Test case to fail this code?

Question

I have to maintain code from a contractor. It has this "interesting" snippet:

String webServicesValue = 
  webResponse.substring(webResponse.indexOf("<" + fieldName + ">") + 
                        fieldName.length() + 2, 
                        webResponse.indexOf("</" + fieldName + ">"));

It took me a few minutes to understand what he's trying to do.

It seems to be a WTF code, but my colleague suggested, "If it ain't broken, don't fix it." I'm wondering if this code will ever fail. It seems to be working so far, and I honestly can't think of a test case to fail this.

Thanks,

Answer 1

Yes. It will throw an Exception if "<fieldname>" is not present in the response. Specifically it will try to call webResponse.substring(fieldName.length() + 1, -1);

"<fieldname/>" will cause similar problems, as will any attributes on the element.

And if you get "<fieldname> ... <fieldname> ... </fieldName> ... </fieldName>", you'll get the wrong answer.

EDIT: in the light of followup discussions, I'd say that this code should be rewritten to use a proper XML parser ... unless you / your team can guarantee that the code will never have to deal with problematic XML. XML simply allows too many valid (and invalid) variations to deal with by ad-hoc string manipulation.

Answer 2

如果您收到<fieldname/>会发生什么？

Answer 3

Instead of manually parsing XML it's better to use a real XML parser. There are all kinds of corner cases that are hard to cover with plain string manipulation. It will be more readable with a real parser also. It's best to consider XML data to be binary data, especially when taking all the possible character encodings into consideration.

Answer 4

In addition to Igor Brejc and Stephen C's responses above, there's CDATA:

<fieldname><![CDATA[ I am not really </fieldname> ]]></fieldname>

or even

<othertag>
  <![CDATA[ I am not really <fieldname> and there is no closing tag ]]>
</othertag>

Answer 5

I would suggest re-writing it in that case. If it is not easy to understand, then it is more difficult to test.

Answer 6

• Nested tags. You'll start at the first open tag, skip any others that exist and stop on the first close tag instead of the matching close tag.
• There is a close tag somewhere before the open tag (you're searching for the close tag from the beginning of the string, not the end of the open tag)