When "select_related" is needed? - django

Question

In my project , Each candidate can takepart in some assessments,each assessment has some tests, each test has some questions in it and candidates should answer the questions
at last scores of the questions are saved in question_score and test_score table
I need to get some values of field and use them
I write a method for question_result table, to get them
but i dont know if it is needed to use select_related or not
if it is needed how can i use it ?
Assessment:
class Assessment(BaseModel):
company = models.ForeignKey(
'company.Company',
on_delete=models.CASCADE,
related_name='assessments',
)
title = models.CharField(max_length=255)
job_role = models.ForeignKey(
JobRole,
on_delete=models.PROTECT,
related_name='assessments',
blank=True,
null=True,
)
tests = models.ManyToManyField(
'exam.Test',
related_name='assessments',
blank=True,
through='TestOfAssessment',
)
candidates = models.ManyToManyField(
'user.User',
related_name='taken_assessments',
blank=True,
through='candidate.Candidate'
)
def __str__(self):
return self.title
Test:
class Test(BaseModel):
class DifficultyLevel(models.IntegerChoices):
EASY = 1
MEDIUM = 2
HARD = 3
company = models.ForeignKey(
'company.Company',
on_delete=models.PROTECT,
related_name='tests',
null=True,
blank=True,
)
questions = models.ManyToManyField(
'question.Question',
related_name='tests',
blank=True,
help_text='Standard tests could have multiple questions.',
)
level = models.IntegerField(default=1, choices=DifficultyLevel.choices)
title = models.CharField(max_length=255)
summary = models.TextField()
def __str__(self):
return self.title
Question :
class Question(BaseModel):
company = models.ForeignKey(
'company.Company',
on_delete=models.SET_NULL,
null=True,
blank=True,
related_name='company_questions',
)
question_text = models.TextField()
def __str__(self):
return truncatewords(self.question_text, 7)
TestResult:
class TestResult(BaseModel):
candidate = models.ForeignKey(
'Candidate',
on_delete=models.CASCADE,
related_name='test_results',
)
test = models.ForeignKey(
'exam.Test',
on_delete=models.CASCADE,
)
test_score = models.DecimalField(default=0.00, max_digits=5, decimal_places=2)
def __str__(self):
return f'{self.candidate.user.email} - {self.test.title}'
Candidate :
class Candidate(BaseModel):
assessment = models.ForeignKey(
'assessment.Assessment',
on_delete=models.CASCADE,
)
user = models.ForeignKey(
'user.User',
on_delete=models.CASCADE,
)
is_rejected = models.BooleanField(default=False)
def __str__(self):
return f'{self.user.email} - {self.assessment.title}'
Company :
class Company(models.Model):
manager = models.ForeignKey('user.User', on_delete=models.CASCADE, related_name='user_companies')
name = models.CharField(max_length=255)
city = models.ForeignKey('company.City', null=True, on_delete=models.SET_NULL)
address = models.CharField(max_length=255, blank=True, null=True)
def __str__(self):
return self.name
QuestionResult :
class QuestionResult(BaseModel):
test = models.ForeignKey(
'TestResult',
on_delete=models.CASCADE,
related_name='question_results',
)
question = models.ForeignKey(
'question.Question',
on_delete=models.CASCADE,
related_name='results',
)
result = models.TextField(
null=True,
blank=True,
)
answer_score = models.DecimalField(default=0.00, max_digits=5, decimal_places=2)
def __str__(self):
return f'{self.test.candidate.user.email} - {self.question}'
def text_variables(self):
email = self.test.candidate.user.email
company_name = self.test.test.company.name
assessment_name = self.test.candidate.assessment.title
candidate_first_name = self.test.candidate.user.first_name
job_name = self.test.candidate.assessment.job_role
user_fullname = User.full_name
data = dict(
job_name=job_name,
company_name=company_name,
email=email,
assessment_name=assessment_name,
candidate_first_name=candidate_first_name,
job_name=job_name,
user_fullname = user_fullname
)
return data
I wrote the def text_variables(self): method to fill the data dictionary and use it somewhere else
it work properly but i dont know if it needed to use selected_related or not
something like this (it does not work)
def text_variables(self):
question_result_object = QuestionResult.objects.filter(id=self.id).select_related(
"test__candidate","test__test__company","test__candidate__assessment")
email = question_result_object.test.candidate.user.email
company_name = question_result_object.test.test.company.name
assessment_name = question_result_object.test.candidate.assessment.title
candidate_first_name = question_result_object.test.candidate.user.first_name
job_name = question_result_object.test.candidate.assessment.job_role
data = dict(
job_name=job_name,
company_name=company_name,
email=email,
assessment_name=assessment_name,
candidate_first_name=candidate_first_name,
job_name=job_name,
user_fullname = user_fullname
)
return data
the error is :
File "E:\work\puzzlelity\talent-backend\candidate\models.py", line 385, in report_to_candidate_email_text_variables
email = question_result_object.test.candidate.user.email
AttributeError: 'QuerySet' object has no attribute 'test'
[03/Jan/2023 17:59:00] "POST /api/v1/candidatures/183f8432-ea81-4099-b211-3b0e6475ffab/submit-answer/ HTTP/1.1" 500 123319
I dont know how should i use the select_related

Answer 1

It's never required. It optimizes querysets, especially in ListViews.
Consider your Assessment model. It has ForeignKey fields company and job_role. If you simply fetch
assessment = Assessment.objects.get( id=something)
and then refer to assessment.company, that causes a second DB query to fetch the company object. And then a third if you refer to assessment.job_role.
You can reduce these three queries to one by using
assessment = Assessment.objects.select_related(
'company', 'job_role') .get( id=something)
which does a more complex query to retrieve all the data.
Where it matters is in a list view where you iterate over a large number of assessment objects in Python or in a template. For example, if object_list is assessment.objects.all() and there are 300 of them, then
{% for assessment in object_list %}
... stuff ...
{{assessment.company.name }}
...
{% endfor %}
Will hit the DB 300 times, once for each company! If you use select_related, all 300 companies linked to the 300 assessments will be retrieved in a single DB query. which will be very noticeably faster.
I'd strongly recommend installing Django Debug Toolbar in your development project. Then click on the SQL option on any view, and you can see what SQL was required, and in particular how many SQL queries were performed and whether there were batches of repetetive queries which mean there's a trivial optimisation to be made.