Understanding preprocessing
David Lippman
I've been using and developing with AsciiMath for years, and finally
have the time and interest to look into integrating AsciiMath with
MathJax. I have seen one successful experiment (http://
www2.ph.ed.ac.uk/mathplayground/asciimath-simple-input-demo), but
wanted to clarify a few things before I proceed:
* In the experiment above, dynamic content is being rendered. This is
done by using AsciiMathML to create a math node, then feeding that
math node to MathJax's queue. Is this the most efficient route, or is
MathJax just going to flatten the math node to a string then re-parse
it?
* For static environments, typically the page is processed onload. Is
there good way to extend MathJax with different pre processors,
ideally while being able to use the CDN version of MathJax? Once
again, is having AsciiMath create math nodes an efficient workflow, or
would there be a better, more direct way for AsciiMath to convert the
content into a format usable by MathJax?
Thanks,
David Lippman
Davide P. Cervone
> have the time and interest to look into integrating AsciiMath with
> MathJax. I have seen one successful experiment (http://
> www2.ph.ed.ac.uk/mathplayground/asciimath-simple-input-demo),
Thanks for sharing your example.
> but wanted to clarify a few things before I proceed:
>
> * In the experiment above, dynamic content is being rendered. This is
> done by using AsciiMathML to create a math node, then feeding that
> math node to MathJax's queue. Is this the most efficient route, or is
> MathJax just going to flatten the math node to a string then re-parse
> it?
Certainly this route works, and a number of sites do use this
approach. But as you feared, MathJax does flatten it and re-parse it
(using the browsers XML parsing machinery, so it is reasonably fast
compared to parsing the tags by hand in javascript). So no, it is not
the most efficient approach.
It would be better to have an ASCIIMathML input jax -- one that takes
the ASCIIMathML input language and converts it to MathJax's internal
format. Since the internal format is essentially MathML (as
javascript objects), this would not be all that hard. In fact, it I
think it should be possible to combine your ASCIIMathML parser with
(part of) MathJax's MathML input jax to do that. It is something that
has been on my to-do list for a while, but hasn't made it to the top
yet.
> * For static environments, typically the page is processed onload. Is
> there good way to extend MathJax with different pre processors,
> ideally while being able to use the CDN version of MathJax? Once
> again, is having AsciiMath create math nodes an efficient workflow, or
> would there be a better, more direct way for AsciiMath to convert the
> content into a format usable by MathJax?
You can write your own preprocessors, and it would be possible to load
them from a remote source while using the CDN (as though they were
config files). There is a call to register a function as a pre-
processor, which you could use to hook up your own so that it will
become part of the MathJax typesetting process. See the tex2jax,
mml2jax, or jsMath2jax extensions for examples. Your code could
convert the ASCIIMathML input to a MathML string, which you would
insert into a <script type="math/mml"> tag at the location of the
mathematics.
But I don't think a preprocessor is the right way to do this. It
would work, but has the double-processing problem that you pointed
out. An input jax is the right way to do that. That could also
potentially be loaded as a config file (like the ones loaded as part
of the combined configuration file you are using). You could use the
MathML input jax as a starting, since it already converts a <math>
node and its children to the MathJax internal format.
Hope that helps.
Davide
David Lippman
It took me some tinkering to figure out the internal format, but I'm
happy to say I got something working: http://dlippman.imathas.com/asciimathjax/
The jax is a bit of a mess still - it should probably be restructured
to better fit the object-oriented style of rest of the code, but at
least it works. If you have any interest in integrating it into the
official MathJax package, go for it :) I don't have copyright to the
AsciiMath parser, but it is licensed LGPL if that's good enough. If
not, Peter Jipsen might be willing to relicense if needed.
In the meantime, is there a way to load a local input jax, but use the
rest of the code off the CDN?
David Lippman
Marc Grober
Hash: SHA1
Thank you David!
How soon could we get something in CDN, Davide?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.16 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iQEcBAEBAgAGBQJOSJm4AAoJEMCi9wxDaNqEzwAH/1s3UGljObwqPk8kbH/emzKd
1EehR1knSjdSBNoBo8g3sr6d0hzBgWyxsIBC7npJTgXUomwnVN/dJrr04LVCjhun
/z1anzMzapk9fIkwECPhoC1xXPwkS5NHmivpdWhwowEd3iO/kJnpgIERqDa9ZNfw
B0cB6P5M7Zy4/rA3TgKDefPWtZTNu1tY7upjTLYF8d7gksQg3rBi1Xcrs9U7ZF8t
hA/F/8KBLV/sRXAXirznEOXTf+wholOFToMSd/g0Dg+XhWbQT1DfDhfxKJmqIthg
IhNnSBFI2mZmtEvNgZ+iQ1U7pCo5eZnabGp6z2i5yzWL1zPBvX2koJM01LpEgCg=
=7Edt
-----END PGP SIGNATURE-----
Davide P.Cervone
You made fast work of this, and I'm impressed with the results. As
you say, it needs restructuring (for example, the global variables and
functions would need to be put into the AsciiMath input jax's object,
as you mention, and the pre-processor probably can be simplified).
