Writing Data to an NSOutputStream in Swift 3
Martin R, thank you for your answer. That was a foundation for a complete solution. Here it is:
extension OutputStream {
/// Write String to outputStream
///
/// - parameter string: The string to write.
/// - parameter encoding: The String.Encoding to use when writing the string. This will default to UTF8.
/// - parameter allowLossyConversion: Whether to permit lossy conversion when writing the string.
///
/// - returns: Return total number of bytes written upon success. Return -1 upon failure.
func write(_ string: String, encoding: String.Encoding = String.Encoding.utf8, allowLossyConversion: Bool = true) -> Int {
if let data = string.data(using: encoding, allowLossyConversion: allowLossyConversion) {
var bytesRemaining = data.count
var totalBytesWritten = 0
while bytesRemaining > 0 {
let bytesWritten = data.withUnsafeBytes {
self.write(
$0.advanced(by: totalBytesWritten),
maxLength: bytesRemaining
)
}
if bytesWritten < 0 {
// "Can not OutputStream.write(): \(self.streamError?.localizedDescription)"
return -1
} else if bytesWritten == 0 {
// "OutputStream.write() returned 0"
return totalBytesWritten
}
bytesRemaining -= bytesWritten
totalBytesWritten += bytesWritten
}
return totalBytesWritten
}
return -1
}
}
NSData
had a bytes
property to access the bytes.
The new Data
value type in Swift 3 has a withUnsafeBytes()
method instead, which calls a closure with a pointer to the bytes.
So this is how you write Data
to an NSOutputStream
(without casting to NSData
):
let data = ... // a Data value
let bytesWritten = data.withUnsafeBytes { outputStream.write($0, maxLength: data.count) }
Remarks:
withUnsafeBytes()
is a generic method:
/// Access the bytes in the data.
///
/// - warning: The byte pointer argument should not be stored and used outside of the lifetime of the call to the closure.
public func withUnsafeBytes<ResultType, ContentType>(_ body: @noescape (UnsafePointer<ContentType>) throws -> ResultType) rethrows -> ResultType
In the above call,
both ContentType
and ResultType
are automatically inferred by
the compiler (as UInt8
and Int
), making additional
UnsafePointer()
conversions unnecessary.
outputStream.write()
returns the number of bytes actually written.
Generally, you should check that value. It can be -1
if
the write operation failed, or less than data.count
when writing
to sockets, pipes, or other objects with a flow control.
Just use this extension:
Swift 5
extension OutputStream {
func write(data: Data) -> Int {
return data.withUnsafeBytes {
write($0.bindMemory(to: UInt8.self).baseAddress!, maxLength: data.count)
}
}
}
And for InputStream
extension InputStream {
func read(data: inout Data) -> Int {
return data.withUnsafeMutableBytes {
read($0.bindMemory(to: UInt8.self).baseAddress!, maxLength: data.count)
}
}
}
Swift 4
extension OutputStream {
func write(data: Data) -> Int {
return data.withUnsafeBytes { write($0, maxLength: data.count) }
}
}
extension InputStream {
func read(data: inout Data) -> Int {
return data.withUnsafeMutableBytes { read($0, maxLength: data.count) }
}
}