Add ByteBuffer methods getUTF8ValidatedString and readUTF8ValidatedString #2973
Conversation
There was a problem hiding this comment.
Thanks for this! I think the API design needs a bit of a tweak.
I also wonder if we can implement an older-Swift fallback using https://developer.apple.com/documentation/swift/string/init(validatingutf8:)-208fn.
Sources/NIOCore/ByteBuffer-aux.swift
Outdated
| } | ||
| return self.withUnsafeReadableBytes { pointer in | ||
| assert(range.lowerBound >= 0 && (range.upperBound - range.lowerBound) <= pointer.count) | ||
| return String(validating: UnsafeRawBufferPointer(fastRebase: pointer[range]), as: Unicode.UTF8.self) |
There was a problem hiding this comment.
This is probably not the spelling we want. Right now it's not possible to tell if this API call fails because there aren't enough readable bytes or because the string is not valid UTF8. We need to distinguish the two cases.
There was a problem hiding this comment.
I could throw an error for invalid strings?
There was a problem hiding this comment.
I think that might be appropriate, yeah.
There was a problem hiding this comment.
ok it is now throwing an error on invalid UTF8
The problem with using the above function is it requires the string to be null terminated. I guess I could copy it out to a separate buffer and add the null termination. |
|
Oh, we could do this one instead: https://developer.apple.com/documentation/swift/string/init(utf8string:)-8qmaq |
Which also requires a null-terminated string |
Add methods to ByteBuffer to read validated UTF8 strings
Motivation:
The current
readStringandgetStringmethods ofByteBufferdo not verify that the string being read is valid UTF8. The Swift 6 standard library comes with a new initialiserString(validating:as:). This PR adds alternative methods to ByteBuffer which uses this instead ofString(decoding:as:).Modifications:
Added
ByteBuffer.getUTF8ValidatedString(at:length:)Added
ByteBuffer.readUTF8ValidatedString(length:)Result:
You can read strings from a ByteBuffer and be certain they are valid UTF8