Allow hyphens anywhere in URL hostnames

[sideshowbarker]

This change brings galimatias’s ToASCII and ToUnicode behavior into conformance with the current URL spec, which requires the ToASCII and ToUnicode algorithms to be performed with the CheckHyphens flag set to false — which means that hyphens are allowed anywhere in the URL hostname, including leading and trailing hyphens.

Fixes validator/validator#720 Thanks @KatieMFritz

[coveralls]

Coverage decreased (-0.3%) to 87.579% when pulling c1d463b on validator:allow-hyphens-in-hostnames into 1e647f6 on smola:master.

[smola]

@sideshowbarker Could you add a test?

[sideshowbarker]

@sideshowbarker Could you add a test?

Sure. Where should it go? In urltestdata_host_whatwg.txt or instead in ParseIssueTest.java?

[smola]

@sideshowbarker Right into ParseIssueTest, as a new test that actually checks the exception message.
(urltestdata_host_whatwg.txt will be removed and replaced with latest official version without modifications).

[sideshowbarker]

@sideshowbarker Right into ParseIssueTest, as a new test that actually checks the exception message.

OK, but now I’m not sure how to write a test for this case — because with this patch applied, if a URL has a hostname containing a leading hyphen or trailing hyphen, then no exception will be thrown.

So does the current test harness have any existing ”this URL should cause not exceptions” test cases? (So that I can use an existing test case as an example for writing the one needed here.)

[smola]

@sideshowbarker Ok, don't worry, I'm in the process of adding such tests. So will add it one for this PR shortly. Merging.