scrapegraphai/scrapegraph-ai is a Python web scraping library that extracts information specified by a prompt from websites and local documents. It does this by leveraging LLMs and graph logic to define the shape of the data it needs and extract it. The latest release listed in the repository is v2.3.1, but the provided record does not list what newly changed in that version.
According to the README summary, it handles web and document content including XML, HTML, JSON and Markdown. It processes single pages and multiple pages through several pipelines, including SmartScraperGraph, and can extract information from search results. It also states that it supports generating Python scripts and audio files. The model choices presented are OpenAI, Groq, Azure, Gemini, MiniMax and Ollama. The repository license is MIT, and the project was created on 2024-01-27. This setup fits cases where you want to specify the items to extract via a prompt and apply it across multiple input formats.
However, this feature description alone cannot tell you about extraction accuracy or processing speed. The provided record contains no performance measurements and does not give the v2.3.1 change history. So the verifiable scope is limited to the inputs and features it claims to support; specific performance or this release's improvements require separate evidence.