PdfEditor/server.js

const fs = require('fs')
const asfs = require('asfs')
const del = require('delete')
const express = require('express')
const multer = require('multer')
const childProcess = require('child_process')
const bodyParser = require('body-parser')
const puppeteer = require('puppeteer')
const uuid = require('uuid')
const parseDataUrl = require('data-urls')
const path = require('path')
const sanitize = require('./lib/sanitize')
const expressLogging = require('express-logging')
const log4js = require('log4js')
const getConfig = require('microservice-config')
const bytes = require('bytes')
const {htmlEncode} = require('htmlencode')
const pixelsPerPoint = 1.3333333333333333

const config = getConfig({
    port: process.env.PORT || 0,
    logLevel: 'info',
    tempPath: 'temp',
    pdf2htmlexPath: `${__dirname}/node_modules/@alancnet/pdf2htmlex/bin-win/pdf2htmlEX.exe`,
    fileSizeLimit: '10mb',
    chromeUserDataDir: path.join(__dirname, 'temp/chrome')
})


const logger = log4js.getLogger('Server')
log4js.getLogger().level = config.logLevel

const chromePromise = (async () => {
    if (await asfs.existsAsync(config.chromeUserDataDir)) {
        logger.info('Cleaning Chrome user data directory...')
        await del.promise(config.chromeUserDataDir)
    }
    logger.info('Launching chrome...')
    const chrome = await puppeteer.launch({
        userDataDir: config.chromeUserDataDir
    })
    logger.info('Chrome is running.')
    return chrome
})()

const upload = ext => multer({
    storage: multer.diskStorage({
        destination: (req, file, cb) => cb(null, config.tempPath),
        filename: (req, file, cb) => {
            file.uuid = uuid()
            cb(null, `${file.uuid}.${ext}`)
        }
    }),
    limits: {
        fileSize: bytes(config.fileSizeLimit)
    }
})

const app = express()
let counter = 0

// Provide a unique id and a logger to each request
app.use((req, res, next) => {
    req.id = counter++
    req.logger = log4js.getLogger(`Request ${req.id}`)
    next()
})

app.use(expressLogging(logger))

const async = (fn) => (req, res, next) => {
    fn(req, res, next)
        .then(next)
        .catch(err => {
            req.logger.error(err)
            res.status(500).send(`<pre>${err}</pre>`)
        })
}

const exec = (logger, commandLine) => new Promise((resolve, reject) => {
    logger.debug(`Executing: ${commandLine}`)
    childProcess.exec(commandLine, (error, stdout, stderr) => {
        if (error) {
            logger.error(`Exec error: ${error}\n${stderr}`)
            reject(stderr)
        } else {
            logger.trace(`Exec success`)
            resolve({stdout, stderr})
        }
    })
})

app.use(bodyParser.urlencoded({extended: false, limit: config.fileSizeLimit}))
app.get('/', (req, res) => res.redirect('/edit'))
app.use(express.static('./public'))
const viewerCommon = mode => async(async (req, res) => {
    const logger = req.logger
    let pdfFile, htmlFile
    if (req.file) {
        logger.debug(`Request includes file: ${req.file.originalname}`)
        pdfFile = req.file.path
        htmlFile = `${config.tempPath}/${req.file.uuid}.html`
    } else if (req.body.url) {
        logger.debug(`Request includes url.`)
        pdfFile = `${config.tempPath}/${uuid()}.pdf`
        htmlFile = `${pdfFile}.html`
        const pdf = parseDataUrl(req.body.url)
        logger.debug(`Writing ${pdfFile}...`)
        await asfs.writeFileAsync(pdfFile, pdf.body)
    } else {
        logger.debug(`Request includes no data.`)
        return res.status(400).send('Requires file or url.')
    }
    /*
        Executes pdf2htmlex[.exe] with:
        - 200 horizontal and vertical DPI
        - TrueType Font format (because woff does not render in PhantomJS)
        - No DRM, overriding any PDF settings forbidding copying or modifying
        - The path to the source PDF file
        - The path to the output HTML file
    */
    await exec(logger, `"${config.pdf2htmlexPath}" --hdpi 200 --vdpi 200 --font-format ttf --no-drm 1 "${pdfFile}" "${htmlFile}"`)

