ox-html: Bug or feature for export of title and meta information?

ox-html: Bug or feature for export of title and meta information?

[Jens Lechtenboerger]

Hi there,

I wonder about the treatment of different pieces of HTML header
information in org-html--build-meta-info: Contents for the title
element are computed by org-export-data.  This interprets Org markup
such as italics or hyperlinks although title elements are not
allowed to contain the resulting sub-elements.  Why is
org-export-data applied?  Why not produce plain text?

Contents for the description and keywords meta attributes are
computed with org-html-encode-plain-text and replacement of
quotation marks.  Seems fine to me.

Contents for the author meta attribute are computed with
org-element-interpret-data, before org-html-encode-plain-text and
quotation mark replacement are applied.  How would author
information look like that benefits from this more complex treatment
(compared to description and keywords)?

Thanks
Jens

Re: ox-html: Bug or feature for export of title and meta information?

[Bastien]

Hi Jens,

it is difficult to justify existing code, it is easier to answer to
bug reports.  If you think there is a bug in this aread, can you tell
what it is?

https://orgmode.org/org.html#Feedback might help.

Thanks!

-- 
 Bastien

Re: ox-html: Bug or feature for export of title and meta information?

[Jens Lechtenboerger]

On 2020-02-11, at 09:16, Bastien wrote:

> Hi Jens,
>
> it is difficult to justify existing code, it is easier to answer to
> bug reports.  If you think there is a bug in this aread, can you tell
> what it is?

Hi Bastien,

many thanks for following up on this!

I see two or more bugs:

1. Export to HTML when the title contains markup, say this one:

#+TITLE: This does *not* work

The HTML title element contains a nested b element, which is
invalid as only text is allowed.

2. The documentation does not explain which headers are exported how
to HTML.  I described different variants in my previous e-mail.
I’m not sure whether there was a reason to treat them differently.
So, this is at least a bug in the documentation (with potentially
more bugs in the code if the current treatment was not desirable).

Best wishes
Jens

Re: ox-html: Bug or feature for export of title and meta information?

[Nicolas Goaziou]

Hello,

Jens Lechtenboerger <jens.lechtenboerger@fsfe.org> writes:

> 1. Export to HTML when the title contains markup, say this one:
>
> #+TITLE: This does *not* work

What does not work?

> The HTML title element contains a nested b element, which is
> invalid as only text is allowed.

Who said that? The above is perfectly valid.

> 2. The documentation does not explain which headers are exported how
> to HTML.

Some keywords have are more important than others, since they have
a special meaning during export, e.g., TITLE vs. RANDOM_FOO.

General export rules are described in ox.el (org-export-options-alist),
then each export back-end define its own specificities (look at the
back-end definition).


Regards,

-- 
Nicolas Goaziou

Re: ox-html: Bug or feature for export of title and meta information?

[Nicolas Goaziou]

Completing myself,

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> General export rules are described in ox.el (org-export-options-alist),
> then each export back-end define its own specificities (look at the
> back-end definition).

This doesn't mean the manual cannot be improved, of course. Improvements
are welcome in that area.

Re: ox-html: Bug or feature for export of title and meta information?

[Bastien]

Hi Nicolas,

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> This doesn't mean the manual cannot be improved, of course. Improvements
> are welcome in that area.

I see org-export-options-alist is not mentioned in the manual - would
it be worth mentioning it in the "Adding Export Back-ends" section?

-- 
 Bastien

Re: ox-html: Bug or feature for export of title and meta information?

[Nicolas Goaziou]

Hello,

Bastien <bzg@gnu.org> writes:

> I see org-export-options-alist is not mentioned in the manual - would
> it be worth mentioning it in the "Adding Export Back-ends" section?

I don't think so. It is a constant, which may only matter when you write
an export-backend. Its value is not even particularly important since
any back-end can ignore it. Only its syntax really matters, as pointed
out in, e.g., `org-export-define-derived-backend' docstring. To sum it
up, I see it is an implementation detail.

The manual might, however, be more rigorous about what keywords each
back-end really interprets. I don't know how, and where, it could fit,
tho.

Regards,

-- 
Nicolas Goaziou

Re: ox-html: Bug or feature for export of title and meta information?

[Bastien]

Hi Nicolas,

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> I don't think so. It is a constant, which may only matter when you write
> an export-backend. Its value is not even particularly important since
> any back-end can ignore it. Only its syntax really matters, as pointed
> out in, e.g., `org-export-define-derived-backend' docstring. To sum it
> up, I see it is an implementation detail.

I see, thanks.

> The manual might, however, be more rigorous about what keywords each
> back-end really interprets. I don't know how, and where, it could fit,
> tho.

