fix: handle short read errors on string #381
Conversation
reader.go
Outdated
| @@ -247,6 +247,10 @@ func (r *Reader) ReadBytes() []byte { | |||
| // ReadString reads a String from the Reader. | |||
| func (r *Reader) ReadString() string { | |||
| b := r.readBytes("string") | |||
| if r.Error != nil { | |||
There was a problem hiding this comment.
IMO, the fix needs to be more fundamental, in r.Read. r.Read is only ever called when an expected number of bytes should be read. Probably something like:
if r.head == r.tail {
if !r.loadMore() {
r.Error = io.ErrUnexpectedEOF
return
}
}
There was a problem hiding this comment.
sounds good, working on it
There was a problem hiding this comment.
hmm looks like the changes broke a bunch of ocf tests somehow
There was a problem hiding this comment.
@nrwiersma how are the files such as testdata/full-snappy.avro generated? I wonder if they have some invalid data in there.
There was a problem hiding this comment.
No, OCF seems to be taking advantage of the no error from before. I can take care of this in a PR if you like.
There was a problem hiding this comment.
Appreciate it! Yeah sorry was stuck on this for a bit.
There was a problem hiding this comment.
thank you so much!
A string is encoded as a long followed by that many bytes of UTF-8 encoded character data. If there aren't that many bytes of encoded character data, then it is an error.
For example,
0x08decodes to long(4). Decoding the bytes[]byte{0x08}to a string should result in an error.This PR wraps the
EOFerror, similar to the PR to fix array/maps.