Community
Participate
Working Groups
The VCard provider has an XML handling bug for the poor people with names and data outside US-ASCII. It's classical abuse of String.getBytes(). How I wish that method was deprecated! A Java string containing international characters is converted to bytes (implicitly using the "native" encoding, perhaps an 8-bit SBCS charset like ISO-8859-15), and is then parsed using XML without the XML declaration which then defaults to UTF-8, and gets the 8 bit SBCS bytes wrong and complains. When this happens, the entire address book of the user I'm logging in as is silently dropped! [Fatal Error] :1:20: Invalid byte 1 of 1-byte UTF-8 sequence. org.xml.sax.SAXParseException: Invalid byte 1 of 1-byte UTF-8 sequence. at com.sun.org.apache.xerces.internal.parsers.DOMParser.parse(DOMParser.java:264) at com.sun.org.apache.xerces.internal.jaxp.DocumentBuilderImpl.parse(DocumentBuilderImpl.java:292) at javax.xml.parsers.DocumentBuilder.parse(DocumentBuilder.java:98) at org.jivesoftware.smackx.provider.VCardProvider._createVCardFromXml(VCardProvider.java:84) at org.jivesoftware.smackx.provider.VCardProvider.parseIQ(VCardProvider.java:76) at org.jivesoftware.smack.PacketReader.parseIQ(PacketReader.java:603) at org.jivesoftware.smack.PacketReader.parsePackets(PacketReader.java:289) at org.jivesoftware.smack.PacketReader.access$0(PacketReader.java:280) at org.jivesoftware.smack.PacketReader$1.run(PacketReader.java:63) [Fatal Error] :1:20: Invalid byte 1 of 1-byte UTF-8 sequence. The patch: It merely converts the XML string to UTF-8 (if available) instead of the native encoding. Hope this gets in the release, since it really messes up a lot of people. The snag is that the bug is in the Smack code. Is that an issue, IP wise?
Created attachment 105367 [details] Patch Smack by relying on UTF-8 instead of native encoding for XML w/intl chars
Thanks for the report and patch. (In reply to comment #0) > The VCard provider has an XML handling bug for the poor people with names and > data outside US-ASCII. > > It's classical abuse of String.getBytes(). How I wish that method was > deprecated! Me too. <stuff deleted> > The patch: It merely converts the XML string to UTF-8 (if available) instead of > the native encoding. Just to verify...should UTF-8 be used, or ISO-8859 (or whatever the default one is)? > > Hope this gets in the release, since it really messes up a lot of people. I'm inclined to move this up to critical, since the fix is very low risk and it does obviously affect many people. >The > snag is that the bug is in the Smack code. Is that an issue, IP wise? > I don't think so, but I will check with the EF legal people about it. So in summary, I will check with EF about getting this patch into ECF 2.0.0/Ganymede release coming up and notify on this bug. Thanks again for the report and the fix. We will try to get this fix in.
Changing bug to 'blocker'...aka 'stop ship'.
(In reply to comment #0) > the entire address book of the user I'm logging in as is silently > dropped! What do you mean by address book? Is your 'Contacts' view empty?
I contacted BFB of Foundation and we jointly determined that the bug is widespread enough (and the fix low enough risk) to be considered stop ship. Thanks Jesper for reporting and the patch. Applied patch to Release_2_0 and HEAD. Tested fine. Resolving as Fixed. I believe the only info we are using from the VCard currently is the avatar image.
Actually changing to fixed. :)
