mm.site/scripts/files2json.js

import fs from 'fs/promises'
import path from 'path'
import { exec } from 'child_process'

// WAVEFORM GENERATION
// TODO@mx use node-js library
async function generateWaveform(audioFile, outputDir) {
  const inputFile = path.join(outputDir, audioFile)
  audioFile = audioFile.replace(/\.[^/.]+$/, '.png')
  const outputPath = path.join(outputDir || '.', audioFile)

  return new Promise((resolve, reject) => {
    exec(
      `audiowaveform -i '${inputFile}' -o  '${outputPath}' --output-format png -z auto -w 640 -h 180 --background-color 00000000 --waveform-color FFFFFF50 --waveform-style bars --bar-style rounded --border-color 00000000 --bar-width 4 --bar-gap 6 --no-axis-labels`,
      (error, stdOut) => {
      if (error) {
        console.error(error)
        reject(error)
      } else {
        resolve(audioFile)
      }
    })
  })
}

async function generateThumb(imgPath, outputDir) {
  const { name, ext } = path.parse(imgPath)

  if (!ext) {
    return 0
  }

  const thumb = path.format({ name: `${name}.thumb`, ext })
  const inputFile = path.join(outputDir, imgPath)
  const outputFile = path.join(outputDir, thumb)

  return new Promise((resolve, reject) => {
    exec(
      `magick "${inputFile}" -resize 250x250\\> "${outputFile}"`,
      (error, stdout) => {
        if (error) {
          console.error(error)
          reject(error)
        } else {
          resolve(thumb)
        }
      }
    )
  })
}

// MAIN FUNCTION
async function main() {
  // variable for incoming variable
  let mediaDir = ''
  let outputDir = ''
  let recurse = false
  let imagesOnly = false
  let dryRun = false
  let formatted = null
  let waveform = false
  let order = false

  if (process.argv.length < 3 || process.argv.includes('-h')) {
    console.log(`
Files to Json converter

Commands :
  -h   : This message
  -dir : The input directory to scan
  -o   : The output directory of the json file. If not set will default to current working directory
  -u   : [NOT IMPLEMENTED]  List of comma separated feilds to update. Will only overwrite that which is set
  -r   : Recurse the directory (once) for sub-fields. Albums and image categories for example.

Other flags
  --images-only  : For a folder with only images, rather than media files with associated thumbs.
  --dry-run      : Output to the console
  --formatted    : Directory or file name, will format on the the undersore : "title_details".
                   This takes the arguments, parent, recurse or both. Defaults to both.
  --gen-waveform : [may break if recursing] If going through audio files can generate waveform images for the file.
  --order        : Take number & number + "." as order of array
    `)
    return
  }

  // get command line arguments
  process.argv.forEach((val, index) => {
    switch (val) {
      case '-dir':
        mediaDir = process.argv[index + 1] || ''
        break
      case '-o':
        outputDir = process.argv[index + 1] || ''
        break
      case '-u':
        update = process.argv[index + 1].match(/(title|details|media|image[s]*|tracks|)/i)
          ? process.argv[index + 1]
              .split(',')
              .filter(a => [
                'title',
                'details',
                'media',
                'image',
                'tracks',
                'images'
              ].includes(a))
          : true
        break
      case '-r':
        recurse = true
        break
      case '--images-only':
        imagesOnly = true
        break
      case '--dry-run':
        dryRun = true
        break
      case '--formatted':
        formatted = [
          'parent',
          'recurse',
          'both' ].includes(
            process.argv[index + 1]
          ) ? process.argv[index + 1] : 'both'
        break
      case '--gen-waveform':
        waveform = true
        break
      case '--order':
        order = true
        break
      default:
        break
    }
  })

  // if no media dir passed jump out
  if (!mediaDir) {
    console.log('no directory passed to script')
    return 0
  }
  //
  // read directory
  const dirList = await fs.readdir(mediaDir)

  // variable to hold media
  let media = []
  let images = []

  // acutal media to playback or see
  if (recurse) {
    media = dirList.filter(i => !i.match(/\.[^/.]{1,4}$/))
  } else {
    media = dirList.filter(i => i.match(/.(mp\d|m\d\w)$/i))
  }

  // create an array of images found in the directory as well
  images = dirList.filter(i => i.match(/.(jp\w*g|png)$/i))

