Initializing a Puppeteer Browser Outside of Scraping Function

Viewed 779

I am very new to puppeteer (I started today). I have some code that is working the way that I want it to except for an issue that I think is making it extremely inefficient. I have a function that links me through potentially thousands of urls that have incremental IDs to pull the name, position, and stats of each player and then inserts that data into a neDB database. Here is my code:

const puppeteer = require('puppeteer');
const Datastore = require('nedb');
const database = new Datastore('database.db');
database.loadDatabase();

async function scrapeProduct(url, id){
  const browser = await puppeteer.launch();
  const page = await browser.newPage();
  await page.goto(url);
  let attributes = [];

  const [name] = await page.$x('//*[@id="ctl00_ctl00_ctl00_Main_Main_name"]');
  const txt = await name.getProperty('innerText');
  const playerName = await txt.jsonValue();
  attributes.push(playerName);

  //Make sure that there is a legitimate player profile before trying to pull a bunch of 'undefined' information.
  if(playerName){
    const [role] = await page.$x('//*[@id="ctl00_ctl00_ctl00_Main_Main_position"]');
    const roleTxt = await role.getProperty('innerText');
    const playerRole = await roleTxt.jsonValue();
    attributes.push(playerRole);

    //Loop through the 12 attributes and pull their values.
    for(let i = 1; i < 13; i++){
      let vLink = '//*[@id="ctl00_ctl00_ctl00_Main_Main_SectionTabBox"]/div/div/div/div[1]/table/tbody/tr['+i+']/td[2]';
      const [e1] = await page.$x(vLink);
      const val = await e1.getProperty('innerText');
      const skillVal = await val.jsonValue();
      attributes.push(skillVal);
    }

    //Create a player profile to be pushed into the database. (I realize this is very wordy and ugly code)
    let player = {
      Name: attributes[0],
      Role: attributes[1],
      Athleticism: attributes[2],
      Speed: attributes[3],
      Durability: attributes[4],
      Work_Ethic: attributes[5],    
      Stamina: attributes[6],   
      Strength: attributes[7],  
      Blocking: attributes[8],
      Tackling: attributes[9],  
      Hands: attributes[10],    
      Game_Instinct: attributes[11],
      Elusiveness: attributes[12],  
      Technique: attributes[13],
      _id: id,
    };

      database.insert(player);
      console.log('player #' + id + " scraped.");
      await browser.close();
  } else {
    console.log("Blank profile");
    await browser.close();
  }
}

//Making sure the first URL is scraped before moving on to the next URL. (i removed the URL because its unreasonably long and is not important for this part).
(async () => {
  for(let i = 0; i <= 1000; i++){
    let link = 'https://url.com/Ratings.aspx?rid='+i+'&section=Ratings';
    await scrapeProduct(link, i);
  }
})();

What I think is making this so inefficient is the fact that everytime scrapeProduct() is called, i create a new browser and create a new page. Instead I believe it would be more efficient to create 1 browser and 1 page and just change the pages URL with

await page.goto(url)

I believe that in order to do what I'm trying to accomplish here, i need to move:

  const browser = await puppeteer.launch();
  const page = await browser.newPage();

outside of my scrapeProduct() function but i cannot seem to get this to work. Anytime I try i get an error in my function saying that page is not defined. I am very new to puppeteer (started today), I would appreciate any guidance on how to accomplish this. Thank you very much!

TL;DR

How do i create 1 Browser instance and 1 Page instance that a function can use repeatedly by only changing the await page.goto(url) function.

2 Answers

To achieve this, you'll just need to separate the browser from your requests, like in a class, for example:

class PuppeteerScraper {
  async launch(options = {}) {
    this.browser = await puppeteer.launch(options);
    // you could reuse the page instance if it was defined here
  }

  /**
   * Pass the address and the function that will scrape your data,
   * in order to mantain the page inside this object
   */
  async goto(url, callback) {
    const page = await this.browser.newPage();
    await page.goto(url);

    /**evaluate its content */
    await callback(page);
    await page.close();
  }

  async close() {
    await this.browser.close();
  }
}

and, to implement it:

/**
 * scrape function, takes the page instance as its parameters
 */
async function evaluate_page(page) {
  const titles = await page.$$eval('.col-xs-6 .star-rating ~ h3 a', (itens) => {
    const text_titles = [];
    for (const item of itens) {
      if (item && item.textContent) {
        text_titles.push(item.textContent);
      }
    }
    return text_titles;
  });
  console.log('titles', titles);
}

(async () => {
  const scraper = new PuppeteerScraper();
  await scraper.launch({ headless: false });

  for (let i = 1; i <= 6; i++) {
    let link = `https://books.toscrape.com/catalogue/page-${i}.html`;
    await scraper.goto(link, evaluate_page);
  }
  scraper.close();
})();

altho, if you want something more complex, you could take a look how they done at Apify project.

About a year ago I tried to a make an React Native Pokemon Go helper app. Since there wasn't an api for pokemon nest and pokestops I created a server that scraped thesilphroad.com and I found the need to implement something like @Arkan said.

