mining.py 文件源码

def calc_frequency(areaNum, word_counts, curr_timeslot_word_counts):
    timeslot_area_percentage = {}

    for i in word_counts:
        for j in curr_timeslot_word_counts:
            #on same word (key) calculate actual percent
            if i == j:
                timeslot_area_percentage[i] = (word_counts[i]/(curr_timeslot_word_counts[j] + 0.00)) * 100
                timeslot_area_percentage[i] = round(timeslot_area_percentage[i], 2)

    return timeslot_area_percentage
    # if areaNum == 1:
    #   return {"area1" : timeslot_area_percentage} # TODO: convert to django friendly using below??
    # elif areaNum == 2:
    #   return {"area2" : timeslot_area_percentage}
    # elif areaNum == 3:
    #   return {"area3" : timeslot_area_percentage}

    #convert site percents to django friendly JSON
    # site1_percentage_json = json.dumps(dict(site1_percentage), cls=DjangoJSONEncoder)
    # site2_percentage_json = json.dumps(dict(site2_percentage), cls=DjangoJSONEncoder)

# takes a url as a string and returns a SET of TUPLES of all of the words
# that are used on that webpage in order of frequency