Question

考虑这个模型：

SoccerMatch(Model):
    """
    FIFA World Cup 2014 match. It could have no teams yet
    if it's scheduled but a shootout stage (they are never
    defined until both previous winner are determined)
    """
    #this is defined beforehand
    date = DateTimeField()
    #this is defined beforehand -for group matches- and when
    #the previous winners are determined -for shootout matches-
    team1 = ForeignKey(Team, related_name="+", null=True)
    team2 = ForeignKey(Team, related_name="+", null=True)
    #these are defined when the match ends
    goals1 = PositiveSmallIntegerField(null=True, blank=True)
    goals2 = PositiveSmallIntegerField(null=True, blank=True)

GroupSoccerMatch(SoccerMatch):
    """
    FIFA World Cup 2014 match for first stage.
    Each group has 4 teams and plays 6 matches.
    """
    group = ForeignKey(Group, related_name="matches")

ShootoutSoccerMatch(SoccerMatch):
    """
    FIFA World Cup 2014 match for second stage.
    Second stage plays 16 matches in rounds of
    8, 4, 2, 1 (3rd and 4th) and 1 (1st and 2nd).
    """

    #... more fields related to the shootout hierarchy
    #....those fields are omitted since they don't matter here

    #this is defined when the match ends
    penalty_shootout = PositiveSmallIntegerField(null=True, blank=True, choices=((1, "Team 1 won"), (2, "Team 2 won")))

在这段代码中（我省略了团队和团队定义，因为它涉及到有关数据和关于问题的微不足道的琐事）我模拟足球比赛世界杯（目前只适用于这个世界杯）和我想要对匹配结果做出预测。

考虑以前的模型有自己的clean()方法。

此应用程序可以配置为具有准确的预测（您必须猜测两个团队的目标和惩罚轮次 IF EXISTS 的获胜者）或不准确的预测（您必须猜测目标中的赢家数量）或者，如果他们绑定了和PENALTY ROUND EXISTS ，则该罚款轮次的获胜者。）

在这两种情况下，我都要考虑罚球，这是不可能的（前48场比赛，根本不存在）。

我的预测模型类是：

class Prediction(Model)
    user = ForeignKey(User, related_name="predictions")
    match = ForeignKey(SoccerMatch, related_name="predictions")
    goals1 = PositiveSmallIntegerField(null=True, blank=True)
    goals2 = PositiveSmallIntegerField(null=True, blank=True)
    penalty_shootout = PositiveSmallIntegerField(null=True, blank=True, choices=((1, "Team 1 won"), (2, "Team 2 won")))

考虑最简单的情况：精确预测 - 一个好的预测有目标1 = match.goals1，goals2 = match.goals2，penalty_shootout = match.penalty_shootout。

我以为我可以解决这样的准确预测：

#1 we consider predictions from a specific user
done_predictions = Prediction.objects.filter(user=anUser)

#2 we consider resolved predictions (i.e. involving matches played)
resolved_predictions = done_predictions.filter(match__date__lt=now() + timedelta(hours=1, minutes=30), match__goals1__isnull=False, match__goals2__isnull=False)

#3 we consider accurate predictions
accurate_predictions = resolved_predictions.filter(goals2=F('match__goals2'), goals1=F('match__goals1'))

现在一切都很好，但我有一个问题：我想考虑最近16场比赛的情况，其中我必须计算点球大战阶段的机会。

问：有没有办法可以修复第二个和第三个查询来考虑这种情况而不是（制作两个）单独的查询？（这意味着：无法查询match__penalty_shootout=，因为SoccerMatch不是那个字段）或者：有没有办法声明一个＆＃34;虚构字段＆＃34; penalty_shootout匹配的组的虚拟值为None？

Answer 1

该死的。找到了解决方法。

Django允许通过查询参数查找遍历派生属性（例如管理器，以及django 1.7中的自定义变换）。

对于我的问题，我的逻辑可以确定是否定义了penalty_shootout（1或2是绑定时射击匹配的唯一允许值）或者是无。

如果匹配不是Shootout而是组匹配，则该匹配记录将不存在Shootout片段。因此，当查询match__shootoutsoccermatch__penalty_shootout时，如果未指定值（即没有平局）或者当Shootout子类记录不存在时，我将得到None。因此，对于我的示例，查询match__shootoutsoccermatch__penalty_shootout__isnull=True和match__goals1=F('match__goals2')我指定了tie，并且由于clean()实现，None不能表示任何内容，但匹配不是Shootout。

Django - 异构查询

1 个答案: