We Keep Coding

How to make more than one fields primary key and remove auto generated id in django models

Suppose in a relational database schema we have a student, a subject and a teacher which connect to each other with a relation teaches. Also, the relation has an attribute time that stores the time of the lesson. This is the most complete yet simplified example I can think to describe my case. Now, the most pythonic and django-wise way I can think of trying to reach a correct solution is, after creating a model class for student, subject and teacher, to create a new class Teaches, which has the foreign keys for the three other classes; also it has the property date field for time. This class would look something like this:
class Teaches(models.Model):
teachers = models.ForeignKey(Teacher, on_delete_models.CASCADE)
subjects = models.ForeignKey(Subject, on_delete_models.CASCADE)
students = models.ForeignKey(Student, on_delete_models.CASCADE)
time = models.DateField
class Meta:
constraints = [
fields=['teachers', 'subjects', 'students']
name='teacher_subject_student_triplet'
]
I added the Meta class because this is what this answer recommends as the correct approach.
The problem is that that in the migrations file I can still see the id field. The only way I've seen there is to remove it is to set another field as Primary Key, but in my case I cannot do that, having more than one keys. Any suggestions?

=========== model.py =============
from django.db import models
class TeacherModel(models.Model):
teacher_code = models.CharField(max_length=255)
def __str__(self):
return self.teacher_code
class SubjectModel(models.Model):
subject_code = models.CharField(max_length=255)
def __str__(self):
return self.subject_code
class StudentModel(models.Model):
student_code = models.CharField(max_length=255)
def __str__(self):
return self.student_code
class Teaches(models.Model):
custom_primary_key = models.SlugField(primary_key=True,blank=True)
teacher = models.ForeignKey(TeacherModel, on_delete=models.CASCADE)
subject = models.ForeignKey(SubjectModel, on_delete=models.CASCADE)
student = models.ForeignKey(StudentModel, on_delete=models.CASCADE)
time = models.DateField
#property
def make_key(self):
new_key = str(self.teacher.teacher_code + self.subject.subject_code + self.student.student_code)
return new_key
def save(self, *args, **kwargs):
self.custom_primary_key = self.make_key
super(Teaches, self).save(*args, **kwargs)
========= Output ==============

You can remove autogenerated id by adding primary_key=True, see below code:
class Person(models.Model):
username = CharField(primary_key=True, max_length=100)
first_name = CharField(null=True, blank=True, max_length=100)
setting a field to primary_key=True automatically makes it unique and not null.

In settings.py:
DEFAULT_AUTO_FIELD = 'django.db.models.BigAutoField'
Controls the automatic generation of primary keys of each model if defined in settings.
Read this article:
Set AutoField or BigAutoField on a per model basis

What is the "instance" being passed to the to_representation function of my ListSerializer?

