在使用ast.literal_eval时出现了格式错误的字符串

Question

我在同一组数据上使用了json dump和json load。该数据为Unicode，因此我将其转换为字符串。使用ast.literal_eval，我尝试将该字符串转换为字典类型，但是出现了“格式错误”的错误。

下面是json load的输出：

('data', u'{\n  "a": "spawning", \n  "addresses": "", \n  "image": "b", \n  "OS-EXT-STS:vm_state": "building", \n  "c:launched_at": null, \n  "d": "e (fgh)", \n  "user_id": "hhh", \n 
    "accessIPv4": "", \n  "accessIPv6": "", \n  "name": "kk", \n  "created": "2017-12-08T07:52:44Z", \n  "z:xyz": []\n}', <type 'unicode'>)

我尝试了什么？

        with open('openstack_list.json', 'w') as e:
            json.dump(check_output(['openstack', 'server', 'show', i, '-f', 'json']), e)
        with open('openstack_list.json', 'r') as a:
            data = json.load(a)
            new_data = data.encode('utf-8')  # output type is unicode
            dict_data = ast.literal_eval(new_data) # output type is string

我希望输出结果是字典类型，但我没有得到它。此外，json加载会给出unicode数据，因此我认为new_data = data.encode('utf-8')是多余的。但如果我不使用编码就使用ast.literal_eval，则会出现格式错误。无论如何，我都无法将数据类型设置为字典。

编辑：

错误：

Traceback (most recent call last):
  File "openstack_resource_list.py", line 84, in <module>
    output = get_resources()
  File "openstack_resource_list.py", line 47, in get_resources
    dict_data = ast.literal_eval(new_data)
  File "/usr/lib64/python2.7/ast.py", line 80, in literal_eval
    return _convert(node_or_string)
  File "/usr/lib64/python2.7/ast.py", line 63, in _convert
    in zip(node.keys, node.values))
  File "/usr/lib64/python2.7/ast.py", line 62, in <genexpr>
    return dict((_convert(k), _convert(v)) for k, v
  File "/usr/lib64/python2.7/ast.py", line 79, in _convert
    raise ValueError('malformed string')
ValueError: malformed string

转换为json之前的数据：

"{\n  \"aaa\": null, \n  \"addresses\": \"inner-net=192.168.0.173, x.x.x.x\", \n  \"image\": \"aaa (aaa)\", 
    \n  \"aaa:vm_state\": \"active\", \n  \"aaa:launched_at\": \"2017-12-08T08:21:45.000000\", \n  \"flavor\": \"aaa4 (aaa)\", 
    \n  \"id\": \"aaa\", \n  \"security_groups\": [\n    {\n      \"name\": \"default\"\n    }\n  ], \n  \"user_id\": \"aaa\", 
    \n  \"OS-DCF:diskConfig\": \"MANUAL\", \n  \"accessIPv4\": \"\", \n  \"accessIPv6\": \"\", \n  \"progress\": 0, \n  \"Oaa:power_state\": 1, \n  \"project_id\": \"aaa\", 
    \n  \"config_drive\": \"\", \n  \"status\": \"ACTIVE\", \n  \"updated\": \"2017-12-08T08:21:45Z\", \n  \"hostId\": \"aaa\", \n  \"OS-SRV-USG:terminated_at\": null, 
    \n  \"key_name\": \"pg_ci\", \n  \"properties\": \"\", \n  \"OS-EXT-AZ:availability_zone\": \"nova\", \n  \"name\": \"taaa\", \n  \"created\": \"2017-12-08T08:21:31Z\", \n 
    \"os-extended-volumes:volumes_attached\": [\n    {\n      \"id\": \"aaa\"\n    }\n  ]\n}"

- Heenashree Khandelwal

---

请问您能展示openstack_list.json的实际内容吗？另外，您正在使用Python 2，对吗？ - poke

看起来你的数据被JSON化了两次...能否发布check_output()的代码或者至少是check_output()调用的原始返回值（在json.dump(...)之前）？ - bruno desthuilliers