There also seems to be a problem with the Fractur, Script, and some
other font selections. But this is an excellent start, and well worth
pursuing.
The license may well be a problem. My reading of it (and I certainly
am not a lawyer) is that if modifications are made, it can be used
only in another library that is also LGPL, which MathJax isn't. I
know Peter and will contact him about it to see what he recommends.
It might be possible to package this component separately as LGPL if
nothing else can be worked out.
As for using it with the CDN, that should be able to be done. Here's
how:
Create a file called AsciiMath.js (or some such thing) that you will
place on your own server. Let's say that it is at http://www.myserver.edu/MathJax/AsciiMath.js
(of course, you will change this to suite your situation). Then put
into this file the following lines:
MathJax.Ajax.Preloading(
"[MathJax]/jax/input/AsciiMath/config.js",
"[MathJax]/jax/input/AsciiMath/jax.js",
"[MathJax]/extensions/asciimath2jax.js"
);
and then insert the contents of your config.js, the jax.js and the
ascimath2jax.js files. (I.e., concatenate them to the file containing
the lines above). Then, add
MathJax.Ajax.loadComplete("http://www.myserver.edu/MathJax/
AsciiMath.js");
but use whatever URL you are using for this file. Finally, change the
line
ASCIIMATH.loadComplete("jax.js");
to
MathJax.Callback.Queue(
MathJax.Hub.Register.StartupHook("mml Jax Ready"),
["loadComplete",ASCIIMATH,"jax.js"]
);
in the copy of the jax.js file so that the input jax will not start up
until after the element jax is loaded (which happens later on).
Then put the file in the right place on your server, and call the CDN as
<script type="text/x-mathjax-config">
MathJax.Hub.Config({
extensions: ["asciimath2jax.js"],
jax: ["input/AsciiMath","output/NativeMML"]
});
</script>
<script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js?config=http://www.myserver.edu/MathJax/AsciiMath.js
"></script>
I would recommend that you use the MMLorHTMLjs configuration file as
well, rather than explicitly setting the NativeMML output jax, since
that will not work with many browsers. To do that, use
<script type="text/x-mathjax-config">
MathJax.Hub.Config({
config: ["MMLorHTML.js"],
extensions: ["asciimath2jax.js"],
jax: ["input/AsciiMath"]
});
</script>
<script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js?config=http://www.myserver.edu/MathJax/AsciiMath.js
"></script>
Anyway, that should do it for you.
Thanks for your work on this!
Davide
Davide P. Cervone
As David pointed out, there are licensing issues to be worked out
before anything can be done with this. There are also some structural
issues with the code; David's work is a good first step, but it needs
to be cleaned up a bit before it can become part of MathJax, even if
the licensing can be worked out. I have also noticed at least one
situation that causes an error, so there is a little more work to be
done there as well.
I'm in the process of trying to get version 1.2 together for release
this fall, and there is not much time before I have to go back to
teaching, so most of my effort is going to that. I don't think this
will become part of that release. We are still trying to work out how
to handle external contributions properly, and how to vet them
appropriately. David's work should provide some pressure to push that
process forward.
In the meantime, you can use the process I suggested to David if you
want to use his code unofficially now.
Davide
David Lippman
I have a feeling the font issues are related to Peter's workarounds
for Mozilla's font mappings. Taking out the remapping should help.
I could use some advice from you or someone else who knows Javascript
OO stuff, since I haven't done much with it...