The goal of this project is to create an API that refreshes hourly with the most up to date betting odds for a list of games that I'll be scraping hourly from the internet. The goal structure for the JSON returned will be each game as the parent object and the nested children will be the top 1 record for each of linesmakers being scraped by updated date. My understanding is that the best way to accomplish this is to modify the to_representation function within the ListSerializer to return the appropriate queryset.
Because I need the game_id of the parent element to grab the children of the appropriate game, I've attempted to pull the game_id out of the data that gets passed. The issue is that this line looks to be populated correctly when I see what it contains through an exception, but when I let the full code run, I get a list index is out of range exception.
For ex.
class OddsMakerListSerializer(serializers.ListSerializer):
def to_representation(self, data):
game = data.all()[0].game_id
#if I put this here it evaluates to 1 which should run the raw sql below correctly
raise Exception(game)
data = OddsMaker.objects.filter(odds_id__in = RawSQL(''' SELECT o.odds_id
FROM gamesbackend_oddsmaker o
INNER JOIN (
SELECT game_id
, oddsmaker
, max(updated_datetime) as last_updated
FROM gamesbackend_oddsmaker
WHERE game_id = %s
GROUP BY game_id
, oddsmaker
) l on o.game_id = l.game_id
and o.oddsmaker = l.oddsmaker
and o.updated_datetime = l.last_updated
''', [game]))
#if I put this here the data appears to be populated correctly and contain the right data
raise Exception(data)
data = [game for game in data]
return data
Now, if I remove these raise Exceptions, I get the list index is out of range. My initial thought was that there's something else that depends on "data" being returned as a list, so I created the list comprehension snippet, but that doesn't resolve the issue.
So, my question is 1) Is there an easier way to accomplish what I'm going for? I'm not using a postgres backend so distinct on isn't available to me. and 2) If not, its not clear to me what instance is that's being passed in or what is expected to be returned. I've consulted the documentation and it looks as though it expects a dictionary and that might be part of the issue, but again the error message references a list. https://www.django-rest-framework.org/api-guide/serializers/#overriding-serialization-and-deserialization-behavior
I appreciate any help in understanding what is going on here in advance.
Edit:
The rest of the serializers:
class OddsMakerSerializer(serializers.ModelSerializer):
class Meta:
list_serializer_class = OddsMakerListSerializer
model = OddsMaker
fields = ('odds_id','game_id','oddsmaker','home_ml',
'away_ml','home_spread','home_spread_odds',
'away_spread_odds','total','total_over_odds',
'total_under_odds','updated_datetime')
class GameSerializer(serializers.ModelSerializer):
oddsmaker_set = OddsMakerSerializer(many=True, read_only=True)
class Meta:
model = Game
fields = ('game_id','date','sport', 'home_team',
'away_team','home_score', 'away_score',
'home_win','away_win', 'game_completed',
'oddsmaker_set')
models.py:
class Game(models.Model):
game_id = models.AutoField(primary_key=True)
date = models.DateTimeField(null=True)
sport=models.CharField(max_length=256, null=True)
home_team = models.CharField(max_length=256, null=True)
away_team = models.CharField(max_length=256, null=True)
home_score = models.IntegerField(default=0, null=True)
away_score = models.IntegerField(default=0, null=True)
home_win = models.BooleanField(default=0, null=True)
away_win = models.BooleanField(default=0, null=True)
game_completed = models.BooleanField(default=0, null=True)
class OddsMaker(models.Model):
odds_id = models.AutoField(primary_key=True)
game = models.ForeignKey('Game', on_delete = models.CASCADE)
oddsmaker = models.CharField(max_length=256)
home_ml = models.IntegerField(default=999999)
away_ml = models.IntegerField(default=999999)
home_spread = models.FloatField(default=999)
home_spread_odds = models.IntegerField(default=9999)
away_spread_odds = models.IntegerField(default=9999)
total = models.FloatField(default=999)
total_over_odds = models.IntegerField(default=999)
total_under_odds = models.IntegerField(default=999)
updated_datetime = models.DateTimeField(auto_now=True)
views.py:
class GameView(viewsets.ModelViewSet):
queryset = Game.objects.all()
serializer_class = GameSerializer
Thanks

To answer the question in the title:
The instance being passed to the Serializer.to_representation() is the instance you pass when initializing the serializer
queryset = MyModel.objects.all()
Serializer(queryset, many=True)
instance = MyModel.objects.all().first()
Serializer(data)
Usually you don't have to inherit from ListSerializer per se. You can inherit from BaseSerializer and whenever you pass many=True during initialization, it will automatically 'becomeaListSerializer`. You can see this in action here
To answer your problem
from django.db.models import Max
class OddsMakerListSerializer(serializers.ListSerializer):
def to_representation(self, data): # data passed is a queryset of oddsmaker
# Do your filtering here
latest_date = data.aggregate(
latest_date=Max('updated_datetime')
).get('latest_date').date()
latest_records = data.filter(
updated_date_time__year=latest_date.year,
updated_date_time__month=latest_date.month,
updated_date_time__day=latest_date.day
)
return super().to_representation(latest_records)

Django Models: Default for Choice Foreign Key

I have a Type model class as follows:
class Type(models.Model):
ENVIRONMENT = 'A'
HUMANENV = 'B'
HUMAN = 'C'
ANIMAL = 'D'
UNKNOWN = 'H'
TYPE_CHOICES = [
(ENVIRONMENT, 'Environment'),
(HUMANENV, "Human-related Environment"),
(HUMAN, 'Human'),
(ANIMAL, 'Animal'),
(UNKNOWN, 'Unknown'),
]
code = models.CharField(max_length=1, choices=TYPE_CHOICES, unique=True)
class Meta:
ordering = ['code']
def __str__(self):
return self.get_code_display()
And another Sample model where one of the fields is a foreign key to the Type model as follows:
class Sample(models.Model):
sample_id = models.CharField(max_length=20, unique=True)
type = models.ForeignKey("Type", on_delete=models.CASCADE, blank=True, default=get_default_type())
class Meta:
ordering = ["sample_id"]
def __str__(self):
return self.sample_id
where get_default_type is a function that returns the pk for the default Type model instance:
def get_default_type():
return Type.objects.get(code="H").id
The problem is when I run Sample.objects.create(sample_id="some_id"), it is giving me the error
IntegrityError: null value in column "type_id" violates not-null constraint
DETAIL: Failing row contains (28113, some_id, null).
As you can see in the second line of the error message, the type_id is null instead of the pk as returned by the get_default_type function.
I have tried setting null=True for the foreign key and when I do that I am able to create the Sample model instance, but with a None type instead of the Unknown type as I wanted. How can I fix this?

