python urllib3 multipart/form-data 的坑

################################3.上传文件时带了个header与urllib3.request.py不兼容

传header的Content-Type时,应区分大小写

def request_encode_body(self, method, url, fields=None, headers=None,

encode_multipart=True, multipart_boundary=None,

**urlopen_kw):

...

extra_kw['body'] = body

extra_kw['headers'] = {'Content-Type': content_type}#这里header中必须是Content-Type,如果用户传来的header中已经有了'content-type', 则最终实际会出现2个这个重复值, 访问接口时会报错

extra_kw['headers'].update(headers)

...

#实际使用的header为:

#'content-type': 'multipart/form-data; boundary=----WebKitFormBoundary0b1047518e88296',

#'Content-Type': 'multipart/form-data; boundary=----WebKitFormBoundary0b1047518e88296'

################################2.上传文件后缀总是为null的问题解决

上传文件后缀总是为null的问题解决

注意这里的example.jpeg只能是文件名, 不能是带路径的, 若写成"\\image\\example.jpeg"上传的文件后缀就成null了.

fields={

'text': ('None', "upload a string",'None'),

'file': ('example.jpeg',file_data,"image/jpeg"),

}

##################################1

在一个项目中用到python urllib3上传文件

urllib3.disable_warnings()

http = urllib3.PoolManager()

with open('example.jpeg','rb') as fp:

file_data = fp.read()

r = http.request(

'POST',

'http://httpbin.org/post',

headers = {'Content-Type':"multipart/form-data"},

fields={

'text': ('None', "upload a string",'None'),

'file': ('example.jpeg',file_data,"image/jpeg"),

})

结果上传错误, 试着打印出urllib3组成的post body, 打开D:\python27\Lib\site-packages\urllib3\request.py的request_encode_body函数, 在return 前面加上print extra_kw, 打印出的body结果如下:

我使用postman发送该接口上传文件都是对的:

点击Send下面一行的红色 Code,HTTP显示的是:

发现这里的boundary与python中的urllib3不一样. 原来urllib3 是post form-data时, 没有传multipart_boundary该参数会自动生成:D:\python27\Lib\site-packages\urllib3\filepost.py

def encode_multipart_formdata(fields, boundary=None):

...

if boundary is None:

boundary = choose_boundary()

def choose_boundary():

"""

Our embarrassingly-simple replacement for mimetools.choose_boundary.

"""

return uuid4().hex#生成的是32为的唯一码

试着在python中修改为以WebKitFormBoundary开头的试着再执行, 还是报错, 用同样的方法, 查找到headers的Content-Type没有带上boundary,在header中明明带了Content-Type:multipart/form-data, 调试发现:header中传了Content-Type后, urllib3生成的带boundary的Content-Type会被传进来的覆盖掉,如下:

D:\python27\Lib\site-packages\urllib3\request.py

def request_encode_body(self, method, url, fields=None, headers=None,

encode_multipart=True, multipart_boundary=None,

**urlopen_kw):

if headers is None:

headers = self.headers

extra_kw = {'headers': {}}

if fields:

if 'body' in urlopen_kw:

raise TypeError(

"request got values for both 'fields' and 'body', can only specify one.")

if encode_multipart:

body, content_type = encode_multipart_formdata(fields, boundary=multipart_boundary)#产生了新的content-type

else:

body, content_type = urlencode(fields), 'application/x-www-form-urlencoded'

extra_kw['body'] = body

extra_kw['headers'] = {'Content-Type': content_type}#也将产生的新的content-type加到headers中了

extra_kw['headers'].update(headers)#一更新headers又还原成传进来的headers了

extra_kw.update(urlopen_kw)

return self.urlopen(method, url, **extra_kw)

总结可行方案:

1.http.request传的参数中带fields时,带上特定的boundary(以WebKitFormBoundary开头的, 原因未知, 知道的朋友可以留言告知),headers中不要带Content-Type;

2.http.request传的参数要带上headers的话,格式应该是:Content-Type: multipart/form-data; boundary=----WebKitFormBoundary7MA4YWxkTrZu0gW

最终的代码修改:

urllib3.disable_warnings()

http = urllib3.PoolManager()

with open('example.jpeg','rb') as fp:

file_data = fp.read()

boundary = '----WebKitFormBoundary7MA4YWxkTrZu0gW'

#方案1

r = http.request(

'POST',

'http://httpbin.org/post',

headers = {},

multipart_boundary = boundary,

fields={

'text': ('None', "upload a string",'None'),

'file': ('example.jpeg',file_data,"image/jpeg"),

})

#方案2

r = http.request(

'POST',

'http://httpbin.org/post',

headers = {'Content-Type':"multipart/form-data;%s"%boundary},

multipart_boundary = boundary,

fields={

'text': ('None', "upload a string",'None'),

'file': ('example.jpeg',file_data,"image/jpeg"),

})