playwright-python VS scraper

TikTok uses quite a lot of JavaScript on its site, both for displaying content and for analyzing user behavior, including detecting and blocking crawlers. Therefore, for crawling TikTok, we'll use a headless browser with Playwright.

Integrating Ollama, Microsoft vision models and Playwright I've made a simple agent that can browse websites and data to answer your query.

Do you have to use Curl? It wouldn't render a lot of sites correctly anyway (anything that uses JS for rendering).

Can you run a puppeteer/playwright instance and add an ad blocker to that? e.g. https://github.com/ghostery/adblocker or https://github.com/microsoft/playwright-python/issues/782

Playwright

An alternative is to use a package like playwright (or Selenium) to run a browser remotely and login.

Using the tools at hand is often the best approach. That said, I've spent most of the last 13 years of my career automating browsers. For years, I used Selenium with a variety of libraries. After switching to Puppeteer/Playwright, I have zero interest in going back lol. Playwright actually has first party Python support. (Puppeteer has a port called Pyppeteer, but it's no longer maintained and the author recommends using Playwright)

https://playwright.dev/python/

This is pretty great for web automation https://playwright.dev/python/

Additional info and source code for generating the dataset, summarizing it and rendering the chart are available at https://github.com/get-set-fetch/scraper/tree/main/datasets/javascript-libs-from-top-1mm-sites

Store the scrape progress (to-be-scraped / in-progress / scraped / in-error URLs) in a database shared by all participants and scrape in parallel with as many machines as the db load permits. Got a connection timeout / IP is blocked on one machine ? Update the scrape status for the corresponding URL and let another machine retry it. https://github.com/get-set-fetch/scraper (written in typescript) follows this idea. Using Terraform from a simple config file you can adjust the number of scraper instances to be deployed in cloud at startup and during the scraping process. In benchmarks a PostgreSQL server running on a DigitalOcean vm with 4vCPU, 8GB memory allows for ~2000 URLs to be scraped per second (synthetic data, no external traffic). From my own experience this is almost never the bottleneck. Obeying robots.txt crawl-delay will surely put you under this limit. Disclaimer: I'm the npm package author.

Written in typescript https://github.com/get-set-fetch/scraper stores scraped content directly in a database (sqlite, mysql, postgresql). Each URL represents a Resource. You can implement your own IResourceStorage and define the exact db columns you need.

You can use https://github.com/get-set-fetch/scraper with a custom plugin based on the mozilla/readability as detailed in https://getsetfetch.org/node/custom-plugins.html (extracting news article content). I think it's a close match to your use case.

I'm currently implementing concurrency conditions at project/proxy/domain/session level in https://github.com/get-set-fetch/scraper . On each level you can define the maximum number of requests and the delay between two consecutive requests.

If you're familiar with nodejs give https://github.com/get-set-fetch/scraper a try. Scraped content can be stored in sqlite, mysql or postgresql. It also supports puppeteer, playwright, cheerio or jsdom for the actual content extraction. No scheduler though.

I'm using this exact strategy to scrape content directly from DOM using APIs like document.querySelectorAll. You can use the same code in both headless browser clients like Puppeteer or Playwright and DOM clients like cheerio or jsdom (assuming you have a wrapper over document API). Depending on the way a web page was fetched (opened in a browser tab or fetched via nodejs http/https requests), ExtractHtmlContentPlugin, ExtractUrlsPlugin use different DOM wrappers (native, cheerio, jsdom) to scrape the content.

[1] https://github.com/get-set-fetch/scraper/blob/main/src/plugi...

https://github.com/get-set-fetch/scraper - I've been working (intermittently :) ) on a nodejs or browser extension scraper for the last 3 years, see the other projects under the get-set-fetch umbrella. Putting a lot more effort lately as I really want to do those Alexa top 1 million analysis like top js libraries, certificate authorities and so on. A few weeks back I've posted on Show:HN as you can do basic/intermediate? scraping with it.

Not capable of handling 1 mil+ pages as it still limited to puppeteer or playwright. Working on adding cheerio/jsdom support right now.