In the code I threw together, all the AsciiMath parser variables and
functions are wrapped inside the big "(function (ASCIIMATH,BROWSER)
{", so doesn't that keep them out of the global scope? Is there a
particular advantage to attaching the variables and methods to the
ASCIIMATH object, rather than doing as currently done, or wrapping the
whole asciimath parser in a closure? (my apologies if that terminology
is misused). I had been considering doing this approach to isolate
the parser:
var AsciiMathParser = (function () {
all the interval asciimath parser variables and methods, then
return function AMparseMath(str) {
}
})();
So that then the code in the jax Parse function would be
this.mml = AsciiMathParser(math);
-David
On Aug 16, 4:52 am, Davide P.Cervone <d...@union.edu> wrote:
> David:
>
> You made fast work of this, and I'm impressed with the results. As
> you say, it needs restructuring (for example, the global variables and
> functions would need to be put into the AsciiMath input jax's object,
> as you mention, and the pre-processor probably can be simplified).
> There also seems to be a problem with the Fractur, Script, and some
> other font selections. But this is an excellent start, and well worth
> pursuing.
>
> The license may well be a problem. My reading of it (and I certainly
> am not a lawyer) is that if modifications are made, it can be used
> only in another library that is also LGPL, which MathJax isn't. I
> know Peter and will contact him about it to see what he recommends.
> It might be possible to package this component separately as LGPL if
> nothing else can be worked out.
>
> As for using it with the CDN, that should be able to be done. Here's
> how:
>
> Create a file called AsciiMath.js (or some such thing) that you will
>
> I would recommend that you use the MMLorHTMLjs configuration file as
> well, rather than explicitly setting the NativeMML output jax, since
> that will not work with many browsers. To do that, use
>
> <script type="text/x-mathjax-config">
> MathJax.Hub.Config({
> config: ["MMLorHTML.js"],
> extensions: ["asciimath2jax.js"],
> jax: ["input/AsciiMath"]
> });
> </script>
Davide P. Cervone
> for Mozilla's font mappings. Taking out the remapping should help.
OK. I haven't looked into the details of how it works, but did note
that you were ending up with empty elements for several of the font
changes (but not for all of them).
> I could use some advice from you or someone else who knows Javascript
> OO stuff, since I haven't done much with it...
> In the code I threw together, all the AsciiMath parser variables and
> functions are wrapped inside the big "(function (ASCIIMATH,BROWSER)
> {", so doesn't that keep them out of the global scope?
I hadn't looked very closely at the code, and so I missed that. You
are right, the closure does avoid the global variable issue that I was
concerned about. But see below.
> Is there a
> particular advantage to attaching the variables and methods to the
> ASCIIMATH object, rather than doing as currently done, or wrapping the
> whole asciimath parser in a closure?
There are two reasons to do it as methods of an object (though neither
is a critical as keeping the global namespace clean, which you have
already done). The first is that the routines and data are NOT
completely inaccessible (as they are in the closure), and so can be
overridden or modified without having to change the input jax itself.
For example, it makes it possible to apply patches to the input jax as
configuration files rather than editing the input jax files, and now
that most people are using the CDN, that is particularly useful, since
they don't have access to modify the original code directly. As an
example, with the TeX input jax, someone complained that \sqrt\frac{a}
{b} gave a parsing error, and so I was able to provide a small patch
that he could include in his configuration that would override the
routine used by \sqrt so that this would work for him. That would not
have been possible if the TeX input jax had used the closure technique
to isolate its internals completely. For your jax, I'm thinking that
the AMsymbols table, for instance, might be something that people
would want to be able to add to, and using the object-oriented
approach, that would be possible, whereas the closure prevents that.
The second reason is that, as an object, your input jax could be
subclassed to provide a similar but modified one. This is less
likely, but still something that has occurred with some of the other
jax (I have done this with output jax for very special purposes).
In any case, it would be good to structure your input jax along the
same lines as the existing ones, if only for consistency.
Hope that makes the issues more clear.
Davide
PS, Peter seems amenable to granting a special license to use the code
in MathJax, though the details haven't been worked out, but it looks
like we will be able to include your jax in the core MathJax
distribution. Very nice!
Murray
that is, the user inputs ASCIIMathML syntax and the browser gets
MathJax and it renders in all browsers.
http://www2.ph.ed.ac.uk/mathplayground/asciimath-simple-input-demo
(Well, of course IE failed, but it was OK on McKain's second page;)
http://www2.ph.ed.ac.uk/mathplayground/asciimath-semantic-input-demo
But your example, David, appears to be ASCIIMathML input and
ASCIIMathML output:
http://dlippman.imathas.com/asciimathjax/
That is, it only renders (for me) properly in FF.
I've read the thread a few times and am not sure what you are trying
to do. Aren't you also trying to achieve ASCIIMathML input and MathJax
output? Or did I miss something?
