#!/usr/bin/perl | |
# Remove HTML codes from data. | |
binmode(STDIN, ":utf-8"); | |
binmode(STDOUT, ":utf-8"); | |
my %map = ( | |
"'" => "'", | |
"," => ",", | |
"&" => "&", | |
">" => ">", | |
"<" => "<", | |
""" => "\"", | |
"ā" => "ā", | |
"ā:" => "ā", | |
"ā " => "ā", | |
"ṇ" => "Ṇ", | |
"ṇ:" => "Ṇ", | |
"ṣ" => "Ṣ", | |
"ṣ:" => "Ṣ", | |
"&#;" => "", | |
); | |
while (my $line = <>) { | |
foreach my $key (keys %map) { | |
$line =~ s/$key/$map{$key}/g; | |
} | |
print $line; | |
} |