Handle datasets that have different types of files

[vhoulbreque]

For example:

• data1.zip
• data2.tar.gz
• data3.csv
• data4.txt

[chrplr]

Yes! Why not make 'type' an attribute of url and add a 'type' option to selectively download the files acting this type among the list of urls? (I wish I had time to write code and propose pull requests rather than just asking for more features)