BF: make data.importConditions() able to import csv with ',' and ';' separator

[AnnaTruzzi]

In the data.importConditions() (lines 276-289) I added a try/except block to allow an automatic loading of csv with either ',' or ';' as separator.
Now, pandas loads the file without raising errors even if only one column is detected (for example when loading a file separated with semicolon with the argument sep = ',') and the presence of weird punctuation in the dataset is only tested later when the _assertValidVarNames() function is called from within the pandasToDictList() function. Therefore the except function only works if pandasToDictList(trialsArr) is run before it.
Right now to avoid changing the code too much I have repeated the same block of code twice (and it is used one more time later on for the excel format). It works but it seems to me a bit redundant.
What do you think?

[coveralls]

Coverage decreased (-0.008%) to 48.396% when pulling 4eda564 on AnnaTruzzi:master into 5df405c on psychopy:master.

[peircej]

I totally hear you. It's tricky. Two things could do with improving ideally :

• the amount of code inside the try/except. The more lines we have in a block like that the more risk we have of trapping also trapping a different (unrelated) ValueError in the future and confusing ourselves by the lack of a message
• the repetition of those lines. The issue then is that if we need to fix the we have to fix in 3 places

Possible solutions:

• make those calls in a loop for sep in [',', ';']:. Use try/except just around the critical line as in:

  try:
    lineMightFail()
    break  # it didn't so end the loop
  except ValueError:
    continue  # it failed so try other option

• make those lines into a private function (as we did with pandasToDictList) and then we can call it multiple times. The try/except could just go around the line that fails inside the private function and it could return a value to indicate that type of failure (or we could create a custom Exception class if we're being fancy)

Sorry, this has turned into a rabbit hole. I'm happy to take the version you submitted already - it does fix the bug and working inelegant code is better than broken elegant code! Up to you whether you want to pretty-it-up further.

[peircej]

Hi Anna, I thnk I'm going to pull this in and then adapt it further if that's OK. I've realised that there is a further issue not handled yet which is that the decimal separator is also going to be variable the reason your spreadsheet uses ; as its delimiter is almost certainly that it's using , as the decimal separator. So I think we do need to switch to using a loop (because there are several possible options now) and I think I will put the code into a separate private function. So I'll share the changes once complete.

[AnnaTruzzi]

Sounds great!