Fix .netrc parsing

[fredrik-eriksson]

Fixes: #7177

It seems that the current Tokenizer-implementation splits the netrc tokens on a (string) separator of " \n\t" instead of splitting on any of the whitespace characters which leads to the issue described in #7177.

This implementation splits the content of netrc on any whitespace character using a regex instead, but should otherwise use the same strategy for parsing. I'm not very familiar with C++ (or QT) and does not completely understand the code, so although this solution seems to work for me, I would appreciate a review 🙂

Thanks!

[mgallien]

thanks for the PR
there would still be failing tests
see

FAIL! : TestNetrcParser::testValidNetrc() Compared values are not the same
Actual (parser.find("frob")) : std::pair(""user", "'space")
Expected (qMakePair(QString("user with spaces"), QString("space pwd"))): std::pair("user with spaces", "space pwd")
Loc: [/home/mgallien/work/nextcloud/desktop/test/testnetrcparser.cpp(60)]

I also needed to remove the QEXPECT_FAIL instances from test/testnetrcparser.cpp file
can you add those changes to the PR ?

[fredrik-eriksson]

I think this is as good as I can do. The only test that still fail is username and passwords containing whitespaces, and I think that would require a bigger refactoring. I've commented out that test for now to enable the rest of them.

[mgallien]

thanks @fredrik-eriksson
can you fix the missing signed off comment and fix my inline comment ?
apart from that, I fully agree that your change can be merged as is and is already a big improvement

[fredrik-eriksson]

Thanks for the feedback, I hope I got it right this time 🙂

[mgallien]

@fredrik-eriksson thanks for the very fast review cycles
my plan is to get this merged now and backported to the current stable branch for a release next week
congratulations for your contribution and many thanks for the fix

[mgallien]

/backport to stable-3.15