dhmit/gender_analysis

"Bad user" testing

Closed this issue · 2 comments

Try to break the system with really terrible inputs, see what happens, make sure we have good errors.

Try to imagine you don't know what you're doing, and see what happens.

e.g.,

  • What happens if a user points Corpus.__init__'s path_to_files to a directory full of Word docs?
  • Malformed CSV files
  • CSV files where the number of records doesn't actually match the number of files

Write lines of code that don't use our functions correctly, and make sure that the resulting errors are reasonable.

e.g., bad arguments to analysis functions

Malformed CSV files should be handled by the CSV reader itself, so I'm calling this done for now.