Thanks for the detailed responses, they helped clarify things a lot. I will try the first option you mentioned about creating the Questions first (and saving them) and then creating the Choices using the Questions that were just created. My back-up plan will be to use a different pk field for Question (like question text) that will be already known when I create Choices for a question.
On Monday, November 28, 2016 at 8:55:02 AM UTC-5, C. Kirby wrote: > > > > On Monday, November 28, 2016 at 3:12:37 PM UTC+2, Michael Viveros wrote: >> >> Thanks for the reply Kirby. >> >> I understand that the Question's pk is null when you pass in the Question >> to the Choice constructor. But when you save the Choice, the Question was >> saved earlier so the Question's pk is not null. But the Choice's >> question_id attribute (representing the Question's pk) is still null when >> you save the Choice. >> >> *1. Should the Choice's question_id have been updated to it's current >> value when saving the Choice?* >> > No. The preparing you linked handles type changes between python and the > db, if the db cannot handle the type you have (like the referenced python > datetime). > What is happening is no different than: > > >>> a = 1 > >>> b = a > >>> a = 2 > >>> b > 1 > >>> a > 2 > > >> I read some more about what happens when you save >> <https://docs.djangoproject.com/en/1.10/ref/models/instances/#what-happens-when-you-save> >> an >> object in Django and it seems like the Choice's question_id should be >> updated when the data is prepared for the db (step 3 from the link). >> I understand I can override the Choice class's save method to do >> this (see code below) but it seems like it should be something that is >> already done by the Django framework. >> >> def save(self, *args, **kwargs): >> self.question_id = self.question.id >> super(Choice, self).save(*args, **kwargs) >> >> >> This won't work for 1 or 2 reasons, depending on versions. save() does > not get called when using bulk_create(). Also, unless you are using > postgres and django 1.10, pk's get saved to the db, but not assigned to the > object when using bulk_create() > >> *2. Is there a better way to bulk create objects in a one-to-many >> relationship *(Questions and Choices)*?* >> >> My idea (from the code snippet after models.py) was to bulk create the >> Questions and then bulk create the Choices but this has the problem >> mentioned above of the Choice's FK (question_id) not being updated even >> after the Questions were bulk created. >> > You pretty much can't do it inline like you want with bulk_create(), I > don't think. If you have the question objects retrieved you can then assign > them to the choices and you are all set. If you really want to do it how > you are trying then your best bet would be to assign a primary_key field > instead of using the AutoField primary key provided. If you use the > question title, or generate a uuid for each question and set that as the > primary_field then you can provide the primary key that you know for a > question to the associated choices. > >> >> On Monday, November 28, 2016 at 6:59:13 AM UTC-5, C. Kirby wrote: >>> >>> The reason it isn't working in your first example is because the pk gets >>> assigned to the object when it is saved to the database. When you supply >>> question=q to the choice, it is (correctly) getting a pk of null because q >>> has not been saved. >>> >>> On Saturday, November 26, 2016 at 2:56:37 AM UTC+2, Michael Viveros >>> wrote: >>>> >>>> I have a Question model and a Choice model where a Question can have >>>> many Choices so the Choice model has a Foreign Key mapping to a Question. >>>> >>>> models.py: >>>> >>>> from django.db import models >>>> >>>> >>>> class Question(models.Model): >>>> text = models.CharField(max_length=200) >>>> title = models.CharField(max_length=200) >>>> total_response_count = models.IntegerField() >>>> topic = models.CharField(max_length=20) >>>> sequence_num = models.IntegerField() >>>> select_multiple = models.BooleanField(default=False) >>>> >>>> def __str__(self): >>>> return self.text >>>> >>>> >>>> class Choice(models.Model): >>>> question = models.ForeignKey(Question, on_delete=models.CASCADE) >>>> text = models.CharField(max_length=200) >>>> response_count = models.IntegerField() >>>> response_percent = models.FloatField() >>>> >>>> def __str__(self): >>>> return self.text >>>> >>>> >>>> I need to parse a csv file containing these models and insert them into >>>> a database. >>>> I want to be efficient and bulk create these models instead of having >>>> to call save() after creating each model. >>>> I can bulk create the Questions but then when I try to bulk create the >>>> Choices I get an IntegrityError about the Choice's foreign key id >>>> attribute >>>> ("question_id") violating a not-null constraint. >>>> After creating the Questions, the Question's id attribute gets updated >>>> but the Choice's question_id attribute does not get updated and it does >>>> not >>>> get re-evaluated when the Choice is created. >>>> >>>> Here's a simplified version of what I want to do: >>>> from censustest.models import Question,Choice >>>> >>>> >>>> q=Question(text="Q1",title="",topic="",sequence_num=0,total_response_count=0) >>>> c=Choice(text="C1",response_count=5,response_percent=0.3,question=q) >>>> print("q.id %s, c.question %s, c.question_id %s" % (q.id, c.question, >>>> c.question_id)) # None,Q1,None >>>> Question.objects.bulk_create([q]) >>>> print("After creating q: q.id %s, c.question %s, c.question_id %s" % ( >>>> q.id, c.question, c.question_id)) # valid id,Q1,None >>>> Choice.objects.bulk_create([c]) # ERROR: IntegrityError >>>> >>>> Output: >>>> >>>> <https://lh3.googleusercontent.com/-FFbndWx_lfU/WDjbHC_0QUI/AAAAAAAACPQ/xnwdyg_mVNM6lS-jytZTN7n0ujR7FzYiwCLcB/s1600/Screen%2BShot%2B2016-11-25%2Bat%2B7.42.20%2BPM.png> >>>> >>>> *Is this a bug with Django *(the foreign key id attribute not being >>>> updated)* or is there a better way to do it?* >>>> >>>> An alternative approach I thought of was to assign the Choice's >>>> Question after creating the Questions and this works. >>>> from censustest.models import Question,Choice >>>> >>>> >>>> q=Question(text="Q1",title="",topic="",sequence_num=0,total_response_count=0) >>>> c=Choice(text="C1",response_count=5,response_percent=0.3) >>>> print("q.id %s" % (q.id)) # None,Q1,None >>>> Question.objects.bulk_create([q]) >>>> c.question=q >>>> print("c.question %s, c.question_id %s" % (c.question, c.question_id)) >>>> # Q1,valid id >>>> Choice.objects.bulk_create([c]) # works, no error >>>> >>>> Output: >>>> >>>> <https://lh3.googleusercontent.com/-pncM3xKl0Gc/WDjboxPCzWI/AAAAAAAACPU/V4I9TczbqnAm6mBnogU-kIVCvWXJ7ebrQCLcB/s1600/Screen%2BShot%2B2016-11-25%2Bat%2B7.43.53%2BPM.png> >>>> >>> -- You received this message because you are subscribed to the Google Groups "Django users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/django-users. To view this discussion on the web visit https://groups.google.com/d/msgid/django-users/2eac5cbb-2f5e-4444-9e60-1fb0c446c183%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
