html2htpy includes unneccessary white space

[pelme]

$ echo \<p\>     \t\t\nHi\n\n\t  \</p\> | html2htpy
from htpy import p

p[""" 		Hi	 """]

it would be nicer of the output was just

p["Hi"]

[OleJoik]

I remember concidering to strip the strings, but decided against it for some reason. Cant quite recall why, but might have had to do with some cases where I actually preferred whitespace to stay as is. Possibly theres some quirks with this in multi line strings.

I can take a look

[pelme]

I am not sure what the best exact rules would be, not all white space are safe to remove. I guess there are html minifiers and similar that could be used as inspiration for how to properly strip some of the whitespace

[OleJoik]

One special case at least would be <pre> tags. It should preserve most spaces and newlines.
https://developer.mozilla.org/en-US/docs/Web/HTML/Element/pre

[OleJoik]

Another thing to consider here; other html tags also can have goofy behavior as relates to css properties such as white-space and white-space-collapse. After reviewing this I'm not sure ever stripping whitespace is a good idea, as in quite a few cases the white space does matter.

To play around a bit, I made a codepen based on this mdn example that showcases a couple of cases where we would not want (all) whitespace trimmed. Below is a little teaser:

<!-- Example: Newlines preserved -->
<h2 style="white-space-collapse: preserve-breaks;">
  
  In this case
  all  the   newlines   are   preserved
  in    the     heading, 
  but    spaces    are    collapsed     .
  
</h2>

One option might be to filter for these properties and handle them in a special way. I'm not sure if I'd recommend it though, as it might quickly get complicated. w3 spec defines a non trivial set of white space processing rules that we should probably try to abide by if we do decide to do some trimmin'.

Edit to add: These properties might also be inherited or assigned with a class as in the code-pen, so it would be impossible to accurately recognize when they're applied to a given string.

[OleJoik]

An option might be to disregard the css props and trim away, though leave pre elements as is. The benefit might outweigh the cost for those few (?) users going crazy with whitespace css attributes

[pelme]

What about:

• text.replace(" ", " ").strip() all nodes except <pre>, <code> and <textarea> (maybe there are a few others too)
• Add --preserve-whitespace to opt-out of this behaviour.

When copy/pasting HTML code from the internet, it is often filled with spaces/white space that makes the HTML document look nice but is just there because it makes the HTML source look good rather then required from a functionality point of view.

[pelme]

To be clear: I do not think we should bother about white-space-collapse: preserve-breaks. I never used it myself AFAIK and think it is pretty rare. html2py is mostly meant as a starting point when converting a bulk of html to htpy but requires manual inspection/fixups.

[OleJoik]

I agree with your comments, and this should be pretty simple to implement I guess. I have some time and can bring a PR tomorrow?

[pelme]

That is very welcome! :)

[OleJoik]

#84