how to extract specific character from a string using regex

Instead of using replace, you might want to use match, which will return a match object.

Mozilla Article

Regex Patterns:

For matching values and units:

([\d,\.]+)\s*(g|kg|l|ml)

More units can be added in the last group.

Example:

"20x0,50l (1 l = 1,70 €) zzgl. 3,10€ Pfand".match(/([\d,\.]+)\s*(g|kg|l|ml)/)

returns
(3) ...:
    0: "0,50l" // full match
    1: "0,50" // value
    2: "l" // unit
    ...

For matching just units (though this is kind of unnecessary, the previous regex matches values in group 1, units in group 2 at the same time):

(?<=[\d,\.]+)\s*(g|kg|l|ml)

For matching quantities:

([\d,\.]+)(?:x|\*)

Example:

"20x0,50l (1 l = 1,70 €) zzgl. 3,10€ Pfand".match(/([\d,\.]+)(?:x|\*)/)

returns
(2) ...:
    0: "20x" // full match
    1: "20" // quantity
    ...

EDIT: to further elaborate on my comment

var units = ["g", "kg", "l", "ml"];
var re = new RegExp(`([\\d,\\.]+)\\s*(${units.join("|")})`);

Then using re for matching:

"20x0,50l (1 l = 1,70 €) zzgl. 3,10€ Pfand".match(re)

Works the same, but is more maintainable.

You could use 2 capturing groups to extract the units and optional 20

(?:(\d+)x\d+(?:,\d+)?)?\d+(k?g|g|m?l)

Capture group 1 optionally matches the quantity like 20 if present, capture group 2 matches the unit.

Explanation

• (?: Non capture group to make the whole part optional
  • (\d+)x Capture group 1, match 1+ digits followed by x
  • \d+(?:,\d+)? Match 1+ digits and an optional decimal part
• )? Close group and make it optional
• \d+ Match 1+ digits
• (k?g|g|m?l) Capture group 2, match any of the listed alternatives

Regex demo

const regex = /(?:(\d+)x\d+(?:,\d+)?)?\d+(k?g|g|m?l)/g;
[
  "0,5g",
  "500l",
  "1000kg",
  "20,5ml",
  "20x0,50l (1 l = 1,70 €) zzgl. 3,10€ Pfand",
].forEach(s => console.log(Array.from(s.matchAll(regex), m => [m[1] ? m[1] : "", m[2]])));

Something like this might do, extracting units, quantities, and doses:

const convert = (input) => {
  const match = /(?:(\d+)x)?\s*([\d\,.]+)([a-z]+)/i .exec (input)
  return match
    ? {
        quantity: Number(match [2].replace(/\,/g, '.')), 
        unit: match [3], 
        ...(match [1] ? {doses: Number(match [1].replace(/,/g, '.'))} : {})
      }
    : {}
}

const inputs = ['500g', '0,5g', '500l', '1000kg', '20,5ml', 
                '20x0,50l (1 l = 1,70 €) zzgl. 3,10€ Pfand']

inputs .forEach (
  input => console .log (`"${input}" --> ${JSON.stringify(convert(input))}`)
)

.as-console-wrapper {max-height: 100% !important; top: 0}

Note that the number handling is naïve. It just looks for any combination of digits, commas and periods, which means that it might also accept '12,345,6.7.8'. I'm sure we could fix that up if it was an issue for your data.

Our regex looks like this:

   /(?:(\d+)x)?\s*([\d\,.]+)([a-z]+)/i
//   \_______/ \_/ \______/ \_____/\_/
//       |      |     |        |    +--- Case insensitive, accepts 'KG' as well as 'kg'
//       |      |     |        +-------- Capturing group for units, composed of letters.
//       |      |     +----------------- Capturing group for quantity, composed of
//       |      |                        digits, commas, and periods.
//       |      +----------------------- Optional space after dosage
//       +------------------------------ Optional non-capturing group with 
//                                         - a capturing group of digits
//                                         - the literal character `x1```

Note, an earlier version, which didn't deal with doses and didn't convert quantities and doses to numbers, was simpler:

const convert = (input) => {
  const match = /([\d\,.]+)([a-z]+)(?![a-z0-9])/i .exec (input)
  return match
    ? {qty: match [1], unit: match [2]}
    : {}
}

And of course, if we simply want the units, we could so something like this:

const convert2 = (input) => /(?:(?:\d+)x)?\s*(?:[\d\,.]+)([a-z]+)/i .exec (input) [1]

const inputs = ['500g', '0,5g', '500l', '1000kg', '20,5ml', 
                '20x0,50l (1 l = 1,70 €) zzgl. 3,10€ Pfand']

inputs .forEach (
  input => console .log (`"${input}" --> ${JSON.stringify(convert2(input))}`)
)

.as-console-wrapper {max-height: 100% !important; top: 0}