3

Currently, I am using DRF version '3.9.2'. I was doing some CSV export from models synchronously. Sample of the code blocks explained as below:

urls.py

from django.urls import path

urlpatterns = [
    path('api/users-csv-export/', UsersExportAsCSV.as_view())
]

views.py

from rest_framework.views import APIView

def get_users_data():
    queryset = User.objects.only('first_name', 'last_name', 'created_at', 'email', 'gender', 'date_of_birth') 
    fields = ['first_name', 'last_name', 'created_at', 'email', 'gender', 'date_of_birth']
    titles = ['First Name', 'Last Name', 'Date Added', 'Email', 'Gender', 'Date of Birth']
    file_name = 'users'
    return queryset, fields, titles, file_name

class UsersExportAsCSV(APIView):
    def get(self, request):
        users = get_users_data()
        data = export_to_csv(queryset=users[0], fields=users[1], titles=users[2], file_name=users[3])
        return data

utils.py

def export_to_csv(queryset, fields, titles, file_name):
    """
    will export the model data in the form of csv file
    :param queryset: queryset that need to be exported as csv
    :param fields: fields of a model that will be included in csv
    :param titles: title for each cell of the csv record
    :param file_name: the exported csv file name
    :return:
    """
    model = queryset.model
    response = HttpResponse(content_type='text/csv')
    # force download
    response['Content-Disposition'] = 'attachment; filename={}.csv'.format(file_name)
    # the csv writer
    writer = csv.writer(response)
    if fields:
        headers = fields
        if titles:
            titles = titles
        else:
            titles = headers
    else:
        headers = []
        for field in model._meta.fields:
            headers.append(field.name)
        titles = headers

    # Writes the title for the file
    writer.writerow(titles)

    # write data rows
    for item in queryset:
        writer.writerow([nested_getattr(item, field) for field in headers])
    return response

As the data grows the request-response cycle becomes heavier. To prevent this request blocking other requests Threading vs Async task which one is preferable? Is there any good idea about optimizing the request-response cycle.

Here the user has to wait until whole the exporting works and download complete. The desired result here is that whenever the user visits the URL then the instant response of "file being generated" and generating and downloading the file in the background.

Any help on this topic will be highly appreciated.

dipesh
  • 763
  • 2
  • 9
  • 27

3 Answers3

2

I think you can achieve with this http://www.celeryproject.org/

You can use "shared_task" to complete task for generating the csv file in background and save it to table (for ex: DownloadFileModel) if it already finish. Then you can download it later.
Response your current view with redirecting to DetailView (DownloadFileModel record with null value in file field) that you prepared to download the file, if the file is unready just give the description to wait until the file is ready (and you can assign the file to the record).

Reiza Judiz
  • 21
  • 1
  • As I explained in the previous answer of aman kumar, I need to pass the class object inside the task function, which was not possible in the celery task. I did the use of threading in python and the Redis cache server to store the record the data temporarily. The concept you gave has inspired me to change the whole logic for the code. Thank you :) – dipesh Jan 15 '20 at 07:44
1

i think, you can use the celery to run async task, it will give the lot of functionality show the progress and when task has been completed. 

https://pypi.org/project/django-celery/

from celery import task

def get_users_data():
    queryset = list(User.objects.values_list('first_name', 'last_name', 'created_at', 'email', 'gender', 'date_of_birth')) 
    fields = ['first_name', 'last_name', 'created_at', 'email', 'gender', 'date_of_birth']
    titles = ['First Name', 'Last Name', 'Date Added', 'Email', 'Gender', 'Date of Birth']
    file_name = 'users'
    return queryset, fields, titles, file_name



@task
def export_to_csv(queryset, fields, titles, file_name):
    """
    will export the model data in the form of csv file
    :param queryset: queryset that need to be exported as csv
    :param fields: fields of a model that will be included in csv
    :param titles: title for each cell of the csv record
    :param file_name: the exported csv file name
    :return:
    """
    model = queryset.model
    response = HttpResponse(content_type='text/csv')
    # force download
    response['Content-Disposition'] = 'attachment; filename={}.csv'.format(file_name)
    # the csv writer
    writer = csv.writer(response)
    if fields:
        headers = fields
        if titles:
            titles = titles
        else:
            titles = headers
    else:
        headers = []
        for field in model._meta.fields:
            headers.append(field.name)
        titles = headers

