how to POST multipart list of JSON/xml files using python requests

Question

In python2.7 I use requests to communicate with a REST endpoint. I can upload single JSON and xml objects to it. To speed things up I want to upload multiple json objects using multipart.

I have a curl command that show how it can be done and this works. I need to translate that in a python requests POST command.

WORKING curl POST:

curl --anyauth --user admin:admin -X POST --data-binary @sample-body 
     -i -H "Content-type: multipart/mixed; boundary=BOUNDARY" 
     "http://localhost:8058/v1/resources/sight-ingest?rs:transform=aireco-transform&rs:title=file1.xml&rs:title=file2.xml&rs:title=file3.xml"

Things to note: I need to send custom parameter list including a list of 'title' parameters, cannot do that by passing a dict? but we can potentially work around this.

my python trail:

import requests
files = {'file1': ('foo.txt', 'foo
contents
','text/plain'), 
          'file2': ('bar.txt', 'bar contents', 'text/plain'),
          'file3': ('baz.txt', 'baz contents', 'text/plain')}

headers = {'Content-Type': 'multipart/mixed','Content-Disposition': 'attachment','boundary': 'GRENS'}
params={'title':'file1','title':'file2','title':'file2'}
r = requests.Request('POST', 'http://example.com', files=files , headers=headers, params=params)
print r.prepare().url
print r.prepare().headers
print r.prepare().body

Gives me:

http://example.com/?title=file2
{'boundary': 'GRENS', 'Content-Type': 'multipart/mixed', 'Content-Length': '471', 'Content-Disposition': 'attachment'}
--7f18a6c1b09f42009228f600b0af35fd
Content-Disposition: form-data; name="file3"; filename="baz.txt"
Content-Type: text/plain

baz contents
--7f18a6c1b09f42009228f600b0af35fd
Content-Disposition: form-data; name="file2"; filename="bar.txt"
Content-Type: text/plain

bar contents
--7f18a6c1b09f42009228f600b0af35fd
Content-Disposition: form-data; name="file1"; filename="foo.txt"
Content-Type: text/plain

foo
contents

--7f18a6c1b09f42009228f600b0af35fd--

Questions:

• Seems the stuff in the header is not used in the body?
• Can i set my own boundary? 'GRENS' is not used in the body?
• Can I pass a list of title parameters (with same key) as in curl example?

Answer 1

Do not use a dictionary, use a list of (key, value) tuples for your query parameters:

params = [('title', 'file1'), ('title', 'file2'), ('title', 'file3')]

otherwise you'll end up with just the one key.

You should not set the Content-Type header; requests will set that correctly for you when you use the files parameter; that way the correct boundary will also be included. You should never set the boundary directly yourself, really:

params = [('title', 'file1'), ('title', 'file2'), ('title', 'file3')]
r = requests.post('http://example.com', 
                  files=files, headers=headers, params=params)

You can set headers per file part by adding a 4th element to the per-file tuple for extra headers, but in your case, you should not try to set the Content-Disposition header yourself; it'll be overwritten anyway.

Introspecting a prepared request object then gives you:

>>> import requests
>>> from pprint import pprint
>>> files = {'file1': ('foo.txt', 'foo
contents
','text/plain'), 
...           'file2': ('bar.txt', 'bar contents', 'text/plain'),
...           'file3': ('baz.txt', 'baz contents', 'text/plain')}
>>> headers = {'Content-Disposition': 'attachment'}
>>> params = [('title', 'file1'), ('title', 'file2'), ('title', 'file3')]
>>> r = requests.Request('POST', 'http://example.com',
...                      files=files, headers=headers, params=params)
>>> prepared = r.prepare()
>>> prepared.url
'http://example.com/?title=file1&title=file2&title=file3'
>>> pprint(dict(prepared.headers))
{'Content-Disposition': 'attachment',
 'Content-Length': '471',
 'Content-Type': 'multipart/form-data; boundary=7312ccd96db94419bf1d97f2c54bbad1'}
>>> print prepared.body
--7312ccd96db94419bf1d97f2c54bbad1
Content-Disposition: form-data; name="file3"; filename="baz.txt"
Content-Type: text/plain

baz contents
--7312ccd96db94419bf1d97f2c54bbad1
Content-Disposition: form-data; name="file2"; filename="bar.txt"
Content-Type: text/plain

bar contents
--7312ccd96db94419bf1d97f2c54bbad1
Content-Disposition: form-data; name="file1"; filename="foo.txt"
Content-Type: text/plain

foo
contents

--7312ccd96db94419bf1d97f2c54bbad1--

If you absolutely must have multipart/mixed and not multipart/form-data, you'll have to build the POST body yourself and set the headers from that. The included urllib3 tools should be able to do this for you:

from requests.packages.urllib3.fields import RequestField
from requests.packages.urllib3.filepost import encode_multipart_formdata

fields = []    
for name, (filename, contents, mimetype) in files.items():
    rf = RequestField(name=name, data=contents,
                      filename=filename)
    rf.make_multipart(content_disposition='attachment', content_type=mimetype)
    fields.append(rf)

post_body, content_type = encode_multipart_formdata(fields)
content_type = ''.join(('multipart/mixed',) + content_type.partition(';')[1:])

headers = {'Content-Type': content_type}
requests.post('http://example.com', data=post_body, headers=headers, params=params)

or you could use the email package to do the same:

from email.mime.multipart import MIMEMultipart
from email.mime.text import MIMEText

body = MIMEMultipart()
for name, (filename, contents, mimetype) in files.items():
    part = MIMEText(contents, _subtype=mimetype.partition('/')[-1], _charset='utf8')
    part.add_header('Content-Disposition', 'attachment', filename=filename)
    body.attach(part)

post_body = body.as_string().partition('

')[-1]
content_type = body['content-type']

headers = {'Content-Type': content_type}
requests.post('http://example.com', data=post_body, headers=headers, params=params)

but take into account that this method expects you to set a character set (I assumed UTF-8 for JSON and XML) and that it'll more than likely use Base64 encoding for the contents:

>>> body = MIMEMultipart()
>>> for name, (filename, contents, mimetype) in files.items():
...     part = MIMEText(contents, _subtype=mimetype.partition('/')[-1], _charset='utf8')
...     part.add_header('Content-Disposition', 'attachment', filename=filename)
...     body.attach(part)
... 
>>> post_body = body.as_string().partition('

')[-1]
>>> content_type = body['content-type']
>>> print post_body
--===============1364782689914852112==
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="baz.txt"

YmF6IGNvbnRlbnRz

--===============1364782689914852112==
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="bar.txt"

YmFyIGNvbnRlbnRz

--===============1364782689914852112==
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="foo.txt"

Zm9vCmNvbnRlbnRzCg==

--===============1364782689914852112==--

>>> print content_type
multipart/mixed; boundary="===============1364782689914852112=="