Scrapegraph-ai/docs
Lorenzo Padoan 739b05ac79
Some checks failed
CodeQL / Analyze (python) (push) Has been cancelled
Release / Build (push) Has been cancelled
Release / Release (push) Has been cancelled
doc: 1$ banner
2025-10-04 16:52:23 +02:00
..
assets doc: 1$ banner 2025-10-04 16:52:23 +02:00
source Update llm.rst 2025-03-16 14:15:56 +01:00
chinese.md feat: update terms 2025-03-10 11:27:33 +01:00
japanese.md feat: update terms 2025-03-10 11:27:33 +01:00
korean.md feat: update terms 2025-03-10 11:27:33 +01:00
make.bat add: readthedocs structure 2024-01-31 16:46:05 +01:00
Makefile add: readthedocs structure 2024-01-31 16:46:05 +01:00
README.md run pre commit 2025-01-12 16:35:31 +01:00
requirements-dev.txt add read the docs 2025-01-17 22:20:23 +01:00
requirements.txt add furo 2025-01-17 22:36:28 +01:00
russian.md feat: update terms 2025-03-10 11:27:33 +01:00
turkish.md feat: update terms 2025-03-10 11:27:33 +01:00

title markmap
ScrapGraphAI Roadmap
colorFreezeLevel maxWidth
2 500

ScrapGraphAI Roadmap

Short-Term Goals

  • Improve the documentation (ReadTheDocs)

  • Create tutorials for the library

Medium-Term Goals

  • Node for handling API requests

  • Make scraping more deterministic

    • Create DOM tree of the website
    • HTML tag text embeddings with tags metadata
    • Study tree forks from root node
    • How do we use the tags parameters?
  • Create scraping folder with report

    • Folder contains .scrape files, DOM tree files, report
    • Report could be a HTML page with scraping speed, costs, LLM info, scraped content and DOM tree visualization
    • We can use pyecharts with R-markdown
  • Scrape multiple pages of the same website

    • Create new node that instantiate multiple graphs at the same time
    • Make graphs run in parallel
    • Scrape only relevant URLs from user prompt
    • Use the multi dimensional DOM tree of the website for retrieval
  • Crawler graph

    • Scrape all the URLs with the same domain in all the pages
    • Build many DOM trees and link them together
    • Save the multi dimensional tree in a file
  • Compare two DOM trees to assess the similarity

    • Save the DOM tree of the scraped website in a file as a sort of cache to be used to compare with future website structure
    • Create similarity metrics with multiple DOM trees (overall tree? only relevant tags structure?)
  • Nodes for handling authentication

    • Use Selenium or Playwright to handle authentication
    • Passes the cookies to the other nodes
  • Nodes that attaches to an open browser

    • Use Selenium or Playwright to attach to an open browser
    • Navigate inside the browser and scrape the content
  • Nodes for taking screenshots and understanding the page layout

    • Use Selenium or Playwright to take screenshots
    • Use LLM to asses if it is a block-like page, paragraph-like page, etc.
    • Issue #88

Long-Term Goals

  • Automatic generation of scraping pipelines from a given prompt

  • Create API for the library