我有一个基本的字典如下:

sample = {}
sample['title'] = "String"
sample['somedate'] = somedatetimehere

当我尝试做jsonify(sample)时,我得到:

TypeError: datetime.datetime(2012, 8, 8, 21, 46, 24, 862000) is not JSON serializable

我该怎么做才能使我的字典样本克服上面的错误呢?

注意:虽然它可能不相关,字典是从mongodb的记录检索中生成的,当我打印出str(sample['somedate'])时,输出是2012-08-08 21:46:24.862000。


当前回答

对于不需要或不想使用pymongo库的其他人。你可以用这个小片段轻松实现datetime JSON转换:

def default(obj):
    """Default JSON serializer."""
    import calendar, datetime

    if isinstance(obj, datetime.datetime):
        if obj.utcoffset() is not None:
            obj = obj - obj.utcoffset()
        millis = int(
            calendar.timegm(obj.timetuple()) * 1000 +
            obj.microsecond / 1000
        )
        return millis
    raise TypeError('Not sure how to serialize %s' % (obj,))

然后像这样使用它:

import datetime, json
print json.dumps(datetime.datetime.now(), default=default)

输出:

'1365091796124'

其他回答

我的解决方案(我认为不那么冗长):

def default(o):
    if type(o) is datetime.date or type(o) is datetime.datetime:
        return o.isoformat()

def jsondumps(o):
    return json.dumps(o, default=default)

然后使用jsondumps而不是json.dumps。它将打印:

>>> jsondumps({'today': datetime.date.today()})
'{"today": "2013-07-30"}'

如果你想,以后你可以添加其他特殊情况,通过一个简单的默认方法。例子:

def default(o):
    if type(o) is datetime.date or type(o) is datetime.datetime:
        return o.isoformat()
    if type(o) is decimal.Decimal:
        return float(o)

Another approach is to adopt a concept from FEEL (the Friendly Enough Expression Language) defined in DMN (Decision Model Notation) - namely @strings. Any string starting with @" and ending with " is decoded separately with FEEL decoding. Of course the sender and the receiver have to agree to this convention, but ... the code below lets you encode lots of other things as well as dates, times, date/times, timedeltas. You can encode year/month durations and ranges (so long as you except a 4 element tuple of chr, expr, expr, chr as being a good representation of a range - where the two chrs are open/close brackets). So, @"P4Y2M" is a duration of 4 years and 2 months. @"P2DT5H" is a timedelta of 2 days and 4 hours, @"(2021-01-02 .. 2021-12-31)" is a year range.

下面的代码可用于序列化和反序列化@strings。

import datetime
import pySFeel

parser = pySFeel.SFeelParser()


def convertAtString(thisString):
    # Convert an @string
    (status, newValue) = parser.sFeelParse(thisString[2:-1])
    if 'errors' in status:
        return thisString
    else:
        return newValue


def convertIn(newValue):
    if isinstance(newValue, dict):
        for key in newValue:
            if isinstance(newValue[key], int):
                newValue[key] = float(newValue[key])
            elif isinstance(newValue[key], str) and (newValue[key][0:2] == '@"') and (newValue[key][-1] == '"'):
                newValue[key] = convertAtString(newValue[key])
            elif isinstance(newValue[key], dict) or isinstance(newValue[key], list):
                newValue[key] = convertIn(newValue[key])
    elif isinstance(newValue, list):
        for i in range(len(newValue)):
            if isinstance(newValue[i], int):
                newValue[i] = float(newValue[i])
            elif isinstance(newValue[i], str) and (newValue[i][0:2] == '@"') and (newValue[i][-1] == '"'):
                newValue[i] = convertAtString(newValue[i])
            elif isinstance(newValue[i], dict) or isinstance(newValue[i], list):
                newValue[i] = convertIn(newValue[i])
    elif isinstance(newValue, str) and (newValue[0:2] == '@"') and (newValue[-1] == '"'):
        newValue = convertAtString(newValue)
    return newValue