    logger.debug(`Reading public/viewer.html...`)
    const editHtml = await asfs.readFileAsync('public/viewer.html')

    logger.debug(`Reading ${htmlFile}...`)
    const html = await asfs.readFileAsync(htmlFile, 'utf8')

    const pageWidth = (/\.w0{width:([\d\.]*)pt/).exec(html)[1] * 1
    const pageHeight = (/\.h0{height:([\d\.]*)pt/).exec(html)[1] * 1
    const pageStyle = `
        html {
            -webkit-print-color-adjust: exact;
        }
        /* Prevent overflow into blank pages */
        .pf {
            max-height: ${pageHeight * pixelsPerPoint / 96 * 0.999}in;
            overflow-y: hidden;
        }
        @page {
            width: ${pageWidth / 96}in;
            height: ${pageHeight / 96}in;
            margin: 0;
        }
    `

    logger.trace(`Page style: ${pageStyle}`)

    res.status(200).send(html
        .replace(
            '</body>', 
            `<input type="hidden" id="viewerMode" value="${htmlEncode(mode)}" />
            <input type="hidden" id="filename" value="${htmlEncode(req.file.originalname)}" />
            ${editHtml}</body>`
        )
        .replace(
            '</head>',
            `<style type="text/css">${pageStyle}</style></head>`
        )
    )

    logger.debug(`Deleting ${htmlFile}...`)
    await asfs.unlinkAsync(htmlFile)

    logger.debug(`Deleting ${pdfFile}...`)
    await asfs.unlinkAsync(pdfFile)

    logger.debug(`Done.`)
})

app.get('/edit', (req, res) => res.sendFile(path.join(__dirname, './public/index.html')))
app.post('/edit', upload('pdf').single('document'), viewerCommon('edit'))

app.get('/view', (req, res) => res.sendFile(path.join(__dirname, './public/index.html')))
app.post('/view', upload('pdf').single('document'), viewerCommon('view'))

app.post('/save', async(async (req, res) => {
    const logger = req.logger
    const tmpUuid = uuid()
    const htmlFile = `${config.tempPath}/${tmpUuid}.html`
    const htmlUrl = `file://${path.join(process.cwd(), htmlFile)}`
    const pdfFile = `${config.tempPath}/${tmpUuid}.pdf`
    const html = sanitize(req.body.html)
    const filename = req.body.filename || 'document.pdf'
    logger.trace({tmpUuid, htmlFile, htmlUrl, pdfFile})

    const pageWidth = /\.w0{width:([\d\.]*)pt/.exec(html)[1]
    const pageHeight = /\.h0{height:([\d\.]*)pt/.exec(html)[1]
    logger.trace({pageWidth, pageHeight})

    logger.debug(`Writing HTML file, ${html.length} chars to ${htmlFile}...`)
    await asfs.writeFileAsync(htmlFile, html, 'utf8')
    
    logger.debug(`Getting chrome...`)
    const chrome = await chromePromise
    
    logger.debug(`Getting new page...`)
    const page = await chrome.newPage()

    logger.debug(`Navigating to ${htmlUrl}...`)
    await page.goto(htmlUrl)

    logger.debug(`Capturing PDF to ${pdfFile}...`)
    await page.pdf({
        path: pdfFile,
        preferCSSPageSize: true
    })

    logger.debug(`Closing page...`)
    await page.close()

    logger.debug(`Sending ${pdfFile} to client...`)
    await new Promise((resolve, reject) => res.download(pdfFile, filename, err => err ? reject(err) : resolve()))

    logger.debug(`Deleting ${htmlFile}...`)
    await asfs.unlinkAsync(htmlFile)

    logger.debug(`Deleting ${pdfFile}...`)
    await asfs.unlinkAsync(pdfFile)

    logger.debug(`Done.`)
}))


const listener = app.listen(config.port, (err) => {
    if (err) logger.error(err)
    else {
        const address = listener.address()
        logger.info(`Listening on http://127.0.0.1:${address.port}`)
    }
})