Ignoring $ $ in buttons and images

[brankok]

Helou,

is it possible to convince MathJax to render LaTeX syntax that is written in HTML buttons and in HTML images? For example

<input type="button" value="$x+y$" />
<img src="...." alt="$x+y$" />

Regards,
Branko

[William F Hammond]

Part of the yoga of html (and, more generally, all sgml and xml
languages) is that attribute values cannot contain markup.

So, absent a major shaking of the foundation, the answer is NO.

--
William F Hammond
http://www.albany.edu/~hammond/
Email: gel...@gmail.com

[David Carlisle]

On 27 June 2012 00:42, William F Hammond <gel...@gmail.com> wrote:
> On Tue, Jun 26, 2012 at 6:40 PM, brankok <branko...@gmail.com> wrote:
>> Helou,
>>
>> is it possible to convince MathJax to render LaTeX syntax that is written in
>> HTML buttons and in HTML images? For example
>>
>> <input type="button" value="$x+y$" />
>> <img src="...." alt="$x+y$" />
>>
>> Regards,
>> Branko
>
> Part of the yoga of html (and, more generally, all sgml and xml
> languages) is that attribute values cannot contain markup.
>
> So, absent a major shaking of the foundation, the answer is NO.
>

But javascript shakes that foundation. I think it's fairly safe to
assume any page using mathjax has access to javascript so any
attribute value can be extracted at will and stuck in some div
somewhere and rendered. The div could be styled like a button and
react to the same events as the original button.



David

[William F Hammond]

> ...

>>> is it possible to convince MathJax to render LaTeX syntax that is written in
>>> HTML buttons and in HTML images? For example
>>>
>>> <input type="button" value="$x+y$" />
>>> <img src="...." alt="$x+y$" />

>> ...

>> Part of the yoga of html (and, more generally, all sgml and  xml
>> languages) is that attribute values cannot contain markup.
>>
>> So, absent a major shaking of the foundation, the answer is NO.
>>
>
> But javascript shakes that foundation. I think it's fairly safe to
> assume any page using mathjax has access to javascript so any
> attribute value can be extracted at will and stuck in some div
> somewhere and rendered. The div could be styled like a button and
> react to the same events as the original button.

The design of markup organizes the way it is processed. Chaotic
markup leads to chaotic processing.

I suppose mathjax could do something like this, but are you saying it should?
That is, are you saying it should search for paired '$' chars inside
quoted attribute value strings ...

Notwithstanding the fact that for the particular example some
conventional markup might accomplish the purpose intended, perhaps
markup reminiscent of

style="color: rgb(555,777,999);"

like

<input type="button" value="mathjax($x+y$)" />

would be tolerable markup, not that I'm suggesting it.

-- Bill

[Frédéric WANG]

Branko,

There is a "button" example at the bottom of the page here:

http://www.mozilla.org/projects/mathml/demo/basics.html

(not sure it is described in any spec, but this page is valid HTML5). So
you can probably do

<button>
$x+y$
</button>

(edit the skipTags tex2jax configuration parameter if necessary)

I'm not sure what you want to do with your image example. The "alt"
parameter is supposed to provide alternate text for an image not a
complex math formula. Do you mean to use a formula as a tooltip instead
(title attribute)? MathML has maction@actiontype=tooltip for that and
there is a MathJax extension "action.js" to write <maction> from LaTeX.
But I'm not sure how they can be mixed with HTML.

BTW, <input value="$x+y$" />

would essentially be

<maction actiontype="input">
<mi>x</mi><mo>+</mo><mi>y</mi>
</maction>

However, maction@actiontype=input is not supported I think. Many people
have asked this kind of editing feature and many people have been able
to write such a MathJax-based editor. We have a basic example here
http://cdn.mathjax.org/mathjax/latest/test/sample-dynamic.html. So I
think it would not be too difficult to implement
maction@actiontype=input and allow it in "action.js".

Frédéric Wang
maths-informatique-jeux.com/blog/frederic

[Frédéric WANG]

On 27/06/2012 09:30, Frédéric WANG wrote:
> Branko,
>
> There is a "button" example at the bottom of the page here:
>
> http://www.mozilla.org/projects/mathml/demo/basics.html
>
> (not sure it is described in any spec, but this page is valid HTML5).
> So you can probably do
>
> <button>
> $x+y$
> </button>
>
> (edit the skipTags tex2jax configuration parameter if necessary)
>
> I'm not sure what you want to do with your image example. The "alt"
> parameter is supposed to provide alternate text for an image not a
> complex math formula. Do you mean to use a formula as a tooltip
> instead (title attribute)? MathML has maction@actiontype=tooltip for
> that and there is a MathJax extension "action.js" to write <maction>
> from LaTeX. But I'm not sure how they can be mixed with HTML.

As an example, the page I mentioned above uses images in formula using a
background-image property on a <mspace> (maybe that can be done via
<mglyph> but Firefox does not implement it and <mglyph> is not really
intended to be the equivalent of <img> anyway).

MathML can be included in HTML and SVG:

http://www.mozilla.org/projects/mathml/demo/extras.html

but AFAIK there isn't a standard way to do the reverse inclusion
(although I think some people use <semantics> to do that).