  def convertOut(thisValue):
      if isinstance(thisValue, datetime.date):
          return '@"' + thisValue.isoformat() + '"'
      elif isinstance(thisValue, datetime.datetime):
          return '@"' + thisValue.isoformat(sep='T') + '"'
      elif isinstance(thisValue, datetime.time):
          return '@"' + thisValue.isoformat() + '"'
      elif isinstance(thisValue, datetime.timedelta):
          sign = ''
          duration = thisValue.total_seconds()
          if duration < 0:
              duration = -duration
              sign = '-'
          secs = duration % 60
          duration = int(duration / 60)
          mins = duration % 60
          duration = int(duration / 60)
          hours = duration % 24
          days = int(duration / 24)
          return '@"%sP%dDT%dH%dM%fS"' % (sign, days, hours, mins, secs)
      elif isinstance(thisValue, bool):
          return thisValue:
      elif thisValue is None:
          return thisValue:
      elif isinstance(thisValue, int):
          sign = ''
          if thisValue < 0:
              thisValue = -thisValue
              sign = '-'
          years = int(thisValue / 12)
          months = (thisValue % 12)
          return '@"%sP%dY%dM"' % (sign, years, months)
      elif isinstance(thisValue, tuple) and (len(thisValue) == 4):
          (lowEnd, lowVal, highVal, highEnd) = thisValue
          return '@"' + lowEnd + str(lowVal) + ' .. ' + str(highVal) + highEnd
      elif thisValue is None:
          return 'null'
      elif isinstance(thisValue, dict):
          for item in thisValue:
              thisValue[item] = convertOut(thisValue[item])
          return thisValue
      elif isinstance(thisValue, list):
          for i in range(len(thisValue)):
              thisValue[i] = convertOut(thisValue[i])
          return thisValue
      else:
          return thisValue

我的快速和肮脏的JSON转储吃日期和一切:

json.dumps(my_dictionary, indent=4, sort_keys=True, default=str)

Default是一个应用于不可序列化对象的函数。 在这种情况下,它是str,所以它只是把所有它不知道的东西转换成字符串。这对于序列化很好,但在反序列化时就不太好了(因此是“快速和肮脏”),因为任何东西都可能在没有警告的情况下被字符串化,例如函数或numpy数组。

通常有几种方法来序列化日期时间,比如:

ISO字符串,短,可以包含时区信息,例如@jgbarah的答案 时间戳(时区数据丢失),例如@JayTaylor的答案 属性字典(包括时区)。

如果您同意最后一种方法,json_tricks包将处理日期、时间和日期时间(包括时区)。

from datetime import datetime
from json_tricks import dumps
foo = {'title': 'String', 'datetime': datetime(2012, 8, 8, 21, 46, 24, 862000)}
dumps(foo)

这使:

{"title": "String", "datetime": {"__datetime__": null, "year": 2012, "month": 8, "day": 8, "hour": 21, "minute": 46, "second": 24, "microsecond": 862000}}

所以你要做的就是

`pip install json_tricks`

然后导入从json_tricks而不是json。

不将其存储为单个字符串、int型或float型的优势体现在解码时:如果你遇到的只是字符串,特别是int型或float型,你需要了解一些关于数据的信息,以知道它是否是一个datetime。作为dict,您可以存储元数据,以便自动解码,这就是json_tricks为您做的。它也很容易为人类编辑。

免责声明:这是我做的。因为我也有同样的问题。

2018年更新

最初的答案适应了MongoDB“date”字段的表示方式:

{" $日期":1506816000000}

如果你想要一个通用的Python解决方案来将datetime序列化为json,请查看@jjmontes的答案,这是一个不需要依赖的快速解决方案。


当你在使用mongoengine(每个注释)而pymongo是一个依赖时,pymongo有内置的实用工具来帮助json序列化: http://api.mongodb.org/python/1.10.1/api/bson/json_util.html

示例用法(序列化):

from bson import json_util
import json

json.dumps(anObject, default=json_util.default)

示例用法(反序列化):

json.loads(aJsonString, object_hook=json_util.object_hook)

Django

Django提供了一个原生的DjangoJSONEncoder序列化器来正确地处理这类问题。

看到https://docs.djangoproject.com/en/dev/topics/serialization/ djangojsonencoder

from django.core.serializers.json import DjangoJSONEncoder

return json.dumps(
  item,
  sort_keys=True,
  indent=1,
  cls=DjangoJSONEncoder
)

我注意到DjangoJSONEncoder和使用自定义默认之间的一个区别:

import datetime
import json

def default(o):
    if isinstance(o, (datetime.date, datetime.datetime)):
        return o.isoformat()

return json.dumps(
  item,
  sort_keys=True,
  indent=1,
  default=default
)

Django剥离了一些数据:

 "last_login": "2018-08-03T10:51:42.990", # DjangoJSONEncoder 
 "last_login": "2018-08-03T10:51:42.990239", # default

所以,在某些情况下,你可能需要小心。