    # Writes the title for the file
    writer.writerow(titles)

    # write data rows
    # here you can save the file at particular path 
    for item in queryset:
        writer.writerow([nested_getattr(item, field) for field in headers])
    return file_path


class UsersExportAsCSV(APIView):
    def get(self, request):
        users = get_users_data()
        task_id = export_to_csv.delay(queryset=users[0], fields=users[1], titles=users[2], file_name=users[3])
        return task_id


using the task id you can get the result of that

another approach is using djnago-channel, it uses the socket connection you don't need to make pooling request for check task is completed or not

aman kumar
  • 3,086
  • 1
  • 17
  • 24
  • I am trying to use celery tasks. It 's quite tough to generate CSV, download it and send response at the same time. Can you provide the celery task on how the above mentioned scenarios can be overcome? – dipesh Jan 14 '20 at 08:58
  • The above code results following error response > kombu.exceptions.EncodeError: Object of type BaseQuerySet is not JSON serializable The **.delay()** in task returns ***celery.result.AsyncResult: Future promise*** – dipesh Jan 14 '20 at 09:26
  • instead of object pass, either list of id , data which can be serialize, not class object – aman kumar Jan 14 '20 at 09:28
  • I used the threading to solve this problem because I need to send the queryset in the function. The function above mentioned was so generalized to work for different scenarios. Thank you for your valuable answer, It helped me to change a few logics in my code. – dipesh Jan 15 '20 at 07:42
1

To solve my problem, I used the python threading module and to store the file record I used the Redis cache server.

At first, the GET method will check whether the exported file record available in cache or not. If it is available then URL to that file is sent in response. The POST method will generate the file and write it in media/temp and store the record in the cache.

Few changes are made in the code blocks and are explained as below:

views.py

from rest_framework.views import APIView

def get_users_data():
    queryset = User.objects.only('first_name', 'last_name', 'created_at', 'email', 'gender', 'date_of_birth') 
    fields = ['first_name', 'last_name', 'created_at', 'email', 'gender', 'date_of_birth']
    titles = ['First Name', 'Last Name', 'Date Added', 'Email', 'Gender', 'Date of Birth']
    return queryset, fields, titles, file_name

class TripHistoryExportAsCSV(APIView):
    file_name = "users_all"
    file_extension = 'csv'

    def post(self, request):
        try:
            queryset = get_users_data()[0]
            fields = get_users_data()[1]
            titles = get_users_data()[2]
            x = threading.Thread(target=export_to_csv, args=(queryset, fields, titles, self.file_name))
            x.start()
            return Response({
                'message': 'CSV file is generating'
            })
        except EmptyResultSet:
            return Response({
                'message': 'Can not create CSV file'
            }, status=status.HTTP_200_OK)

    def get(self, request):
        data = check_export_data_in_cache(self.file_name, self.file_extension)
        if data:
            return Response({
                'url': data.get('report_url')
            })
        else:
            return Response({
                'message': 'Generation of new files required'
            }, status=status.HTTP_204_NO_CONTENT)

utils.py

def nested_getattr(obj, attribute, split_rule='__'):
    """
    This function is responsible for getting the nested record from the given obj parameter
    :param obj: whole item without splitting
    :param attribute: field after splitting
    :param split_rule:
    :return:
    """
    split_attr = attribute.split(split_rule)
    for attr in split_attr:
        if not obj:
            break
        obj = getattr(obj, attr)
    return obj


def export_to_csv(queryset, fields, titles, file_name):
    """
    will export the model data in the form of csv file
    :param queryset: queryset that need to be exported as csv
    :param fields: fields of a model that will be included in csv
    :param titles: title for each cell of the csv record
    :param file_name: the exported csv file name
    :return:
    """
    model = queryset.model
    import os
    from yatruadminbackend.settings import MEDIA_ROOT
    if fields:
        headers = fields
        if titles:
            titles = titles
        else:
            titles = headers
    else:
        headers = []
        for field in model._meta.fields:
            headers.append(field.name)
        titles = headers

    with open(os.path.join(MEDIA_ROOT, f'temp/{file_name}.csv'), 'w', newline='') as file:
        # Writes the title for the file
        writer = csv.writer(file)
        writer.writerow(titles)
        # write data rows
        for item in queryset:
            writer.writerow([nested_getattr(item, field) for field in headers])
        set_cache_for_export_file(file_name, 'csv')


def set_cache_for_export_file(filename, extension):
    generated_date = timezone.now()
    export_file_name = f'{filename}_{extension}'
    record_in_cache = {
        'key': export_file_name,
        'report_url': f'{BACKEND_URL}media/temp/{filename}.csv',
        'generated_on': generated_date
    }
    cache.set(export_file_name, record_in_cache, 300)
    print(cache.get(export_file_name))


def check_export_data_in_cache(file_name, file_extension):
    cache_key = f'{file_name}_{file_extension}'
    print(cache_key)
    if cache.get(cache_key):
        return cache.get(cache_key)
dipesh
  • 763
  • 2
  • 9
  • 27