process_reviews.py

process_reviews.py

# Top level attributes
PLACE_ID_A                    = 'place_id'
FS_ID_A                       = 'fs_id'
YELP_ID_A                     = 'yelp_id'
PHONE_A                       = 'phone'
NAME_A                        = 'name'
LOCATION_A                    = 'loc'                   # dict, see sub attributes below
REVIEW_A                      = 'review'                # single incoming review; a dict, see sub attributes below
ALL_REVIEWS_A                 = 'all_reviews'           # list of raw reviews
RECOS_A                       = 'recos'                 # dict of processed reviews
YELP_RATING_COUNT_A           = 'yp_rating_count'
YELP_REVIEW_COUNT_A           = 'yp_review_count'
CREATED_A                     = 'created'
UPDATED_A                     = 'updated'

# Sub attributes under location dictionary
ADDRESS_SA                    = 'address'
LAT_SA                        = 'lat'
LON_SA                        = 'lon'

# Sub attributes under review dictionary and wifi, exp, plugs are first level under recommendation data
WIFI_SA                       = 'wifi'
EXP_SA                        = 'exp'
PLUGS_SA                      = 'plugs'
REVIEWER_SA                   = 'reviewer'
CREATED_REV_SA                = 'created_review'

# Second level under recommendation data for wifi, exp, plugs keys
LOW_SA                        = 'low'                     # wifi = None, experience = Poor, plugs = None
MED_SA                        = 'med'                     # wifi = Limited, experience = Okay
HIGH_SA                       = 'high'                     # wifi = Unlimited, experience = Great
REVIEW_COUNT_SA               = 'total_count'
RECO_SA                       = 'reco'
OVERALL_RECO_SA               = 'overall_reco'         # first level key, overall recommendation


'''
Sample dictionaries

 # ignoring timestamp and sender for now
 
 review = {
            'wifi': 'low'
            'plugs': 'low'
          }
 
 reco_dict = {
              'wifi' : {
                          'low' : 1
                          'high' : 3
                          'total_count': 4
                          'reco' : 'high'
              },
              
              'plugs' : {
                          'low':1
                          'total_count': 1
                          'reco' : 'low'
              }
 }
 
 '''



def process_reviews_and_update_recommendation_data(review={}, reco_dict={}):
  for (review_attribute, review_score) in review.iteritems():
    
    # ensure score is a valid key in reco_dict i.e., ignore reviewer and created data in review
    if score_is_valid_reco_dict_key(review_score):
    
      # get the attribute to update in the reco_dict
      reco_attribute_dict = reco_dict.get(review_attribute)
    
      # historical data exists for this attribute in the reco_dict
      if reco_attribute_dict:
      
        # increment the current count for this score
        if reco_attribute_dict.has_key(review_score):
           reco_attribute_dict[review_score] += 1
           reco_attribute_dict[REVIEW_COUNT_SA] += 1
                        
        # otherwise create this score with count 1
        else:
           reco_attribute_dict[review_score] = 1
           curr_count = reco_attribute_dict.get(REVIEW_COUNT_SA)
           if curr_count:
             reco_attribute_dict[REVIEW_COUNT_SA] += 1
           else:
             reco_attribute_dict[REVIEW_COUNT_SA] = 1
                   
        # update the highest frequency score if necessary
        curr_reco = reco_attribute_dict.get(RECO_SA)
        if reco_attribute_dict[review_score] > reco_attribute_dict.get(curr_reco):
           reco_attribute_dict[RECO_SA] = review_score
            
        # update the reco_dict
        reco_dict[review_attribute] = reco_attribute_dict
    
      # otherwise create this attribute and this score with count 1 in the reco_dict
      else:
        print review_attribute
        reco_dict[review_attribute] = {}
        reco_dict[review_attribute][review_score] = 1
        reco_dict[review_attribute][REVIEW_COUNT_SA] = 1
        reco_dict[review_attribute][RECO_SA] = review_score

  return reco_dict
  
def score_is_valid_reco_dict_key(score=None):
    if score==HIGH_SA or score==MED_SA or score==LOW_SA:
        return True
    else:
        return False