From 3c9f3e036bf6a40bb1055cb41e94583189678e7d Mon Sep 17 00:00:00 2001
From: aland <aland>
Date: Mon, 8 Aug 2005 20:22:14 +0000
Subject: [PATCH] 	Script to format dictionaries in a consistent manner,
 to avoid 	hand-formatting issues

---
 share/Makefile  |  15 +++++
 share/format.pl | 176 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 191 insertions(+)
 create mode 100644 share/Makefile
 create mode 100755 share/format.pl

diff --git a/share/Makefile b/share/Makefile
new file mode 100644
index 0000000..ab342b4
--- /dev/null
+++ b/share/Makefile
@@ -0,0 +1,15 @@
+#
+#	Scripts to format dictionary files.
+#
+#	$Id$
+#
+.PHONY: format
+
+#
+#  This should only be run by hand, and then sanity checked by hand!
+#
+format: dictionary*
+	@for x in dictionary* ; do \
+		cat $$x | ./format.pl > tmp; \
+		mv tmp $$x; \
+	done
diff --git a/share/format.pl b/share/format.pl
new file mode 100755
index 0000000..be4caf8
--- /dev/null
+++ b/share/format.pl
@@ -0,0 +1,176 @@
+#!/usr/bin/env perl
+#
+#  Format the dictionaries according to a standard scheme.
+#
+#  Usage: cat dictionary | ./format.pl > new
+#
+#  We don't over-write the dictionaries in place, so that the process
+#  can be double-checked by hand.
+#
+#  This is a bit of a hack.
+#
+#  FIXME: get lengths from variables, rather than hard-coding.
+#
+#  $Id$
+#
+
+$begin_vendor = 0;
+$blank = 0;
+
+while (<>) {
+    #
+    #  Clear out trailing whitespace
+    #
+    s/[ \t]+$//;
+
+    #
+    #  Suppress multiple blank lines
+    #
+    if (/^\s+$/) {
+	next if ($blank == 1);
+	$blank = 1;
+	print "\n";
+	next;
+    }
+    $blank = 0;
+
+    #
+    #  Remember the vendor
+    #
+    if (/^VENDOR\s+([\w-]+)\s+(\d+)/) {
+	$name=$1;
+	$len = length $name;
+	if ($len < 32) {
+	    $lenx = 31 - $len;
+	    $lenx += 7;		# round up
+	    $lenx /= 8;
+	    $lenx = int $lenx;
+	    $tabs = "\t" x $lenx;
+	} else {
+	    $tabs = " ";
+	}
+	print "VENDOR\t$name$tabs$2\n";
+	$vendor = $name;
+	next;
+    }
+
+    #  
+    #  Remember if we did begin-vendor.
+    #
+    if (/^BEGIN-VENDOR\s+([\w-]+)/) {
+	$begin_vendor = 1;
+	print "BEGIN-VENDOR\t$vendor\n";
+	next;
+    }
+
+    #
+    #  Get attribute.
+    #
+    if (/^ATTRIBUTE\s+([\w-]+)\s+(\d+)\s+(\w+)(.*)/) {
+	$name=$1;
+	$len = length $name;
+	if ($len < 40) {
+	    $lenx = 40 - $len;
+	    $lenx += 7;		# round up
+	    $lenx /= 8;
+	    $lenx = int $lenx;
+	    $tabs = "\t" x $lenx;
+	    if ($tabs eq "") {
+		$tabs = " ";
+	    }
+	} else {
+	    $tabs = " ";
+	}
+
+	$value = $2;
+	$type = $3;
+	$stuff = $4;
+
+	#
+	#  See if it's old format, with the vendor at the end of
+	#  the line.  If so, make it the new format.
+	#
+	if ($stuff =~ /$vendor/) {
+	    if ($begin_vendor == 0) {
+		print "BEGIN-VENDOR\t$vendor\n\n";
+		$begin_vendor = 1;
+	    }
+	    $stuff =~ s/$vendor//;
+	    $stuff =~ s/\s+$//;
+	}
+
+	print "ATTRIBUTE\t$name$tabs$value\t$type$stuff\n";
+	next;
+    }
+
+    #
+    #  Values.
+    #
+    if (/^VALUE\s+([\w-]+)\s+([\w-]+)\s+(\d+)/) {
+	$attr=$1;
+	$len = length $attr;
+	if ($len < 32) {
+	    $lenx = 32 - $len;
+	    $lenx += 7;		# round up
+	    $lenx /= 8;
+	    $lenx = int $lenx;
+	    $tabsa = "\t" x $lenx;
+	    if ($tabsa eq "") {
+		$tabsa = " ";
+		$len += 1;
+	    } else {
+		$len -= $len % 8;
+		$len += 8 * length $tabsa;
+	    }
+	} else {
+	    $tabsa = " ";
+	    $len += 1;
+	}
+
+	#
+	#  For the code below, we assume that the attribute lengths
+	#
+	if ($len < 32) {
+	    $lena = 0;
+	} else {
+	    $lena = $len - 32;
+	}
+
+	$name=$2;
+	$len = length $name;
+	if ($len < 24) {
+	    $lenx = 24 - $lena - $len;
+	    $lenx += 7;		# round up
+	    $lenx /= 8;
+	    $lenx = int $lenx;
+	    $tabsn = "\t" x $lenx;
+	    if ($tabsn eq "") {
+		$tabsn = " ";
+	    }
+	} else {
+	    $tabsn = " ";
+	}
+
+	print "VALUE\t$attr$tabsa$name$tabsn$3\n";
+	next;
+    }
+
+    #
+    #  Remember if we did this.
+    #
+    if (/^END-VENDOR/) {
+	$begin_vendor = 0;
+    }
+
+    #
+    #  Everything else gets dumped out as-is.
+    #
+    print;
+}
+
+#
+#  If we changed the format, print the end vendor, too.
+#
+if ($begin_vendor) {
+    print "\nEND-VENDOR\t$vendor\n";
+}
-- 
2.1.4