How to filter django model by its objects in many-to-many field (exact match)?

陌路散爱 提交于 2019-12-18 19:01:20

问题


I have this model in my code:

class Conversation(models.Model):
    participants = models.ManyToManyField(User, related_name="message_participants")

and I need to filter this "Conversation" model objects by the "participants" many-to-many field. meaning: I have for example 3 User objects, so I want to retrieve the only "Conversation" objects that has this 3 Users in it's "participants" field.

I tried doing this:

def get_exist_conv_or_none(sender,recipients):
    conv = Conversation.objects.filter(participants=sender)
    for rec in recipients:
        conv = conv.filter(participants=rec)

where sender is a User object and "recipients" is a list of User objects. it won't raise error but it gives me the wrong Object of Conversation. Thanks.

edit: A more recent try lead me to this:

def get_exist_conv_or_none(sender,recipients):
    participants=recipients
    participants.append(sender)
    conv = Conversation.objects.filter(participants__in=participants)
    return conv

which basically have the same problem. It yields Objects which has one or more of the "participants" on the list. but what Im looking for is exact match of the many-to-many object. Meaning, an Object with the exact "Users" on it's many-to-many relation.

edit 2: My last attempt. still, won't work.

def get_exist_conv_or_none(sender,recipients):
    recipients.append(sender)
    recipients = list(set(recipients))
    conv = Conversation.objects.annotate(count=Count('participants')).filter(participants=recipients[0])
    for participant in recipients[1:]:
        conv.filter(participants=participant)
    conv.filter(count=len(recipients))
    return conv

回答1:


Ok so I found the answer: In order to make an exact match I have to chain-filter the model and then make sure it has the exact number of arguments it needs to have, so that the many-to-many field will have in it all the objects needed and no more.

I will check for the objects number using annotation: ( https://docs.djangoproject.com/en/dev/topics/db/aggregation/ )

ended up with this code:

def get_exist_conv_or_none(recipients):
    conv = Conversation.objects.annotate(count=Count('participants')).filter(participants=recipients[0])
    for participant in recipients[1:]:
        conv = conv.filter(participants=participant)
    conv = conv.filter(count=len(recipients))
    return conv



回答2:


For fast search using database index, I use this code:

class YandexWordstatQueue(models.Model):
    regions = models.ManyToManyField(YandexRegion)
    regions_cached = models.CharField(max_length=10000, editable=False, db_index=True)
    phrase = models.ForeignKey(SearchPhrase, db_index=True)
    tstamp = models.DateTimeField(auto_now_add=True)

class YandexWordstatRecord(models.Model):
    regions = models.ManyToManyField(YandexRegion)
    regions_cached = models.CharField(max_length=10000, editable=False, db_index=True)
    phrase = models.ForeignKey(SearchPhrase, db_index=True)
    Shows = models.IntegerField()
    date = models.DateField(auto_now_add=True)

@receiver(m2m_changed, sender=YandexWordstatRecord.regions.through)
@receiver(m2m_changed, sender=YandexWordstatQueue.regions.through)
def yandexwordstat_regions_changed(sender, **kwargs):
    if kwargs.get('action') in ['post_add', 'post_remove']:
        instance = kwargs.get('instance')
        l = list(instance.regions.values_list('RegionID', flat=True))
        l.sort()
        instance.regions_cached = json.dumps(l)
        instance.save()

This adds overhead when saving, but now I can perform fast filter with this snippet:

region_ids = [1, 2, 3] # or list(some_queryset.values_list(...))
region_ids.sort()
regions_cahed = json.dumps(region_ids)
YandexWordstatQueue.objects.filter(regions_cached=regions_cached)


来源:https://stackoverflow.com/questions/13586507/how-to-filter-django-model-by-its-objects-in-many-to-many-field-exact-match

易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!