Org-special-blogs does not make well-formed xhtml

Org-special-blogs does not make well-formed xhtml

[Christian Moe]

Hi,

XHTML produced with Org-special-blocks is not well-formed; <div> tags 
get wrapped in <p> tags. Example:

    Some text.

    #+begin_sidebar
      Some details left out of the main text.

      Some more details.
    #+end_sidebar

    Some more text.

This results in the following html, which causes XML processors to fail.

    <p>
    <div class="sidebar">
      Some details left out of the main text.
    </p>
    <p>
      Some more details.
    </div>
    </p>
    <p>
    Some more text.
    </p></div>

The problem seems to be fixed by un-commenting the fifth line in the 
below function in org-special-blocks.el:

    (defun org-special-blocks-convert-html-special-cookies ()
      "Converts the special cookies into div blocks."
      ;; Uses the dynamically-bound variable `line'.
      (when (string-match "^ORG-\\(.*\\)-\\(START\\|END\\)$" line)
    ;    (org-close-par-maybe)
        (message "%s" (match-string 1))
        (if (equal (match-string 2 line) "START")
	   (insert "<div class=\"" (match-string 1 line) "\">\n")
	 (insert "</div>\n"))
        (throw 'nextline nil)))

Does anyone know if that was commented out for a reason?


Yours,
Christian

Re: Org-special-blogs does not make well-formed xhtml

[Sebastien Vauban]

Hi Christian Moe,

Christian Moe wrote:
> Hi,
>
> XHTML produced with Org-special-blocks is not well-formed; <div> tags 
> get wrapped in <p> tags. Example:
>
>     Some text.
>
>     #+begin_sidebar
>       Some details left out of the main text.
>
>       Some more details.
>     #+end_sidebar
>
>     Some more text.
>
> This results in the following html, which causes XML processors to fail.
>
>     <p>
>     <div class="sidebar">
>       Some details left out of the main text.
>     </p>
>     <p>
>       Some more details.
>     </div>
>     </p>
>     <p>
>     Some more text.
>     </p></div>
>
> The problem seems to be fixed by un-commenting the fifth line in the 
> below function in org-special-blocks.el:
>
>     (defun org-special-blocks-convert-html-special-cookies ()
>       "Converts the special cookies into div blocks."
>       ;; Uses the dynamically-bound variable `line'.
>       (when (string-match "^ORG-\\(.*\\)-\\(START\\|END\\)$" line)
>     ;    (org-close-par-maybe)
>         (message "%s" (match-string 1))
>         (if (equal (match-string 2 line) "START")
> 	   (insert "<div class=\"" (match-string 1 line) "\">\n")
> 	 (insert "</div>\n"))
>         (throw 'nextline nil)))
>
> Does anyone know if that was commented out for a reason?

git blame suggests it was there since that file org-special-blocks.el has been
added (in contrib/lisp, by Carsten, on 2009-05-07 13:53)...

Best regards,
  Seb

-- 
Sebastien Vauban

Re: Org-special-blogs does not make well-formed xhtml

[Carsten Dominik]

On Oct 3, 2011, at 3:26 PM, Sebastien Vauban wrote:

> Hi Christian Moe,
> 
> Christian Moe wrote:
>> Hi,
>> 
>> XHTML produced with Org-special-blocks is not well-formed; <div> tags 
>> get wrapped in <p> tags. Example:
>> 
>>    Some text.
>> 
>>    #+begin_sidebar
>>      Some details left out of the main text.
>> 
>>      Some more details.
>>    #+end_sidebar
>> 
>>    Some more text.
>> 
>> This results in the following html, which causes XML processors to fail.
>> 
>>    <p>
>>    <div class="sidebar">
>>      Some details left out of the main text.
>>    </p>
>>    <p>
>>      Some more details.
>>    </div>
>>    </p>
>>    <p>
>>    Some more text.
>>    </p></div>
>> 
>> The problem seems to be fixed by un-commenting the fifth line in the 
>> below function in org-special-blocks.el:
>> 
>>    (defun org-special-blocks-convert-html-special-cookies ()
>>      "Converts the special cookies into div blocks."
>>      ;; Uses the dynamically-bound variable `line'.
>>      (when (string-match "^ORG-\\(.*\\)-\\(START\\|END\\)$" line)
>>    ;    (org-close-par-maybe)
>>        (message "%s" (match-string 1))
>>        (if (equal (match-string 2 line) "START")
>> 	   (insert "<div class=\"" (match-string 1 line) "\">\n")
>> 	 (insert "</div>\n"))
>>        (throw 'nextline nil)))
>> 
>> Does anyone know if that was commented out for a reason?
> 
> git blame suggests it was there since that file org-special-blocks.el has been
> added (in contrib/lisp, by Carsten, on 2009-05-07 13:53)...
> 
> Best regards,
>  Seb
> 
> -- 
> Sebastien Vauban
> 
> 

