• file
• revisions
• annotate
• diff
• comparison
• raw

intl/uconv/tools/parse-mozilla-encoding-table.pl@b8a032363ba2 (annotated)

intl/uconv/tools/parse-mozilla-encoding-table.pl

Thu, 22 Jan 2015 13:21:57 +0100

author

Michael Schloh von Bennewitz <michael@schloh.com>

date

Thu, 22 Jan 2015 13:21:57 +0100

branch

TOR_BUG_9701

changeset 15

b8a032363ba2

permissions

-rw-r--r--

Incorporate requested changes from Mozilla in review:
https://bugzilla.mozilla.org/show_bug.cgi?id=1123480#c6

michael@0	1	#!/usr/bin/perl
michael@0	2	# parse-mozilla-encoding-table.pl, version 0.6
michael@0	3	#
michael@0	4	# Script to deassemble existing Mozilla *.uf or *.ut files
michael@0	5	# back to source conversion tables.
michael@0	6	# by Anthony Fok <anthony@thizlinux.com>, ThizLinux Laboratory Ltd., 2002/11/27
michael@0	7	# License: GNU General Public License, version 2 or newer
michael@0	8	#
michael@0	9	# Used for verifying HKSCS-1999 hkscs.uf and hkscs.ut so that I can make
michael@0	10	# new ones for HKSCS-2001. This script is quick-and-dirty and not very
michael@0	11	# robust, so if the debug output of fromu/tou ever changes, this script
michael@0	12	# will need to be modified too. :-)
michael@0	13
michael@0	14	my %data = ();
michael@0	15	my $mappingPos = 0;
michael@0	16	my $filename = shift;
michael@0	17	my $mode;
michael@0	18	if ($filename =~ /\.(ut|uf)$/) {
michael@0	19	print $filename, "\n";
michael@0	20	$mode = $1;
michael@0	21	} else {
michael@0	22	die;
michael@0	23	}
michael@0	24
michael@0	25	open(INFILE, "<$filename") or die;
michael@0	26
michael@0	27	# Quick-and-dirty routine to populate %data
michael@0	28	while (<INFILE>) {
michael@0	29	if (/^Begin of Item ([[:xdigit:]]+)/) {
michael@0	30	die if defined($itemId) and hex($itemId) + 1 != hex($1);
michael@0	31	$itemId = $1;
michael@0	32	<INFILE> =~ /Format ([012])/ or die;
michael@0	33	$format = $1;
michael@0	34	<INFILE> =~ /srcBegin = ([[:xdigit:]]+)/ or die;
michael@0	35	$srcBegin = $1;
michael@0	36
michael@0	37	if ($format == 0) { # Range
michael@0	38	<INFILE> =~ /srcEnd = ([[:xdigit:]]+)/ or die;
michael@0	39	$srcEnd = $1;
michael@0	40	<INFILE> =~ /destBegin = ([[:xdigit:]]+)/ or die;
michael@0	41	$destBegin = $1;
michael@0	42
michael@0	43	for ($i = hex($srcBegin); $i <= hex($srcEnd); $i++) {
michael@0	44	$data{sprintf("%04X",$i)} = sprintf("%04X",
michael@0	45	hex($destBegin) + $i - hex($srcBegin));
michael@0	46	}
michael@0	47
michael@0	48	<INFILE> =~ /^End of Item $itemId\s*$/ or die;
michael@0	49	}
michael@0	50	elsif ($format == 1) { # Mapping
michael@0	51	<INFILE> =~ /srcEnd = ([[:xdigit:]]+)/ or die;
michael@0	52	$srcEnd = $1;
michael@0	53	<INFILE> =~ /mappingOffset = ([[:xdigit:]]+)/ or die;
michael@0	54	$mappingOffset = hex($1);
michael@0	55	die unless $mappingOffset == $mappingPos;
michael@0	56	<INFILE> =~ /Mapping =\s*$/ or die;
michael@0	57	until ($_ = <INFILE>, /^End of Item/) {
michael@0	58	chop;
michael@0	59	for $i (split ' ') {
michael@0	60	$key = sprintf("%04X", hex($srcBegin) - $mappingOffset + $mappingPos++);
michael@0	61	next if $i eq "FFFD";
michael@0	62	if (defined($data{$key})) {
michael@0	63	print "Error: doubly defined. $key was $data{$key}, and now $i.\n";
michael@0	64	} else {
michael@0	65	$data{$key} = $i;
michael@0	66	}
michael@0	67	}
michael@0	68	}
michael@0	69	die unless $mappingPos - $mappingOffset == hex($srcEnd) - hex($srcBegin) + 1;
michael@0	70	/^End of Item $itemId\s*$/ or die;
michael@0	71	}
michael@0	72	else { # Single ($format == 2)
michael@0	73	<INFILE> =~ /destBegin = ([[:xdigit:]]+)/ or die;
michael@0	74	$destBegin = $1;
michael@0	75	$data{$srcBegin} = $destBegin;
michael@0	76	<INFILE> =~ /^End of Item $itemId\s*$/ or die;
michael@0	77	}
michael@0	78	}
michael@0	79	}
michael@0	80
michael@0	81	# Generate conversion table
michael@0	82	for $key (sort keys %data) {
michael@0	83	if ($mode eq "ut") {
michael@0	84	print "0x$key\t0x$data{$key}\n";
michael@0	85	} elsif ($mode eq "uf") {
michael@0	86	print "0x$data{$key}\t0x$key\n";
michael@0	87	} else {
michael@0	88	die;
michael@0	89	}
michael@0	90	}
michael@0	91
michael@0	92	close INFILE;