Replace substring with string from second csv file - perl

Question

Earlier I was working on a loop within a loop and if a match was made it would replace the entire string from the second loop file. Now i have a slightly different situation. I'm trying to replace a substring from the first loop with a string from the second loop. They're both csv files and semicolon delimited. What i'm trying to replace are special characters: from the numerical code to the character itself The first file looks like:
1;2;bla&#322blabla &#261bla;7;8
3;4;bl&#261blabla;9;10
2;3;blablabla&#261ał8;9
and the second file has the numerical code and the corresponding character:
Ą;Ą
ą;ą
Ǟ;Ǟ
Á;Á
á;á
Â;Â
ł;ł
The first semicolon in the second file belongs to the numerical code of the corresponding character and should not be used to split the file. The result should be:
1;2;blałblabla ąbla;7;8
3;4;bląblabla;9;10
2;3;blablablaąał;8;9
This is the code I have. How can i fix this?
use strict;
use warnings;
my $inputfile1 = shift || die "input/output!\n";
my $inputfile2 = shift || die "input/output!\n";
my $outputfile = shift || die "output!\n";
open my $INFILE1, '<', $inputfile1 or die "Used/Not found :$!\n";
open my $INFILE2, '<', $inputfile2 or die "Used/Not found :$!\n";
open my $OUTFILE, '>', $outputfile or die "Used/Not found :$!\n";
my $infile2_pos = tell $INFILE2;
while (<$INFILE1>) {
s/"//g;
my #elements = split /;/, $_;
seek $INFILE2, $infile2_pos, 0;
while (<$INFILE2>) {
s/"//g;
my #loopelements = split /;/, $_;
#### The problem part ####
if (($elements[2] =~ /\&\#\d{3}\;/g) and (($elements[2]) eq ($loopelements[0]))){
$elements[2] =~ s/(\&\#\d{3}\;)/$loopelements[1]/g;
print "$2. elements[2]\n";
}
#### End problem part #####
}
my $output_line = join(";", #elements);
print $OUTFILE $output_line;
#print "\n"
}
close $INFILE1;
close $INFILE2;
close $OUTFILE;
exit 0;

Answer 1

Assuming your character codes are standard Unicode entities, you are better off using HTML::Entities to decode them.
This program processes the data you show in your first file and ignores the second file completely. The output seems to be what you want.
use strict;
use warnings;
use HTML::Entities 'decode_entities';
binmode STDOUT, ":utf8";
while (<DATA>) {
print decode_entities($_);
}
__DATA__
1;2;bla&#322blabla &#261bla;7;8
3;4;bl&#261blabla;9;10
2;3;blablabla&#261ał8;9
output
1;2;blałblabla ąbla;7;8
3;4;bląblabla;9;10
2;3;blablablaąał8;9

Answer 2

You split your #elements at every occurrence of ;, which is then removed. You will not find it in your data, the semicolon in your Regexp can never match, so no substitutions are done.
Anyway, using seek is somewhat disturbing for me. As you have a reasonable number of replacement codes (<5000), you might consider putting them into a hash:
my %subst;
while(<$INFILE2>){
/^&#(\d{3});;(.*)\n/;
$subst{$1} = $2;
}
Then we can do:
while(<$INFILE1>){
s| &# (\d{3}) | $subst{$1} // "&#$1" |egx;
# (don't try to concat undef
# when no substitution for our code is defined)
print $OUTFILE $_;
}
We do not have to split the files or view them as CSV data if replacement should occur everywhere in INFILE1. My solution should speed things up a bit (parsing INFILE2 only once). Here I assumed your input data is correct and the number codes are not terminated by a semicolon but by length. You might want to remove that from your Regexes.(i.e. m/&#\d{3}/)
If you have trouble with character encodings, you might want to open your files with :uft8 and/or use Encode or similar.