blob: 9a7865c47ba69b96367ab5b202169422e06e1256 [file] [log] [blame]
#!/usr/bin/perl
# Remove HTML codes from data.
binmode(STDIN, ":utf-8");
binmode(STDOUT, ":utf-8");
my %map = (
"'" => "'",
"," => ",",
"&" => "&",
">" => ">",
"&lt;" => "<",
"&quot;" => "\"",
"&#257;" => "ā",
"&#257:" => "ā",
"&#257 " => "ā",
"&#7751;" => "Ṇ",
"&#7751:" => "Ṇ",
"&#7779;" => "Ṣ",
"&#7779:" => "Ṣ",
"&#;" => "",
);
while (my $line = <>) {
foreach my $key (keys %map) {
$line =~ s/$key/$map{$key}/g;
}
print $line;
}