Question

I’m trying to transmit UTF-8 strings in complex data structures with SOAP::Lite. However, as it turns out, SOAP::Lite quietly converts all UTF-8 strings into base-64-encoded octets. The problem with that is that the deserializing does not revert the conversion and only does a straight base64 decode.

This leaves me confused as to how a user is supposed to ensure that they get UTF-8 data from the SOAP::Lite response. Walking the tree and running decode_utf8 on all strings seems wasteful.

Any suggestions?

Edit: In a nutshell, how do i make this test pass without monkey-patching?

Answer 1

Use of is_utf8 (line 278) is evil and wrong. As we can’t trust SOAP::Lite with encoding character data properly (to be fair, this code was likely written before word got around in the community how to do this particular kind of string processing), we shall give it octet data only and therefore have to handle encoding/decoding ourself. Pick a single encoding, apply it before handing off data to S::L, reverse it after receiving data.

use utf8;
use strictures;
use Encode qw(decode encode);
use SOAP::Lite qw();
use Test::More;

my $original = 'mü';
my $xml      = SOAP::Serializer->envelope(
    freeform => encode('UTF-8', $original, Encode::FB_CROAK | Encode::LEAVE_SRC)
);
my ($roundtrip) = map {
    decode('UTF-8', $_, Encode::FB_CROAK | Encode::LEAVE_SRC)
} values %{SOAP::Deserializer->deserialize($xml)->body};

is(length($original), length($roundtrip),
    'Perl character string round-trips without changing length');
done_testing;