我正在尝试在Django的多对多字段自引用上实现所谓的“同义”关系。
以该模型为例(实际上,我不使用真实词,而是使用类别标签):
class Word(models.Model):
name = models.CharField(max_length=30, unique=True)
synonymous = models.ManyToManyField('self', blank=True, related_name='synonymous')
def __str__(self):
return self.name
我要实现的是,当我有3个对象并将它们的任何组合添加到同义词字段中时,我希望所有这些对象都可以连接。
# Create synonymous words
bunny = Word.objects.create(name='bunny')
hare = Word.objects.create(name='hare')
rabbit = Word.objects.create(name='rabbit')
# Set synonymous words to rabbit
rabbit.synonymous.set([bunny, hare])
现在,当我获得Rabbit的同义词对象时,它具有我想要的:
(Pdb) rabbit.synonymous.all()
<QuerySet [<Word: bunny>, <Word: hare>]>
但是当我把兔子和野兔的同义词当作对象时,它们只会返回兔子。
(Pdb) bunny.synonymous.all()
<QuerySet [<Word: rabbit>]>
(Pdb) hare.synonymous.all()
<QuerySet [<Word: rabbit>]>
我想要实现的是所有同义对象,都是“对称的”。现在,m2m字段已经是对称的,但是它仅停留在一个对象上,而不是所有给定的同义对象。
所以,理想的结果是这样:
# Create synonymous words
bunny = Word.objects.create(name='bunny')
hare = Word.objects.create(name='hare')
rabbit = Word.objects.create(name='rabbit')
# Set synonymous ON ONE WORD
rabbit.synonymous.set([bunny, hare])
# And now ALL the objects, which have at least ONE related synonym, would automatically be assigned to the other words as well
(Pdb) rabbit.synonymous.all()
<QuerySet [<Word: bunny>, <Word: hare>]>
(Pdb) hare.synonymous.all()
<QuerySet [<Word: bunny>, <Word: rabbit>]>
(Pdb) bunny.synonymous.all()
<QuerySet [<Word: rabbit>, <Word: hare>]>
我希望这很清楚。
我想知道最干净的方法是什么?也许有一些方法可以通过ORM来做到这一点,但我对此表示怀疑。
我最好只是写一个信号来手动管理那些关系吗?
答案 0 :(得分:1)
我认为最简单的方法是通过以下代码:
for word in synonymous:
word.synonymous.set(synonymous.exclude(pk=word.pk))
修改
放置此代码的最佳位置是您的views.py
之内。如果您使用的是Django Admin,则应使用save_related。
答案 1 :(得分:1)
无法像这样一次创建所有对称关系的直接方法,因为:
rabbit.synonymous
是一个描述符(ManyToManyDescriptor
),它实际上返回有关实例属性访问的ManyRelatedManager
由于多对多管理器将关系附加到单个对象,因此ManyRelatedManager
没有任何预期的方法,因为它是单个实例的属性(rabbit
这种情况)不同于适用于行/实例集合
models.Manager
访问的objects
)
要获得所需的内容,可以创建一个辅助函数来创建传递的对象之间的所有相互关系:
from itertools import combinations
def create_m2m_inter_relations(*objs):
if len(objs) < 2:
raise ValueError(
'There must be at least two objects '
'passed to create relationship.'
)
if len(objs) == 2:
objs[0].set([objs[1]])
return
objs = set(objs)
relations_map = {
next(iter(objs.difference(comb))): comb
for comb in combinations(objs, len(objs) - 1)
}
for instance, relations in relations_map.items():
instance.synonymous.set(relations)
注意事项:
由于Django M2M关系是对称的,因此对于在循环的早期迭代中已建立的关系,以上内容重复了相同的set
操作。读者可以自己删除重复的set
操作。
当您与所有对象创建关系时,ManyToManyDescriptor
-synonymous
的名称相同,因此该函数在函数中进行了硬编码,在这种情况下可以正常工作。但是,如果您想要更通用的解决方案,请考虑到这一点并进行相应的修改。
答案 2 :(得分:0)
与heemayl mentioned一样,无法直接通过ORM进行操作。
所以我最终完成了两个功能,一个用于获取/查找所有同义词关系,另一个用于设置/同步它们
获取/查找:
def get_all_synonymous_words(word):
# Start with the word's own synonymous words, and their synonymous words
found_syn_ids = set(word.synonymous.values_list('id', flat=True)) | set(word.synonymous.values_list('synonymous', flat=True))
while found_syn_ids:
words = Word.objects.filter(id__in=found_syn_ids)
batch_syn_ids = set(words.values_list('synonymous', flat=True))
# If the batch syn ids are equal to last batch ids, all relationships found
if batch_syn_ids == found_syn_ids:
return batch_syn_ids
found_syn_ids = batch_syn_ids
用于设置/同步:
def set_all_synonymous_words(word):
syn_ids = get_all_synonymous_words(word)
if syn_ids:
syn_qs = Word.objects.filter(id__in=syn_ids)
for word in syn_qs:
# Exclude current word to not self reference
excluded_qs = syn_qs.exclude(id=word.id)
if not set(word.synonymous.exclude(id=word.id)) == excluded_qs:
word.synonymous.set(excluded_qs)
这样,当您更新/创建单词时,您可以找到所有组合并进行更新,而无需预先定义同义词列表。
当我使用django-rest-framework时,我通过覆盖perform_update / perform_create方法在视图中将其调用:
def perform_update(self, serializer):
serializer.save()
# Get the updated word, and sync the synonymous words to be all symmetrical
set_all_synonymous_words(Word.objects.get(id=serializer.data['id']))
def perform_create(self, serializer):
serializer.save()
# Get the updated word, and sync the synonymous words to be all symmetrical
set_all_synonymous_words(Word.objects.get(id=serializer.data['id']))
或者您可以手动将其调出:
# Create synonymous words
bunny = Word.objects.create(name='bunny')
hare = Word.objects.create(name='hare')
rabbit = Word.objects.create(name='rabbit')
# Set synonymous
rabbit.synonymous.set([bunny, hare])
# Sync synonymous
set_all_synonymous_words(bunny)