On Sat, 21 Mar 2026 09:00:58 GMT, Eirik Bjørsnøs <[email protected]> wrote:
>> Please review this PR which improves validation of unmappable characters in
>> names in the `ZipFileSystem` and `ZipFileOutputStream` APIs.
>>
>> Currently, `ZipFileSystem::getPath` and `ZipFileOutputStream:putNextEntry`
>> both throw `IllegalArgumentException` when rejecting a path or entry name
>> which cannot be encoded with the given charset.
>>
>> This PR fixes `ZipFileSystem::getPath` to instead throw
>> `InvalidPathException` as specified. Similarly,
>> `ZipOutputStream::putNextEntry` is updated to throw `ZipException` a
>> specified.
>>
>> Related, `ZipOutputStream::putNextEntry` is updated to reject unmappable
>> ZipEntry comments in a similar fashion.
>>
>> This change effectively means that `ZipOutputStream` now encodes names and
>> comments twice, once in `putNextEntry` and second time in `writeCEN` when
>> the stream is closed. An alternative would be to capture the encoded byte
>> arrays in the `XEntry`, however this would increase retained heap memory for
>> large number of entries.
>>
>> New tests are added in the ZipFS and ZipFileOutputStream area to verify that
>> these APIs throw exceptions according to their specifications when faced
>> with unmappable characters.
>
> Eirik Bjørsnøs has updated the pull request incrementally with one additional
> commit since the last revision:
>
> Align ZipFS ZipCoder.toString and ZipCoder.getBytes implementations with
> that in java.util.zip
src/java.base/share/classes/java/util/zip/ZipCoder.java line 104:
> 102: try {
> 103: ByteBuffer bb = encoder().encode(CharBuffer.wrap(s));
> 104: int pos = bb.position();
Slightly unrelated cleanup:
CharsetEncoder::encode(CharBuffer) is specified to return a ByteBuffer always
positioned at zero, so we can simplify here and use ByteBuffer.remaining()
instead.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/30319#discussion_r2969338722
