>I have a scalar variable containing HTML that needs to be converted
>to XML. It's not the best HTML so it has invalid characters (like
>smart quotes, 1/2 character, etc.). I need to determine if these
>characters exist in the data and throw an error if they do. What
>is the best way to do this? I can't use an XML parser because it's
>not really XML.
Welp, ultimately, if you were using an XML parser, it would
choke on the bad data. For instance, this code:
use XML::Simple;
my $data = eval { XMLin( $xml_data ); };
if ($@) { print $@; }
Would produce error messages like:
There was an error loading morelikethisweblog.xml: not
well-formed at line 72, column 28, byte 4001 at
C:/Perl/site/lib/XML/Parser.pm line 168.
There was an error loading hackintheboxorg.xml: mismatched tag
at line 168, column 2, byte 4751 at
C:/Perl/site/lib/XML/Parser.pm line 168.
A cheat would be to:
my $invalid_data_check = "<data>$real_data</data>";
And then XMLin on $invalid_data_check, as above. Another option is to HTML
encode all the data before passing it off to the XML creator/parsing code:
use HTML::Entities qw( %char2entity );
$real_data =~ s/([^\s!\#\$&%\'-;=?-~<>"])/$char2entity{$1}/g;
(note, in this example, I'm importing the char2entity hash myself, which
allows me to define exactly what characters I DO NOT want turned into
entities (the first part of the regexp). Check the man page for the defaults.
With the above in hand, my XML parsing usually runs like this:
use XML::Simple;
my $data = eval { XMLin( $xml_data ); };
if ($@) {
print "$@, attempting to repair.";
$xml_data =~ s/([^\s!\#\$&%\'-;=?-~<>"])/$char2entity{$1}/g;
eval { XMLin( $xml_data ); }
if ($@) { print "Nope. Still an error."; }
}
You can probably modify that to your use.
--
Morbus Iff ( softcore vulcan pr0n rulezzzzz )
http://www.disobey.com/ && http://www.gamegrene.com/
please me: http://www.amazon.com/exec/obidos/wishlist/25USVJDH68554
icq: 2927491 / aim: akaMorbus / yahoo: morbus_iff / jabber.org: morbus
--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
