Magogenie questions(related to MathML) not rendering properly onto content curation server about studio

[harishbalachandran]

@rtibbles @aronasorman can you put a ETA on this? Its a blocker for the MG channel.

[rtibbles]

This should be achievable using a combination of https://github.com/transpect/mml2tex and then leveraging the xslt file in Python: http://stackoverflow.com/questions/16698935/how-to-transform-an-xml-file-using-xslt-in-python#answer-16699042

[jamalex]

The Perseus exercise renderer does not support MathML. Math formulas will need to be converted into latex format, and then inserted into the exercise text surrounded by $ symbols.

For example, if you put in $ f(x) = 2 ^ 3 $, it will be rendered as:

(the tools that @rtibbles mentiones above may be helpful for doing this in an automated way)

[yogeshmhaskule]

@rtibbles @aronasorman @jamalex For more clarification, will this issue be handled in ricecooker or we need to handle it in our code (i.e sushi chef). I remember @aronasorman @jayoshih had taken care of the same issue before in ricecooker itself.

[jayoshih]

@yogeshmhaskule See #222 You'll need to convert it to latex to bypass the html escaping using the tools @rtibbles mentioned

[yogeshmhaskule]

@jayoshih @aronasorman If you provide sample example of convert mathml data to latex in "Sample program file". So we can take reference from it. It will helpful to us to move forward.

[jamalex]

Again, as this isn't something we support in the ricecooker, it's the responsibility of the sushi chef to convert into usable formats. You can use whatever tools you want to do the conversion, before passing it into the ricecooker.

Here's an example:

• Install xsltproc (e.g., on Ubuntu, sudo apt-get install xsltproc)
• Download and unzip these xslt specifications
• Run a file with MathML in it through the converter:

> cat test.mml
<math xmlns="http://www.w3.org/1998/Math/MathML">
  <mrow>
    <mfrac>
      <mrow><mi>x</mi></mrow>
      <mrow><mi>y</mi></mrow>
    </mfrac>
  </mrow>
</math>

> xsltproc mmltex.xsl test.mml > test.tex

> cat test.tex
$ \frac{x}{y}$

You can either use a tool like that and run it from Python via subprocess, or you may be able to plug those xslt definitions into the tool @rtibbles linked to do it directly within Python.

[yogeshmhaskule]

@jamalex @aronasorman I went through the xsltproc and it's working for me. But few of mathml questions are not rendering properly onto kolibri. This issue is specially to mathml which contains text. Latex conversion of same question is $ Convert the improper fraction into whole number.\phantom{\rule{0ex}{0ex}}\frac{12}{3}$. And same data shown on cc server. After importing the questions onto kolibri. It removes the white spaces between question text (Converttheimproperfractionintowholenumber.).

Screenshot of cc

Screenshot of kolibri

[jamalex]

@yogeshmhaskule Can you post the full source HTML & MathML for that source question into the issue here? It seems as if the first sentence should be left outside the $, as it's not a math formula.

[yogeshmhaskule]

@jamalex Below is the response of same question.

{
	'48113': {
		'success': True,
		'question': {
			'answer_type': 'number',
			'id': 48113,
			'topic_id': 2890,
			'unit': None,
			'difficulty_level': 1,
			'content': '<math xmlns="http://www.w3.org/1998/Math/MathML"><mi>C</mi><mi>o</mi><mi>n</mi><mi>v</mi><mi>e</mi><mi>r</mi><mi>t</mi><mo>&#160;</mo><mi>t</mi><mi>h</mi><mi>e</mi><mo>&#160;</mo><mi>i</mi><mi>m</mi><mi>p</mi><mi>r</mi><mi>o</mi><mi>p</mi><mi>e</mi><mi>r</mi><mo>&#160;</mo><mi>f</mi><mi>r</mi><mi>a</mi><mi>c</mi><mi>t</mi><mi>i</mi><mi>o</mi><mi>n</mi><mo>&#160;</mo><mi>i</mi><mi>n</mi><mi>t</mi><mi>o</mi><mo>&#160;</mo><mi>w</mi><mi>h</mi><mi>o</mi><mi>l</mi><mi>e</mi><mo>&#160;</mo><mi>n</mi><mi>u</mi><mi>m</mi><mi>b</mi><mi>e</mi><mi>r</mi><mo>.</mo><mspace linebreak="newline"/><mfrac><mn>12</mn><mn>3</mn></mfrac></math>'
		},
		'possible_answers': [{
			'question_id': 48113,
			'content': '4',
			'id': 114224,
			'is_correct': True
		}]
	}
}

Here is the another mathml question where xsltproc working properly also it shows the content properly onto kolibri.

{
	'100755': {
		'success': True,
		'question': {
			'answer_type': 'number',
			'id': 100755,
			'topic_id': 2890,
			'unit': None,
			'difficulty_level': 1,
			'content': 'Convert <math xmlns="http://www.w3.org/1998/Math/MathML"><mfrac><mn>30</mn><mn>3</mn></mfrac></math> into a whole number.'
		},
		'possible_answers': [{
			'question_id': 100755,
			'content': '10',
			'id': 192953,
			'is_correct': True
		}]
	}
}

Screenshot of working mathml conversion onto kolibri

[jamalex]

The MathML in the first example doesn't seem quite proper. It splits each letter of the non-math content into separate symbols. When I render it using an online MathML renderer, I get:

The format of the second example (which renders properly) looks correct -- non-math text is outside of the <math> block.

[harishbalachandran]

@jamalex There are quite a few questions on MagoGenie front, that are of the first type (The entire question with the text declared as Math ML). Is there a way to work around for these set of questions.

For the second one, its not a problem since only the equations or the formulas are declared as MathML and we can import them without road block.

[jamalex]

It will probably require a pre-processing stage that detects and extracts the non-math text from the MathML block, before going through xsltproc. Depending on how consistently this content is marked up, it may be possible to do this with regular expressions, or else through parsing with an XML parser like lxml or maybe BeautifulSoup.

To more easily test final output and how it will look in Kolibri, you can use this tool:
http://khan.github.io/perseus/
(Put the output text in the Question box, and click "view rendered")