how to parse HTTP POST(file upload) stream?

Question

I am using actionscript engine to upload a file, the engine will select the file and send the file over network thru HTTP POST command, the document says the POST message is like:

POST /handler.cfm HTTP/1.1
  Accept: text/*
  Content-Type: multipart/form-data; 
  boundary=----------Ij5ae0ae0KM7GI3KM7ei4cH2ei4gL6 
  User-Agent: Shockwave Flash 
  Host: www.example.com 
  Content-Length: 421 
  Connection: Keep-Alive 
  Cache-Control: no-cache

  ------------Ij5GI3GI3ei4GI3ei4KM7GI3KM7KM7
  Content-Disposition: form-data; name="Filename"

  MyFile.jpg
  ------------Ij5GI3GI3ei4GI3ei4KM7GI3KM7KM7
  Content-Disposition: form-data; name="photo"; filename="MyFile.jpg"
  Content-Type: application/octet-stream

  FileDataHere
  ------------Ij5GI3GI3ei4GI3ei4KM7GI3KM7KM7
  Content-Disposition: form-data; name="Upload"

  Submit Query

  ------------Ij5GI3GI3ei4GI3ei4KM7GI3KM7KM7--

In server side I have a C++ program listens on port 80 and parse the POST message. I only want the file name and file data. How to decode the file data using c++, is it base64 encoded and is there a library can do it for me? I want to decode the binary, and write it to the file, thanks!

Solution 1

No, there is no encoding. The body of each subpart of the multipart message is included as verbatim bytes. Consequently you have to be careful to choose a boundary string that doesn't occur anywhere in the file data.

To parse a multipart/form-data form submission you will enough of a MIME parser to parse the headers, pick out parameters you want like boundary and name, and split the received message by the boundary string. This is not completely trivial, so you may want to consider existing libraries.

(Unfortunately, what browsers actually do in HTTP differs a bit from the standard MIME rules outlined in RFC 1341. In particular, field name and filename parameters tend to include non-ASCII characters and unescaped quotes. But if you are generating the POST yourself hopefully you can steer clear of these areas of contention.)

Solution 2

In the absence of a "Content_Transfer_Encoding" header, it is assumed data is encoded in "7bit" (RFC 1521; RFC 1867; RFC 2616).

I don't know if tere's a C library to parse / decode HTTP POSTs. I'm pretty sure there are though :)

Comments

[pmg]

If you don't give a name to the submit button on the HTML, it won't get POSTed to the server (the "Upload" section) :)

[Arthur Kushman]

Oh, dude I really came here to see not C++, but particularly C lib for parsing boundary data and I'm also pretty sure there are some.