@alordash/parse-word-to-number

$ npm i @alordash/parse-word-to-number

Description

Parses string and returns numbers written as words inside it.
It uses my realization of Damerau-Levenshtein algorithm to properly parse words even if they are written with mistakes.
Supports Russian and English language.

Usage

Function parseWord(string, errorLimit):{Array.<Number>}

Arguments

1. string {String} — source string.
2. errorLimit {Number} — From 0.0 to 1.0, the less — the less results. Used for recognizing words with mistakes.
   Parses all words in that string into numbers.
   Returns all found numbers.

Usage example:

const { parseWord } = require('@alordash/parse-word-to-number');

//Parse single word
let parsedWord = parseWord("twonty-one");
console.log(parsedWord[0].value);
//=> 20
console.log(parsedWord[1].value);
//=> 1

parsedWord = parseWord("читырэ");
console.log(parsedWord[0].value);
//=> 4

//You can specify mistakes multiplication from 0.0 and on with second argument, where
//0 — do not accept words with mistakes,
//1 — accept words if error < error limit for that word
//List of limits for all words is located in /lib/expressions/*.csv files
parsedWord = parseWord("hundrid", 1);
console.log(parserWord[0].value);
//=> 100

parsedWord = parseWord("hundrid", 0);
console.log(parserWord[0]);
//=> undefined

Function parseString(string, errorLimit):{String}

Arguments

1. string {String} — source string.
2. errorLimit {Number} — From 0.0 to 1.0, the less — the less results. Used for recognizing words with mistakes.
   Parses all words in that string into numbers and combines them.
   Returns string with parsed numbers.

Usage example:

const { parseString } = require('@alordash/parse-word-to-number');

console.log(parseString("four-huntred-sevinty-six balloons"));
//=> 476 balloons

console.log(parseString("двести дивяносто пять тысоч ложек сто восмьдесят три тарелки"));
//=> 295000 ложек 183 тарелки

//Mistakes multiplication
console.log(parseString("four-huntred-sevinty-six balloons", 0));
//=> 4 balloons

console.log(parseString("двести дивяносто пять тысоч ложек сто восмьдесят три тарелки", 0));
//=> 200 дивяносто 5 тысоч ложек 100 восмьдесят 3 тарелки

Getting array of ConvertedWords

Class ConvertedWord

class ConvertedWord {
   //@type {String}
   text;                    //Text of word
   //@type {Array.<Number>}
   indexes;                 //Indexes of used words from original string
}

Function arrayParseString(string, errorLimit):{Array.}

Works the same as parseString function and accepts same arguments, except it returns array of converted words.

Usage example:

const { parseString } = require('@alordash/parse-word-to-number');

let result = arrayParseString("four huntred sevinty-six balloons");
console.log(JSON.stringify(result));
//=> [{"text":"476","indexes":[0,1,2]},{"text":"balloons","indexes":[3]}]


result = arrayParseString("двести дивяносто пять тысоч ложек сто восмьдесят три тарелки");
console.log(JSON.stringify(result));
//=> [{"text":"295000","indexes":[0,1,2,3]},{"text":"ложек","indexes":[4]},{"text":"183","indexes":[5,6,7]},{"text":"тарелки","indexes":[8]}]

Adding custom expressions

You can add new expressions for parsing more cases by creating .csv file inside lib/expressions folder. Fill new .csv file with following format:

META;;;;
separators;;;;
%separators_symbols% (for example I'm using "-" as separator for English);;;;
text;value;multiply level;errors limit;rank
String;Number;Number;Number;Number

For better understanding see example .csv files located in lib/expressions folder.