Musicbrainz - Combine Catalog Numbers

mbrainz-combine-catnum.user.js

// ==UserScript==
// @name        Musicbrainz - Combine Catalog Numbers
// @namespace   com.toadking.mb.catalog
// @match       https://musicbrainz.org/*
// @grant       none
// @version     1.4
// @author      Toad King
// ==/UserScript==

const catNumRegExp = /^(.*?)(\d+)([^\d]*)$/

function parseCatNums(catalogNumbers) {
  let splitCatNums = []
  for (const catNum of catalogNumbers) {
    const catnumSearch = catNum.match(catNumRegExp)
    if (!catnumSearch) {
      splitCatNums.push({ prefix: catNum, suffix: '' })
    } else {
      const [, prefix, num, suffix] = catnumSearch
      splitCatNums.push({ prefix, num, suffix })
    }
  }

  let parsedCatNums = []
  while (splitCatNums.length > 0) {
    const { prefix, suffix, num } = splitCatNums[0]
    // skip non-matching catalog numbers or number-only ones
    if ((prefix == '' && suffix === '') || num === undefined) {
      parsedCatNums.push(prefix + (num !== undefined ? num : '') + suffix)
      splitCatNums.shift()
    } else {
      const matchedPrefixes = splitCatNums.filter(c => c.num !== undefined && c.prefix === prefix && c.suffix === suffix).sort((a, b) => Number(a.num) - Number(b.num))
      splitCatNums = splitCatNums.filter(c => c.num === undefined || c.prefix !== prefix)
      const padSize = matchedPrefixes[0].num.length

      const ranges = []
      let currentRange = { start: Number(matchedPrefixes[0].num), end: Number(matchedPrefixes[0].num) }
      for (const catNum of matchedPrefixes.slice(1)) {
        const num = Number(catNum.num)
        if (num === currentRange.end + 1) {
          currentRange.end = num
        } else {
          ranges.push(currentRange)
          currentRange = { start: num, end: num }
        }
      }
      ranges.push(currentRange)

      for (const { start, end } of ranges) {
        let endString = String(end).padStart(padSize, '0')
        if (start === end) {
          parsedCatNums.push(prefix + endString)
        } else {
          // use endString.length to handle cases like 99~100
          let startString = String(start).padStart(endString.length, '0')

          while (startString.length > 0) {
            if (startString[0] !== endString[0]) {
              break
            }
            startString = startString.substring(1)
            endString = endString.substring(1)
          }

          parsedCatNums.push(prefix + String(start).padStart(padSize, '0') + '~' + endString + suffix)
        }
      }
    }
  }

  return parsedCatNums
}

const catalogLists = new Set(Array.from(document.querySelectorAll('.catalog-number')).map(e => e.parentElement))

for (const listElem of catalogLists) {
  const catalogNumbers = new Set(Array.from(listElem.querySelectorAll('.catalog-number')).map(e => e.textContent))
  // skip elements that contain only one catalog number (also includes label lists on release pages)
  if (catalogNumbers.size <= 1) {
    continue
  }

  const parsedCatNums = parseCatNums(catalogNumbers)

  const newElems = parsedCatNums.map(c => {
    const span = document.createElement('span')
    span.className = 'catalog-number'
    span.textContent = c
    return span
  })
  listElem.replaceChildren(newElems.shift())
  for (const elem of newElems) {
    listElem.appendChild(document.createTextNode(', '))
    listElem.appendChild(elem)
  }
}

// Tests

// const tests = [
//   // Long list
//   {
//     input: 'SBPS-0076, SBPS-0077, SBPS-0078, SBPS-0079, SBPS-0080, SBPS-0081, SBPS-0082, SBPS-0083, SBPS-0084, SBPS-0085, SBPS-0086, SBPS-0087, SBPS-0088, SBPS-0089, SBPS-0090, SBPS-0091, SBPS-0092, SBPS-0093, SBPS-0094, SBPS-0095'.split(', '),
//     expected: ['SBPS-0076~95'],
//   },
//   // Reverse order
//   {
//     input: 'VPCG-84243, VPCG-84242, VPCG-84241, VPCG-84240, VPCG-84239, VPCG-84238, VPCG-84237, VPCG-84236, VPCG-84235, VPCG-84234'.split(', '),
//     expected: ['VPCG-84234~43'],
//   },
//   // No dashes
//   {
//     input: 'LC2493, LC2494, LC2495, LC2496, LC2497, LC2498, LC2499, LC2500, LC2501, LC2502, LC2503, LC2504, LC2505'.split(', '),
//     expected: ['LC2493~505'],
//   },
//   // Dashes
//   {
//     input: 'GFCA-525, GFCA-526, GFCA-527, GFCA-528, GFCA-529, GFCA-530, GFCA-531, GFCA-532, GFCA-533, GFCA-534, GFCA-535, GFCA-536, GFCA-537'.split(', '),
//     expected: ['GFCA-525~37'],
//   },
//   // Split in ranges
//   {
//     input: 'TEST-01, TEST-02, TEST-04, TEST-05'.split(', '),
//     expected: ['TEST-01~2', 'TEST-04~5'],
//   },
//   // Padding increases
//   {
//     input: 'VTJL-10, VTJL-9'.split(', '),
//     expected: ['VTJL-9~10'],
//   },
//   // Numbers only
//   {
//     input: '1, 2, 3'.split(', '),
//     expected: ['1', '2', '3'],
//   },
//   // no catalog numbers
//   {
//     input: [],
//     expected: [],
//   },
//   // no numbers
//   {
//     input: 'foo, bar'.split(', '),
//     expected: ['foo', 'bar'],
//   },
//   // Prefix and Suffix
//   {
//     input: 'KBCI-00008A, KBCI-00009A, KBCI-00010A, KBCI-00011A, KBCI-00012A, KBCI-00013A'.split(', '),
//     expected: ['KBCI-00008~13A'],
//   },
//   // Suffix only
//   {
//     input: '00008A, 00009A, 00010A'.split(', '),
//     expected: ['00008~10A'],
//   },
//   // Two number areas
//   {
//     input: '1-00008, 1-00009, 1-00010'.split(', '),
//     expected: ['1-00008~10'],
//   },
//   // Two number areas with suffix
//   {
//     input: '1-00008A, 1-00009A, 1-00010A'.split(', '),
//     expected: ['1-00008~10A'],
//   },
// ]
// for (const { input, expected } of tests) {
//   const output = parseCatNums(input)
//   if (JSON.stringify(output) !== JSON.stringify(expected)) {
//     throw new Error(`test failed (expected ${expected}, got ${output})`)
//   }
// }