我有一长串链接,我使用下面的代码,总投票,提交,通常的东西吐出但我不是100%如何确定当前登录的用户是否已在链接上投票或不。我知道如何在我的视图中执行此操作,但是我是否需要更改下面的视图代码,还是可以使用模板的工作方式来确定它?
我已阅读Django Vote Up/Down method但我不太明白发生了什么(并且不需要任何新闻)。
模型(摘录):
class Link(models.Model):
category = models.ForeignKey(Category, blank=False, default=1)
user = models.ForeignKey(User)
created = models.DateTimeField(auto_now_add=True)
modified = models.DateTimeField(auto_now=True)
url = models.URLField(max_length=1024, unique=True, verify_exists=True)
name = models.CharField(max_length=512)
def __unicode__(self):
return u'%s (%s)' % (self.name, self.url)
class Vote(models.Model):
link = models.ForeignKey(Link)
user = models.ForeignKey(User)
created = models.DateTimeField(auto_now_add=True)
def __unicode__(self):
return u'%s vote for %s' % (self.user, self.link)
观点(摘录):
def hot(request):
links = Link.objects.select_related().annotate(votes=Count('vote')).order_by('-created')
for link in links:
delta_in_hours = (int(datetime.now().strftime("%s")) - int(link.created.strftime("%s"))) / 3600
link.popularity = ((link.votes - 1) / (delta_in_hours + 2)**1.5)
if request.user.is_authenticated():
try:
link.voted = Vote.objects.get(link=link, user=request.user)
except Vote.DoesNotExist:
link.voted = None
links = sorted(links, key=lambda x: x.popularity, reverse=True)
links = paginate(request, links, 15)
return direct_to_template(
request,
template = 'links/link_list.html',
extra_context = {
'links': links,
})
上述观点实际上实现了我的需要,但我认为这是一种非常低效的方式。这导致了可怕的n + 1个查询,因为它代表了对包含29个链接的页面的33个查询,而最初我只有4个查询。我真的更喜欢使用Django的ORM或至少.extra()。
有什么建议吗?
修改
@Gabriel Hurley
我正在尝试重新创建你的答案,我的结果好坏参半,让我告诉你我得到了什么。
views.py
links = Link.objects.select_related().extra(
select={
'votes': 'COUNT(links_vote.id)',
'voted': 'SELECT COUNT(links_vote.id) FROM links_vote WHERE links_vote.user_id = 1 AND links_vote.link_id = links_link.id',
},
tables = ['links_vote']
)
models.py
class Vote(models.Model):
link = models.ForeignKey(Link)
user = models.ForeignKey(User)
created = models.DateTimeField(auto_now_add=True)
class Meta:
unique_together = ('link', 'user')
def __unicode__(self):
return u'%s vote for %s' % (self.user, self.link)
但是它返回了一个错误:
subquery uses ungrouped column "links_link.id" from outer query
第1行:... E links_vote.user_id = 1 AND links_vote.link_id = links_link ...
生成的查询看起来像(完全)这样:
SELECT (SELECT COUNT(links_vote.id) FROM links_vote WHERE links_vote.user_id = 1 AND links_vote.link_id = links_link.id) AS "voted", "links_link"."id", "links_link"."category_id", "links_link"."user_id", "links_link"."created", "links_link"."modified", "links_link"."url", "links_link"."name", "links_category"."id", "links_category"."name", "auth_user"."id", "auth_user"."username", "auth_user"."first_name", "auth_user"."last_name", "auth_user"."email", "auth_user"."password", "auth_user"."is_staff", "auth_user"."is_active", "auth_user"."is_superuser", "auth_user"."last_login", "auth_user"."date_joined" FROM "links_link" INNER JOIN "links_category" ON ("links_link"."category_id" = "links_category"."id") INNER JOIN "auth_user" ON ("links_link"."user_id" = "auth_user"."id") , "links_vote"
我正在使用PostgreSQL,我知道喜欢GROUP BY,但我不是100%就如何纠正这个问题。
编辑2(重大进展)
links = Link.objects.select_related()。annotate(votes = Count('vote'))。extra( 选择= { #'voteed':'SELECT COUNT()FROM links_vote WHERE links_vote.user_id =%s AND links_vote.link_id = links_link.id'%(request.user.id), #'voteed':''%(request.user.id), #'voteed':'SELECT CASE WHEN links_vote.user_id =%s THEN 1 ELSE 0 END'%(request.user.id), #'voteed':'SELECT COUNT()FROM links_vote WHERE links_vote.link_id = links_link.id AND links_vote.user_id =%s'%(request.user.id), }, where = ['links_link.id = links_vote.link_id'], ).order_by( ' - 创建')
*这仅适用于从此处(http://code.djangoproject.com/ticket/11916)
应用错误修补程序后我非常接近于找到我需要确定用户是否投票的最后一篇文章......
答案 0 :(得分:2)
之前我已经解决了这个问题,并且extra
或多或少地解决了这个问题:
# annotate whether you've already voted on this item
table = Vote._meta.db_table
select = 'SELECT COUNT(id) FROM %s' %table
where1 = 'WHERE ' + table + '.user_id = %s'
where2 = 'AND ' + table + '.item_id = appname_item.id'
items = items.extra(
select={'votes':" ".join((select, where1, where2,))},
select_params=(request.user.id,)
)
实际上,这将获取项目的QuerySet,并使用0或一些投票数注释每个项目。在我的系统中,我在投票上使用unique_together = ('link', 'user')
来确保每个用户只能投票一次,因此带注释的数据是0或1(实际上是布尔值)。它运作良好,避免了n + 1问题。