[CD Reduce Toil] Auto-translate/handle translation cards that only consist of digits

[sagangwee]

With the enabling of translating interactions, multiple choice questions can be translated, but for math lessons, frequently a multiple choice option will be only a digit (see below screenshot) that doesn't need to be translated. We should be handling these types of contents automatically in some way.

Additionally, it may be helpful to group all the choices together in one translation card.

[seanlip]

Just a note -- not all numbers can be systematically translated (but whole numbers should be OK). I just learned that 3.67 is 3,67 in several countries, and vice versa.

[seanlip]

(We can probably handle those automatically too but just noting that there's a little subtlety here.)

[sagangwee]

Oh interesting, thanks! I'll make sure to address that case.

[sagangwee]

Job request to run the ExplorationOpportunitySummaryModelRegenerationJob for regenerating the exploration opportunity models (to update the translation and content counts) submitted.

[sagangwee]

We probably want to do this for unicode translations and also account for the case where the translation is wrapped in an HTML tag, e.g. <p>1000</p>

[sagangwee]

Some feedback from translators:

"In Portuguese (and for other languages) the whole numbers (not the decimals) and the operators and equations are exactly the same as in English; Having to literally just copy and paste from the original text, and than using the reviewers time to review something that didn't even need to be translated, makes the whole process more slow and time/energy consuming (ex: 45, 3 + 5, 6 * 6, 6 - 8 + 6 -- the division symbol "/" however is more commonly used as ÷ in Brazil;"

[seanlip]

Just a note -- there's some follow-up work here: see https://github.com/oppia/oppia/commit/9579f238656dbe453b20db6aa8792350ad11fed5#r87013502

[sagangwee]

@qinghaoyang Do you mind dropping a comment outlining the work that's left here?

[qinghaoyang]

@sagangwee Here is the remaining work.

1. Auto-translate math expressions for some languages. For example, math expressions in Portuguese are the same as in English (https://github.com/oppia/oppia/issues/13055#issuecomment-1071419701).

2. Currently, the HTML tags are stripped to check if cards only consist of digits. We should not strip tags including meaningful content, e.g., <img/> (https://github.com/oppia/oppia/commit/9579f238656dbe453b20db6aa8792350ad11fed5#r87013502).

3. Group all the choices together in one translation card.

[seanlip]

Hi @qinghaoyang @sagangwee -- I don't think we should do (3). Otherwise we'll end up having some multiple-choice options all grouped in one translation card, and others in separate translation cards. Could we drop that part of this issue?

[sagangwee]

Yup, let's drop 3.