Aegisub/devel/traydict/unicodereplace.pl
Amar Takhar aa506bfe40 SVN Transition Step 6/7
1. svn mv OverLua SSATool athenasub avisynth_prs kanamemo \
     motiontracker prs traydict unit_test vsfilter devel/

* See r2749 for full description.

Originally committed to SVN as r2755.
2009-03-08 08:31:54 +00:00

31 lines
946 B
Perl

#!/usr/bin/perl
#############################
# unicodereplace.pl - replaces high codepoints in infile with \uHEX escapes
####
# Usage: unicodereplace.pl infile outfile
#############################
# oneliner version:
# perl -C7 -ne'print map {$c=ord; ($c < 127 or $c == 0xFEFF or $c == 0xFFFE) ? $_ : sprintf "\\u%04x", $c} split //;' infile > outfile
use warnings;
use strict;
use utf8;
my ($infile, $outfile) = @ARGV; # get arguments
open(INFILE, "<:utf8", $infile) or die("Can't open $infile: $!");
open(OUTFILE, ">:utf8", $outfile) or die("Can't open $outfile: $!");
# loop over lines in the infile
while (<INFILE>) {
for (split('', $_)) { # loop over characters in the line
my $cp = ord;
if ($cp < 127 or $cp == 0xFFFE or $cp == 0xFEFF) { # is it a normal ASCII codepoint or a BOM?
print OUTFILE $_; # then pass through unchanged
}
else {
print OUTFILE "\\u", sprintf("%04x", $cp); # otherwise print as \uHEX
}
}
}