I see no reason to not uncomment this line.  Shall we just do this?



- Carsten

Re: Org-special-blogs does not make well-formed xhtml

[Christian Moe]

>>
>
> I see no reason to not uncomment this line.  Shall we just do this?
>
>
>
> - Carsten

Yes, would you, please?

Yours,
Christian

Re: Org-special-blogs does not make well-formed xhtml

[Carsten Dominik]

Done...


- Carsten
On Oct 4, 2011, at 2:45 PM, Christian Moe wrote:

> 
>>> 
>> 
>> I see no reason to not uncomment this line.  Shall we just do this?
>> 
>> 
>> 
>> - Carsten
> 
> Yes, would you, please?
> 
> Yours,
> Christian
> 

- Carsten

Re: Org-special-blogs does not make well-formed xhtml

[Jambunathan K]

Carsten Dominik <carsten.dominik@gmail.com> writes:

> On Oct 3, 2011, at 3:26 PM, Sebastien Vauban wrote:
>
>> Hi Christian Moe,
>> 
>> Christian Moe wrote:
>>> Hi,
>>> 
>>> XHTML produced with Org-special-blocks is not well-formed; <div> tags 
>>> get wrapped in <p> tags. Example:
>>> 
>>>    Some text.
>>> 
>>>    #+begin_sidebar
>>>      Some details left out of the main text.
>>> 
>>>      Some more details.
>>>    #+end_sidebar
>>> 
>>>    Some more text.
>>> 
>>> This results in the following html, which causes XML processors to fail.
>>> 
>>>    <p>
>>>    <div class="sidebar">
>>>      Some details left out of the main text.
>>>    </p>
>>>    <p>
>>>      Some more details.
>>>    </div>
>>>    </p>
>>>    <p>
>>>    Some more text.
>>>    </p></div>
>>> 
>>> The problem seems to be fixed by un-commenting the fifth line in the 
>>> below function in org-special-blocks.el:
>>> 
>>>    (defun org-special-blocks-convert-html-special-cookies ()
>>>      "Converts the special cookies into div blocks."
>>>      ;; Uses the dynamically-bound variable `line'.
>>>      (when (string-match "^ORG-\\(.*\\)-\\(START\\|END\\)$" line)
>>>    ;    (org-close-par-maybe)
>>>        (message "%s" (match-string 1))
>>>        (if (equal (match-string 2 line) "START")
>>> 	   (insert "<div class=\"" (match-string 1 line) "\">\n")
>>> 	 (insert "</div>\n"))
>>>        (throw 'nextline nil)))
>>> 
>>> Does anyone know if that was commented out for a reason?
>> 
>> git blame suggests it was there since that file org-special-blocks.el has been
>> added (in contrib/lisp, by Carsten, on 2009-05-07 13:53)...
>> 
>> Best regards,
>>  Seb
>> 
>> -- 
>> Sebastien Vauban
>> 
>> 
>
> I see no reason to not uncomment this line.  Shall we just do this?

Shouldn't a paragraph be opened/closed while entering/leaving the div.

,---- See org-xhtml-format-environment
|     (center
|      (case beg-end
|        (BEGIN
| 	(org-lparse-end-paragraph)
| 	(insert "\n<div style=\"text-align: center\">")
| 	(org-lparse-begin-paragraph))
|        (END
| 	(org-lparse-end-paragraph)
| 	(insert "\n</div>")
| 	(org-lparse-begin-paragraph))))
`----

We can always open a paragraph gratis, because empty paragraphs are
pruned at the end of export.

ps: I haven't gotten myself to understand when paragraphs are opened and
closed in the html exporter. So we can take a leaf out of the centering
environment and mimic the same behaviour with the special blocks as
well.

