Support for Base64 encoding has been added to Pipeline: Shared Groovy Libraries 2.10 and Pipeline: Basic Steps 2.8.1 (or 2.10 if you are on Jenkins 2.121).

quas The problem is that UTF-8 is a variable length encoding so any byte whose highest bit is set will not round-trip correctly. You might be able to use a fixed-length encoding like Windows-1252 in your most recent snippet to get things working, but libraryResource currently doesn't take an encoding parameter so you would have to change your system default to use that as a workaround. It seems better to explicitly support this use case.

Would be great if we could support this, the resource (png) in my case have to be versioned, I could have separate project for them and expose as artifacts, but its quite an overhead.

Another idea, as I see readFile/writeFile are exposing the encoding parameter, so maybe there is a way to set it in a way to maintain consistency?

def fileContents = readFile file: "test.txt", encoding: "UTF-8"
fileContents = fileContents.replace("hello", "world")
echo fileContents
writeFile file: "test.txt", text: fileContents, encoding: "UTF-8"

If we want to add support for this, the best option to me seems like adding an encoding option to libraryResource, and then updating both libraryResource and writeFile to support Base64 as a valid encoding option to be used for binary data.

It looks like libraryResource reads the contents of the file into a String using the default system encoding, and writeFile encodes the given text using the system default encoding if no other encoding is specified, so doing this with binary files totally mangles the data. It doesn't look like there are any options that would make those steps work for your use case.

The best workarounds that I can think of are if the image is available at some static URL that Jenkins has access to you could download and write it to the desired location in a shell step using curl/wget, or if the image is in the repo itself you could use a shell step to copy it as desired.