tidy -i decides which line a comment belongs on

[jidanni]

Perhaps tidy -i should better respect that this comment was not
associated with any line:

 <p>A</p>

 <!-- COMMENT -->

 <p>B</p>

Which alas becomes

  <p>A</p><!-- COMMENT -->
  <p>B</p>

At least it should be like plain tidy,

  <p>A</p>
  <!-- COMMENT -->
  <p>B</p>

tidy 5.6 HTML 5.

[jidanni]

E.g., we want to say via our comment that THE NERDSBERG PAPERS begin below. But instead that is swung onto the end of the previous line, destroying its "dignity" and making it no longer noticeable unless we have the habit of looking at the end of each line for important things. Yes maybe tidy only promises not to cause changes when HTML is rendered. But still...

[geoffmcl]

@jidanni thanks for the issue... although I think the title is maybe mis-leading...

If you refer to, something more like -

The indent option changes output line of a comment.

i.e. if you feed in 1 line -

   <p>A</p><!-- COMMENT --><p>B</p>

The output of Tidy, without indent, will be 3 lines -

<p>A</p>
<!-- COMMENT -->
<p>B</p>

Ok, if you accept that... the comment should have its own line... then...

But if you ADD the indent option, -i, the output will be 2 lines -

  <p>A</p><!-- COMMENT -->
  <p>B</p>

Although, it might not be to everyone's liking, I would certainly prefer something like -

  <p>A</p>
  <!-- COMMENT -->
  <p>B</p>

I have noted this Pretty Print contrariness for a long time now...

Please confirm this is what you are referring to, or provide futher explanation, examples, etc...

I would love a code pointer, patch, PR, fixing this small anomoly... thanks...

[jidanni]

Hmmm, tidy's long lost relative from Transylvania, perltidy, has $ man perltidy Comment Controls Perltidy has a number of ways to control the appearance of both block comments and side comments. [300 lines omitted] OK that's overkill. Anyway, for tidy, in your example there were to <P>'s on the same line. OK, I never thought about that kind of input. For now all I know is if a comment is surrounded by newlines, on either side, just make sure those newlines survive unscathed no matter -i or not.

[geoffmcl]

@jidanni thank you for the feedback...

Given the present coding, tidy can NOT make sure those newlines survive unscathed...

Except for a few tags, like <script>, <pre>, ... most newlines in the input, are discarded, as a DOM like tree of nodes is built up, so when it comes to the output, tidy has no memory, information, knowledge, that a comment is surrounded by newlines, so can in no way influence the output...

Tidy has some 20 Pretty Print Options, and many other options that influence the style of the output, and given a use case, and a clear specification, a new option could be considered...

But for sure, if a comment is surrounded by newlines, on either side, just make sure those newlines survive unscathed no matter -i or not does not yet rise to this level... sorry...

And the option spec has to deal with comments within a tag, as well as outside, like -

<p>A
 <!-- comment --> a
a</p>
<p>B
 <!-- comment -->
 b
</p>

And every variation there of... not an easy task...

I did try to take a look at my small -i anomaly, but did not spot an easy fix... although I can see the indent option certainly changes the code path chosen... but to me is quite minor...

Look forward to further feedback, comments, patches, etc, to help in formalizing this issue, else will consider closing it ... thanks...

[jidanni]

Ah! On the man page we/you could say: "If you want the whitespace/newlines surrounding your comments preserved, consider placing it/them **within** the comment."

[noloader]

Add a mee too. We have the same problem in out cleanup script. HTML Tidy performs an unwanted folding of comments into the previous line.

For example, this:

  <!-- ************** -->
  <!--  Some comment about list 1 -->
  <!-- ************** -->
  <ul>
    <li>...</li>
  </ul>
  <!-- ************** -->
  <!--  Some comment about list 2 -->
  <!-- ************** -->
  <ul>
    <li>...</li>
  </ul>

Is transformed into:

  ... </p><!-- ************** -->
  <!--  Some comment about list 1 -->
  <!-- ************** -->
  <ul>
    <li>...</li>
  </ul><!-- ************** -->
  <!--  Some comment about list 2 -->
  <!-- ************** -->
  <ul>
    <li>...</li>
  </ul>

It is very unsightly and makes it harder to read the comments in a tidied document because it is so distracting.

Also see #870.