Django：子查询的注释

我正在尝试注释一个查询集Station与id最近邻的Station使用 Django 2.0.3 和 PostGIS (GeoDjango) 功能。

简化版Station model:

class Station(models.Model):
    name = models.CharField(max_length=128)
    location = models.PointField()
    objects = StationQuerySet.as_manager()

我遇到的问题是尝试计算最近距离，这涉及注释一个引用子查询location在外部查询集中。

from django.db.models import OuterRef, Subquery
from django.contrib.gis.db.models.functions import Distance

class StationQuerySet(models.QuerySet):

    def add_nearest_neighbour(self):
        '''
        Annotates each station with the id and distance of the nearest neighbouring station
        '''
        # Get Station model
        Station = self.model

        # Calculate distances to each station in subquery
        subquery_with_distance = Station.objects.annotate(distance=Distance('location', OuterRef('location')) / 1000)

        # Get nearest from subquery
        nearest = subquery_with_distance.order_by('distance').values('id')[0]

        return self.annotate(
            nearest_station_id=Subquery(nearest)
        )

线路distance = Station.objects.annotate(distance=Distance('location', OuterRef('location')) / 1000)结果出现如下错误：

from apps.bikeshare.models import Station
stations = Station.objects.add_nearest_neighbour()

Error:

Traceback (most recent call last):
  File "/home/gbrown/Envs/bikeshare-dev/lib/python3.5/site-packages/IPython/core/interactiveshell.py", line 2847, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-3-cb35ea6d5d8b>", line 1, in <module>
    stations = Station.objects.add_nearest_neighbour()
  File "/home/gbrown/Envs/bikeshare-dev/lib/python3.5/site-packages/django/db/models/manager.py", line 82, in manager_method
    return getattr(self.get_queryset(), name)(*args, **kwargs)
  File "/home/gbrown/Development/transit_bikeshare/apps/bikeshare/querysets.py", line 162, in add_nearest_neighbour
    subquery_with_distance = Station.objects.annotate(distance=Distance('location', OuterRef('location')) / 1000)
  File "/home/gbrown/Envs/bikeshare-dev/lib/python3.5/site-packages/django/db/models/manager.py", line 82, in manager_method
    return getattr(self.get_queryset(), name)(*args, **kwargs)
  File "/home/gbrown/Envs/bikeshare-dev/lib/python3.5/site-packages/django/db/models/query.py", line 997, in annotate
    clone.query.add_annotation(annotation, alias, is_summary=False)
  File "/home/gbrown/Envs/bikeshare-dev/lib/python3.5/site-packages/django/db/models/sql/query.py", line 975, in add_annotation
    summarize=is_summary)
  File "/home/gbrown/Envs/bikeshare-dev/lib/python3.5/site-packages/django/db/models/expressions.py", line 452, in resolve_expression
    c.lhs = c.lhs.resolve_expression(query, allow_joins, reuse, summarize, for_save)
  File "/home/gbrown/Envs/bikeshare-dev/lib/python3.5/site-packages/django/contrib/gis/db/models/functions.py", line 58, in resolve_expression
    source_fields = res.get_source_fields()
  File "/home/gbrown/Envs/bikeshare-dev/lib/python3.5/site-packages/django/db/models/expressions.py", line 349, in get_source_fields
    return [e._output_field_or_none for e in self.get_source_expressions()]
  File "/home/gbrown/Envs/bikeshare-dev/lib/python3.5/site-packages/django/db/models/expressions.py", line 349, in <listcomp>
    return [e._output_field_or_none for e in self.get_source_expressions()]
AttributeError: 'ResolvedOuterRef' object has no attribute '_output_field_or_none'

想出了一个解决办法，使用原始查询来查找最近的车站并从子查询中选择 id 和距离，奖金解释如下：

class StationQuerySet(models.QuerySet):

    def nearest_neighbour(self):
        '''
        Creates a RawQuerySet of each station with the id and distance of the nearest neighbouring station
        '''
        # Have to execute the query in order to get the list of ids to inject
        ids = tuple(self.values('id').values_list('id', flat=True))

        return self.raw('''
               SELECT
                 A0.id   as id,
                 SUB.closest_id,
                 SUB.closest_distance
               FROM "bikeshare_station" A0
                 CROSS JOIN LATERAL (
                            SELECT
                              B0.id   as closest_id,
                              st_distance_sphere(A0.location, B0.location) as closest_distance
                            FROM "bikeshare_station" B0
                            WHERE A0.id != B0.id
                            ORDER BY A0.location <-> B0.location
                            limit 1
                            ) SUB
               WHERE A0.id IN %s;
           ''', [ids])

Usage

您可以将查询集调用链接在一起以在查找最近邻居之前过滤查询集：

query = Station.objects.filter(name='Albert Gate, Hyde Park')
closest_stations = query.nearest_neighbour()
station = closest_stations[0]
station.name
[out]: 'Albert Gate, Hyde Park'
station.closest_distance
[out]: 133.52459069
station.closest_id
[out]: 6369

SQL解释

这种类型的子查询称为相关子查询，因为它引用外部查询中的列。另外，我需要选择最近车站的多条信息（id, distance, etc.).

子查询被放置在FROM子句，允许选择多列。 ALATERAL需要 join 来允许子查询引用同级表FROM列表。当子查询返回单行时，CROSS可以应用联接来形成基于笛卡尔积而不是共享列的联接表。

子查询使用PostGIS<->运算符，根据车站之间的距离对桌子进行排序的效率要高得多，并且st_distance_sphere，进行点之间的精确距离计算。