scrapeghost

scrapeghost is an experimental library for scraping websites using OpenAI's GPT API.

Source: https://github.com/jamesturk/scrapeghost

Documentation: https://jamesturk.github.io/scrapeghost/

Issues: https://github.com/jamesturk/scrapeghost/issues

Use at your own risk. This library makes considerably expensive calls ($0.36 for a GPT-4 call on a moderately sized page.) Cost estimates are based on the OpenAI pricing page and not guaranteed to be accurate.

Features

The purpose of this library is to provide a convenient interface for using GPT for the purpose of web scraping.

Python-based schema definition - Define the shape of the data you want to extract as any Python object.

• Future versions will support optional validation that the response matches the schema.

Token Reduction - Fewer tokens means lower costs, faster responses, and staying under the API's token limits.

• Automatic HTML cleaning - Remove unnecessary HTML tags and attributes to reduce the size of the HTML sent to the model.
• CSS and XPath selectors - Pre-filter the HTML to send to the model by writing a single CSS or XPath selector.
• Auto-splitting - Optionally split the HTML into multiple calls to the model, each of a specified length.

Cost Controls - Scrapers keep running totals of how many tokens have been sent and received, so costs can be tracked.

• Future versions will allow setting a budget and stopping the scraper if the budget is exceeded.

Model Options - Works with GPT-3.5-Turbo or GPT 4, and allows passing additional parameters to the model to customize behavior.

• Support for automatic fallbacks (e.g. use cost-saving GPT-3.5-Turbo by default, fall back to GPT-4 if needed.)

Error Handling & Logging - Detailed logging and error handling to help debug issues.