Symbols inside numbers in currencies

[Ekleog]

Hello,

So I am trying to add the Yen as a currency in hledger. I thus put this:

commodity 円
    format 1億0,000万0,000円

which is the common way of writing Yen in japanese.

However, hledger then replies:

unexpected '0'
expecting end of input, newline, or the rest of right-symbol amount

As hledger appears to support UTF-8 pretty well (as eg. a format 1,0000,0000円 works), I guess this is due to having a non-digit non-dot-or-comma symbol inside the number.

Is there a way to have symbols other than dot-and-comma be considered for inclusion as digit separators?

As usual, thank you for your work!

[simonmichael]

Wow. :) Yes your guess is correct. Where could we learn more about this usage, the allowed symbols and their meanings etc, eg on wikipedia ?

[simonmichael]

/me reads https://en.wikipedia.org/wiki/Japanese_numerals / http://www.guidetojapanese.org/learn/grammar/numbers .. mind blown.

I would be glad to have more support for this in hledger, if: a developer steps up to design and build it, verify all tests pass, and stick around to resolve the issues that will show up later.

[Ekleog]

Hmm, according to https://ja.wikipedia.org/wiki/%E5%86%86_(%E9%80%9A%E8%B2%A8)#%E5%86%86%E3%81%AE%E6%B5%81%E9%80%9A%E9%AB%98 the total amount of currency of Yen is 96兆0,377億円 ; so I guess for Yen it should be enough to add 兆, 億 and 万 as dot-and-comma-like symbols.

Do you think this would be an acceptable solution, or that it'd be better to have something more generic, in order to accommodate other currencies that may have “weird” (non-pejorative menaing) numbering schemes?

[simonmichael]

That sounds ok, though it seems unlikely to be the last word on the matter.

[alerque]

Having symbols other than comma, period, and space inside the currency is not unique to the Yen. Swiss Francs would be another example.

I just tried to import my books from a year when I had some extra travels (from Money Manager Ex) and ran into lots of issues with currencies. I even thought I could make it work by adopting only ISO-4217 symbols, but that's a not quite a go either. Basically it doesn't seem possible to get properly localized currency displays yet.

• KZT (Kazakhstan Tenge) adopts spaces for thousands separators only if there are ten or more thousands (₸1000,00 but ₸10 000,00). Space separators work, but it's all or nothing.
• BGN (Bulgarian Lev), TND (Tunisian dinar), and CHF (Swiss Franc) all officially have a dot in their currency symbol (лв., د.ت, and Fr.). None of these are acceptable to hledger at all.
• CHF uses an apostrophe as a thousands separator (Fr.1'000.00). This flunks on two counts (period in symbol and apostrophe in number).
• I'm not done going through old records yet...

[simonmichael]

Nice additional examples, thanks. This remains a tricky problem that will require a champion [and careful cost-benefit analysis].

[alerque]

Another example @bradyt dredged up, the Cape Verdean escudo:

Amounts are generally written by using the $ as the decimal separator, such as 20$00 for 20 escudos, or 1.000$00 for 1000.

Genius. :warning:

[alerque]

Just for future reference, here are some things related to this issue:

• currencyFormatter.js — a Javascript library with localized number formats for most world currencies including ones with crazy separator "issues". I realize JS isn't what we're going for but reference implementations might be useful to review, esp. for where they get their format data. Also similar:
  • dinero.js
  • accounting.js
• currency-list — a flat data set in a lot of different formats with localized currency names (not number formats, just language based currency names.