From 5a79d757b02f1b232f651f0ef8296395723789e3 Mon Sep 17 00:00:00 2001 From: Mike Taylor Date: Fri, 2 Mar 2007 11:56:15 +0000 Subject: [PATCH] Hack for dealing with ASCII, Latin-1 and UTF-8 character-sets. Other Latin-n character sets might work, more or less. --- lib/ZOOM/IRSpy/Test/Ping.pm | 25 +++++++++++++++++++++++-- 1 file changed, 23 insertions(+), 2 deletions(-) diff --git a/lib/ZOOM/IRSpy/Test/Ping.pm b/lib/ZOOM/IRSpy/Test/Ping.pm index 09e6781..5128302 100644 --- a/lib/ZOOM/IRSpy/Test/Ping.pm +++ b/lib/ZOOM/IRSpy/Test/Ping.pm @@ -1,4 +1,4 @@ -# $Id: Ping.pm,v 1.20 2007-02-23 15:03:44 mike Exp $ +# $Id: Ping.pm,v 1.21 2007-03-02 11:56:15 mike Exp $ # See the "Main" test package for documentation @@ -13,6 +13,9 @@ our @ISA = qw(ZOOM::IRSpy::Test); use ZOOM::IRSpy::Utils qw(isodate); +use Text::Iconv; +my $conv = new Text::Iconv("LATIN1", "UTF8"); + sub start { my $class = shift(); @@ -52,7 +55,25 @@ sub maybe_connected { foreach my $opt (qw(serverImplementationId serverImplementationName serverImplementationVersion)) { - $conn->record()->store_result($opt, value => $conn->option($opt)); + # There doesn't seem to be a reliable way to tell what + # character set the server uses for these. At least one + # server (z3950.bcl.jcyl.es:210/AbsysCCFL) returns an ISO + # 8859-1 string containing an o-acute, which breaks the + # XML parser if we just insert it naively. It seems + # reasonable, though, to guess that the great majority of + # servers will use ASCII, Latin-1 or Unicode. The first + # of these is a subset of the second, so that brings it to + # down to two. The strategy is simply this: assume it's + # ASCII-Latin-1, and try to convert to UTF-8. If that + # conversion works, fine; if not, assume it's because the + # string was already UTF-8, so use it as is. + my $val = $conn->option($opt); + my $maybe = $conv->convert($val); + if (defined $conv->retval() && $maybe ne $val) { + $conn->log("irspy", "converted '$val' from Latin-1 to UTF-8"); + $val = $maybe; + } + $conn->record()->store_result($opt, value => $val); } } -- 1.7.10.4