> - Carsten
>
>
>
>
>

-- 

Re: Org-special-blogs does not make well-formed xhtml

[Christian Moe]

Hi,

Carsten, thanks.

Jambunathan's probably right, though, that this fix (while necessary) 
is incomplete. Sorry, I hadn't thought it through.

Now, when the first paragraph of a special block comes right after the 
#+BEGIN_WHATEVER line, it will just come as the first text node of the 
<div> block, and will not be wrapped in a <p> element. So among other 
things, any special CSS styling for <p>s will not be applied to it.

It's not a big deal -- if you need a <p> (and often you really don't) 
the workaround is as simple as leaving an empty line between 
#+BEGIN_WHATEVER and the first paragraph.

But I guess it would be more consistent/expected/helpful behavior for 
special-blocks to start a new paragraph -- this time *after* the 
opening <div> tag and not around it!

Yours,
Christian


On 10/4/11 4:36 PM, Jambunathan K wrote:
>
> Carsten Dominik<carsten.dominik@gmail.com>  writes:
>> I see no reason to not uncomment this line.  Shall we just do this?
>
> Shouldn't a paragraph be opened/closed while entering/leaving the div.
>
> ,---- See org-xhtml-format-environment
> |     (center
> |      (case beg-end
> |        (BEGIN
> | 	(org-lparse-end-paragraph)
> | 	(insert "\n<div style=\"text-align: center\">")
> | 	(org-lparse-begin-paragraph))
> |        (END
> | 	(org-lparse-end-paragraph)
> | 	(insert "\n</div>")
> | 	(org-lparse-begin-paragraph))))
> `----
>
> We can always open a paragraph gratis, because empty paragraphs are
> pruned at the end of export.

Re: Org-special-blogs does not make well-formed xhtml

[Carsten Dominik]

Hi Christian,

for now I have reverted the patch.  WOuld you like to work on a better one?

- Carsten

On 4.10.2011, at 21:01, Christian Moe wrote:

> Hi,
> 
> Carsten, thanks.
> 
> Jambunathan's probably right, though, that this fix (while necessary) is incomplete. Sorry, I hadn't thought it through.
> 
> Now, when the first paragraph of a special block comes right after the #+BEGIN_WHATEVER line, it will just come as the first text node of the <div> block, and will not be wrapped in a <p> element. So among other things, any special CSS styling for <p>s will not be applied to it.
> 
> It's not a big deal -- if you need a <p> (and often you really don't) the workaround is as simple as leaving an empty line between #+BEGIN_WHATEVER and the first paragraph.
> 
> But I guess it would be more consistent/expected/helpful behavior for special-blocks to start a new paragraph -- this time *after* the opening <div> tag and not around it!
> 
> Yours,
> Christian
> 
> 
> On 10/4/11 4:36 PM, Jambunathan K wrote:
>> 
>> Carsten Dominik<carsten.dominik@gmail.com>  writes:
>>> I see no reason to not uncomment this line.  Shall we just do this?
>> 
>> Shouldn't a paragraph be opened/closed while entering/leaving the div.
>> 
>> ,---- See org-xhtml-format-environment
>> |     (center
>> |      (case beg-end
>> |        (BEGIN
>> | 	(org-lparse-end-paragraph)
>> | 	(insert "\n<div style=\"text-align: center\">")
>> | 	(org-lparse-begin-paragraph))
>> |        (END
>> | 	(org-lparse-end-paragraph)
>> | 	(insert "\n</div>")
>> | 	(org-lparse-begin-paragraph))))
>> `----
>> 
>> We can always open a paragraph gratis, because empty paragraphs are
>> pruned at the end of export.

Re: Org-special-blogs does not make well-formed xhtml

[Christian Moe]

[-- Attachment #1: Type: text/plain, Size: 2203 bytes --]

Hi, Carsten,

I've submitted a patch separately. It seems to play nice with the 
enclosed test file.

