* Add RequestID to the RequestScope
* Expose RequestID to auth & request loggers
* Use the RequestID in templated HTML pages
* Allow customizing the RequestID header
* Document new Request ID support
* Add more cases to scope/requestID tests
* Split Get vs Generate RequestID funtionality
* Add {{.RequestID}} to the request logger tests
* Move RequestID management to RequestScope
* Use HTML escape instead of sanitization for Request ID rendering
* Use a specialized ResponseWriter in middleware
* Track User & Upstream in RequestScope
* Wrap responses in our custom ResponseWriter
* Add tests for logging middleware
* Inject upstream metadata into request scope
* Use custom ResponseWriter only in logging middleware
* Assume RequestScope is never nil
Add the Prometheus http.Handler to serve metrics at MetricsPath ("/metrics"
by default). This allows Prometheus to scrape metrics from OAuth2 Proxy.
Add a new middleware NewRequestMetrics and attach it to the preAuth
chain. This will collect metrics on all requests made to OAuth2 Proxy
Collapse some calls to Prinf() and os.Exit(1) to Fatalf as they are
equivalent. main() has a strict 50 lines limit so brevity in these
calls appreciated
* Use comma separated value for multiple claims
* Fix lint error
* Fix more tests
* Fix one more test
* Always flatten the headers
* Ensure we test the real multi-groups
* Only update map when necessary
* Update CHANGELOG
* Move to the right location of change log
* Fix blank line
Reorganized the structure of the Request Utils due to their widespread use
resulting in circular imports issues (mostly because of middleware & logger).
* Add req.host to targetURL when redirecting to https
The req.URL.Host might not be present when redirecting to https if the
given req.URL is something like "/". In such scenario, the req.Host is
still present and valid.
This commit adds the original req.Host to the targetURL before returning
the 308 status, to avoid having a `Location: https:///` in the response.
* Bring back empty line
* Wrap the setting of targetURL.Host in a condition
* Add a comment to the test explaining why the redirectURL includes example.com
* Add changelog entry
A blank user agent is considered == to an empty string. When no -ping-user-agent option is specified, this is considered to be an empty string.
This reveals two problems:
- When no ping-user-agent is specified, main.go sets up a health check user agent of ""
- When no user agent is specified, the empty string is still checked against the health check user agents.
Now the health check middleware ignores blank user agents and paths in order to sanitise it's input to avoid this issue.
Additional tests have been added to verify these situations.
Moves the logic for redirecting to HTTPs to a middleware package and adds tests for this logic.
Also makes the functionality more useful, previously it always redirected to the HTTPS address of the proxy, which may not have been intended, now it will redirect based on if a port is provided in the URL (assume public facing 80 to 443 or 4180 to 8443 for example)
