corp/sql/fixnews.pl

#!/usr/local/bin/perl

use DBI;

my $dbh = DBI->connect("DBI:Pg:dbname=corp","","") || die $DBI::errstr;

sub nukehtml {
        my $foo = $_[0];
        $foo=~s/<\/*[^>]+>//g;
        $foo=~s/^\s+//g;
        $foo=~s/\s+$//g;
        return $foo;
}

while(<>) {
        chomp;
        tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/;      # 1250 -> iso8859-2
        if (/^(-*\d+)\t(.+)$/) {
                ($t_id,$t_title) = ($1,$2);
                $body=~s/\s+/ /g;
                $body=~s/'/\\'/g;
                $title=~s/'/\\'/g;
                print "id: $id\ntitle: $title\ndate: $date\n$body\n----------\n";
                if ($title) {
                        $dbh->do("insert into news (title,id,date,body) values ('$title',$id,'$date','$body')") || die $dbh->errstr();
                }
                $title=nukehtml($t_title);
                $id=$t_id;
                $body="";
        } elsif (/Zagreb/i && (/199[89]/ || /200[01]/)) {
                $date=nukehtml($_);

#               if ($date=~/,\s*(\d+).*\s(\w+)\s+(\d+)/) {
#                       $date.="-- $1 $2 $3 --";
#               }

        } else {
                $body.=$_;
        }
}
1	dpavlin	1.1	#!/usr/local/bin/perl
2
3			use DBI;
4
5			my $dbh = DBI->connect("DBI:Pg:dbname=corp","","") \|\| die $DBI::errstr;
6
7			sub nukehtml {
8			my $foo = $_[0];
9			$foo=~s/<\/*[^>]+>//g;
10			$foo=~s/^\s+//g;
11			$foo=~s/\s+$//g;
12			return $foo;
13			}
14
15			while(<>) {
16			chomp;
17			tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2
18			if (/^(-*\d+)\t(.+)$/) {
19			($t_id,$t_title) = ($1,$2);
20			$body=~s/\s+/ /g;
21			$body=~s/'/\\'/g;
22			$title=~s/'/\\'/g;
23			print "id: $id\ntitle: $title\ndate: $date\n$body\n----------\n";
24			if ($title) {
25			$dbh->do("insert into news (title,id,date,body) values ('$title',$id,'$date','$body')") \|\| die $dbh->errstr();
26			}
27			$title=nukehtml($t_title);
28			$id=$t_id;
29			$body="";
30			} elsif (/Zagreb/i && (/199[89]/ \|\| /200[01]/)) {
31			$date=nukehtml($_);
32
33			# if ($date=~/,\s(\d+).\s(\w+)\s+(\d+)/) {
34			# $date.="-- $1 $2 $3 --";
35			# }
36
37			} else {
38			$body.=$_;
39			}
40			}