I guess a tutorial on how to write a new export backend will give us a
hint on where and how to mention this. I am not sure it belongs to the
manual.

Best,

-- 
 Bastien

Re: ox-html: Bug or feature for export of title and meta information?

[Jens Lechtenboerger]

On 2020-02-14, at 20:31, Nicolas Goaziou wrote:

> Hello,
>
> Jens Lechtenboerger <jens.lechtenboerger@fsfe.org> writes:
>
>> 1. Export to HTML when the title contains markup, say this one:
>>
>> #+TITLE: This does *not* work
>
> What does not work?
>
>> The HTML title element contains a nested b element, which is
>> invalid as only text is allowed.
>
> Who said that? The above is perfectly valid.

Good morning (over here),

the W3C Markup Validator [1] disagrees with this output from
ox-html (which I reduced to the essential parts):

#+begin_src html
<?xml version="1.0" encoding="utf-8"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en">
<head>
<title>This does <b>not</b> work</title>
</head>
<body><p></p></body></html>
#+end_src

Best wishes
Jens

[1] https://validator.w3.org/#validate_by_input

Re: ox-html: Bug or feature for export of title and meta information?

[Nicolas Goaziou]

Hello,

Jens Lechtenboerger <jens.lechtenboerger@fsfe.org> writes:

> the W3C Markup Validator [1] disagrees with this output from
> ox-html (which I reduced to the essential parts):

OK. I was talking about the Org syntax. I misread your initial report.

I won't comment about the generated HTML. If needed, the HTML exporter
can prevent parsing TITLE keyword.

Regards,

-- 
Nicolas Goaziou

Re: ox-html: Bug or feature for export of title and meta information?

[Jens Lechtenboerger]

Hi there!

On 2020-02-15, at 15:02, Nicolas Goaziou wrote:

> Jens Lechtenboerger <jens.lechtenboerger@fsfe.org> writes:
>
>> the W3C Markup Validator [1] disagrees with this output from
>> ox-html (which I reduced to the essential parts):
>
> OK. I was talking about the Org syntax. I misread your initial report.
>
> I won't comment about the generated HTML. If needed, the HTML exporter
> can prevent parsing TITLE keyword.

I do not know what is needed.  Function org-html--build-meta-info
should return valid XHTML (also in case of HTML5, I believe that
Org syntax is preferable over ignored [1] HTML tags).

Based on the treatment of meta elements for author and description
in that function, alternatives might use org-element-interpret-data
(author) or not (description).  I do not understand the role of
org-element-interpret-data to generate author information.  Which
“non exportable objects” can be skipped by that function (as
mentioned in a comment in org-html--build-meta-info)?  Should they
also be skipped for description or title?

Best wishes
Jens

[1] https://developer.mozilla.org/en-US/docs/Web/HTML/Element/title

Re: ox-html: Bug or feature for export of title and meta information?

[Nicolas Goaziou]

Hello,

Jens Lechtenboerger <jens.lechtenboerger@fsfe.org> writes:

> Based on the treatment of meta elements for author and description
> in that function, alternatives might use org-element-interpret-data
> (author) or not (description).  I do not understand the role of
> org-element-interpret-data to generate author information.  

`org-element-interpret-data' means: "down with HTML, go back to Org
syntax". IOW, it mean the author is supplied as-is.

> Which “non exportable objects” can be skipped by that function (as
> mentioned in a comment in org-html--build-meta-info)? Should they also
> be skipped for description or title?

That non-exportable part is confusing. I think

  (org-element-interpret-data auth)

is sufficient. I pushed a change in that direction.

Regards,

-- 
Nicolas Goaziou

Re: ox-html: Bug or feature for export of title and meta information?

[Jens Lechtenboerger]

Hi there!

On 2020-02-17, at 10:47, Nicolas Goaziou wrote:

> Jens Lechtenboerger <jens.lechtenboerger@fsfe.org> writes:

>> Which “non exportable objects” can be skipped by that function (as
>> mentioned in a comment in org-html--build-meta-info)? Should they also
>> be skipped for description or title?
>
> That non-exportable part is confusing. I think
>
>   (org-element-interpret-data auth)
>
> is sufficient. I pushed a change in that direction.

Thank you!

The function org-element-interpret-data seems to return the empty
string for nil.  Is that by contract or accident?  In the former
case, maybe use

(org-element-interpret-data (plist-get info :author))

instead of the let statement?

What do you think about applying org-element-interpret-data (instead
of org-export-data) when let-binding title, like the following?

(org-html-encode-plain-text
 (org-element-interpret-data (plist-get info :title)))

As far as I can tell, this would create valid (X)HTML.

Best wishes
Jens