surveymonkey_repustate.py

# -*- coding: UTF-8 -*-
"""
Combining the Repustate API with the SurveyMonkey API to understand open ended
responses.
"""
import requests
import json

SM_HOST = 'https://api.surveymonkey.net'

# This token is the one you get after doing the OAUTH 2 handshake.
SM_TOKEN = 'FOO_TOKEN'

# Get this key after signing up with SurveyMonkey.
SM_API_KEY = 'SURVEYMONKEY_API_KEY'

# Sign up at https://www.repustate.com to get a free API key.
REPUSTATE_API_KEY = 'REPUSTATE_API_KEY'

def chunks(a_list, N):
    """
    Create evenly sized chunks of a list.
    """
    for i in xrange(0, len(a_list), N):
        yield a_list[i:i+N]

def run(survey_id, question_ids):
    client = requests.session()
    client.headers = {
        "Authorization": "bearer %s" % SM_TOKEN,
        "Content-Type": "application/json"
    }


    post_args = {
        'survey_id':SM_SURVEY_ID,
        'page':1 # defaults to the first 1000 respondents.
        # NB: If you have lots of respondents, you'll have to page through them by
        # passing increasing values for N.
    }

    respondent_data = client.post(SM_HOST + '/v2/surveys/get_respondent_list?api_key=%s' % SM_API_KEY, post_args).json()

    # Get the individual respondent_ids.
    respondent_ids = []
    for respondent in respondent_data['data']['respondents']:
        respondent_ids.append(respondent['respondent_id'])

    # Now get our responses for each of our respondents. The SurveyMonkey API let's
    # you get at most 100 at a time, so you'll have to page through your list of IDs.
    
    for groups_of_ids in chunks(respondent_ids, 100):
        
        post_args = {
            'survey_id':survey_id,
            'respondent_ids':groups_of_ids
        }
        
        response_data = client.post(SM_HOST + '/v2/surveys/get_responses?api_key=%s' % SM_API_KEY, post_args).json()
        
        for response_set in response_data['data']:
            # Get the ID for this person.
            respondent_id = response_set['respondent_id']
            for question in response_set['questions']:
                
                # Loop over each question, looking for the ones that match the
                # IDs were interested in. When we find them, extract the open
                # ended response and then send to Repustate for analysis.
                
                if question['question_id'] in question_ids:
                    
                    # Take a look at how answers are formatted:
                    # https://developer.surveymonkey.com/mashery/get_responses
                    answer_text = question['answers'][0]['text']

                    # Here you can persist the text to your database or
                    # whatever, but we'll analyze it real-time.
                    repustate_response = requests.post(
                        'https://api.repustate.com/v2/%s/named_entities.json' % REPUSTATE_API_KEY, {'text':text}
                    ).json()

                    entities = repustate_response['entities']

                    # Now do what you will with the entities. Run them through
                    # Repustate's topical sentiment analysis, or generate the
                    # n-grams or adjectives surrounding this term. Lots of
                    # possibilities here.


if __name__ == '__main__':
    
    # The survey we want to analyze.
    # We know the survey ID we want, but if you didn't, you can access your surveys via this API call:
    # surveys = client.post(SM_HOST+'/v2/surveys/get_survey_list', params).json()
    survey_id = 'YOUR_SURVEY_ID'
    
    # These would be the IDs of the open ended questions were interested in.
    question_ids = set(['1', '2', '3', '4'])
    
    # Run it!
    run(survey_id, question_ids)