Node API needs inspection wrt attributes about dryscrape HOT 2 CLOSED

Rather than parsing the source HTML, you should parse the current DOM HTML:

doc = sess.document()
lxml_node = doc.xpath(node.path())[0]

This seems a bit complicated for such a simple operation, but I think it's preferrable over a separate attributes() method for several reasons:

• All the other methods of the Driver and Node classes communicate with the server to get the data from the live DOM
• To make this work with the live DOM, we would either create a new document object every time (which is time-consuming) or implement smart caching, so that the document is only recreated if the DOM was changed (which would only be fast if the server could notify us about changes, which is not possible)
• dryscrape's strength is navigation to a final destination even if there are "obstacles", not extracting data from websites (there is Scrapy for that)
• It's rarely need

I also can't make Node.xpath('@*') behave as you expect without adapting the webkit_server to allow this. In the current implementation, the node representation on the server side is expecting HTML tag nodes, not attributes or even text.

I realize that in the current implementation, it's not a simple task to extend dryscrape, so probably I will add some kind of factory pattern here to allow you to add your own helper methods to Node.

from dryscrape.

Thanks very much for adding the session document method, which allowed me to complete my little project. It was sufficient in my case, because I didn't need to manipulate the live DOM after the attribute names were known. Navigating two trees in parallel could be unpleasant if I needed to navigate across ajax pages based on those unknown attributes.

I appreciate that data extraction is not dryscrape's strength and Scrapy should be used for more complex requirements, however Scrapy's abilities are not useful when it cannot easily navigate past said obstacles!

Though inspection of a node to find its parent, children, attributes and text could be considered very basic tree functionality, it is a shame the implementation is so complex. Thanks for considering it.

I'll close this issue because there is a workaround (though not the ideal solution).

from dryscrape.