fuzzy-substring.js

const levenshtein = require('damerau-levenshtein');

export function matchTransaction(name1, name2) {
  // It's important that target is the longer of the two strings,
  // since we'll try to do a fuzzy substring match of src into target
  const target = scrub(name1.length > name2.length ? name1 : name2);
  const src = scrub(name1.length > name2.length ? name2 : name1);

  // Run a levenshtein distance on the src string and a substring of
  // the target string that is about equal length of src. If it fails,
  // keep walking through target up to each word boundary, and repeat
  let searchStr = target;
  while (true) {
    const subStr = searchStr.slice(0, src.length + 2);
    const result = levenshtein(src, subStr);

    if (result.similarity > 0.6) {
      return true;
    }

    const idx = searchStr.indexOf(' ');
    if (idx === -1) {
      break;
    }
    searchStr = searchStr.slice(idx + 1);
  }

  return false;
}

scrub.js

function scrub(str) {
  return str.replace(/[^a-zA-Z ]/g, '').toLowerCase();
}

tests.js

expect(matchTransaction('Macsy wkout', 'macys weekout')).toBe(true);
expect(matchTransaction("Papa John's", 'papa john')).toBe(true);
expect(
  matchTransaction("#001 fenn st Macy's is good 33333 EMX", 'macys good')
).toBe(true);
expect(
  matchTransaction("#001 fenn st Macy's is good 33333 EMX", 'good place')
).toBe(false);