Module talk:Strip to numbers

Current problems

Fixed

Obvious known issues:

Matches all numbers, hyphen-minuses, and dots, and will:
- produce an invalid number from input like font-size: -10% (resolves to --10) or approx. 53.5 (resolves to .53.5)
- the opposite of the desired number for cases like margin-left: 2.5em (resolves to -2.5)
- probably-undesired concatenation from input like 42 chickens on 12 farms (resolves to 4212)

It needs to only match on - and . when preceding a numeral, and to stop parsing and return a value when it reaches the first end of numeric data. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 10:35, 18 July 2015 (UTC)[reply]

WOSlinker's rewrite fixed all this. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 16:53, 18 July 2015 (UTC)[reply]

For later development

A more advanced version that could be invoked directly or required by other modules, might do something like this:

For any arbitrary input, trim all leading material until it hits any of:
- a numeral; or
- a . followed by a numeral; or
- a - or the proper negative/minus glyph − followed by either of:
  - a numeral, or by
  - . followed by a numeral
then retain that character;
proceed to next character, and retain it if matches either:
a numeral, or
a ., unless one was matched earlier;
repeat until that fails (i.e. a second . is found, or any other non-numeral is found);
then trim everything after that;
and do all this for multiple values passed,
and even do it for multiple numbers found in the same value (or discard any found after the first, or something)
In the division function:
- Accept arbitrary divisors;
- Round the results to arbitrary decimal places if necessary (default: two, as in 23.48).

A later super-badass version could add:

recognize ^ or e, and x or × or * when found in a context that indicates an exponent (23.5x10^8), and a few other such cases (e.g. characters used to indicate a truncated long/endless decimal)
find multiple numbers per input string, and separate them by something (e.g. a space or a comma) depending on which function is invoked
recognize simple formulae
recognize English words for numbers and convert to use digits

— SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 10:35, 18 July 2015 (UTC)[reply]