If org-special-blocks is supposed to become part of the core (it's 
moved out of contrib, but we still have to require it), it might make 
sense to just move the html and latex formatting code into org-html.el 
and org-latex.el respectively, together with the handling of 
blockquotes, verses, centering etc.

Yours,
Christian

On 10/4/11 11:33 PM, Carsten Dominik wrote:
> Hi Christian,
>
> for now I have reverted the patch.  WOuld you like to work on a better one?
>
> - Carsten
>
> On 4.10.2011, at 21:01, Christian Moe wrote:
>
>> Hi,
>>
>> Carsten, thanks.
>>
>> Jambunathan's probably right, though, that this fix (while necessary) is incomplete. Sorry, I hadn't thought it through.
>>
>> Now, when the first paragraph of a special block comes right after the #+BEGIN_WHATEVER line, it will just come as the first text node of the<div>  block, and will not be wrapped in a<p>  element. So among other things, any special CSS styling for<p>s will not be applied to it.
>>
>> It's not a big deal -- if you need a<p>  (and often you really don't) the workaround is as simple as leaving an empty line between #+BEGIN_WHATEVER and the first paragraph.
>>
>> But I guess it would be more consistent/expected/helpful behavior for special-blocks to start a new paragraph -- this time *after* the opening<div>  tag and not around it!
>>
>> Yours,
>> Christian
>>
>>
>> On 10/4/11 4:36 PM, Jambunathan K wrote:
>>>
>>> Carsten Dominik<carsten.dominik@gmail.com>   writes:
>>>> I see no reason to not uncomment this line.  Shall we just do this?
>>>
>>> Shouldn't a paragraph be opened/closed while entering/leaving the div.
>>>
>>> ,---- See org-xhtml-format-environment
>>> |     (center
>>> |      (case beg-end
>>> |        (BEGIN
>>> | 	(org-lparse-end-paragraph)
>>> | 	(insert "\n<div style=\"text-align: center\">")
>>> | 	(org-lparse-begin-paragraph))
>>> |        (END
>>> | 	(org-lparse-end-paragraph)
>>> | 	(insert "\n</div>")
>>> | 	(org-lparse-begin-paragraph))))
>>> `----
>>>
>>> We can always open a paragraph gratis, because empty paragraphs are
>>> pruned at the end of export.
>
>
>


[-- Attachment #2: special-blocks-test.org --]
[-- Type: text/plain, Size: 1612 bytes --]

#+style: <style>.abstract:before { content: "Abstract: "; font-weight: bold; display: inline}
#+style: .abstract {font-size: 10pt; margin-left: 3em}
#+style: .sidebar {border: 1px solid black; padding: 1em; background: yellow; counter-increment: box}
#+style: .sidebar:before {content: "Box " counter(box); font-weight: bold; display: block}
#+style: p {font-family: arial}</style

Special blocks test


#+begin_src emacs-lisp
  (require 'org-special-blocks)
#+end_src

#+results:
: org-special-blocks


#+begin_abstract
  This is a special =abstract= block.
#+end_abstract

Some text.

#+begin_sidebar
  This is a sidebar with additional detail. Check that the first
  paragraph is styled the same as the others.

  It extends over several paragraphs.
#+end_sidebar

Some more text.

#+begin_sidebar

  This is another example with lines above/below.

#+end_sidebar

Text again.

#+begin_center
  This is centered text. Like special blocks, it's wrapped in a =div=
  element in HTML.
#+end_center


New code:


#+begin_src emacs-lisp
  (defun org-special-blocks-convert-html-special-cookies ()
  "Converts the special cookies into div blocks."
  ;; Uses the dynamically-bound variable `line'.
  (when (string-match "^ORG-\\(.*\\)-\\(START\\|END\\)$" line)
    (message "%s" (match-string 1))
    (when (equal (match-string 2 line) "START")
      (org-close-par-maybe)
      (insert "\n<div class=\"" (match-string 1 line) "\">")
      (org-open-par))
    (when (equal (match-string 2 line) "END")
      (org-close-par-maybe)
      (insert "\n</div>")
      (org-open-par))
    (throw 'nextline nil)))
#+end_src