Python 3.4 -使用标准库的多部分帖子

Question

有没有人有一个例子，如何在Python 3.4中做一个多部分的帖子，而不使用第三方库，如请求?

我在将旧的Python 2代码移植到Python 3.4时遇到问题。

下面是python 2的编码代码:

def _encode_multipart_formdata(self, fields, files):
    boundary = mimetools.choose_boundary()
    buf = StringIO()
    for (key, value) in fields.iteritems():
        buf.write('--%srn' % boundary)
        buf.write('Content-Disposition: form-data; name="%s"' % key)
        buf.write('rnrn' + self._tostr(value) + 'rn')
    for (key, filepath, filename) in files:
        if os.path.isfile(filepath):
            buf.write('--%srn' % boundary)
            buf.write('Content-Disposition: form-data; name="%s"; filename="%s"rn' % (key, filename))
            buf.write('Content-Type: %srn' % (self._get_content_type3(filename)))
            file = open(filepath, "rb")
            try:
                buf.write('rn' + file.read() + 'rn')
            finally:
                file.close()
    buf.write('--' + boundary + '--rnrn')
    buf = buf.getvalue()
    content_type = 'multipart/form-data; boundary=%s' % boundary
    return content_type, buf

我发现我可以用以下代码替换mimetools.choose_boundary():

import email.generator
print (email.generator._make_boundary())

对于_get_content_type3()方法，我正在做以下操作:

def _get_content_type(self, filename):
        return mimetypes.guess_type(filename)[0] or 'application/octet-stream'

当我在使用Python3.4中将StringIO更改为BytesIO时，数据似乎从未被放入POST方法中。

有什么建议吗?

Answer 1

是的，email.generator._make_boundary()可以工作:

import email.generator
import io
import shutil
def _encode_multipart_formdata(self, fields, files):
    boundary = email.generator._make_boundary()
    buf = io.BytesIO()
    textwriter = io.TextIOWrapper(
        buf, 'utf8', newline='', write_through=True)
    for (key, value) in fields.items():
        textwriter.write(
            '--{boundary}rn'
            'Content-Disposition: form-data; name="{key}"rnrn'
            '{value}rn'.format(
                boundary=boundary, key=key, value=value))
    for (key, filepath, filename) in files:
        if os.path.isfile(filepath):
            textwriter.write(
                '--{boundary}rn'
                'Content-Disposition: form-data; name="{key}"; '
                'filename="{filename}"rn'
                'Content-Type: {content_type}rnrn'.format(
                    boundary=boundary, key=key, filename=filename,
                    content_type=self._get_content_type3(filename)))
            with open(filepath, "rb") as f:
                shutil.copyfileobj(f, buf)
            textwriter.write('rn')
    textwriter.write('--{}--rnrn'.format(boundary))
    content_type = 'multipart/form-data; boundary={}'.format(boundary)
    return content_type, buf.getvalue()

使用io.TextIOWrapper()对象来简化标头的格式化和编码(bytes对象不支持格式化操作;你必须等待Python 3.5加入%支持)。

如果你坚持使用email包的整个工作，考虑到你将需要两倍的内存;一次保存email.mime对象，另一次保存写入结果:

from email.mime import multipart, nonmultipart, text
from email.generator import BytesGenerator
from email import policy
from io import BytesIO
def _encode_multipart_formdata(self, fields, files):
    msg = multipart.MIMEMultipart('form-data')
    for (key, value) in fields.items():
        part = text.MIMEText(value)
        part['Content-Disposition'] = 'form-data; name="{}"'.format(key)
        msg.attach(part)
    for (key, filepath, filename) in files:
        if os.path.isfile(filepath):
            ct = self._get_content_type3(filename)
            part = nonmultipart.MIMENonMultipart(*ct.split('/'))
            part['Content-Disposition'] = (
                'form-data; name="{}"; filename="{}"'.format(
                    key, filename))
            with open(filepath, "rb") as f:
                part.set_payload(f.read())
            msg.attach(part)
    body = BytesIO()
    generator = BytesGenerator(
        body, mangle_from_=False, policy=policy.HTTP)
    generator.flatten(msg)
    return msg['content-type'], body.getvalue().partition(b'rnrn')[-1]

结果基本上是相同的，除了一些MIME-Version和Content-Transfer-Encoding头。