Don't include extra whitespace in compiled text

[jbdietrich]

Currently Curly ends up adding extra newlines after components like conditional or context block tags, which can make the resulting markup hard to read.

With this PR the following output:

<section class="knowledge-base">

  <div class="category-tree">

      <section class="category">




            <section class="section">

is shortened to:

<section class="knowledge-base">

  <div class="category-tree">

      <section class="category">

            <section class="section">

Alternatively Curly could strip text output altogether, but that may make the markup unreadable for lack of whitespace.

/cc @dasch

[dasch]

I'm not quite sure I agree with the way it's implemented. I agree that for block components, it would make sense to remove the entire line in the output, iff there's only whitespace and the component itself on that line, e.g.

<div>
{{@post}}
  <ul>
  {{*comments}}
    <li>{{body}}</li>
  {{/comments}}
  </ul>
{{/post}}
</div>

... should output

<div>
  <ul>
    <li>one</li>
    <li>two</li>
    <li>three</li>
  </ul>
</div>

I'm just not sure what the best way to do that is. Maybe add tokens for end-of-line in the scanner, and collapse leading and trailing whitespace in the parser?

[jbdietrich]

Yeah the method proposed in this PR is pretty heavy-handed. I was also unsure what the expectation was for preserving whitespace, so thanks for clarifying that.

I'll take a look into a smarter way to to this in the scanner/parser.

[dasch]

In theory, we could have a stream modifier that post-processes the token stream (which is output by the scanner) by looking for NEWLINE, WHITESPACE, BLOCK COMPONENT START/END, WHITESPACE, NEWLINE and replacing with BLOCK COMPONENT START/END, NEWLINE, but I'm not sure it's worth it.