Django CSV 导入速度变慢

Question

我正在将一个 csv 城市列表导入我的 Django 应用程序。我是 Django 和 Python 的新手，导入运行相当快，前 25,000 行大约需要 5 分钟，接下来的 25,000 行需要 2 小时。我停止导入并从停止的地方重新开始，接下来的 25,000 次大约需要 4 分钟。显然我做错了，因为每次插入它都会减慢速度。

任何帮助都会很棒，我这样做主要是为了学习而不只是导入数据，目前直接导入 postgresql 更快，这样我就可以继续我的项目，但我想知道我做错了什么，所以我可以通过 Django / Python.

变得更好

蒂亚

from myapp import Country, State, City

def add_country(isocode, name):
    c = Country.objects.get_or_create(name=name.strip().replace('"', ''), isocode=isocode.strip())[0]
    return c


def add_state(country, isocode, name, statetype):
    country_model = Country.objects.get(isocode=country.strip().lower())
    s = State.objects.get_or_create(name=name.strip().replace('"', ''), isocode=isocode.strip().lower().replace('"', ''), country=country_model, statetype=statetype.strip().replace('"', ''))[0]
    return s


def add_city(country, state, name):
    country_model = Country.objects.get(isocode=country.strip().lower().replace('"', ''))
    try:
        state_model = State.objects.get(name=state.strip().replace('"', ''), country=country_model)
    except State.DoesNotExist:
        state_model = None
    ci = City.objects.get_or_create(name=name.strip().replace('"', ''), state=state_model, postcode='')[0]
    return ci


with open('country.csv', 'rb') as csvfile:
    myreader = csv.reader(csvfile, delimiter=',', quotechar='"')
    for counrow in myreader:
        add_country(counrow[0], counrow[1])


with open('state.csv', 'rb') as csvfile:
    myreader = csv.reader(csvfile, delimiter=',', quotechar='"')
    for counrow in myreader:
        add_state(counrow[0], counrow[1], counrow[2], counrow[3])


with open('city1.csv', 'rb') as csvfile:
    myreader = csv.reader(csvfile, delimiter=',', quotechar='"')
    for counrow in myreader:
        add_city(counrow[0], counrow[1], counrow[2])

with open('city2.csv', 'rb') as csvfile:
    myreader = csv.reader(csvfile, delimiter=',', quotechar='"')
    for counrow in myreader:
        add_city(counrow[0], counrow[1], counrow[2])

with open('city3.csv', 'rb') as csvfile:
    myreader = csv.reader(csvfile, delimiter=',', quotechar='"')
    for counrow in myreader:
        add_city(counrow[0], counrow[1], counrow[2])

更新：
所以我更改了代码以使用批量插入，第一组城市现在只有两分钟多一点，第二组是 10 分钟，几个小时后我还没有完成第三组。一定有某种垃圾收集过程或我遗漏的东西，因为我什至已经切换了文件，每个文件在运行第一时花费相同的时间。

新代码如下所示：

def add_country(isocode, name, created_by, changed_by, country_list):
    country_list.append(Country(name=name.strip().replace('"', ''), isocode=isocode.strip()))

def add_state(country, isocode, name, statetype, created_by, changed_by, state_list):
    country_model = Country.objects.get(isocode=country.strip().lower())
    state_list.append(State(name=name.strip().replace('"', ''), isocode=isocode.strip().lower().replace('"', ''), country=country_model, statetype=statetype.strip().replace('"', '')))

def add_city(country, state, name, created_by, changed_by, city_list):
    country_model = Country.objects.get(isocode=country.strip().lower().replace('"', ''))
    try:
        state_model = State.objects.get(name=state.strip().replace('"', ''), country=country_model)
    except State.DoesNotExist:
        state_model = None
    city_list.append(City(name=name.strip().replace('"', ''), state=state_model, postcode=''))

    country_list = []
    state_list = []
    city_list = []

    print "Countries"
    print time.strftime("%H:%M:%S")
    with open('country.csv', 'rb') as csvfile:
        myreader = csv.reader(csvfile, delimiter=',', quotechar='"')
        for counrow in myreader:
            add_country(counrow[0], counrow[1], adminuser, adminuser, country_list)

    Country.objects.bulk_create(country_list)

    print "States"
    print time.strftime("%H:%M:%S")
    with open('state.csv', 'rb') as csvfile:
        myreader = csv.reader(csvfile, delimiter=',', quotechar='"')
        for counrow in myreader:
            add_state(counrow[0], counrow[1], counrow[2], counrow[3], adminuser, adminuser, state_list)

    State.objects.bulk_create(state_list)

    print "Cities 1"
    print time.strftime("%H:%M:%S")
    with open('city1.csv', 'rb') as csvfile:
        myreader = csv.reader(csvfile, delimiter=',', quotechar='"')
        for counrow in myreader:
            add_city(counrow[0], counrow[1], counrow[2], adminuser, adminuser, city_list)

    City.objects.bulk_create(city_list)

    print "Cities 2"
    print time.strftime("%H:%M:%S")
    city_list = []
    with open('city2.csv', 'rb') as csvfile:
        myreader = csv.reader(csvfile, delimiter=',', quotechar='"')
        for counrow in myreader:
            add_city(counrow[0], counrow[1], counrow[2], adminuser, adminuser, city_list)

    City.objects.bulk_create(city_list)

    print "Cities 3"
    print time.strftime("%H:%M:%S")
    city_list = []
    with open('city3.csv', 'rb') as csvfile:
        myreader = csv.reader(csvfile, delimiter=',', quotechar='"')
        for counrow in myreader:
            add_city(counrow[0], counrow[1], counrow[2], adminuser, adminuser, city_list)

    City.objects.bulk_create(city_list)

Answer 1

您应该更新您的代码以使用批量操作 - bulk_create 在更新过程的最后。

Django CSV 导入速度变慢

Django CSV Import Slows Down

python

csv

django

postgresql

postgresql-copy