Portfolio Blog Snippets

Scrape(General|Single|Node): Node Attributes + Text

(() => {
	const getText = (el) => {

var text = el.innerText ? el.innerText : el.value ? el.value : el.content ? el.content : el.nodeValue ? el.nodeValue : "";
       return  text.replace(/\.+/gim,".")
                .replace(/\?+/gim,"?")
                .replace(/!+/gim,"!")
                .replace(/[ \t]+/gim," ")
                .replace(/[\n\r]+/gim,"\n").trim()
}
	const getElAttributes = el => [...el.attributes]
     .reduce((atts,att) => {
          atts[att.nodeName] = att.value
          return atts
     },{});
	const getAttributes = (report,el,i) => {
     try{
		var attributes = getElAttributes(el);
		if(Object.keys(attributes).length){
				attributes.index = i;
               	attributes.text = getText(el);
				attributes.length = attributes.text.length;
				report[el.nodeName] ? 
				report[el.nodeName].push(attributes) : 
				report[el.nodeName] = [attributes]
			}
     }catch(e){}
		return report;
   };
	var report  = [...document.querySelectorAll("*")].reduce(getAttributes,{})
console.log({report})
})()