  // clean up passed directory for json
  let parent_dir = mediaDir
    .match(/(?<=\/)(\w|\d)+\/*$/i)[0]
    .replace(/\/$/i, '')

  // set an output directory if not set by flags
  if (!outputDir) {
    outputDir = '.'
  }

  // THE MAIN EVENT
  const obj = await Promise.all(media.map(async m => {

    // setup structure
    const _r = {
      ...genMetadata(m, formatted == 'parent' || formatted == 'both', order),
      parent_dir,
      media: m,
      image: '',
      order: 0
    }

    if (order) {
      const idx = m.match(/\d+/) ? m.match(/\d+/)[0] : 0
      _r.order = idx
      _r.title = _r.title.replace(/^\d+\W*_*/, '')
    }

    // if we are scanning for images only, remove the 'image' property
    // as it will be saved in the media property
    if (imagesOnly) {
      delete _r.image
    } else {
      if (waveform) {
        // if waveform generate waveform
        _r.image = await generateWaveform(m, mediaDir)
      } else {
        // if not compare the media name with the image name
        // and if a match assign it to 'image' variable from earlier
        _r.image = linkImage(m.replace(/\.[^/.]+$/, ''), images)
      }
    }

    // if recurse (again) now we are looking for the media in the folders
    if (recurse) {
      // create full path from the passed dir, and the recuring dir
      const rDir = path.join(mediaDir, m)
      // read the dir
      let files = await fs.readdir(rDir)
      // filterout any rogue folders
      // generate the metadata for each file
      files = files.filter(f => f.match(/(?<!\.thumb)\.[^.]+$/))
      files = files.map(f => {
        const { title, details } = genMetadata(f, formatted == 'recurse' || formatted == 'both', false)
        return { title, details, media: f }
      })

      // sort files by leading nuber (if exists)
      files.sort((a, b) => {
        const valA = a.title.match(/\d+/) ? a.title.match(/\d+/)[0] : 0
        const valB = b.title.match(/\d+/) ? b.title.match(/\d+/)[0] : 0
        return parseInt(valA) - parseInt(valB)
      })

      // remove leading number from tile &
      // sanitize title for html unicode characters
      files.forEach(f => {
        f.title = f.title.replace(/^\d+(\s+|_)/, '')
        f.title = f.title.replaceAll('&#47', '&#34')
      })

      // if images only
      if (imagesOnly) {

        // setup object per image in array
        _r.images = files.filter(f => !f.media.match(/\.thumb\.[^/.]{1,4}$/i))
      } else {
        _r.album = true
        _r.tracks = files
      }
    }

    if (imagesOnly) {
      _r.images = await Promise.all(_r.images.map(async i => {
        return { ...i, thumb: await generateThumb(i.media, path.join(mediaDir, _r.media)) }
      }))
    }

    // return the result to the 'obj' variable
    return _r
  }))

  if (order) {
    obj.sort((a, b) => {
      return a.order - b.order
    })
  }

  // turn the obj variable into JSON
  const json = JSON.stringify(obj, null, 2)
  // set the output filename
  let outputFile = `${parent_dir}.json`

  // check if file already exists at location
  // const dataFile = await fs.readFile(path.join(outputDir, outputFile), 'utf8')
  // console.log(JSON.parse(dataFile) || 'no data file exists')
  // update only new fields

  // if dryrun
  if (dryRun) {
    // just console it out
    console.log(json)
  } else {
    // write it to a file
    fs.writeFile(path.join(outputDir, outputFile), json)
  }
}

function genMetadata(path, formatted) {
  let title, details

  const cleanPath = path.replace(/^\d+(_|.)/, '')

  // if formatted flag set
  if (formatted) {
    // split the incoming name from the map on the '_'
    // and set the first element to title, and the second to details
    let [ t, _, d ] = cleanPath.split(/(,|_)(.+)/s)
    // assign to return object
    title = t
    details = d ? d.replace(/\.+(mp\d|m\dv)$/i, '').trim() : ''
  } else {
    // else just set title to filename without extensions
    const name = cleanPath.replace(/\.\S{3,4}$/, '')
    title = name.replace(/_/g, ' ')
  }

  title = title.replace(/\.\w+$/, '')
  details = details ? details.replace(/\.\w+$/, '') : undefined

  return { title, details }
}

function linkImage(name, images) {
  return images.find(i => {
    return i.replace(/\.[^/.]+$/, '') == name
  })
}


main()