【发布时间】：2020-06-08 00:32:47
【问题描述】：
TL;DR 我想让长注释查询（在下面的项目对象管理器部分）更加优雅和高效，并且能够在相关对象更新后自行重新计算。
背景 我正在建立一个市场 - 每个客户都应该能够创建一个包含零件的项目。零件是提供定价的对象。
每个部分都包含一个状态枚举字段 - 因此我们可以知道它是否符合投标或已经在工作中。
我正在使用
- Python 3.8.1
- Django-rest-framework 3.10.3
- django-url-filter 0.3.14
- Postgres 12.1
我需要的功能
- 项目应按其包含的部件状态过滤 - 具有特定条件。
- 项目状态更新应更新所有相关部分。
- 可以使用同一个对象管理器创建、更新和删除项目。
零件模型 只有代码的相关部分：
class PartStatuses(Enum):
Draft = "Saved but not published"
PendingBID = "It's BIDing time!"
Proposal = "All BIDs are set"
PendingPO = "Waiting for vendor to approve PO"
WorkInProgress = "Vendor has accepted a PO"
OnItsWay = "The part is ready and now await to be delivered"
Delivered = "Delivery process has ended"
Disputed = "Open for Dispute"
Closed = "Part has received"
Paid = "Vendor received the payment"
@classmethod
def choices(cls):
return [(key.name, key.value) for key in cls]
class Part(models.Model):
id = models.AutoField(primary_key=True)
title = models.CharField(max_length=100)
project = models.ForeignKey('Project', on_delete=models.CASCADE, related_name='part')
status = models.CharField(choices=PartStatuses.choices(),
max_length=100,
default=PartStatuses.Draft)
...
项目模型 起初，我尝试使用属性方法用于项目状态字段和设置方法用于部件状态更新。
class Project(models.Model):
id = models.AutoField(primary_key=True)
owner = models.ForeignKey(User, on_delete=models.CASCADE, editable=False,
limit_choices_to={'is_vendor': False})
created_at = models.DateTimeField(auto_now_add=True)
updated_at = models.DateTimeField(auto_now=True)
...
def got_parts(self):
return self.part.count()
@property
def _status(self):
if self.got_parts():
all_parts: object = self.part.values_list('status')
# Project is done and paid
if all_parts.filter(status=PartStatuses.Paid.value).count() == all_parts.count():
return str(PartStatuses.Paid.name)
# At least one parts are open for disputed
if all_parts.filter(status=PartStatuses.Disputed.value).count() > 0:
return str(PartStatuses.Disputed.name)
# Some parts are draft
if all_parts.filter(status=PartStatuses.Draft.value).count() > 0:
return str(PartStatuses.Draft.name)
# Some parts are pending to bid
if all_parts.filter(status=PartStatuses.PendingBID.value).count() > 0:
return str(PartStatuses.PendingBID.name)
# Some parts are on proposal
if all_parts.filter(status=PartStatuses.Proposal.value).count() > 0:
return str(PartStatuses.Proposal.name)
# Some parts are pending to vendor approve PO
if all_parts.filter(status=PartStatuses.PendingPO.value).count() > 0:
return str(PartStatuses.PendingPO.name)
# Some parts are in progress of working
if all_parts.filter(status=PartStatuses.WorkInProgress.value).count() > 0:
return str(PartStatuses.WorkInProgress.name)
# Some parts are on their way
if all_parts.filter(status=PartStatuses.OnItsWay.value).count() > 0:
return str(PartStatuses.OnItsWay.name)
# Some parts has been marked by delivery guys as delivered
if all_parts.filter(status=PartStatuses.Delivered.value).count() > 0:
return str(PartStatuses.Delivered.name)
# Some parts has been marked by customer as delivered
if all_parts.filter(status=PartStatuses.Closed.value).count() > 0:
return str(PartStatuses.Closed.name)
# Error with the parts status mapping
return "Other"
else: # No parts - this projects is draft
return str(PartStatuses.Draft.name)
@_status.setter
def _status(self, status):
if any(k[0] == status for k in PartStatuses.choices()) and self.got_parts():
for part in self.part.all():
part.status = PartStatuses[status].value
part.save()
def __str__(self):
return self.title
很快我意识到没有选项可以将此属性作为字段查询，引用docs：
查找中指定的字段必须是模型字段的名称
所以我必须添加我自己的自定义对象管理器，其条件与状态模型属性相同：
class ProjectManager(models.Manager):
"""QuerySet manager for Project class to add non-database fields."""
