Suggestion: Roman numerals support

[uliska]

I would like to add support for roman numerals because that's a common idiom for indexing variable names without arabic numbers.

I have added a basic module with the conversion routines int_to_roman() and roman_to_int() to my fork (https://github.com/uliska/frescobaldi), and invite anybody who is interested in this to pull the romannumerals branch to check it out.

Create a new snippet with this text (updated incorporating the later commits):

-*- python; selection: yes, keep;

from romannumerals import int_to_roman

number = int(text)
newtext = []

for i in range(1, number + 1):
  newtext.append(int2roman(i) + ' = \\relative c\' {\n\n}\n\n')

text = newtext

enter and select a (preferrably small) positive integer in an empty document and run the snippet.

This is only a demonstration to get the idea, there is no error checking etc.

---

Before continuing I would like to get some feedback if it's interesting and worth the effort. And I'd like to get some ideas/suggestions for functions to be implemented.

As a start here are a few ideas:

• Create a set of music variables, similar to the above example but with real names:
  Input: name+int (e.g. melody6)
  Output: music blocks with names melodyI upt to melodyVI
• Split a music definition to two at the cursor
  (I need to do this sometimes with transpositions that need enharmonic changes somewhere)

• Create a series of variable uses to concatenate a series of indexed variables like

  concat = {
  \I % or \melodyI
  \II
  \III
  \IV
  %etc.
  }

[wbsoft]

A good idea indeed. As of the implementation, there is already frescobaldi_app/ly/util.py which has int2text, int2roman and int2letter. The score wizard also uses this functionality. Maybe it is nice to add the conversion routine roman2int there.

[uliska]

Oops. Good that I didn't implement this myself :-) I'm afraid such things will happen more often because of course I don't know the sources intimately ...

For now I'll keep the stuff in my additional module to see where it leads to (and how much it will become). In the end I can incorporate it in an existing file or not.

Have removed int_to_roman and renamed roman_to_int into roman2int. Unfortunately there are now two parallel structures _roman_numerals (in util.py) and _roman_letters (in romannumerals), but I have the impression that merging them would make one of the two algorithms much more complicated.

Also added a few helper routines.

[uliska]

Example snippet (working with the latest commits from romannumerals):

-*- python; selection: yes;

from romannumerals import *

base, index = romansuffix2int(text)
if index >= 0:
    text = base + int2roman(index+1)

select a variable name like melodyII and run the snippet. Also test with melody and melodyXD (invalid numeral).

This isn't a real-world application but an example where this could be going.