fix: replace regex used to split ranges

[lukekarrys]

Found in #433, the regex used to split ranges on || was padded by \s*
on either side causing a decrease in performance when used on range
strings with a lot of spaces. Since the result of the split is
immediately trimmed, we can just split on the string instead.

[lukekarrys]

I ran some benchmarks locally using the following script which passes different arguments to range.split for ranges with differing amounts of whitespace. The difference between /\|\|/ and '||' seemed trivial, so I went with '||'.

const Range = require('./classes/range')
const a = process.argv.slice(2)
const s = ' '.repeat(Number(a[0]))
const r = `${s}1.1.1${s}`
const v = `${r}||${r}`.repeat(Number(a[1]))
const split = ({
  poly: /\s*\|\|\s*/,
  regex: /\|\|/,
  string:  '||',
})[a[2]]
new Range(v, {}, split)

5 spaces x 10 versions

❯ hyperfine -w 3 -L r poly,regex,string -L s 5 -L v 10 'node regex.js {s} {v} {r}'
Benchmark 1: node regex.js 5 10 poly
  Time (mean ± σ):      55.1 ms ±   2.0 ms    [User: 42.0 ms, System: 13.1 ms]
  Range (min … max):    51.5 ms …  61.8 ms    48 runs
 
Benchmark 2: node regex.js 5 10 regex
  Time (mean ± σ):      54.3 ms ±   1.9 ms    [User: 41.6 ms, System: 12.4 ms]
  Range (min … max):    50.6 ms …  58.1 ms    47 runs
 
Benchmark 3: node regex.js 5 10 string
  Time (mean ± σ):      54.6 ms ±   2.1 ms    [User: 41.9 ms, System: 12.5 ms]
  Range (min … max):    50.5 ms …  61.9 ms    50 runs
 
Summary
  'node regex.js 5 10 regex' ran
    1.01 ± 0.05 times faster than 'node regex.js 5 10 string'
    1.01 ± 0.05 times faster than 'node regex.js 5 10 poly'

100 spaces x 100 versions

❯ hyperfine -w 3 -L r poly,regex,string -L s 100 -L v 100 'node regex.js {s} {v} {r}'
Benchmark 1: node regex.js 100 100 poly
  Time (mean ± σ):      57.4 ms ±   1.5 ms    [User: 44.5 ms, System: 12.5 ms]
  Range (min … max):    54.0 ms …  59.8 ms    46 runs
 
Benchmark 2: node regex.js 100 100 regex
  Time (mean ± σ):      54.7 ms ±   2.2 ms    [User: 41.8 ms, System: 12.6 ms]
  Range (min … max):    51.3 ms …  64.3 ms    47 runs
 
Benchmark 3: node regex.js 100 100 string
  Time (mean ± σ):      54.4 ms ±   2.0 ms    [User: 41.7 ms, System: 12.5 ms]
  Range (min … max):    50.9 ms …  62.8 ms    49 runs
 
Summary
  'node regex.js 100 100 string' ran
    1.01 ± 0.06 times faster than 'node regex.js 100 100 regex'
    1.05 ± 0.05 times faster than 'node regex.js 100 100 poly'

1000 spaces x 1000 versions

❯ hyperfine -w 3 -L r poly,regex,string -L s 1e3 -L v 1e3 'node regex.js {s} {v} {r}'
Benchmark 1: node regex.js 1e3 1e3 poly
  Time (mean ± σ):      2.179 s ±  0.011 s    [User: 2.152 s, System: 0.018 s]
  Range (min … max):    2.166 s …  2.202 s    10 runs
 
Benchmark 2: node regex.js 1e3 1e3 regex
  Time (mean ± σ):      68.4 ms ±   2.1 ms    [User: 53.6 ms, System: 14.2 ms]
  Range (min … max):    66.6 ms …  80.1 ms    40 runs
 
  Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet PC without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
 
Benchmark 3: node regex.js 1e3 1e3 string
  Time (mean ± σ):      68.5 ms ±   2.2 ms    [User: 53.5 ms, System: 14.3 ms]
  Range (min … max):    65.9 ms …  80.3 ms    40 runs
 
  Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet PC without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
 
Summary
  'node regex.js 1e3 1e3 regex' ran
    1.00 ± 0.04 times faster than 'node regex.js 1e3 1e3 string'
   31.85 ± 1.00 times faster than 'node regex.js 1e3 1e3 poly'

[nlf]

this is very clearly correct 👍