WebFrame::_stringByEvaluatingJavaScriptFromString methods don't handle nil string

[Michael Saboff]

The various flavors of WebFrame::_stringByEvaluatingJavaScriptFromString don't properly handle nil string.  Instead such nil strings are passed down to the JavaScript engine where it may be a crash.

An if (!string) return @"": should be added as appropriate.

[Michael Saboff]

Created attachment 145346 [details]
Patch

[Simon Fraser (smfr)]

Comment on attachment 145346 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=145346&action=review

> Tools/TestWebKitAPI/Tests/mac/StringByEvaluatingJavaScriptFromString.mm:60
> +    EXPECT_EQ(nil, result);

Don't you expect an empty string, not a nil string?

[Geoffrey Garen]

Comment on attachment 145346 [details]
Patch

Expectation should be empty string. Does nil pass?

[Michael Saboff]

Created attachment 145369 [details]
Patch with fix

(In reply to comment #3)
> (From update of attachment 145346 [details])
> Expectation should be empty string. Does nil pass?

I was thinking about this as I was out at lunch, realizing that the test was wrong.  The test was "passing", but shouldn't.  I was assuming that run-test-webkit-api built the tests before running.  Therefore I wasn't testing the new code.  Built and tested this patch.  It works as expected.

[Geoffrey Garen]

Comment on attachment 145369 [details]
Patch with fix

r=me

[Michael Saboff]

Committed r119280: <http://trac.webkit.org/changeset/119280>

[Radar WebKit Bug Importer]

<rdar://problem/11620927>