Flask应用程序文件上传-获取文件内容时出错

Question

I am developing a flask application which uploads a file to IBM Bluemix Cloudant DB. I need to save the contents of the file as a key value pair in Cloudant. If I try to save a text file, it reads the content correctly. For other type of files it does not work. Following is my flask REST API CODE:

@app.route('/upload', methods=['POST'])
def upload_file():
    file_to_upload = request.files['file_upload'];
    response = CloudantDB().upload_file_to_db(file_to_upload);

//tHE FUNCTION upload_file under CloudantDB is as shown below.
file_name = file.filename;
uploaded_file_content = file.read();
data = {
            'file_name': file_name,
            'file_contents': uploaded_file_content,
            'version': version
            }
my_doc = self.database.create_document(data);

I know the error is because "uploaded_file_content" is in a different format (ie For PDFs, JPGs etc). Is there anyway I can overcome this?

Thanks!

Answer 1

The difference is that text files contain ordinary text whereas JPG, PNG etc. contain binary data.

Binary data should be uploaded as an attachment with a mime type, and you need to base64 encode the data. You don't show what create_document() is doing, but it's unlikely that it is able to treat binary data as an attachment. This might fix it for you:

from base64 import b64encode

uploaded_file_content = b64encode(file.read());
data = {
    'file_name': file_name,
    'version': version,

    '_attachments': {
        file_name : {
            'content-type': 'image/png',
            'data': uploaded_file_content
        }
    }
}

my_doc = self.database.create_document(data);

It should also be possible with your current code to simply base64 encode the file content and upload it. So that you know what type of data is stored should you later retrieve it, you will need to add another key value pair to store the mime type as content-type does above.

Attachments have advantages in that they can be individually addressed, read, deleted, updated without affecting the containing document, so you are probably better off using them.