简体   繁体   English

Python,App Engine:HTTP多部分POST到Vk.Ads API

[英]Python, App Engine: HTTP multipart POST to Vk.Ads API

I tried to sole this problem myself for a few days searching on examples and documentations, also it wasn't solved on ruSO . 我花了几天的时间尝试寻找示例和文档来解决这个问题, 但是在ruSO上并没有解决 So, I hope that solution will be found here, on enSO. 因此,我希望能在enSO上找到解决方案。

I develop a service for automatic creation of ads at Vk social network using Python and Google App Engine. 我开发了一项服务,可使用Python和Google App Engine在Vk社交网络上自动创建广告。 Initially, pictures for ads are loaded to my server ( part 1 ), then they are uploaded to Vk server at some time ( parts 2.1 and 2.2 ). 最初,广告图片会加载到我的服务器( 第1部分 ),然后有时将它们上传到Vk服务器(第2.1和2.2部分 )。 It seems that pictures are loaded and stored on my server correctly (I downloaded them and compared with original ones — every byte is the same). 似乎图片已正确加载并存储在我的服务器上(我下载了图片并与原始图片进行了比较-每个字节都相同)。 But I attach the part 1 code just in case. 但是我附上第1部分的代码以防万一。

To upload a picture to Vk.Ads firstly I need to get a URL — this is simple, so skip it. 要将图片上传到Vk.Ads,首先我需要获取一个URL ,这很简单,因此请跳过它。 Secondly, I need to send a POST request to this link with field file with binary content of the photo ( API documentation ). 其次,我需要使用带有照片二进制内容的字段file向该链接发送POST请求( API文档 )。 I created two ways for that ( 2.1 and 2.2 ), but both of them returns errcode: 2 which means corrupted file . 我为此创建了两种方法( 2.1和2.2 ),但是它们都返回errcode: 2表示corrupted file To my mind, the problem is about the requests, but I don't exclude the possibility of it files uploading/storage on my server, or some strange work of the API. 在我看来,问题在于请求,但我不排除文件在服务器上上载/存储或API的某些奇怪工作的可能性。 I'll appreciate any answers and comments. 我将不胜感激任何答案和评论。

1. Uploading to my server 1.上传到我的服务器

import webapp2
from google.appengine.ext import ndb

# stores pictures on the server
class Photo(ndb.Model):
    name = ndb.StringProperty()
    img = ndb.BlobProperty()

    @staticmethod
    def get(name):
        retval = Photo.query(Photo.name == name).get()
        return retval

    @staticmethod
    def create(name, blob):
        retval = Photo()
        retval.name = name
        retval.img = blob
        return retval

class PhotosPage(webapp2.RequestHandler):
    def get(self):
        # general content of the page:
        html = '''<form action="/photos" method="post" enctype="multipart/form-data">
            <input type="file" name="flimg"/>
            <input value="new_pic" name="flname"/>
            <input type="submit" value="Upload"/> </form>'''

    def post(self):
        n = str(self.request.get('flname'))
        f = self.request.get('flimg')
        p = Photo.get(n)
        if p:
            p.img = f
        else:
            p = Photo.create(n, f)
        p.put()

2.1. 2.1。 POST to API, approach #1, using urlfetch и poster : 使用urlfetchposter将方法发布到API,方法#1:

from poster.encode import multipart_encode, MultipartParam
from google.appengine.api import urlfetch

name = 'file'
content = ... # file binary content
where = ... # gotten URL

options = {
    'file': MultipartParam(
        name=name,
        value=content,
        filename=name,
        filetype='image/png',
        filesize=len(content))
}

data, headers = multipart_encode(options)
pocket = "".join(data)

result = urlfetch.fetch(
    url=where,
    payload=pocket,
    method=urlfetch.POST,
    headers=headers)

2.2. 2.2。 POST to API, approach #2, using requests : 使用requests POST到API,方法2:

import requests

name = 'file'
content = ... # file binary content
where = ... # gotten URL

# I also tried without this dict; is it necessary?
data = {
    'fileName': name,
    'fileSize': len(content),
    'description': 'undefined',
}

result = requests.post(where, files={name: StringIO(content)}, data=data)

