Fix codeAsInSource

[HugoGranstrom]

Let's see if this works better now...

It took a day but I think I've found a somewhat non-chaotic implementation that seems to work. I added a few tests but I will add a few more and manually go through the docs to see if it looks OK.

fixes #97

TODO

• [x] Handle comments at end of code blocks
• [x] Go through the docs and check for errors
  • [x] Penguin
  • [x] numerical
  • [x] cheatsheet
  • [x] mostaccio
  • [x] interactivity
  • [x] counter
  • [x] caesar
  • [x] hello
  • [x] files
  • [x] pythno
  • [x] nolan

[HugoGranstrom]

Ok, that seems to have fixed it but why did it work locally without any errors... 🤷

[pietroppeter]

Nice! Can you try also to remove the when defined(nimibCodeFromAst) in t nimib.nim?

[HugoGranstrom]

Something smells weird here, when the defined(nimibCodeFromAst) was included no errors. But when it was removed it gives an error at the very end (after the last test has been run with OK). I read somewhere that a plain Error: execution of an external program failed: means that the program returned a non-zero error code but I have idea how that could be dependent on this when-statement :/

[HugoGranstrom]

To make it even stranger, running it with nim r gives this error but doing `nim c and then running the executable doesn't give any errors

[HugoGranstrom]

Oh wait I'm just blind, there was a failed test 🤦

[HugoGranstrom]

OK seems like codeFromAst gives a different result than codeAsInSource for this one:

nbTextWithCode: """
hi
how are you?
"""

codeFromAst rewrites it as """hi removing the newline.

[pietroppeter]

ah, right, it is to be expected! In nim multiline string literal remove the first newline so CodeFromAst does not see it, while CodeAsInSource sees it as it is written. They are both right! Probably the easieast way to fix it would be a short template/proc used only in tests and added to the multiline strings involved here:

template normalizeMultilineStringLiteral(s: string) =
  when not defined(CodeFromAst):
    if s[0] == "\n":
      return s[1 ..< s.len]

[pietroppeter]

didn't see you already had the fix. fine as it is for me!

[HugoGranstrom]

ah, right, it is to be expected! In nim multiline string literal remove the first newline so CodeFromAst does not see it, while CodeAsInSource sees it as it is written. They are both right! Probably the easieast way to fix it would be a short template/proc used only in tests and added to the multiline strings involved here:

TIL :D

didn't see you already had the fix. fine as it is for me!

It is the only case where it matters thus far so making a template only feels warranted when we have a few test that needs it 👍

[HugoGranstrom]

I've gone over the docs now and to my eyes, it looks like it's working but I would appreciate a second opinion on that 😅

[pietroppeter]

I also give it a pass on all docs and do not find anything strange. We could at some point create a karax app that does a html diff between old and previous docs ;)

I noticed a couple of things while reading docs:

• in numerical.nim we find the only occurence of nbTextWithCode and the source is:

  nbTextWithCode: # note that we need to start the multi line on next line
  fmt"""

  we could change it back to:

  nbTextWithCode: fmt"""

• in interactivity.nim it took me a moment to understand where the second code block was coming from (the two echo hellos generated by nbCodeToJsShowSource()), maybe we could add a short explanation text before that, something like: "This is the nim code that is written to a temporary file and is compiled to javascript"

[HugoGranstrom]

I also give it a pass on all docs and do not find anything strange. We could at some point create a karax app that does a html diff between old and previous docs ;)

🙏 Haha 🤣😎

Fixed your remarks