I wanted the server to be able to take multiple request, so I decided to initialize the browser when the server is booted up. When a request is received, the server checks to see if MAX_TABS have been reached. If reached, it waits, if not a new tab is opened and the scrape is performed

Here's the scraper.js

const puppeteer = require ('puppeteer')
const fs = require('fs')
const Page = require('./Page')
const exec = require('child_process').exec
const execSync = require('util').promisify(exec)
module.exports = class scraper {
    constructor(){
        this.browser = null
        this.getPages = null
        this.getTotalPages = null
        this.isRunning = false
        //browser permissions
        this.permissions = ['geolocation']
        this.MAX_TABS = 5
        //when puppeteer launches
        this.useFirstTab = true
    }
    async init(config={}){
        let headless = config.headless != undefined ? config.headless : true
        this.permissions = this.permissions.concat(config.permissions || [])
        //get local chromium location
        let browserPath = await getBrowserPath('firefox') || await getBrowserPath('chrome')
        this.browser = await puppeteer.launch({
            headless:headless,
            executablePath:browserPath,
            defaultViewport:null,
            args:[
                '--start-maximized',
            ]
        })
        this.getPages = this.browser.pages
        this.getTotalPages = ()=>{
            return this.getPages().then(pages=>pages.length).catch(err=>0)
        }
        this.isRunning = true
    }
    async waitForTab(){
        let time = Date.now()
        let cycles = 1
        await new Promise(resolve=>{
            let interval = setInterval(async()=>{
                let totalPages = await this.getTotalPages()
                if(totalPages < this.MAX_TABS){
                    clearInterval(interval)
                    resolve()
                }
                if(Date.now() - time > 100)
                    console.log('Waiting...')
                if(Date.now() - time > 20*1000){
                    console.log('... ...\n'.repeat(cycle)+'Still waiting...')
                    cycle++
                    time = Date.now()
                }
            },500)
        })
    }

    //open new tab and go to page
    async openPage(url,waitSelector,lat,long){
        await this.waitForTab()
        let pg
        //puppeteer launches with a blank tab, use this
        // if(this.useFirstTab){
        //  let pages = await this.browser.pages()
        //  pg = pages.pop()
        //  this.useFirstTab = false
        // }
        // else
        pg =    await this.browser.newPage()
        if(lat && long){
            await this.setPermissions(url)
        }
        let page = await new Page()
        await page.init(pg,url,waitSelector,lat,long)
        return page

    }
    async setPermissions(url){
        const context = this.browser.defaultBrowserContext();
        await context.overridePermissions(url,this.permissions)
    }
}

// assumes that the browser is in path
async function getBrowserPath(browserName){
    return execSync('command -v chromium').then(({stdout,stderr})=>{
        if(stdout.includes('not found'))
            return null
        return stdout
    }).catch(err=>null)
}

The scraper imports Page.js, which is just wrapper for a puppeteer Page object with the functions I used most made available

const path = require('path')
const fs = require('fs')
const userAgents = require('./staticData/userAgents.json')
const cookiesPath = path.normalize('./cookies.json')

// a wrapper for a puppeteer page with pre-made functions
module.exports = class Page{
    constuctor(useCookies=false){
        this.page = null
        this.useCookies = useCookies
        this.previousSession = this.useCookies && fs.existsSync(cookiesPath)
    }
    async close (){
        await this.page.close()
    }
    async init(page,url,waitSelector,lat,long){
        this.page = page
        let userAgent = userAgents[Math.floor(Math.random()*userAgents.length)]
        await this.page.setUserAgent(userAgent)
        await this.restoredSession()
        if(lat && long)
            await this.page.setGeolocation({
                latitude: lat || 59.95, longitude:long || 30.31667, accuracy:40
            })
        await this.page.goto(url)
        await this.wait(waitSelector)
        
    }
    async screenshotElement(selector='body',directory='./screenshots',padding=0,offset={}) {
      const rect = await this.page.evaluate(selector => {
        const el = document.querySelector(selector)
        const {x, y, width, height} = el.getBoundingClientRect()
        return {
            left: x, 
            top: y, 
            width, 
            height, 
            id: el.id
        }
      }, selector)
      
      let ext = 'jpeg'
      let filename = path.normalize(directory+'/'+Date.now())
      return await this.page.screenshot({
        type:ext,
        path:filename+' - '+selector.substring(5)+'.'+ext,
        clip: {
          x: rect.left - padding+(offset.left || 0),
          y: rect.top - padding+(offset.right || 0),
          width: rect.width + padding * 2+(offset.width||0),
          height: rect.height + padding * 2+ (offset.height||0)
        },
        encoding:'base64'
      })
    }
    async restoredSession(){
        if(!this.previousSession)
            return false
        let cookies = require(cookiesPath)
            for(let cookie of cookies){
                await this.page.setCookie(cookie)
            }
            console.log('Loaded previous session')
            return true
    }
    async saveSession(){
        //write cookie to file
        if(!this.useCookies)
            return
        const cookies = await this.page.cookies()
        fs.writeFileSync(cookiesPath,JSON.stringify(cookies,null,2))
        console.log('Wrote cookies to file')
    }
    //wait for text input elment and type text
    async type(selector,text,options={delay:150}){
        await this.wait(selector)
        await this.page.type(selector,text,options)
    }
    //click and waits
    async click(clickSelector,waitSelector=500){
        await this.page.click(clickSelector)
        await this.wait(waitSelector)
    }
    //hovers over element and waits
    async hover(selector,waitSelector=500){
        await this.page.hover(selector)
        await this.wait(1000)
        await this.wait(waitSelector)
    }
    //waits and suppresses timeout errors
    async wait(selector=500, waitForNav=false){
        try{
            //waitForNav is a puppeteer's waitForNavigation function
            //which for me does nothing but timeouts after 30s
            waitForNav && await this.page.waitForNavigation()
            await this.page.waitFor(selector)
        } catch (err){
            //print everything but timeout errors
            if(err.name != 'Timeout Error'){
                console.log('error name:',err.name)
                console.log(err)
                console.log('- - - '.repeat(4))
            }
            this.close()
        }
    }   
}
``
Related