Alternative to temporary trim (SPP_TEMPORARY_TRIM) for finding Selenium WebElement? about fb-contrib

[mebigfatguy]

Yeah, that's an unfortunate false positive.

So lets see. Why is that a valid case. Well it's because you are not using the result of getText() again. So I guess really we should look to see if the thing that is being trimmed, and thrown away, is being used again.

What this would mean is that the trim() call is probably in a conditional expression, and doesn't show up again in the if or else of the conditional.

Ok, that seems right. Let's see what can be done.

[mebigfatguy]

The issue is of course the trim() may be called on a string that is produced thru a long chain of methods, (yours is a chain, albeit not too long). Finding that pattern again is probably hard, which is probably why the FP exists today. Maybe it's good enough to look for the last method that produces the string to be trimmed, in this case, look for a call to WebElement.getText() regardless of what web element it was called on. sure that is broken too, but probably reduces the false positives to close to 0 (in exchange for a higher FN rate).

[ThrawnCA]

Well, in this case, there's just a void method call and then return; is that a useful pattern for distinguishing false positives?

Maybe, if there is no further string access (no access to string variables, no method calls returning string) in one branch of the conditional, then it's considered legitimate? I guess that might still mask actual errors in the case where someone should have persisted the trimmed version, but didn't...although that should only matter for instance/static variables.

[mebigfatguy]

yes that would be good enough to rule out, although, there would still be possible FPs. But that would be better.

[mebigfatguy]

@ThrawnCA could you see if that resolves your issue?

[ThrawnCA]

Well, it broke the FindBugs check, which is only enforced in the Maven build :D. But it was a deliberate thing (switch fall-through), so I just added a suppression annotation. The trim false positive is fixed :).