def get_queryset(self):
"""Overrides the models.Manager method"""
qs = super().get_queryset().annotate(
parts_num=Count(F('part'), distinct=True),
parts_paid=Count(F('part'), filter=Q(part__status__exact=PartStatuses.Paid.value), distinct=True),
parts_disputed=Count(F('part'), filter=Q(part__status__exact=PartStatuses.Disputed.value), distinct=True),
parts_draft=Count(F('part'), filter=Q(part__status__exact=PartStatuses.Draft.value), distinct=True),
parts_pending_bid=Count(F('part'), filter=Q(part__status__exact=PartStatuses.PendingBID.value), distinct=True),
parts_proposal=Count(F('part'), filter=Q(part__status__exact=PartStatuses.Proposal.value), distinct=True),
parts_workin_progress=Count(F('part'), filter=Q(part__status__exact=PartStatuses.WorkInProgress.value),
distinct=True),
parts_pending_PO=Count(F('part'), filter=Q(part__status__exact=PartStatuses.PendingPO.value), distinct=True),
parts_on_its_way=Count(F('part'), filter=Q(part__status__exact=PartStatuses.OnItsWay.value), distinct=True),
parts_delivered=Count(F('part'), filter=Q(part__status__exact=PartStatuses.Delivered.value), distinct=True),
parts_closed=Count(F('part'), filter=Q(part__status__exact=PartStatuses.Closed.value), distinct=True),
).annotate(
status=Case(
When(parts_num=0, then=Value(PartStatuses.Draft.name)),
When(parts_paid=F('parts_num'), then=Value(PartStatuses.Paid.name)),
When(parts_disputed__gt=0, then=Value(PartStatuses.Disputed.name)),
When(parts_draft__gt=0, then=Value(PartStatuses.Draft.name)),
When(parts_pending_bid__gt=0, then=Value(PartStatuses.PendingBID.name)),
When(parts_proposal=F('parts_num'), then=Value(PartStatuses.Proposal.name)),
When(parts_workin_progress__gt=0, then=Value(PartStatuses.WorkInProgress.name)),
When(parts_pending_PO__gt=0, then=Value(PartStatuses.PendingPO.name)),
When(parts_on_its_way__gt=0, then=Value(PartStatuses.OnItsWay.name)),
When(parts_delivered__gt=0, then=Value(PartStatuses.Delivered.name)),
When(parts_closed__gt=0, then=Value(PartStatuses.Closed.name)),
default=Value("Other"),
output_field=CharField()
)
)
return qs
将经理添加到项目模型中：
class Project(models.Model):
with_status = ProjectManager()
objects = models.Manager()
id = models.AutoField(primary_key=True)
owner = models.ForeignKey(User, on_delete=models.CASCADE, editable=False,
limit_choices_to={'is_vendor': False})
created_at = models.DateTimeField(auto_now_add=True)
updated_at = models.DateTimeField(auto_now=True)
title = models.CharField(max_length=100)
...
并在我的项目视图
上使用它from url_filter.integrations.drf import DjangoFilterBackend
from rest_framework import viewsets
class ProjectsViewSet(viewsets.ModelViewSet):
"""
API endpoint that allows projects to be viewed or edited.
"""
serializer_class = ProjectSerializer
permission_classes = (IsAuthenticated, IsVendorStaffOrOwner)
queryset = Project.with_status.prefetch_related('part').all()
filter_backends = [DjangoFilterBackend]
filter_fields = ('status',)
但是我遇到了一个错误：django.core.exceptions.FieldDoesNotExist: Project has no field named 'status'，所以我尝试解决它并编写了自己的过滤器类：
from url_filter.filtersets import FilterSet
from django.db.models import Q
class ProjectsDynamicFilters(FilterSet):
def filter(self):
all_filters = Q()
if 'status' in self.data:
all_filters &= Q(status=self.data['status'])
if len(all_filters):
return self.queryset.filter(all_filters).distinct()
else:
return self.queryset
并在项目视图中使用它代替我的filter_field 属性：filter_class = ProjectsDynamicFilters。
此时，在项目对象管理器接受很长很笨拙的注解查询，我无法更新或创建对象，因为我的项目序列化程序里面有一个状态字段，而创建或更新不能简单地返回一个实例.
更新函数具体是在更新相关部件对象后不重新计算状态字段。
项目序列化器
class ProjectSerializer(serializers.HyperlinkedModelSerializer):
owner = serializers.PrimaryKeyRelatedField(queryset=User.objects.all(),
default=serializers.CurrentUserDefault())
created_at = serializers.DateTimeField(format="%d-%m-%Y %H:%M", read_only=True)
updated_at = serializers.DateTimeField(format="%d-%m-%Y %H:%M", read_only=True)
title = serializers.CharField()
description = serializers.CharField()
status = serializers.ChoiceField(choices=PartStatuses.choices(), default=PartStatuses.Draft.name)
parts = NestedHyperlinkedRelatedField(many=True, source="part", parent_lookup_kwargs={'project_pk': 'project__pk'},
view_name="parts-detail", read_only=True)
...
def update(self, instance, validated_data):
for data in validated_data:
if data == "status":
instance._status = validated_data.get(data)
else:
setattr(instance, data, validated_data.get(data))
instance.save()
return Project.with_status.get(pk=instance.id)
def create(self, validated_data):
validated_data.pop('status', None)
project = Project.objects.create(**validated_data)
return Project.with_status.get(pk=project.id)
回到问题
- 有没有办法让项目
with_status经理查询更加优雅高效？ - 目前我只保留
_status模型属性，用于更新项目状态的setter函数。有没有办法在模型级别使用此功能而无需再次进行此计算？ - 一旦我更改了部件状态，应该重新计算项目状态字段。我通过在更新序列化方法结束时返回
Project.with_status.get(pk=instance.id)来解决它。有没有更明智的方法？
【问题讨论】：
-
我不明白“在模型级别提供此功能而无需再次进行此计算”。在一个请求期间您需要访问多少次？既然是计算出来的值，你只需要在需要显示的时候计算一下。
-
只是一个探索的方向：由于如果零件具有该状态（按特定顺序），您的计算状态似乎是您点击的第一个状态，您可能希望明确地为状态分配数值并获取最小值：
Project.objects.annotate(status=Min('part__status'))。但是，以后添加新状态会很困难。 -
@dirkgroten 关于第一条评论：意图是：如何在没有
_status属性内容的情况下使用设置器功能（与对象管理器查询相同的计算） -
您想要一个 setter 但没有 getter？这是不可能的。只需调用方法
set_status()即可。
标签： python sql django postgresql