Avoid crash in tidyCleanAndRepair if document was not loaded

[nijel]

This can happen after attempt to load nonexisting file.

This used to work fine with old libtidy, but no longer works with html-tidy, so I think it is a regression. I'm not sure the fix is really correct, but it seems to do what is expected for me.

[geoffmcl]

@nijel ok, what I understand is that your app calls tidyDocCleanAndRepair(doc) even when no document loaded! And this causes a crash in the current library, what version?, while it was harmless in an older library version, what version?

Well, that would certainly be a regression bug. Thanks...

But looking at your patch, I do not think this is quite right, even though as you say it works for you...

We need to look at why we need an extra guard for CheckHTML5 and CheckHTMLTagsAttribsVersions... These calls should be harmless, like earlier List2BQ, and several others... with or without a loaded doc...

I think it might be the test if (doc->lexer->versionEmitted & VERS_HTML5) maybe the crash. Maybe there is no lexer if no document loaded...

It will take me a little time to setup a test for this... be back soonest... but look forward to any further information you can supply... thanks...

[nijel]

It's uTidyLib which is doing this call even when file loading fails. It worked just fine with original Tidy and now leads to segfault.

To reproduce just invoke tidyParseFile on nonexisting file and then tidyCleanAndRepair.

[geoffmcl]

@nijel discussion and patch move to #429 ...

Meantime closing this as not being the correct patch... but thanks for pointing us to it...