Request logging reads chunked bodies fully into memory #213

simplehttpserver version:

dev

Current Behavior:

The request logging layer decides whether to dump the body from Content-Length. A chunked request has no Content-Length, so its whole body is read into memory for the log, with no size limit. This happens before the request reaches the handler (also before basic auth), and also when no log output needs the body.

Expected Behavior:

The body kept for logging has a size limit (-max-dump-body-size, with a default when it is not set), and the body is read only when a log needs it.

Steps To Reproduce:

1. Run simplehttpserver -v.
2. Send a large request with Transfer-Encoding: chunked.

Anything else: