; then
 	echo "Usage: $0 <unicode string>"