In addition, for the second approach I extracted the content of my request: 另外,对于第二种方法,我提取了请求的内容:

POST
https://pu.vk.com/c.../upload.php?act=ads_add&mid=...&size=m&rdsn=1&hash_time=...&hash=...&rhash=...&api=1

Content-Length: 15946
Content-Type: multipart/form-data; boundary=b4b260eace4e4a7082a99753b74cf51f

--b4b260eace4e4a7082a99753b74cf51f
Content-Disposition: form-data; name="description"
undefined

--b4b260eace4e4a7082a99753b74cf51f
Content-Disposition: form-data; name="fileSize"
15518

--b4b260eace4e4a7082a99753b74cf51f
Content-Disposition: form-data; name="fileName"
file

--b4b260eace4e4a7082a99753b74cf51f
Content-Disposition: form-data; name="file"; filename="file" 
< File binary content >

--b4b260eace4e4a7082a99753b74cf51f-- 

UPDATE. 更新。

Thanks to SwiftStudier , I found the origin of the problem: StringIO and BytesIO don't behave identical to file open . 感谢SwiftStudier ,我找到了问题的根源: StringIOBytesIO行为与file open不一致。 If I use just open the code works well, but it doesn't with virtual file. 如果我只使用open代码,效果很好,但不能用于虚拟文件。 How it can be solved? 如何解决?

import requests
from io import BytesIO

with open('path.to/file.png', 'rb') as fin:
    content = BytesIO(fin.read())

token = '...'
url = 'https://api.vk.com/method/ads.getUploadURL?access_token=' + token + '&ad_format=2'
upload_url = requests.get(url).json()['response']

post_fields = {
    'access_token': token
}

data_fields = {
    # This works:
    # 'file': open('path.to/file.png', 'rb')

    # But this does not:
    'file': content
}

response = requests.post(upload_url, data=post_fields, files=data_fields)
print(response.text)

Not sure if it can help, but I'll post it anyway 不确定是否可以提供帮助,但是我还是会贴出来

I used requests to upload an image to ads 我使用requests将图片上传到ads

import requests

token = '***'
url = f'https://api.vk.com/method/ads.getUploadURL?access_token={token}&ad_format=1' # I set add_format randomly just to avoid an error of this parameter was missing
upload_url = requests.get(url).json()['response']

post_fields = {
    'access_token': token
}

data_fields = {
    'file': open('/path/to/image.png', 'rb')
}

response = requests.post(upload_url, data=post_fields, files=data_fields)
print(response.text)

The result looks like valid photo upload, the data received can be used in further actions with ad API. 结果看起来像是有效的照片上传,接收到的数据可用于广告API的进一步操作。

After a lot of experiments and investigating of different HTTP requests content I found out the only difference between wrong and working code. 经过大量实验并研究了不同的HTTP请求内容,我发现了错误代码与有效代码之间的唯一区别。 It was about 4 bytes only: file name MUST contain the extension. 它只有大约4个字节:文件名必须包含扩展名。 Vk API even ignores Content-Type: image/png , but needs .png or similar in filename. Vk API甚至会忽略Content-Type: image/png ,但需要.png或类似的文件名。 So, this doesn't work: 因此,这不起作用:

requests.post(upload_url, files={
    'file': BytesIO('<binary file content>')
})

But this option works properly: 但是此选项正常工作:

requests.post(upload_url, files={
    'file': ('file.png', BytesIO('<binary file content>'), 'image/png')
})

Just like this one, which is not available for GAE: 就像这样,GAE不可用:

requests.post(upload_url, files={
    'file': open('/path/to/image.png', 'rb')
})

Both StringIO and StringIO are appropriate for that task. StringIOStringIO均适合该任务。 As mentioned, Content-Type does not matter, it can be just multipart/form-data . 如前所述, Content-Type无关紧要,它可以只是multipart/form-data

声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM