Re: need two runs

I ran into this same problem myself while attempting to clean HTML
produced by (shudder) Microsoft Word.

Theoretically, tidy should work like this:
1.  Load the "questionable" HTML into a DOM
2.  Clean the DOM
3.  Write the DOM back out as "good" HTML/XHTML

As usual, the reality is somewhat more complex than the theory.  As tidy
is currently implemented, some of the cleanup occurs both in the input
step and the output step.  In your case, the code that discards empty
paragraphs (or converts them to <br /><br />) is implemented in the
input step.

While you did not provide us with your orignal HTML file, I suspect it
contains a construct similar to the following:

<p><small>[some junk that's going to get trimmed into
oblivion]</small></p>

In this case, tidy sees that the paragraph is not empty, so it creates a
node for it in the DOM.  Once the document is loaded, cleanup routines
are called, which trim away what was in the paragraph; in my case this
was DropSections().  Now in cleaning away unused nodes, the program
calls CanPrune() to see if a node can be trimmed.  This routine returns
false if the node has content, _even if the content is a zero length
text node_! The empty paragraph never gets trimmed, and so prints out
<p></p> which should have been trimmed.  The next time through, tidy
trims the empty paragraph.

I'll put a bug in someone's ear, and send them some diffs to see if we
can get this cleared up.

Dan Jacobson wrote:
> 
> Firstly i would like to commend tidy on being such a critical piece of
> software [if something goes wrong then "my web pages are ruined" :-(
> ], OK,
> 
> It seems two runs in a row are required until I've got a file tidyed
> till it can't be tidyed anymore.  "This thing doesn't have a self
> adjusting spin cycle, so I had to run it again before I got the
> whitest whites" as Suzy Homemaker would say.
> 
> you see, first I strip all the big5 chars out of a file to make an
> alternative english version...
> 
> make english
> b5strip mapping.htm|sed '1,/content="zh-tw">/{/<META/,/content="zh-tw">/d;}'>mapping_en.html
> tidy01nov01 -qm mapping_en.html
> mapping_en.html:59:10: Warning: trimming empty <em>
> mapping_en.html:176:7: Warning: trimming empty <h3>
> mapping_en.html:330:6: Warning: trimming empty <p>
> make: [mapping_en.html] Error 1 (ignored)
> tidy01nov01 -q mapping_en.html|diff mapping_en.html -
> mapping_en.html:57:6: Warning: trimming empty <p>
> mapping_en.html:309:6: Warning: trimming empty <p>
> mapping_en.html:311:13: Warning: trimming empty <small>
> mapping_en.html:311:21: Warning: trimming empty <p>
> mapping_en.html:325:6: Warning: trimming empty <p>
> mapping_en.html:327:6: Warning: trimming empty <p>
> 57,58d56
> <   <P></P>
> <
> 308,311d305
> <
> <   <P></P>
> <
> <   <P><SMALL></SMALL></P>
> 324,327d317
> <
> <   <P></P>
> <
> <   <P></P>
> make: [mapping_en.html] Error 1 (ignored)
> tidy01nov01 -qm mapping_en.html
> mapping_en.html:57:6: Warning: trimming empty <p>
> mapping_en.html:309:6: Warning: trimming empty <p>
> mapping_en.html:311:13: Warning: trimming empty <small>
> mapping_en.html:311:21: Warning: trimming empty <p>
> mapping_en.html:325:6: Warning: trimming empty <p>
> mapping_en.html:327:6: Warning: trimming empty <p>
> make: [mapping_en.html] Error 1 (ignored)
> tidy01nov01 -q mapping_en.html|diff mapping_en.html -
> tidy01nov01 -qm mapping_en.html
> tidy01nov01 -q mapping_en.html|diff mapping_en.html -
> --
> http://www.geocities.com/jidanni/ Tel+886-4-25854780

Received on Monday, 17 December 2001 11:56:48 UTC