1a.py

# -*- coding: utf-8 -*-
"""
Created on Fri May  6 11:01:25 2016

@author: xinruyue
"""
from pymongo import MongoClient

db = MongoClient("10.8.8.111:27017")["eventsV35"]
events = db['eventV35']

def video_ids(eventKey,num):
    pipeline = [
    {"$match":{"eventKey":eventKey}},
    {"$group": {"_id":None, "videoId": {"$push": "$eventValue.videoId"}}}]
    
    data = list(events.aggregate(pipeline))[0]
    videoId = data["videoId"]
    
    sum_video = {}
    for i in videoId:
        sum_video[i] = videoId.count(i)
    
    sortSV = sorted(sum_video.iteritems(), key=lambda d:d[1], reverse = True)
    videoIds = []
    for each in sortSV:
        if sortSV.index(each) < num:
            videoIds.append(each[0])
    return(videoIds)


EA_lists = ["enterVideo","startVideo",["finishVideo","clickVideoExit"],
           {"enterTAVideoExitPoll":["refuse","quit"]}]
EP_lists = ["enterVideo","startVideo",["finishVideo","clickVideoExit"],
           {"enterTPVideoExitPoll":["refuse","quit"]}]

#get video_ids
videoA_ids = video_ids("enterTAVideoExitPoll",10)
videoP_ids = video_ids("enterTPVideoExitPoll",5)

def group_users(eventKey,eventValue,input_var,step_users,device_mark):
    pipeline = [
    {"$match": {"eventKey": eventKey, "eventValue.videoId": 
        {"$in": input_var['video_ids']},"user":{"$in":step_users}}},
    {"$group": {"_id": None, "users": {"$addToSet": "$user"}}}]
    
    if step_users == None:
        del pipeline[0]["$match"]["user"]
    
    if device_mark:
        del pipeline[0]["$match"]["user"]
        pipeline[0]["$match"]["device"]= {"$in":step_users}               
    
    if eventValue:
        pipeline.insert(1,{"$match":{"eventValue."+eventKey:eventValue}})
        
    full_users = list(events.aggregate(pipeline))[0]['user']
    return(full_users)

def caculate_data(input_var,device_mark):
    #第一个漏斗，进入视频
    first_step = input_var["event_lists"][0]
    full_users = group_users(first_step,None,input_var,None,device_mark)     
    result = {first_step:len(full_users)}
    #其他漏斗
    for each_step in input_var['event_lists'][1:]:
        if type(each_step) == list:
            for each in each_step:
                each_step = each
                this_step_users = group_users(each_step,None,input_var,full_users,device_mark)           
                result[each_step]= len(this_step_users)  
                if each_step[-1]:
                    full_users = this_step_users 
        if type(each_step) == dict:         
            this_event_key = each_step.keys()
            event_values = each_step.values()[0]
            for each in event_values:
                this_event_value = each
                this_step_users = group_users(this_event_key,this_event_value,input_var,full_users,device_mark)
                result[this_event_key+'.'+this_event_value]=len(this_step_users)
        else:
            this_step_users = group_users(each_step,None,input_var,full_users,device_mark)           
            result[each_step]= len(this_step_users)           
            full_users = this_step_users         
    return(result)

def integrate_data(event_lists,videoX_ids):
    input_var = {"event_lists":event_lists,"video_ids":videoX_ids}
    caculate_data(input_var,device_mark)

if :
device_mark = True
    
result_A = integrate_data(EA_lists,videoA_ids)
result_P = integrate_data(EP_lists,videoP_ids)