Two solutions:
Override the manager
From this response you could use get_by_natural_key in your manager.
In managers.py
from django.db import models
class TypeManager(models.Manager):
"""Enable fixtures using self.sigla instead of `id`"""
def get_by_natural_key(self, code):
return self.get(code=code)
class Type(models.Model):
#.... Declare your model here
objects = Type()
or...
Change your pk!
class Type(models.Model):
#.... Declare your model here
code = models.CharField(max_length=1, choices=TYPE_CHOICES, unique=True, primary_key=True)
Either way, in your related model declaration:
class Sample(models.Model):
type = models.ForeignKey("Type", on_delete=models.CASCADE, blank=True, default='H')
class Meta:
ordering = ["sample_id"]
def __str__(self):
return self.sample_id
As a side note: Please take care about type.
It is a protected keyword and should not be used.

Searching by related fields in django admin

I've been looking at the docs for search_fields in django admin in the attempt to allow searching of related fields.
So, here are some of my models.
# models.py
class Team(models.Model):
name = models.CharField(max_length=255)
class AgeGroup(models.Model):
group = models.CharField(max_length=255)
class Runner(models.Model):
"""
Model for the runner holding a course record.
"""
name = models.CharField(max_length=100)
agegroup = models.ForeignKey(AgeGroup)
team = models.ForeignKey(Team, blank=True, null=True)
class Result(models.Model):
"""
Model for the results of records.
"""
runner = models.ForeignKey(Runner)
year = models.IntegerField(_("Year"))
time = models.CharField(_("Time"), max_length=8)
class YearRecord(models.Model):
"""
Model for storing the course records of a year.
"""
result = models.ForeignKey(Result)
year = models.IntegerField()
What I'd like is for the YearRecord admin to be able to search for the team which a runner belongs to. However as soon as I attempt to add the Runner FK relationship to the search fields I get an error on searches; TypeError: Related Field got invalid lookup: icontains
So, here is the admin setup where I'd like to be able to search through the relationships. I'm sure this matches the docs, but am I misunderstanding something here? Can this be resolved & the result__runner be extended to the team field of the Runner model?
# admin.py
class YearRecordAdmin(admin.ModelAdmin):
model = YearRecord
list_display = ('result', 'get_agegroup', 'get_team', 'year')
search_fields = ['result__runner', 'year']
def get_team(self, obj):
return obj.result.runner.team
get_team.short_description = _("Team")
def get_agegroup(self, obj):
return obj.result.runner.agegroup
get_agegroup.short_description = _("Age group")

The documentation reads:
These fields should be some kind of text field, such as CharField or TextField.
so you should use 'result__runner__team__name'.

django, table inheritance: insert data

I have these tables:
class OpeDatos(models.Model):
id_dato = models.IntegerField(primary_key=True)
id_usuario = models.ForeignKey(SisUsuarios, db_column='id_usuario')
id_region = models.ForeignKey(SisRegiones, db_column='id_region')
titulo = models.CharField(max_length=70, blank=True)
class Meta:
managed = False
db_table = 'ope_datos'
class OpeProductos(OpeDatos):
id_producto = models.IntegerField(primary_key=True)
iddato = models.OneToOneField(OpeDatos, primary_key=True, db_column="id_dato", parent_link=True)
id_producto_tipo = models.ForeignKey(DefProductosTipos, db_column='id_producto_tipo')
class Meta:
managed = False
db_table = 'ope_productos'
I want to insert data :
from apps.inicio.models import SisUsuarios, SisRegiones, OpeDatos
usuario = SisUsuarios.objects.get(pk=1)
region = SisRegiones.objects.get(pk=1)
datos = OpeDatos()
datos.id_usuario = usuario
datos.id_region = region
datos.save()
producto = OpeProductos()
producto.iddato = datos.id_dato
producto.save()
displays this message:
ValueError at /productos/add/
Cannot assign None: "OpeProductos.iddato" does not allow null values.
can you help me, please.

When creating an id manually you should use AutoField instead of IntegerField
id_dato = models.AutoField(primary_key=True)
https://docs.djangoproject.com/en/1.6/ref/models/fields/#autofield
What is happening is that since you are not explicitly defining the 'datos' object's id it doesnt have one, and then producto complains because the key can't have an empty value.
AutoField should fix this

I am surprised it doesn't fail at the earlier line: datos.save() because you have not supplied a value for the primary key datos.id_dato
Normally in Django you would need to use an AutoField to get an auto-incrementing primary key.
Also you should not be specifying primary_key=True on the OpeProductos.iddato field, you can only have one primary key per model.
Then the error you are seeing is due to the fact that datos.id_dato is None since you did not provide any value for it before saving the datos instance.