| Summary: | buggy read() in ChunkedCipherInputStream | ||
|---|---|---|---|
| Product: | POI | Reporter: | Tim Allison <tallison> |
| Component: | POIFS | Assignee: | POI Developers List <dev> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | ||
| Priority: | P2 | ||
| Version: | 4.0.x-dev | ||
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | All | ||
Definitely looks like a bug to me. +1 for merging. Although ChunkedCipherInputStream::read(byte[], int, int, boolean) returns -1, if no more data can be fetched and so there's no result size of 0, I would rewrite the original code:
if (read(b) == 1) {
return b[0];
}
return -1;
to:
return (read(b) == 1) ? b[0] : -1;
|
Over on TIKA-2873, we were just hit with this. In ChunkedCipherInputStream, read() is guaranteed to return -1 if there was something in the stream: @Override public int read() throws IOException { byte[] b = \{ 0 }; // FIXME: compare against -1 or 1? (bug 59893) return (read(b) == 1) ? -1 : b[0]; } This is related to 59893...and probably should be: @Override public int read() throws IOException { byte[] b = \{ 0 }; // FIXME: compare against -1 or 1? (bug 59893) return (read(b) == -1) ? -1 : b[0]; } I just made the change and all tests pass...any reason not to fix this?