I'm looking to remove Ktor from my multiplatform project where it's now basically only used to encod...

# multiplatform

[Stefan Oltmann]

I'm looking to remove Ktor from my multiplatform project where it's now basically only used to encode bytes to ISO 8859-1 strings decode ISO 8859 byte arrays to String objects. The regular myByteArray.decodeToString() decodes to an UTF-8 string, which is what I need most of the time. In some places I need a decodeToString() that does the same with ISO 8859-1.

How to achieve that?

myByteArray.map { it.toChar() }.joinToString("") as ChatGPT suggests is wrong according to my unit tests.

[Sebastian Schuberth]

What about String(myByteArray)?

[Stefan Oltmann]

That's not multiplatform API. String(myCharArray) is the only available option.

[Jan Holešovský]

From your description, it is a bit unclear what do you actually want to achieve? Given that 'String' is a CharSequence, and Char is a 16-bit unicode character, what do you mean by 'encode bytes to ISO 8859-1 strings'?
Even the decodeToString() you've mentioned converts from UTF-8 representation to a String (ie. to 16-bit unicode) - so do you actually want to convert from ByteArray that represents ISO8859-1 to a 'normal' String (16-bit unicode), or something else?
Maybe it might help to understand if you share your original code?

[Stefan Oltmann]

I want to replace the implementation of this method:
https://github.com/Ashampoo/kim/blob/cf686df2be6d77578b9de2b17fd15fdbee8a66aa/src/commonMain/kotlin/com/ashampoo/kim/common/ByteArrayExtensions.kt#L44

Here is the corresponding test:
https://github.com/Ashampoo/kim/blob/cf686df2be6d77578b9de2b17fd15fdbee8a66aa/src/commonTest/kotlin/com/ashampoo/kim/common/ByteArrayExtensionsTest.kt#L143

The use case is that PNG text chunks contain ISO encoded text and PNG itxt chunks contain UTF8.

[Stefan Oltmann]

Ktor currently has no WASM support and so I wondered if there is an easier solution to this so that I can drop Ktor entirely.

[Stefan Oltmann]

And yes, now that you mention it… my description is confusing and wrong. 😅

[Stefan Oltmann]

Also the method name is wrong. It must be „decodeToStringFromIsoBytes()“ or something like that. The result is always a UTF8 string, because it’s Kotlin 😅

[Jan Holešovský]

No worries, thanks for the explanation! 🙂

[Jan Holešovský]

I'm not aware of a function that does that out of the box, so what if you just have an array of chars that correspond to the 0-255, and use something like the ChatGPT's recommended myByteArray.map { mapping[it] }.joinToString("") where the 'mapping' is the array of characters?

[Jan Holešovský]

[Not tested etc. - but hope you get the idea :-)]

[Stefan Oltmann]

Yes, sounds reasonable. I hope this exists somewhere 😄

[Jan Holešovský]

[and not sure if this is the most performant way to write that either - but seems you are not on a critical path there?]

[Stefan Oltmann]

Performance is important. Ashampoo Photos is supposed to read thousands of images very fast. 👀

[Stefan Oltmann]

For Windows Kim on JVM is currently 6 times faster than calling ExifTool using System.exec()

[Jan Holešovský]

OK - then I'd profile first if the conversion function even appears in the profile 😉. If it does, I'd check how the myByteArray.map { mapping[it] }.joinToString("") performs against a normal for loop, and whether it is better to have the mapping as an array, or use a when(...) { ... }.

[Jan Holešovský]

Wrt. "hope this exists somewhere", ChatGPT has generated it for me with this prompt: "Please forget the above, the following is a new problem: In Kotlin, please give me a CharArray that has 256 values, and each value is a iso8859-1 character represented as Kotlin Char." followed by "Please do generate me such a array, instead of writing me how to generate it." 😉 Let me DM you with the result I've got - I don't mean to check the correctness...

[Stefan Oltmann]

That’s a good prompt 👍