Regards
Murray
Marc Grober
Hash: SHA1
right click on displayed equation and change output
-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.16 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iQEcBAEBAgAGBQJOU8UiAAoJEMCi9wxDaNqEoLoIALAY4an+kvy3hY8pLTU7Vhtf
gKqJbTmSi9hs594b7k4LU6BKRRbrIN+krgEImI2dwXMKpmUol73PsTBIlKQiMH3u
uUjPBhF0Gh+eNIvw1O9PmGmGgI+Ke+puZgDuI+UQtZjXx1P/dOFaH1q70wqzV4oG
6c50ugYHSxTS2FnbGddRpLazsak/z00jZ7S5UMCWGcUB1fT93aJ19xaQvpy5SXGe
wC33qyKKS288Z5Ym6JZdBC/v/ZNHlH9qzFiDVkh30dvHqYsBiYIbPtpv8txNi5z4
OmRp7mTKyqcuR5GMF5miEkqLYtB7CYlfE8LzKaGeeDifKs8ZV+PGDhQWDqN8E+E=
=/i4q
-----END PGP SIGNATURE-----
David Lippman
of MathJax, I didn't want to have to upload all the image fonts to my
server, so I set the default output to MathML rather than the HTML/CSS
output. That was sufficient for testing whether the jax was working,
but obviously won't work outside of FF.
Davide: I probably won't have time for a few weeks, but will try to
get the jax restructured so it can hopefully eventually make it into
the codebase.
Davide P. Cervone
> get the jax restructured so it can hopefully eventually make it into
> the codebase.
I understand. I go back to teaching next week myself, so things will
slow down with MathJax for a bit.
I'd like to look more closely at what you have done, and will make
further recommendations about structure after that. It may be that in
order to get the license stuff to work out properly, we need to do it
the way you have done (so that the original code is changed as little
as possible).
Davide
Murray
I've "borrowed" David McCain's scripts and produced a page which works
(the same as his) such that the user can enter AsciiMath in a form
text box and out comes MathJax in a nearby DIV. No problem.
But he stated on his page
"Note that my demo only handles ASCIIMath math expressions (i.e. the
bits you would put between `...` in the ASCIIMath Editor demo). It
wouldn't be hard to extend this to do that too."
Well, I'm still trying to do that and several attempts have failed.
This is the page that I'm trying to get to work (and nothing works at
this point, except the JS does render the page, but no output):
http://intmath.com/asciijax
The (gray) text has id="asciiMathInputControl" name="asciiMathInput"
and the output div is
<div id="rawRendering">
<math xmlns="http://www.w3.org/1998/Math/MathML"></math>
</div>
Here is the un-minified JS that it uses:
http://intmath.com/asciijaxjs
I'm hoping there is a simple function that I can add somewhere that
will process AsciiMath within backticks, so I can have a mix if text
and AsciiMath with MathJax output.
Any help would be great.
Regards
mauno
David's files seem to work well in all browsers - even all versions of
IE - if output is set to HTML-CSS, I was testing them with a moodle 2
configuration that looks like this
<script type="text/x-mathjax-config">
MathJax.Hub.Config({
extensions:
jax: ["input/AsciiMath","input/TeX","input/MathML","output/HTML-
CSS"],
tex2jax: {displayMath: [['<?php echo $CFG->filter_math_mathjaxd1; ?
>','<?php echo $CFG->filter_math_mathjaxd2; ?>']], inlineMath: [['<?
php echo $CFG->filter_math_mathjaxi1; ?>','<?php echo $CFG-
>filter_math_mathjaxi2; ?>']] }
});
</script>
<script type="text/javascript" src="<?php echo $CFG->httpswwwroot; ?>/
lib/MathJax/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>
that is using asciimath2jax.js for asciimathml input and configurable
delimiters (by settings of moodle) for latex input and mathml input
but all output is done as HTML-CSS since mathml output has problems
with printing some symbols.
I did not use any meta tags for IE and do not have mathPlayer
installed - there are only a few css issues as far as I can see so
this seems to be a cross browser compatible solution to output
asciimathml through mathjax.
I would like to modify David's jax/input/AsciiMath/jax.js to use more
symbols that are supported by MathJax and basicly I could of course
simply add new symbols to var AMsymbols = [...] in that file but it
would be nice to be able to use php files to add configurable settings
to mathjax files as well - is it possible to use php files instead of
js files or does mathjax always use a standard structure to render
input/output javascript only files?
