Answers to corner cases

[facorread]

These two commits respond to some shapefiles that contain the ** characters as well as possibly empty shapes. This proposal should be innocuous in all other shapefiles.

[riggsd]

The first commit from this PR, 49edb13, should not be merged upstream. It catches only the case where two '' characters are present. This would only apply if the field happened to have been defined with size=2. A field with size=5 would have the value '****' in the same (NULL) situation.

The PR I submitted today, #16 , handles this as a general case.

I cannot comment on the second commit in this PR other than to say that unrelated changes should probably more appropriately go into multiple PRs.

[facorread]

OK

[karimbahgat]

Closing as the null values were fixed in #16.

As for the null shapes, not sure if I understand it correctly, but since null shapes have a type value of 0 I don't think they would ever end up in those parts of the code. But make a PR explaining more if still relevant.