Mssql script fixes

[johnjaylward]

This PR fixes 2 issues I ran into while trying to debug connection issues with a legacy SQL2000 server.

First commit corrects the return type expectation for a call made to Helper.GetDiscoveredInstances. The actual return is a single value, while the original expected return by the caller was a tuple. I corrected the caller to use only the single value.

The second commit fixes an issue I saw where the SqlServer returned no rows of data. The loop was being entered even though the pos var was already past the length of the data.

[correctmeifimwrong33]

Any reason this is not being merged?

[nnposter]

There is a more straightforward way how to accomplish a test for an empty table, using next(). This has been committed as r38943.

[nnposter]

The second commit fixes an issue I saw where the SqlServer returned no rows of data. The loop was being entered even though the pos var was already past the length of the data.

The way I understand this, the real issue is that the loop above that, which is tasked with finding the first ROW token, is eating up the DONE token, which otherwise causes the subsequent row-populating loop to terminate. Therefore it seems to me that a cleaner way to rectify the issue would be to bail out from the first loop not just when encountering the ROW token but also on DONE.

This has been committed as r38945. Please test it if possible.

As a side note, having true in expression true and pos < data:len() serves no purpose.

[nnposter]

First commit corrects the return type expectation for a call made to Helper.GetDiscoveredInstances. The actual return is a single value, while the original expected return by the caller was a tuple. I corrected the caller to use only the single value.

This change has been committed as r38948, along with fixing another similar bug. Thank you for contributing to Nmap!