srt-to-text.js

import "@johnlindquist/kit"
    
// Name: srt to txt
// Author: Taylor Bell
// Description: Breaks an srt file into readable text blocks

let srtParser2 = await npm('srt-parser-2')

let parser = new srtParser2()

let srtFiles = await drop()

function formatTimeString(str) {
  let [h, m, s] = str.split(':')
  if (h == '00') {
    return `${m}:${s}`
  }
  
  return `${h}:${m}:${s}`
}

for (let f of srtFiles) {
  let text = await readFile(`${f.path}`, 'utf-8')
  
  let result = parser.fromSrt(text)


  let withTimes = result.map(line => {return {...line, totalSeconds: line.endSeconds - line.startSeconds}})

  let timeLimitInSeconds = 20
  let currentTimeInSeconds = 0
  let transcribedSentencesCount = 0

  let arrayByTimes = []
  let tempArray = []


  withTimes.forEach((x, i) => {
    if (currentTimeInSeconds + x.totalSeconds >= timeLimitInSeconds) {
      arrayByTimes.push(tempArray)
      tempArray = []
      currentTimeInSeconds = 0
      transcribedSentencesCount = 0
    }
    
    if (currentTimeInSeconds === 0) {
      tempArray.push(`[${formatTimeString(x.startTime.split(',')[0])}] ${x.text}`)
    } else {
      tempArray.push(x.text)
    }
    currentTimeInSeconds += x.totalSeconds
    transcribedSentencesCount++
  })

  arrayByTimes.push(tempArray)

  let transcript = arrayByTimes.map(x => x.join(' ')).flat().join('\n\n')

  await writeFile(`${f.path}-transcript.txt`, transcript)


}