[solved] Inline Python Script Capabilities

Xophmeister

I feel that OpenSesame primitives don't expose the functionality I am looking for in the experiment I am developing. However, being able to drop into Python should resolve that, providing the following are possible:

1. Time logging at various points within an item: I want to show an item with multiple states, which the user advances through by mouse clicks, where I want to record the time spent on each state. (AFAIK, I can't do this with an OpenSesame loop because the number of states available, each time this item is shown, will be variable.) Is it possible to programatically log a time on-demand?

2. When I have progressed through all the states in said item, I want to then proceed to the next item (which will just be a regular OpenSesame item); again by mouse click. What command is used to advance to the next item?

3. Will I be able to open a SQLite database from the OpenSesame file pool? I feel like experiment.get_file should do the job -- and import sqlite3 appears to work -- but I'm interested to know if anyone has tried this.

I'm sure I'll have more questions (sorry!)...

Many thanks;
Christopher

sebastiaan

Hi Christopher,

Time logging at various points within an item: I want to show an item with multiple states, which the user advances through by mouse clicks, where I want to record the time spent on each state. (AFAIK, I can't do this with an OpenSesame loop because the number of states available, each time this item is shown, will be variable.) Is it possible to programatically log a time on-demand?

The specific solution would depend on what exactly you want to do, but in general it is very easy to log a timestamp using an inline_script object. For example:

exp.set('my_custom_timestamp', self.time())

This will create an experimental variable my_custom_timestamp which holds the current time.

When I have progressed through all the states in said item, I want to then proceed to the next item (which will just be a regular OpenSesame item); again by mouse click. What command is used to advance to the next item?

Again, the question is a bit too generic for a specific answer. In general, the way to pause an experiment until a mouse click is to insert a mouse_response item at the point that you want to pause.

Will I be able to open a SQLite database from the OpenSesame file pool? I feel like experiment.get_file should do the job -- and import sqlite3 appears to work -- but I'm interested to know if anyone has tried this.

Sure. OpenSesame uses a regular Python interpreter, so you can use sqlite3 in an inline_script item in the same way as you would in any normal Python script.

Good luck with your experiment. As a tip: If you have further questions, please try to be as specific as possible, that way it's easier to give a useful answer!

Cheers,
Sebastiaan

Xophmeister

Thanks, Sebastiaan

My first two queries are related, so let me try to be a bit more specific.

I'm trying to create a self-paced reading task. The way I can see this working is to have a single item that displays successive words, controlled by mouse click. The reason for a single item (rather than multiple; one for each word) is that the sentences will be different lengths and (AFAIK) I can't dynamically create items or control loop variables, on-the-fly. (There are other considerations, such as positioning and various counterbalancing measures, hence preferring to use SQLite to store the data.)

So, let's say the item gets the sentence data as this| is| my| sentence, where the | delimit the word boundaries. I can easily split this out into a list and then I'd like to show each element, one at a time, then proceed to the next OpenSesame item after the last element is shown. Like so:

//== ITEM 1 =================\\
||                           ||
||  +---------------------+  ||
||  |                     |  ||
||  | this                |  ||
||  |                     |  ||
||  +---------------------+  ||
||             | click       ||
||             V             ||
||  +---------------------+  ||
||  |                     |  ||
||  |      is             |  ||
||  |                     |  ||
||  +---------------------+  ||
||             | click       ||
||             V             ||
||  +---------------------+  ||
||  |                     |  ||
||  |         my          |  ||
||  |                     |  ||
||  +---------------------+  ||
||             | click       ||
||             V             ||
||  +---------------------+  ||
||  |                     |  ||
||  |            sentence |  ||
||  |                     |  ||
||  +---------------------+  ||
||             | click       ||
\\============ | ============//
               |
//== ITEM 2 == | ============\\
||             V             ||
||  +---------------------+  ||
||  |                     |  ||
||  |  [Some other item]  |  ||
||  |                     |  ||
||  +---------------------+  ||
||                           ||
\\===========================//

Where the "double-lined boxes" represent OpenSesame items and the "single-lined boxes" represent screen state.

With regard to the timing, if I use exp.set('my_custom_timestamp', self.time()) on each state change, then that will override the variable. As such, I'd need a different variable for each screen state (presumably I can't create a list here). Would it be possible to instead trigger a logger event programmatically, so that I can store the timestamp into a single variable, but each screen state gets its own line in the output CSV? (Otherwise, my solution would be to "fake" a list by concatenating the time into the same variable and then postprocessing the CSV to deserialise it.)

I hope that's more clear. Let me know if I can clarify anything further.

Thanks again
Chris

Josh

Hi Chris,

Like you said - each trial you create a list with all the words of a given sentence (e.g. sentence_list = ["this", "is", "an", "example"]). Now, we can run the trial by using a for-loop (in the run-phase of an inline script), where the logic would be something like this:

   from openexp.mouse import mouse
   my_mouse = mouse(exp)

   sentence_responsetimes = []
   mouse_clicked = False
   starting_coordinate = 50 # (or something)
   for word in range(0, len(sentence_list)):
           canvas.clear()
           canvas.text(sentence_list[word]), x=starting_coordinate)
           starting_coordinate += canvas.text_size(sentence_list[word]+" ") 
           canvas.show()
           start_time = self.time()
           while mouse_clicked == False:
                 if my_mouse.flush() == True:
                       responsetime = self.time() - start_time
                       sentence_responsetimes.append(responsetime)
                       mouse_clicked = True

Hope this helps.

Cheers,

Josh

Xophmeister

Thank you, Josh: That's very very helpful

Would I then just call exp.set('someResponseVar', sentence_responsetimes) after the for-loop? That is, will the list be serialised to a string automatically by the logger, so I get a CSV that looks like (for example):

var1  var2  var3  someResponseVar
----- ----- ----- ----------------
foo   bar   quux  [10,12,9,14]

(i.e., the CSV row looks like "foo","bar","quux","[10,12,9,14]")

Thanks;
Chris

Josh

No problem

And exactly; you'll have to use exp.set after the for-loop to have the list logged. Of course you could also divide the list into separate columns for each word, by setting separate variables (e.g. word1 word2) instead of the list - but yeah you get the idea.

Good luck and have fun!

Josh

Xophmeister

I just tried your code out, Josh and, modulo a few typos, it worked a charm Thank you again!

For others' sake, the diff on the typos is as follows:

---        canvas.text(sentence_list[word]), x=starting_coordinate)
+++        canvas.text(sentence_list[word], x=starting_coordinate)

---        starting_coordinate += canvas.text_size(sentence_list[word]+" ")
+++        starting_coordinate += canvas.text_size(sentence_list[word]+" ")[0]

Also, the mouse_clicked=False statement needs to appear inside the for-loop (e.g., as the first statement), otherwise it doesn't get reset after the first click and cycles through the remaining words in the list instantaneously.

Anyway, this is great Cheers!

Josh

Great that you were able to filter out the typo's yourself! I just wrote up something quickly, but should've double-checked at least, I admit