你能发布错误堆栈跟踪吗？ - Akshay Tilekar

@Heena，同时请发布您正在使用的正确的JSON字符串。 - Akshay Tilekar

@brunodesthuilliers，请检查我的编辑。 - Heenashree Khandelwal

Answer 1

check_output 返回的数据已经是 JSON 格式，所以不需要再使用 json.dump 转换成 JSON。你可以直接将其写入文件中，这个文件就是一个有效的 JSON 文件。然后你可以使用 json.loads 将其加载为 Python 对象：

import json
from pprint import pprint

s = """{\n  \"aaa\": null, \n  \"addresses\": \"inner-net=192.168.0.173, x.x.x.x\", \n  \"image\": \"aaa (aaa)\", 
    \n  \"aaa:vm_state\": \"active\", \n  \"aaa:launched_at\": \"2017-12-08T08:21:45.000000\", \n  \"flavor\": \"aaa4 (aaa)\", 
    \n  \"id\": \"aaa\", \n  \"security_groups\": [\n    {\n      \"name\": \"default\"\n    }\n  ], \n  \"user_id\": \"aaa\", 
    \n  \"OS-DCF:diskConfig\": \"MANUAL\", \n  \"accessIPv4\": \"\", \n  \"accessIPv6\": \"\", \n  \"progress\": 0, \n  \"Oaa:power_state\": 1, \n  \"project_id\": \"aaa\", 
    \n  \"config_drive\": \"\", \n  \"status\": \"ACTIVE\", \n  \"updated\": \"2017-12-08T08:21:45Z\", \n  \"hostId\": \"aaa\", \n  \"OS-SRV-USG:terminated_at\": null, 
    \n  \"key_name\": \"pg_ci\", \n  \"properties\": \"\", \n  \"OS-EXT-AZ:availability_zone\": \"nova\", \n  \"name\": \"taaa\", \n  \"created\": \"2017-12-08T08:21:31Z\", \n 
    \"os-extended-volumes:volumes_attached\": [\n    {\n      \"id\": \"aaa\"\n    }\n  ]\n}"""

d = json.loads(s)
pprint(d)

输出

{'OS-DCF:diskConfig': 'MANUAL',
 'OS-EXT-AZ:availability_zone': 'nova',
 'OS-SRV-USG:terminated_at': None,
 'Oaa:power_state': 1,
 'aaa': None,
 'aaa:launched_at': '2017-12-08T08:21:45.000000',
 'aaa:vm_state': 'active',
 'accessIPv4': '',
 'accessIPv6': '',
 'addresses': 'inner-net=192.168.0.173, x.x.x.x',
 'config_drive': '',
 'created': '2017-12-08T08:21:31Z',
 'flavor': 'aaa4 (aaa)',
 'hostId': 'aaa',
 'id': 'aaa',
 'image': 'aaa (aaa)',
 'key_name': 'pg_ci',
 'name': 'taaa',
 'os-extended-volumes:volumes_attached': [{'id': 'aaa'}],
 'progress': 0,
 'project_id': 'aaa',
 'properties': '',
 'security_groups': [{'name': 'default'}],
 'status': 'ACTIVE',
 'updated': '2017-12-08T08:21:45Z',
 'user_id': 'aaa'}

如果你想将其转换为干净的JSON格式，只需将该Python对象传递给 json.dump 或json.dumps

print(json.dumps(d, indent=4))

输出

{
    "aaa": null,
    "addresses": "inner-net=192.168.0.173, x.x.x.x",
    "image": "aaa (aaa)",
    "aaa:vm_state": "active",
    "aaa:launched_at": "2017-12-08T08:21:45.000000",
    "flavor": "aaa4 (aaa)",
    "id": "aaa",
    "security_groups": [
        {
            "name": "default"
        }
    ],
    "user_id": "aaa",
    "OS-DCF:diskConfig": "MANUAL",
    "accessIPv4": "",
    "accessIPv6": "",
    "progress": 0,
    "Oaa:power_state": 1,
    "project_id": "aaa",
    "config_drive": "",
    "status": "ACTIVE",
    "updated": "2017-12-08T08:21:45Z",
    "hostId": "aaa",
    "OS-SRV-USG:terminated_at": null,
    "key_name": "pg_ci",
    "properties": "",
    "OS-EXT-AZ:availability_zone": "nova",
    "name": "taaa",
    "created": "2017-12-08T08:21:31Z",
    "os-extended-volumes:volumes_attached": [
        {
            "id": "aaa"
        }
    ]
}

在原始的JSON中，键按字母顺序排序。要在清理后的JSON中实现这一点，只需将sort_keys=True传递给json.dumps即可。

- PM 2Ring

---

完成了。学会了不要将数据重复转换为JSON格式... :) - Heenashree Khandelwal