How to get human language-specific numerals?

Question

TL;DR: how can I use Hindi digits (١٢٣...) when writing in Arabic?

---

Emacs is arguably hailed as the best editor, but when it comes to using numerals in the context of writing there are no satisfactory solutions to this problem for multilingual users compared to that afforded by Microsoft Word, namely the contextual numeral writing.

In Microsoft Word 2010 go to drop down menu --> More Commands --> Advanced --> under show document content you can select context for Numeral out of (Arabic or Hindi or Context or System). The context option allows you write numerals according to the context of your writing if it were English numerals would be Arabic 1,2,3,4,5,6,7,8,9,0 -- sounds strange! but it is so named (a misnomer maybe), and if you write in Arabic then numerals will be Hindi numerals, written like this:

For Windows, I found that changing the input-method while writing is the way to go if I am to write a multilingual document in Emacs say English and Arabic or any other language with a totally different alphabet. Changing language or keyboard layout from the language bar in Windows IMHO won't give you the same effect since that will destroy all of your key bindings of your Emacs editor (especially in case of two different languages like English and Arabic). The solution is posted here.

In Microsoft Word, after setting numerals to context you can seamlessly write English with Arabic numerals, and when you shift to write in Arabic by cycling through (ALT+R SHFT) you will write Arabic text with Hindi numerals in no time as shown in the image below. In Emacs after shifting the input-method to Arabic language, numerals will still be Arabic (1,2,3,etc) and not the desired Hindi numerals.

I don't think solution to this problems is related to Microsoft Windows as changing the language of regional settings will change the numerals of the Windows interface as well and will be system-wide undesired effect.

I do believe that the extension potential of Emacs holds the key to this problem. To be more specific; I suggest Emacs should detect the input-method and if it is in Arabic it should automatically remap all Arabic digits into Hindi ones. OR somehow customizing the Arabic keyboard layout to have Hindi instead of Arabic numerals.

Listed blow are some related posts to this problem but I couldn't translate these bits of info into an actual solution suitable to the specifics of this post:

https://tex.stackexchange.com/q/191040/26295

Below is an approach to extend the input-method in Emacs. https://emacs.stackexchange.com/a/3404/2443

Typing Hindi numerals can be a tedious task in Emacs if you want to enter the Hindi digits one at a time C-x 8 RET these are mapped in Unicode to the following codes: 0661 for number 1, 0662 for number 2, 0663 for number 3, 0664 for number 4, 0665 for number 5 and so on.

Question
What is the best customization to Emacs 24 and shortest keystrokes in order to replicate the seamless contextual writing experience provided by Microsoft Word?

Answer 1

So the requirement is: When I'm using the arabic input method, and I type a digit, I want Emacs to insert the Hindi character for that digit.

To do this, we can adapt your linked approach to extend an input-method like so:

(eval-after-load "quail/arabic"
  '(let ((quail-current-package (assoc "arabic" quail-package-alist)))
     (quail-define-rules ((append . t)) ;; don't clobber the existing rules
                         ("0" "\u0660")
                         ("1" "\u0661")
                         ("2" "\u0662")
                         ("3" "\u0663")
                         ("4" "\u0664")
                         ("5" "\u0665")
                         ("6" "\u0666")
                         ("7" "\u0667")
                         ("8" "\u0668")
                         ("9" "\u0669"))))

Then when you use C-uC-\ arabic (or just C-\ if it's your default), and type a digit, it inserts the specified translation instead.