Seems the problem is that the cdata section does not include a space before the ]]> closure. After adding a space it is properly handled by the xml decoder.

cc @rsc

I assume that there are no issues on Go code, but the reproduction code on playground might have an encoding bug.

• As we know, Go's source code is supposed to be encoded by UTF-8.
• The original reproduction code contains Chinese sentences. Though they are encoded by UTF-8, the xml tag says it is encoded by GBK. So xml decoder fails to decode properly.
• In the reproduction code, the decoding failure caused a lack of the CDATA's closing tag (see below). As a result, xml decoder skipped it and failed to parse xml behind the CDATA.

s := "润]]>"
r, _ := charset.NewReaderLabel("gbk", strings.NewReader(s))
b, _ := io.ReadAll(r)
fmt.Println(string(b))
// Output: 娑�]>

https://go.dev/play/p/TXhWsqyaSPY

How was the original problem? If the xml body is properly encoded by GBK, the xml decoder with gbk reader works well like below.

https://go.dev/play/p/JkyF8rOfWFU

Oh I see, that makes sense. Thanks for the explanation.