plugin: extract-object

[cldellow]

TODO: name and schema is up in the air.

      // TODO: determine schema for this
      "extract-selectors": {
        // TODO: flag to indicate whether we should mark up the source to
        //       make it more amenable for CSS extraction

        // TODO: how to do invariant things, eg if we wanted to build
        //       out a join table of categories, we'd extract the category
        //       from the same root HTML element... how to infer that?

        // TODO: how to do many-to-many, if such a thing exists? maybe it's
        //       an advanced case and can be ignored.
        "url-regex": "/garments/",

        // optional
        "database": "dbname",
        "extractors": {
          "patterns": {
            "selector": ".w-full.container",
            "attributes": {
              "url!": [
                "h3 a",
                { "attribute": "href" }
              ],
              "name": [
                "h3 a",
                { "text": true }
              ],
              "designer_name": [
                "h3 + p a",
                { "text": true }
              ]
            }
          },
          "pattern_categories": {
            "selector": ".w-full.container",
            "attributes": {
              "url!": [
                "h3 a",
                { "attribute": "href" }
              ],
              "category!": [
                "html h1",
                { "text": true }
              ]
            }
          }
        ]
      }
    }

Needs https://github.com/cldellow/datasette-scraper#extract_from_responsescraper-config-url-response