The wrapped response prevents use after AsyncListener onError or completion
to ensure compliance with Servlet Spec 2.3.3.4.
The wrapped response is applied in RequestMappingHandlerAdapter.
The wrapped response raises AsyncRequestNotUsableException that is now
handled in DefaultHandlerExceptionResolver.
See gh-32341
Based on feedback from several members of the community, we have
decided to revert the caching of the Content-Type header that was
introduced in ContentCachingResponseWrapper in 375e0e6827.
This commit therefore completely removes Content-Type caching in
ContentCachingResponseWrapper and updates the existing tests
accordingly.
To provide guards against future regressions in this area, this commit
also introduces explicit tests for the 6 ways to set the content length
in ContentCachingResponseWrapper and modifies a test in
ShallowEtagHeaderFilterTests to ensure that a Content-Type header set
directly on ContentCachingResponseWrapper is propagated to the
underlying response even if content caching is disabled for the
ShallowEtagHeaderFilter.
See gh-32039
See gh-32317
Closes gh-32321
Commit 375e0e6827 introduced a regression in
ContentCachingResponseWrapper (CCRW). Specifically, CCRW no longer
honors Content-Type and Content-Length headers that have been set in
the wrapped response and now incorrectly returns null for those header
values if they have not been set directly in the CCRW.
This commit fixes this regression as follows.
- The Content-Type and Content-Length headers set in the wrapped
response are honored in getContentType(), containsHeader(),
getHeader(), and getHeaders() unless those headers have been set
directly in the CCRW.
- In copyBodyToResponse(), the Content-Type in the wrapped response is
only overridden if the Content-Type has been set directly in the CCRW.
Furthermore, prior to this commit, getHeaderNames() returned duplicates
for the Content-Type and Content-Length headers if they were set in the
wrapped response as well as in CCRW.
This commit fixes that by returning a unique set from getHeaderNames().
This commit also updates ContentCachingResponseWrapperTests to verify
the expected behavior for Content-Type and Content-Length headers that
are set in the wrapped response as well as in CCRW.
See gh-32039
See gh-32317
Closes gh-32321
Before this commit, the JDK and Jetty connectors do not have any
safeguards against multiple body subscriptions. Such as check has now
been added.
See gh-32100
Closes gh-32102
Prior to this commit, the `RestTemplate` observation instrumentation
would only record `RestClientException` and `IOException` as errors in
the observation. Other types of errors can be thrown by custom
components, such as `ResponseErrorHandler` and in this case they aren't
recorded with the observation.
Also, the current instrumentation does not create any observation scope
around the execution. While this would have a limited benefit as no
application code is executed there, developers could set up custom
components (such as, again, `ResponseErrorHandler`) that could use
contextual logging with trace ids.
This commit ensures that all `Throwable` are recorded as errors with the
observations and that an observation `Scope` is created around the
execution of the client exchange.
Fixes gh-32063
This commit ensures that setting the Content-Length through
setHeader("Content-Length", x") has the same effect as calling
setContentLength in the ShallowEtagHeaderFilter. It also filters out
Content-Type headers similarly to Content-Length.
See gh-32039
Closes gh-32050
This commit adds CORS support for Private Network Access
by adding an Access-Control-Allow-Private-Network response
header when the preflight request is sent with an
Access-Control-Request-Private-Network header and that
Private Network Access has been enabled in the CORS
configuration.
See https://developer.chrome.com/blog/private-network-access-preflight/
for more details.
Closes gh-31975
(cherry picked from commit 318d460256)
Prior to this commit, the `ForwardedHeaderFilter` and the forwarded
header utils would throw `IllegalArgumentException` and
`IllegalStateException` when request headers are invalid and cannot be
parsed for Forwarded handling.
This commit aligns the behavior with the WebFlux counterpart by
rejecting such requests with HTTP 400 responses directly.
Fixes gh-31894
This commit fixes a bug in DefaultPartHttpMessageReader's
MultipartParser, due to which the last token in a part window was not
properly indicated.
See gh-30953
Closes gh-31766
Prior to this commit, regressions were introduced with gh-31417:
1. the observation keyvalues would be inconsistent with the HTTP response
2. the observation scope would not cover all controller handlers, causing
traceIds to be missing
The first issue is caused by the fact that in case of error signals, the
observation was stopped before the response was fully committed, which
means further processing could happen and update the response status.
This commit delays the stop event until the response is committed in
case of errors.
The second problem is caused by the change from a `contextWrite`
operator to using the `tap` operator with a `SignalListener`. The
observation was started in the `doOnSubscription` callback, which is too
late in some cases. If the WebFlux controller handler is synchronous
non-blocking, the execution of the handler is performed before the
subscription happens. This means that for those handlers, the
observation was not started, even if the current observation was
present in the reactor context. This commit changes the
`doOnSubscription` to `doFirst` to ensure that the observation is
started at the right time.
Fixes gh-31715
Fixes gh-31716
Prior to this commit, the Jackson 2.x encoders, in case of encoding a
stream of data, would first release the `ByteArrayBuilder` and then the
`JsonGenerator`. This order is inconsistent with the single value
variant (see `o.s.h.codec.json.AbstractJackson2Encoder#encodeValue`) and
invalid since the `JsonGenerator` uses internally the
`ByteArrayBuilder`.
In case of a CSV Encoder, the codec can buffer data to write the column
names of the CSV file. Writing an empty Flux with this Encoder would not
fail but still log a NullPointerException ignored by the reactive
pipeline.
This commit fixes the order and avoid such issues at runtime.
Fixes gh-31656
Prior to this commit, HTTP server observations for Spring WebFlux could
be recorded twice for a single request in some cases. The "COMPLETE" and
"CANCEL" signals would race in the reactive pipeline and would trigger
both the `doOnComplete()` and ` `doOnCancel()` operators, each calling
`observation.stop()` on the current observation.
This would in fact publish two different observations for the same
request.
This commit ensures that the instrumentation uses the `Mono#tap`
operator to guard against this case and only call `Observation#stop`
once for each request.
Fixes gh-31417
Prior to this commit, a cancelled exchange would always result in an
`"status":"UNKNOWN"` KeyValue. This only applied to reactive variants,
as cancelled exchanges are not currently detected for Servlet
implementations.
In some cases, exchanges can be cancelled by clients before they are
completed, but the response was actually received by the client. The
response status information has been set by the application and the
response has been committed. For those cases, we shouldn't assume an
"UNKNOWN" value.
This commit assumes that committed responses have a response status set
by the application and that the observations should reflect that. From
now on, we only assume an "UNKNOWN" status if the response has not been
commited.
Fixes gh-31388
