unf 0.5.1

Universal Numeric Fingerprints

unf contains Python code for Universal Numeric Fingerprints (UNFs)

http://guides.dataverse.org/en/latest/developers/unf/index.html

Version 0.5.0 contains two important design decisions:

First, lists and tuples containing anything but primitives are not allowed. The UNF specification does not handle this case, presumably since it was written with R in mind and nested vectors are not allowed in R. Versions of this package before 0.5.0 normalized subsequences in place, but because normalized values are simply concatenated in sequences, the normalization of (1, (2, 3)) would then be the same as the normalization of (1, 2, 3) and the two would have the same UNF.

Second, numpy arrays of greater than one dimension are handled the same way that R data frames and matrices are handled, but without the sorting step. The UNFs for the components of an array are calculated, the results are put in a list, and the UNF of the resulting list is calculated. numpy arrays with one dimension are treated as simple sequences.