Changelog mentiones "Support for complex JSONPath, JSON-CSS, and Microdata extraction" ?

[sandermanneke]

I would like to extract json fields from json strings that are in the html source, for example microdata with application/ld+json.

The changelog there is mention of Improved JSON Extraction: Support for complex JSONPath, JSON-CSS, and Microdata extraction.

https://github.com/unclecode/crawl4ai/blob/79328e42925c9ce8c030a1cadfe68c88cbe02c36/CHANGELOG.md#0424x-2024-12-31

This suggests there is some way to parse a json string and then return values by its path.

However I cant find any reference in the docs or in the source to that extent.

Is this an oversight in the changelog or am I missing something ?

[aravindkarnam]

@sandermanneke Yes. It's a feature by the name of JsonCssExtractionStrategy. Here's the docs for it!
https://docs.crawl4ai.com/extraction/no-llm-strategies/

We have also recently introduced a utility function that makes use of an LLM to generate the schema to map the target data fields to the css/xpath in html.

schema = JsonCssExtractionStrategy.generate_schema(
            result.html,
            llm_config=LLMConfig(
                provider="gemini/gemini-2.0-flash", api_token="env:GEMINI_API_KEY"
            ),
            target_json_example={
                "name": "McDonald's",
                "outlet": "Malleshwaram",
                "delivery_time": "30-40 mins",
                "menu": [
                    {
                        "name": "Chicken Surprise Burger",
                        "price": 76,
                        "rating": 3.3,
                        "rating_count": 3,
                        "coupon": "Buy 1 Get 1",
                        "description": "Introducing the new Chicken Surprise Burger which has the perfect balance of a crispy fried chicken patty, the crunch of onions and the richness of creamy sauce."
                    },
                    {
                        "name": "Chicken Surprise Burger Combo",
                        "price": 238,
                        "rating": 3.3,
                        "rating_count": 6,
                        "coupon": "Buy 1 Get 1",
                        "description": "Introducing the new Chicken Surprise Burger which has the perfect balance of a crispy fried chicken patty, the crunch of onions and the richness of creamy sauce."
                    },
                ],
            },
        )

I also recently put together a gist where I show how to get McDonalds menu from an online food delivery site. Check out this example and run it to get a full understanding.
https://gist.github.com/aravindkarnam/021e6c4e3da0da3ca795b40d615ebe9f

Let us know how it goes and if you need any additional help, post your code snippet. I'll help you.