One workaround is of course to use modified local files like
"localasciimath2jax.js" but ideally I would like to allow moodle users
to select if they want to use files from CDN or local files
Davide P. Cervone
> delimiters (by settings of moodle) for latex input and mathml input
> but all output is done as HTML-CSS since mathml output has problems
> with printing some symbols.
>
> I did not use any meta tags for IE and do not have mathPlayer
> installed - there are only a few css issues as far as I can see so
> this seems to be a cross browser compatible solution to output
> asciimathml through mathjax.
Thanks for the information and details on how you are doing this. It
is nice to see that MathJax is working well with Moodle. I have been
following the discussions on the Moodle forums, and you have certainly
done yeoman service in that respect.
> I would like to modify David's jax/input/AsciiMath/jax.js to use more
> symbols that are supported by MathJax and basicly I could of course
> simply add new symbols to var AMsymbols = [...] in that file
One of my concerns about that current arrangement is that this is not
easy to do via configuration parameters. It is certainly something
that would be nice to be able to do.
> it would be nice to be able to use php files to add configurable
> settings
> to mathjax files as well - is it possible to use php files instead of
> js files or does mathjax always use a standard structure to render
> input/output javascript only files?
I'm not certain I understand the question. Are you asking that
MathJax load its components from a .php file so that the contents of
those file can be generated on the fly by php? I don't understand
what you mean by "use a standard structure to render input/output
javascript only files". Certainly MathJax is set up to load
javascript files when it needs to include additional functionality. I
do not think it is necessary to make these be generated by php,
because MathJax already has methods of modifying its configuration
(and even its code) via the configuration blocks like you have already
used. You can use php to include such configuration in your main file
(as you have done already), and that could whatever modifications you
need for the components of MathJax, in general. (The structure of the
current asciimath implementation prevents this, but it is one of the
things I have suggested be changed about it).
Anyway, I may be misunderstanding your request, but I think the needed
functionality is already available.
> One workaround is of course to use modified local files like
> "localasciimath2jax.js" but ideally I would like to allow moodle users
> to select if they want to use files from CDN or local files
I agree, making locally modified versions of the MathJax files isn't
the way to go.
Davide
Davide P. Cervone
again for me, and my teaching and grading are now my primary focus for
the next few months.
It looks to me like you have used the wrong version of the David's
asciimath code (and it is David Lippman, not David McCain). There is
a version that implements asciimath as an input jax for MathJax, and
includes an asciimath preprocessor to locate the backtics (see http://dlippman.imathas.com/asciimathjax/)
. That is the one you want, and if you use it, you can just use
MathJax.Hub.Queue(["Typeset",MathJax.Hub]);
when you want the page to be processed again. I gave instructions at
http://groups.google.com/group/mathjax-users/msg/704b16901f51ccb3?hl=en
how to package this up for use with the CDN copy of MathJax if you
want to do that rather than host your own installation.
In any case, that should do what you need.
Davide
mauno
Yes, I have been testing lately a sort of extended/modified version of
David Lippman's asciimathjax extension & asciimath input together with
html output (set to prefer for all browsers) although I did try david
McCain's parser too.
This combination of CDN & local config is exactly the thing I was
missing and it's ok to use plain javascript files for mathjax. The
reason why I asked about php files is that in some cases it might be
interesting to allow each site to modify input / output for their
local needs without touching original files - with configurable
settings, database and php. For example if we have users that are not
english speakers and do not even have latin characters in their
keyboard we could have additional language files for input to render
mathematical characters. But like you said it can be done with
javascript as well.
Mauno
On 14 syys, 13:26, "Davide P. Cervone" <d...@union.edu> wrote:
> Sorry for the delay in getting back to you. Classes have started up
> again for me, and my teaching and grading are now my primary focus for
> the next few months.
>
> It looks to me like you have used the wrong version of the David's
> asciimath code (and it is David Lippman, not David McCain). There is
> a version that implements asciimath as an input jax for MathJax, and
>
> - Näytä siteerattu teksti -
David Lippman
issues resolved.
I tried restructuring the AsciiMath parser as an object using the
MathJax object functions, and I just couldn't get it to work. I don't
think I know enough about OOP in Javascript, and particularly the
object functions in MathJax, to be able to make it work without
spending a bunch more time than I'm willing to, so I'm going to call
my contribution to this effort done :) If someone with more
experience wants to restructure it, you're welcome to it.
Davide P. Cervone
from here. Can you package up your current versions and send me an
archive? Thanks.
Davide
Davide P. Cervone
and the code is at
https://github.com/dpvc/MathJax/tree/v2.0-candidate/unpacked/jax/input/AsciiMath