Cloud Functions for Firebase - Converting PDF to image

Node modules can install native code that is in the same directory as the Cloud Function's source code. I found that some node libraries on github that do this for ghostscript which is a very useful library for PDF processing:

• Node library that wraps Ghostscript command line: https://github.com/sina-masnadi/node-gs
• Compiled Ghostscript which is used via git submodule: https://github.com/sina-masnadi/lambda-ghostscript

I put lambda-ghostscript into a sub-directory of my functions directory, then add the node-gs as a dependency in my package file like this:

{
  "name": "functions",
  "dependencies": {
    "@google-cloud/storage": "^1.3.1",
    "child-process-promise": "^2.2.1",
    "firebase-admin": "~5.4.0",
    "firebase-functions": "^0.7.2",
    "gs": "https://github.com/sina-masnadi/node-gs/tarball/master"
  }
}

Then in my index.js file I can just require the node library to easily use ghostscript from JavaScript. Here's the complete code for the Cloud Function that uses a Google Cloud Storage trigger:

const functions = require('firebase-functions');
const gcs = require('@google-cloud/storage')();
const spawn = require('child-process-promise').spawn;
const path = require('path');
const os = require('os');
const fs = require('fs');
var   gs = require('gs');

exports.makePNG = functions.storage.object().onChange(event => {

  // ignore delete events
  if (event.data.resourceState == 'not_exists') return false;

  const filePath = event.data.name;
  const fileDir = path.dirname(filePath);
  const fileName = path.basename(filePath);
  const tempFilePath = path.join(os.tmpdir(), fileName);
  if (fileName.endsWith('.png')) return false;
  if (!fileName.endsWith('.pdf')) return false;

  const newName = path.basename(filePath, '.pdf') + '.png';
  const tempNewPath = path.join(os.tmpdir(), newName);


  // // Download file from bucket.
  const bucket = gcs.bucket(event.data.bucket);

  return bucket.file(filePath).download({
    destination: tempFilePath
  }).then(() => {
    console.log('Image downloaded locally to', tempFilePath);

    return new Promise(function (resolve, reject) {
        gs()
          .batch()
          .nopause()
          .option('-r' + 50 * 2)
          .option('-dDownScaleFactor=2')
          .executablePath('lambda-ghostscript/bin/./gs')
          .device('png16m')
          .output(tempNewPath)
          .input(tempFilePath)
          .exec(function (err, stdout, stderr) {
              if (!err) {
                console.log('gs executed w/o error');            
                console.log('stdout',stdout);            
                console.log('stderr',stderr);            
                resolve();
              } else {
                console.log('gs error:', err);
                reject(err);
              }
          });
    });

  }).then(() => {
    console.log('PNG created at', tempNewPath);

    // Uploading the thumbnail.
    return bucket.upload(tempNewPath, {destination: newName});
  // Once the thumbnail has been uploaded delete the local file to free up disk space.
  }).then(() => {
    fs.unlinkSync(tempNewPath);
    fs.unlinkSync(tempFilePath);
  }).catch((err) => {
    console.log('exception:', err);
    return err;
  });

});

Here's the project on github: https://github.com/ultrasaurus/ghostscript-cloud-function

Disclaimer: This is using compiled native code and I verified experimentally that works for this case, so it is probably fine. I didn't look into the specific compile options and validate if they exactly correct for the Cloud Functions environment.

WORKING SOLUTION

Thank you @Ultrasaurus for pointing out this approach! However, for me it did not work and in your Github repo your also stated I haven't tested them. I modified your solution a little bit and got the following code, which is 100% working for me:

{
  "dependencies": {
    "@google-cloud/firestore": "^4.4.0",
    "@google-cloud/storage": "^5.3.0",
    "ghostscript": "https://github.com/musubu/node-ghostscript/tarball/master",
    "pdf-image": "^2.0.0",
    "rimraf": "^3.0.2",
    "uuid": "^8.3.1"
  }
}

The function is triggered by a Firestore event:

const Storage = require('@google-cloud/storage')
const fs = require('fs')
const rimraf = require('rimraf')
const os = require('os')
const gs = require('ghostscript')

const GOOGLE_PROJECT_ID = 'MY_GOOGLE_PROJECT_ID'
const GOOGLE_STORAGE_BUCKET_NAME = 'MY_GOOGLE_STORAGE_BUCKET_NAME'

const storage = new Storage.Storage({
  projectId: GOOGLE_PROJECT_ID
})

exports.createImage = async (event) => {
  let {
    appointment,
    name
  } = event.value.fields

  name = getFileName(name.stringValue)
  appointment = appointment.stringValue

  console.log(`Processing document ${name} in appointment ${appointment}`)

  const tempDir = createTempDir(appointment)

  const tmpDocumentPath = await downloadPdf(tempDir, name, appointment)
  const imagePath = await convertPdfToImage(tmpDocumentPath)
  await uploadImage(imagePath, appointment)

  deleteDir(tempDir)
}

function getFileName (name) {
  const nameParts = name.split('/')
  return nameParts[nameParts.length - 1]
}

function createTempDir (appointment) {
  const tempDir = `${os.tmpdir()}/${appointment}_${Math.random()}`
  fs.mkdirSync(tempDir)
  console.log(`Created dir ${tempDir}`)
  return tempDir
}

async function downloadPdf (tempDir, name, appointment) {
  const destination = `${tempDir}/${name}`
  await storage.bucket(GOOGLE_STORAGE_BUCKET_NAME).file(`${appointment}/${name}`).download({ destination })
  console.log(`Successfully downloaded document ${name}`)
  return destination
}

async function convertPdfToImage (pdfPath) {
  const imagePath = pdfPath.replace('pdf', 'png')

  return new Promise(function (resolve, reject) {
    try {
      gs()
        .batch()
        .nopause()
        .device('png16m')
        .output(imagePath)
        .input(pdfPath)
        .exec(function (err, stdout, stderr) {
          if (!err) {
            console.log('gs executed w/o error')
            console.log('stdout', stdout)
            console.log('stderr', stderr)
            resolve(imagePath)
          } else {
            console.log('gs error:', err)
            reject(err)
          }
        })
    } catch (error) {
      console.log(error)
    }
  })
}

async function uploadImage (imagePath, appointment) {
  const imagePathParts = imagePath.split('/')
  const imageName = imagePathParts[imagePathParts.length - 1]

  console.log(`Starting upload for ${imageName} at ${imagePath} to storage ${appointment}/${imageName}`)

  await storage.bucket(GOOGLE_STORAGE_BUCKET_NAME).upload(imagePath, {
    destination: `${appointment}/${imageName}`,
    metadata: {
      metadata: { appointment }
    }
  })

  console.log(`Successfully uploaded image for appointment ${appointment}`)
}

function deleteDir (dir) {
  rimraf.sync(dir)
}