Bad Performance in my code

I'm having a performance problem in my code. I'm a little new to Python and I can not think of a better way to do the following code.

I have an external database with a table called "cdr" that is not a part of a django project and I need to do some calculations with the rows. To get my variables' values I'm making a query for every row in my cdr table and this is making my code very slow.

Here is my class in view.py

def cdr_adm(request):
        cursor = connections['cdr'].cursor()
        cursor.execute("SELECT calldate, dst, billsec, accountcode, disposition, userfield FROM cdr where calldate >= '%s' and calldate < '%s' and disposition like '%s' and accountcode like '%s' and dst like '%s' and userfield like '%s'" %(start_date, end_date, status, customer, destino, provider))
        result = cursor.fetchall()
        time = 0
        price = 0
        price_prov = 0
        count = 0
        time_now = 0
        ANS = 0
        asr = 0
        rate_cust = 0
        rate_prov = 0

        for call in result:
                if call[3]:
                        #These 2 lines are the problem - It's very slow to run for all rows.
                        rate_cust = User.objects.get(username = call[3])
                        rate_prov = Provider.objects.get(name = call[5])
                        time_now = call[2] / 60
                        time =  time + time_now
                        count = count + 1
                        price = price + (float(rate_cust.first_name) * time_now)
                        price_prov = price_prov + (float(rate_prov.rate) * time_now)
                        if call[4] == "ANSWERED":
                                ANS = ANS + 1
        time = float(time)
        lucro = price - price_prov
        lucro = float(lucro)
        if count > 0:
                asr = (ANS / count)  * 100
        return render_to_response("cdr_all.html",
                {'calls':result,'price':price,'time':time,'count':count,'customers':customers, 'providers':providers,'price_prov':price_prov, 'lucro':lucro, 'asr':asr }, context_instance=RequestContext(request))

I was thinking about creating a dictionary and search in it but I'm not sure about it too.

Upvotes: 1

Views: 157

Answers (1)

David Robinson
David Robinson

Reputation: 78600

You can create a dictionary of all User and Provider objects, indexed by what you're interested in, like this:

users = dict([(u.username, u) for u in User.objects.all()])
providers = dict([(p.name, p) for p in Provider.objects.all()])

(Make sure you do this outside the for call in result: for loop!) You can then change your slow queries to:

                    rate_cust = users[call[3]]
                    rate_prov = provided[call[5]]

I'm guessing there are considerably fewer users and providers than calls, which means that keeping them in a dictionary will be much faster to access than making one query for each call.

Upvotes: 2

Related Questions