shashank created CAMEL-25248:
--------------------------------
Summary: camel-undertow - a String body is always sent as UTF-8,
even when the Content-Type declares another charset
Key: CAMEL-25248
URL: https://issues.apache.org/jira/browse/CAMEL-25248
Project: Camel
Issue Type: Bug
Components: camel-undertow
Reporter: shashank
The Undertow consumer writes a response body with
{code:java}
ByteBuffer bodyAsByteBuffer = tc.mandatoryConvertTo(ByteBuffer.class, body);
// UndertowConsumer.sendResponse
{code}
and the producer a request body with {{tc.tryConvertTo(ByteBuffer.class,
body)}} ({{UndertowProducer.process}}). Both convert *without the exchange*, so
{{NIOConverter.toByteBuffer(String, Exchange)}} gets {{null}} and encodes a
{{String}} in the default charset (UTF-8), whatever the message says:
* a route that sets {{Content-Type: text/plain; charset=ISO-8859-1}} and a
{{String}} body: {{Grüße aus Köln}} (14 characters) is sent as 17 bytes of
UTF-8 labelled ISO-8859-1, and the client reads {{Grü...}};
* an echo, or any route that keeps the request {{Content-Type}}: the consumer
reads an ISO-8859-1 request correctly (it sets {{CamelCharsetName}} from the
request charset), but the {{String}} reply under the same {{Content-Type}} is
UTF-8;
* the producer: {{to("undertow:http://...";)}} with a {{String}} body and
{{Content-Type: text/plain; charset=ISO-8859-1}} sends UTF-8 bytes under that
header.
{{byte[]}}, {{InputStream}} and other non-{{String}} bodies, and ASCII text,
are not affected. The other HTTP components honour the charset:
camel-servlet/camel-jetty ({{DefaultHttpBinding}}) write a {{String}} in
{{CamelCharsetName}}, camel-http uses {{CamelCharsetName}} or the
{{Content-Type}} charset, camel-netty-http converts with the exchange, and
camel-platform-http-vertx uses the {{Content-Type}} charset since CAMEL-25217.
h3. Reproduction
{code:java}
from("undertow:http://localhost:8080/latin1";)
.setHeader(Exchange.CONTENT_TYPE, constant("text/plain;
charset=ISO-8859-1"))
.setBody(constant("Grüße aus Köln"));
from("undertow:http://localhost:8080/echo";)
.convertBodyTo(String.class);
{code}
{{GET /latin1}} returns 17 bytes (UTF-8) instead of 14; {{POST /echo}} with
{{Content-Type: text/plain; charset=ISO-8859-1}} and the 14 ISO-8859-1 bytes
returns 17 bytes;
{{template.sendBodyAndHeader("undertow:http://localhost:8080/received";, "Grüße
aus Köln", "Content-Type", "text/plain; charset=ISO-8859-1")}} delivers 17
bytes. A unit test with these three cases fails on main ({{array lengths
differ, expected: <14> but was: <17>}}); a control with {{Content-Type:
text/plain}} (UTF-8 expected) passes.
h3. Proposed fix
A {{String}} body is encoded in the charset declared by the {{Content-Type}} of
the request or response being sent (new {{UndertowHelper.toByteBuffer(String,
String)}}, used by the consumer and the producer), as CAMEL-25217 does for
camel-platform-http-vertx. Without a declared charset (or with an unknown one)
the body is converted as today (UTF-8), and other bodies are converted as
today. {{CamelCharsetName}} is deliberately not used as a fallback: Undertow
cannot declare it in the header afterwards (camel-servlet does, with
{{setCharacterEncoding}}), so a response to an ISO-8859-1 request with
{{Content-Type: application/json}} would be written in ISO-8859-1 while the
client reads UTF-8.
With the fix the new test (5 cases, including that last one) and the whole
camel-undertow suite pass (196 tests, 1 skipped). Only messages that declare a
charset other than UTF-8 change, so the change gets an upgrade guide note.
Affected: 4.14.x, 4.18.x (same code) and main. The fix does not read
{{CamelCharsetName}}, so it can be backported on its own (on 4.14.x/4.18.x the
consumer still sets it to ISO-8859-1 for a request without charset; 892be63447
changed that on main).
Duplicate check (2026-10-01): JIRA component camel-undertow with "charset",
"encoding" or "UTF-8", and summary "undertow" with
"charset"/"encoding"/"ISO-8859-1": only CAMEL-11392 (ByteBuffer overflow with
multibyte characters, fixed 2017), CAMEL-12247, CAMEL-10803; nothing about the
charset of a String body. GitHub pull requests "undertow charset": none.
_Filed with Claude Code on behalf of allthingssecurity._
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
