Extract with type = computed the "selector" is not available as input ?

[sandermanneke]

I trying to extract a field using type computed - for example use python to remove redundant spaces.

I was expecting something like this to work:

    "fields":[
        {
            "name":"product_name",
            "selector":".//h1",
            "type":"computed",
            "expression": "\" \".join(str(product_name).split())"
        },

However after looking looking at

crawl4ai/crawl4ai/extraction_strategy.py

Line 994 in 79328e4

def _compute_field(self, item, field):

I printed the variable 'item' and noticed that product_name is not available.

I got around it by first creating the field and then overwriting it:

        {
            "name":"product_name",
            "selector":".//h1",
            "type":"text",
        },
        {
            "name":"product_name",
            "type":"computed",
            "expression": "\" \".join(str(product_name).split())"
        },

That works but it feels redundant and clunky. I was expecting the data from the selecter to be available.

Should I approach this differently ?

Also - I could not find any documentation on this. Did I overlook ?