Parsing invalid ampersands with Android's XmlPullParsers

I am writing a little screen-scraping app that consumes some XHTML - it goes without saying that the XHTML is invalid: ampersands aren't escaped as &.

I am using Android's XmlPullParser and it spews out the following error upon the incorrectly encoded value:

org.xmlpull.v1.XmlPullParserException: unterminated entity ref 
(position:START_TAG <a href='/Fahrinfo/bin/query.bin/dox?ld=0.1&n=3&i=9c.0323581.1266265347&rt=0&vcra'>
@55:134 in java.io.InputStreamReader@43b1ef70) 

How do I get around this? I have thought about the following solutions:

1. Wrapping the InputStream in another one that replaces the ampersands with entity refs
2. Configuring the Parser so it magically accepts the incorrect markup

Which ones i开发者_StackOverflow社区s likely to be more successful?

I was stuck on this for about an hour before figuring out that in my case it was the "&" that couldn't be resolved by the XML PULL PARSER, so i found the solution. So Here is a snippet of code which totally fix it.

void ParsingActivity(String r) {
    try {
        parserCreator = XmlPullParserFactory.newInstance();
        parser = parserCreator.newPullParser();
        // Here we give our file object in the form of a stream to the
        // parser.
        parser.setInput(new StringReader(r.replaceAll("&", "&")));
        // as a SAX parser this will raise events/callback as and when it
        // comes to a element.
        int parserEvent = parser.getEventType();
        // we go thru a loop of all elements in the xml till we have
        // reached END of document.
        while (parserEvent != XmlPullParser.END_DOCUMENT) {
            switch (parserEvent) {
            // if u have reached start of a tag
            case XmlPullParser.START_TAG:
                // get the name of the tag
                String tag = parser.getName();

pretty much what I'm doing I'm just replacing the & with & since I was dealing with parsing a URL. Hope this helps.

I would go with your first option, replacing the ampersands seems more of a fit solution than the other. The second option seems more of a hack to get it to work by accepting incorrect markup.