--
Frédéric Wang
maths-informatique-jeux.com/blog/frederic

[David Carlisle]

On 27 June 2012 08:37, Frédéric WANG <fred...@free.fr> wrote:

> MathML can be included in HTML and SVG:
>
> http://www.mozilla.org/projects/mathml/demo/extras.html
>
> but AFAIK there isn't a standard way to do the reverse inclusion (although I
> think some people use <semantics> to do that).
>

MathML3 suggests that the token elements are extended in any compound
document type to allow the host markup.
HTML5 does this and most inline html is allowed inside <mtext> so
<mfrac><mtext><svg>....</svg></mtext><mtext><img...</mtext></mfrac>
is valid html5 with html and svg in mathml.

Of course, saying it is valid isn't necessarily the same as saying it
works everywhere:-)

David

[Davide P. Cervone]

MathJax doesn't allow the mixing of HTML or SVG within MathML (except
within <semantics> elements). Perhaps it can be done in the future,
but currently it isn't.

Davide

[Davide P. Cervone]

Fred beat me to the <button> suggestion. That should work for you,
and does not require any change to skipTags.

@Fred, you are right that actiontype="input" isn't implemented, mostly
because it is difficult to support in all input modes. For example,
in NativeMML mode it would require the native MathML implementations
to support it, and I don't think any do yet. For SVG output, it would
require the use of foreignObject in order to embed some HTML form
elements within the SVG, but IE doesn't support foreignObject in its
implementation (and apparently isn't going to), and as I recall there
were problems using it in Opera. So it is not clear to me how easy it
would be to add support for actiontype="input" at this point.

There is an experimental extension at

https://github.com/leathrum/mathjax-ext-contrib/tree/master/forminput

that provides this type of functionality in the browsers that support
it.

Davide

[Frédéric WANG]

On 27/06/2012 10:30, David Carlisle wrote:
> MathML3 suggests that the token elements are extended in any compound
> document type to allow the host markup. HTML5 does this and most
> inline html is allowed inside <mtext> so
> <mfrac><mtext><svg>....</svg></mtext><mtext><img...</mtext></mfrac> is
> valid html5 with html and svg in mathml. Of course, saying it is valid
> isn't necessarily the same as saying it works everywhere:-) David

Thank you for the information, David. I think that works in Firefox. The
Mozilla MathML demo pages have recently been rewritten to remove the
invalid markup, but I guess this could now be fixed with the MathML3
syntax you mentioned.

https://bugzilla.mozilla.org/show_bug.cgi?id=700433

--
Fr�d�ric Wang
maths-informatique-jeux.com/blog/frederic

[David Carlisle]

> Thank you for the information, David. I think that works in Firefox. The
> Mozilla MathML demo pages have recently been rewritten to remove the invalid
> markup, but I guess this could now be fixed with the MathML3 syntax you
> mentioned.
>
> https://bugzilla.mozilla.org/show_bug.cgi?id=700433
>
> --

> Frédéric Wang
> maths-informatique-jeux.com/blog/frederic
>


Possibly a bit off topic here but you can validate mathml/svg/html5
documents here:

http://validator.w3.org/nu/



note this is an instance of the validator.nu custom java (but
relaxng based) validator not the classic SGML DTD based validator at
http://validator.w3.org

David

[Frédéric WANG]

On 28/06/2012 13:46, Davide P. Cervone wrote:
> Fred beat me to the <button> suggestion. That should work for you,
> and does not require any change to skipTags.
>
> @Fred, you are right that actiontype="input" isn't implemented, mostly
> because it is difficult to support in all input modes. For example,
> in NativeMML mode it would require the native MathML implementations
> to support it, and I don't think any do yet. For SVG output, it would
> require the use of foreignObject in order to embed some HTML form
> elements within the SVG, but IE doesn't support foreignObject in its
> implementation (and apparently isn't going to), and as I recall there
> were problems using it in Opera. So it is not clear to me how easy it
> would be to add support for actiontype="input" at this point.
>
> There is an experimental extension at
>
> https://github.com/leathrum/mathjax-ext-contrib/tree/master/forminput
>
> that provides this type of functionality in the browsers that support it.
>
> Davide

OK, thanks for the information. Yes, now I believe I remember a message
on this list about Thomas' extension. That looks great.

[Frédéric WANG]

> Possibly a bit off topic here but you can validate mathml/svg/html5
> documents here:
>
> http://validator.w3.org/nu/
>
>
>
> note this is an instance of the validator.nu custom java (but
> relaxng based) validator not the classic SGML DTD based validator at
> http://validator.w3.org
>
> David
>

Yes, I saw this message on another list. That will be useful, thanks.

[Frédéric WANG]

David,

The old mixing HTML-in-MathML demos are back. There is also a new
SVG-in-MathML demo here:

http://www.mozilla.org/projects/mathml/demo/extras.html
http://www.mozilla.org/projects/mathml/demo/basics.html

The pages validate with http://validator.w3.org/nu/ except for the title
attribute that should be changed to maction@tooltip when Firefox
supports it.