Turkish character Key error

turco

Hi,
I am having trouble with running Turkish characters. The experiment presents a number of trials with many Turkish characters without problem, however at the end of a sequence it gives Key error and states the Unicode of the character that caused the bug. Here is the detail provided by the program.

item-stack: experiment[run].runBlock[run].blockSequence[run].untilCurrentStackIsEmpty[run].selectAndPresentItem[prepare].selectItemAndDetermineTrialType[prepare]
exception type: KeyError
exception message: u'g\xfcne\u015f'
item: selectItemAndDetermineTrialType
time: Fri Apr 22 15:03:36 2016
phase: prepare

sebastiaan

This error comes from your own inline script, so could you please post that script here?

Cheers,
Sebastiaan

turco

Hey, sorry for the late reply, here is my inline script I think the problem occurs.

{"main-item-name": "selectItemAndDetermineTrialType", "type": "item-snippet", "items": [{"item-type": "inline_script", "item-name": "selectItemAndDetermineTrialType", "script": "define inline_script selectItemAndDetermineTrialType\n\tset description \"Executes Python code\"\n\tset _run \"\"\n\t___prepare__\n\t\n\t

## Select the current item:\n\tcurItem = curItems[0]\n\t\n\t
## Remove this item for the to be studied items, as the participant might\n\t
## answer it correctly.\n\tcurItems.remove(curItem)\n\t\n\t
## Figure out whether we already presented this item. If so, present a test\n\t
## trial, if not, present a study trial *and* make sure that next time round\n\t
## we'll present it as a study trial.\n\t

if (curItem in itemsPresented):\n\t\t
trialType = 'test'\n\telse:\n\t\ttrialType = 'study'\n\t\t

curItems.append(curItem)\n\t\t

itemsPresented.append(curItem)\n\t\n\t
var.cue = curItem\n\tvar.response = allItems[curItem]\n\t

var.trialType = trialType\n\t

__end__\n"}]}

turco

I guess I managed to learn pastebin, so this is the inline script that causes trouble.

http://pastebin.com/pKpgmr6X

Thanks in advance for any help offered.

sebastiaan

Hi,

In general, the KeyError indicates that you're trying to get something from a dict or a list using a key that doesn't exist. It seems likely that one of these lines is the culprit:

curItem = curItems[0]

or

var.response = allItems[curItem]

However, where the error comes from exactly is not clear from this script, because the experiment presumably contains more scripts in which curItems, itemsPresented, etc. are defined.

I have the impression that you did not create this experiment yourself, is that correct? If not, perhaps you could ask the original creator what the problem is? Or you could post the full experimental script here, and we can try to figure out the problem.

Cheers,
Sebastiaan

turco

My supervisor created the scripts, unfortunately I have very limited knowledge on programming. The problem is, he is away for 2 weeks and the study begins next week. Well, here is the full script including the explanation of what the program is supposed to do. I owe you big time.

Regards,
Turco

http://pastebin.com/Bk53rjTt

turco

Maybe I was not clear enough but the script runs and the words in the list are presented multiple times successfully. Error occurs when a new trial is about to be defined and it seems to occur completely at random, for example for items that are presented successfully in an earlier trial and for different items each time. It happens only at the end of trials when the presentation and response to 6 items for that trial is completed.

turco

This is the screenshot of the problem when experiment stops, maybe it can give a clue.

https://www.dropbox.com/s/25rdhkw8nybhsjj/Key-error.png?dl=0

Jarik

Hi turco,

Turn all the words into unicode by prefixing u (see image below).

More info by Sebastiaan here:
http://www.cogsci.nl/blog/miscellaneous/245-a-simple-explanation-of-character-encoding-in-python

Best,
Jarik

sebastiaan

Hi Turco,

@jarik is correct. What happens is the following:

• You have defined the keys of the dictionary as str objects; that is, the strings do not have the u suffix. When working with non-ascii characters, this is asking for trouble. Golden rule: always prefix text with u!
• Then, when the participant makes a mistake, the to-be-repeated item is set as an experimental variable.
• OpenSesame automatically converts this variable from a str into a unicode object.
• Finally, when using the unicode key to look up a value in the dictionary, the key doesn't match, because the dictionary has str keys. Result: a KeyError.

Long story short: Make sure that you're only working with unicode objects by prefixing a u to all strings in your script.

Cheers!
Sebastiaan

turco

Thank you both very much Jarik and Sebastiaan!
I converted the items and now it runs smoothly. Now I can collect data on time. I wish you both a very nice week.

Kind regards,
Turco