Parsing POST request

[fsitedev]

As a continuation of the issue https://github.com/tarantool/nginx_upstream_module/issues/76 I can suggest to parse the POST body and transfer resulting arguments into lua procedure. It seems quite logical for modern web application to have such common things like request headers, GET-POST params and etc. not only in raw form but in commonly-used parsed variant. As an Idea I can suggest to alter some options of "tnt_pass_http_request" directive:

• Rename "pass_body" to "pass_raw_body" only to show an explicit behaviour when no parsing applied in this case.
• Make "parse_body" (as analogue of parse_args). This option could enable automatic encoding resolvation. There are 3 common html-form encodings for now:
  • x-www-urlencoded (default if skipped)
  • multipart/form-data (no encoding, split by boundaries)
  • text-plain (spaces replaced with "+")

So as not to die from so much work, we can start from only one encoding and add others one by one. I'm sure this feature will make "tnt/nginx_upstream" much more convinient for end-level lua applications, which still suffer from heavy text-parsing tasks and will let them focus on it's logic only

[dedok]

x-www-urlencoded (default if skipped) in 1f057360b6a2425e31bd15fe890397e49687d748

[dedok]

You can found details in readme, here is a link https://github.com/tarantool/nginx_upstream_module#tnt_pass_http_request, check the parse_urlencoded

[snoopcatt]

Added pull request that allow access to raw POST body even with parse_urlencoded enabled. In my branch with parse_urlencoded there are new request variable — args_urlencoded, and you still have access to raw body value.

[dedok]

Yeah, I guess we can close this one. Thank you so much!