From: Andrew Ruthven <andrew@etc.gen.nz>
Date: Thu, 29 Nov 2007 17:46:57 +0000 (+1300)
Subject: Clean out the bad UTF-8 characters.
X-Git-Tag: 0.6.0~27
X-Git-Url: http://git.etc.gen.nz/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=5690b0b624e6cc2dccf2ab5c0218b3bff40ee724;p=mythtv-status.git

Clean out the bad UTF-8 characters.
---

diff --git a/ChangeLog b/ChangeLog
index 87f6367..e03f14c 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,6 @@
 2007-11-28	Andrew Ruthven
 	Be more wary about processing what the backend has sent us.
+		This includes cleaning up some invalid UTF-8 characters.
 	Add support for reading XML from a file.
 	Be a bit more forgiving on the XML we're receiving.
 	Show how much disk space is used - currently only total.
diff --git a/bin/mythtv-status b/bin/mythtv-status
index 68192ae..aa5665e 100755
--- a/bin/mythtv-status
+++ b/bin/mythtv-status
@@ -355,6 +355,8 @@ sub load_xml {
   $parser->recover_silently(1)
     unless $verbose;
 
+  clean_xml(\$status);
+
   my $xml = eval { $parser->parse_string( $status ) };
 
   if ($@) {
@@ -381,6 +383,20 @@ sub load_perl_api {
   return $myth;
 }
 
+# We are sometimes passed dodgy XML from MythTV, make some attempts to clean
+# it.
+sub clean_xml {
+  my ($xml) = shift;
+
+  # Deal to invalid Unicode.
+  for my $bad ("&#xdbff;", "&#xdee9;") {
+    if ($$xml =~ s/$bad/?/g) {
+      warn "Found and replaced: $bad\n"
+        if $verbose;
+    }
+  }
+}
+
 sub process_xml {
   my ($block, $xml) = @_;