如何在views.py上优化Django SQL SELECT Query?

Posted

技术标签:

【中文标题】如何在views.py上优化Django SQL SELECT Query?【英文标题】:How to optimize Django SQL SELECT Query on views.py? 【发布时间】:2013-06-25 17:29:27 【问题描述】:

这些是视图中涉及的模型:

class Movie(models.Model):
    title_original = models.CharField(max_length=300,
                                      db_column='titulooriginal')
    title_translated = models.CharField(max_length=300, blank=True,
                                       db_column='titulotraducido')
    thumbnail = models.CharField(max_length=300, blank=True,
                                 db_column='imagen')
    length = models.CharField(max_length=75, db_column='duracion')
    rating = models.CharField(max_length=75, db_column='censura')
    genre = models.CharField(max_length=75, db_column='genero')
    country_of_origin = models.CharField(max_length=150, db_column='pais')
    trailer = models.CharField(max_length=300, blank=True, db_column='trailer')
    synopsis = models.CharField(max_length=3600, blank=True,
                                db_column='sinopsis')
    cast = models.CharField(max_length=300, blank=True, db_column='elenco')
    director = models.CharField(max_length=150, db_column='director')
    slug = models.SlugField(unique=True, blank=True)
    tsc = models.DateTimeField(auto_now=False, auto_now_add=True, db_column='tsc', null=True)
    tsm = models.DateTimeField(auto_now=True, auto_now_add=True, db_column='tsm', null=True)

    class Meta:
        db_table = u'peliculas'

    def __unicode__(self):
        return self.title_original


class ShowTime(models.Model):
    LANGUAGE_ESPANOL = 'Espanol'
    LANGUAGE_SUBTITLED = 'Subtitulada'
    LANGUAGE_CHOICES = (
        (LANGUAGE_ESPANOL, LANGUAGE_ESPANOL),
        (LANGUAGE_SUBTITLED, LANGUAGE_SUBTITLED))

    movie = models.ForeignKey(Movie, db_column='idpelicula')
    theater = models.ForeignKey(Theater, db_column='idcine',
                                      null=True)
    time = models.TimeField(null=True, db_column='hora')
    type_3d = models.NullBooleanField(db_column=u'3d')
    type_xd = models.NullBooleanField(null=True, db_column='xd')
    type_gtmax = models.NullBooleanField(null=True, db_column='gtmax')
    type_vip = models.NullBooleanField(null=True, db_column='vip')
    type_imax = models.NullBooleanField(null=True, db_column='imax')
    language = models.CharField(max_length=33, blank=True, db_column='idioma',
                                choices=LANGUAGE_CHOICES,
                                default=LANGUAGE_ESPANOL)
    city = models.ForeignKey(City, db_column='idciudad')
    start_date = models.DateField(blank=True, db_column='fecha_inicio')
    end_date = models.DateField(blank=True, db_column='fecha_fin')
    visible = models.NullBooleanField(null=True, db_column='visible')
    tsc = models.DateTimeField(auto_now=False, auto_now_add=True, db_column='tsc', null=True)
    tsm = models.DateTimeField(auto_now=True, auto_now_add=True, db_column='tsm', null=True)

    class Meta:
        db_table = u'funciones'

    def __unicode__(self):
        return (unicode(self.theater) + " " + unicode(self.movie.title_original) + " " +         unicode(self.time))

另外,这是查询模型的视图:

def list_movies(request, time_period):
    city = utils.get_city_from_request(request)

    if time_period == 'proximas-dos-horas':
        (start_date,
         end_date,
         hours_start,
         hours_end) = utils.get_datetime_filters_from_time_period(time_period)
    else:
        (start_date,
         end_date) = utils.get_datetime_filters_from_time_period(time_period)

    if not start_date:  # 404 for invalid time periods
        raise Http404

    movies = Movie.objects.filter(
        showtime__city=city.code,
        showtime__visible=1,
        ).order_by('-tsm').distinct()

    if time_period == 'proximas-dos-horas':
        movies = movies.filter(
            showtime__start_date__lte=start_date,
            showtime__end_date__gte=end_date,
            showtime__time__range=(hours_start, hours_end))

    elif time_period == 'proximos-estrenos':
        movies = movies.filter(
            showtime__start_date=None,
            showtime__end_date=None,
            )

    else:
        movies = movies.filter(
            showtime__start_date__lte=start_date,
            showtime__end_date__gte=end_date)

    context = 'movies': movies, 'city': city, 'time_period': time_period
    return render_to_response('list_movies.html', context,
                              context_instance=RequestContext(request))

目前此视图在数据库中消耗大量资源,这导致浏览器上 Web 事务的响应时间较短。我将应用程序与 New Relic 监控软件连接,以分析应用程序和数据库级别的事务。这是我得到的跟踪细节:

我想得到一些关于优化这个视图的建议,以便数据库资源的消耗降到最低

非常感谢!

【问题讨论】:

请注意,您可能想将showtime__start_date=None 更改为showtime__start_date__isnull=True showtime__start_date 或 showtime__end_date 是否有任何索引。 new relic 是否允许您在查询中执行和解释计划。如果是这样,请发布它,因为它肯定有助于诊断问题 【参考方案1】:

我不知道您的问题的背景,但我的 2 条建议是:

首先,考虑您的数据模型中的一些变化。 ShowTime 类有一个指向City 的外键和一个指向Theater 的外键,看起来有点多余。就个人而言,我更喜欢去规范化地址,例如:

class Theater(models.Model):
    name = models.CharField(max_length=48)
    # Location fields.
    geo_country = models.CharField(max_length=48)
    geo_country_code = models.CharField(max_length=2)
    geo_locality = models.CharField(max_length=48)
    geo_street = models.CharField(max_length=48, blank=True, default="")
    geo_address = models.CharField(max_length=120, blank=True, default="")


class ShowTime(models.Model):
    theater = models.ForeignKey(Theater) 

这样,您可以将 JOIN 保存到 City 表中,并且可以从您的数据模型中删除。

其次,阅读Django ORM的select_related feature。

祝你玩得开心。

【讨论】:

以上是关于如何在views.py上优化Django SQL SELECT Query?的主要内容,如果未能解决你的问题,请参考以下文章

Django views.py 版本的 SQL Join 与多表查询

django框架学习六:优化views.py文件,使用rest_framework中的APIVew和Response返回

如何通过 Ajax jQuery 将简单数据发送到 Django 中的views.py?

Django 过滤和查询:在views.py、模板或过滤器中进行?

如何在 Views.py 中运行子进程——Django

如何在 Django 中重复调用 views.py 中的函数?