Added writeUtf8 and writeUtf8Lines

[TonioGela]

Hello!
That's my first contribution to fs2, so I hope everything is where it should be.
I added two helpers to Files[F] that mirror the ones already present for readUtf8 and readUtf8Lines, shamelessly stealing the idea from this @armanbilge comment.
I haven't found any unit tests for readUtf8 and readUtf8Lines but for their components, and since writeUtf8 and writeUtf8Lines are built on the same ones, I haven't added any.

It may be worth adding these new combinators to fs2 2.5.x and adding an example like Stream.emit("foo").through(Files[IO].writeUtf8Lines(Path(...))) to the documentation.

[armanbilge]

It may be worth adding these new combinators to fs2 2.5.x

FS2 2.5.x is EOL (like CE 2.x series). I can't remember when we last merged a PR there :)

adding an example like Stream.emit("foo").through(Files[IO].writeUtf8Lines(Path(...))) to the documentation.

Sure! There may already be an existing example that can be made less verbose with the new helper :)

[TonioGela]

FS2 2.5.x is EOL (like CE 2.x series). I can't remember when we last merged a PR there :)

I got distracted by: https://github.com/typelevel/fs2/blob/main/.github/pull_request_template.md?plain=1#L3

Sure! There may already be an existing example that can be made less verbose with the new helper :)

I'll keep these changes in a separate PR to add them once these features are released

[diesalbla]

As @armanbilge mentioned, if the goal here is to insert a newline between strings, that would be in => writeUtf8(path, flags)(text.intersperse("\n")

[armanbilge]

Not sure, does that add a newline at the end? Seems like each string should terminate like that, so that ultimately the file terminates in a newline.

Also, does text.lines strip newlines? The reason I ask is because ideally these methods would be inverses of each other.

[TonioGela]

I cried for almost 1 hour in front of failing unit tests and then I saw it:

def writeUtf8Lines(path: Path): Pipe[F, String, Nothing] = writeUtf8(path, Flags.Write)
                                                                    ^

Thank god you pushed for them @armanbilge :D

For what is worth, I think that the logic might take in consideration being on windows probably: text.lines implementation takes in consideration \r too.

Not sure, does that add a newline at the end? Seems like each string should terminate like that, so that ultimately the file terminates in a newline.

Can we make it pluggable with a parameter maybe?

Also, does text.lines strip newlines? The reason I ask is because ideally these methods would be inverses of each other.

It does

[TonioGela]

Also, does text.lines strip newlines? The reason I ask is because ideally these methods would be inverses of each other.

Now that I'm thinking about it, is it the case to add roundtrip property-based tests like 👇 ?

forAll { xs: List[String] =>
  Stream.emits(xs).through(Files[IO].writeUtf8Lines(path)) ++
      Files[IO].readUtf8Lines(path).compile.toList.assertEquals(xs)
}

[TonioGela]

For what is worth, I think that the logic might take in consideration being on windows probably: text.lines implementation takes in consideration \r too.

Now it's handled

[armanbilge]

I really feel like we should be appending a line-separator to each string, instead of interspersing.

A sequence of zero or more non- characters plus a terminating <newline> character.

https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap03.html#tag_03_206

[TonioGela]

Would in.flatMap(s => Stream(s, lineSeparator)) do?

[armanbilge]

Yes, exactly!

[armanbilge]

Nice, thanks!