AutoMate/backend/pdfScan.js

const pdfExtract = require('pdf-extract')
const fs = require('fs')
// const path = '/home/tomi/Downloads/2230G-900-01A_Jun_2018_User.pdf'
const path = '/home/tomi/Downloads/MDO4000-B-MSO-DPO4000B-and-MDO3000-Oscilloscope-Programmer-Manual-Rev-A.pdf'
const options = {
  type: 'text' // or 'ocr'
}

function getSCPICommands (pages) {
  const scpiCommon = /(\*\w+\??)/g
  // const header = ``
  const scpi = /((?:\*\w+|(?:\[?\w+\]?)(?=:\w+)\]?)(?:\[?:\w+\]?)*\??)(?:\s+(<?\w+>?)(?:[,|]\s*(<?\w+>?))*)?/g

  const scpiLines = []
  pages.map((page, pageIndex) => {
    const lines = page.split('\n')
    lines.map((line, lineIndex) => {
      const matches = line.match(scpi)
      if (matches) {
        console.log(pageIndex, lineIndex, matches)
        return matches
      }
    })
  })
  fs.writeFile('/home/tomi/Downloads/MDO4000.txt', pages.join('\n'), error => {
    if (error) console.log(error)
    console.log('file was saved.')
  })
}

const processor = pdfExtract(path, options, error => {
  if (error) return error
})
processor.on('complete', data => {
  console.log(data.text_pages)
  getSCPICommands(data.text_pages)
})
processor.on('error', error => {
  console.error(error)
})