[IO-716] ReaderInputStream enter infinite loop for too small buffer sizes

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.8.0
• Fix Version/s: 2.12.0
• Component/s: Streams/Writers
• Assignee: Unassigned
• Reporter: Marcono1234
• Created: 03/Feb/21 03:18
• Updated: 26/Sep/21 16:13
• Resolved: 26/Sep/21 15:28

Description

org.apache.commons.io.input.ReaderInputStream enters an infinite loop when it was constructed with a too small buffer size (0 or 1) and one of its reading methods is called.

Example:

int bufferSize = 1;
try (InputStream in = new ReaderInputStream(new StringReader("\uD800"), StandardCharsets.UTF_8, bufferSize)) {
    in.read();
}

This could be solved by making sure that the buffer size is always >= 2. This should solve this issue because as far as I know all charsets only report UNDERFLOW without encoding any chars if the only char in the buffer is a surrogate.
However, the saner solution might be to increase the buffer when UNDERFLOW is encountered and no chars from encoderIn have been consumed.

---

Similarly for sanety it would be good to throw an AssertionError (or increase the size of encoderOut) if OVERFLOW is returned. I don't think there are any encodings which require more than 128 bytes (used by ReaderInputStream for the output buffer) for encoding a single char or flushing, but would be good to at least handle that case in a saner way than entering an infinite loop anyways.