1 files changed, 102 insertions, 0 deletions
diff --git a/scraper/content-script/check.js b/scraper/content-script/check.js
new file mode 100644
index 00000000..0644084d
--- /dev/null
+++ b/scraper/content-script/check.js
@@ -0,0 +1,102 @@
+(function(){
+
+	var listening = false, loaded = false
+	var TYPES = { Status: 0, Connected: 1, Disconnected: 2, Image: 3, Text: 4 }
+	var PUNCTUATION_REGEX = /[\[\]\{\}]/g  // i.e. resembles code
+	var NULL_ARRAY = []
+	var seen = {}
+	const toArray = (a) => Array.prototype.slice.call(a)
+	const $ = (s) => document.querySelector(s)
+	const $$ = (s) => document.querySelectorAll(s)
+
+	function init () {
+		if (window.location.href.indexOf("lvh.me") !== -1 || window.location.href.indexOf("localhost") !== -1) return
+		bind()
+	}
+	function bind () {
+		chrome.extension.onMessage.addListener(onMessage)
+		chrome.extension.sendMessage({ type: TYPES.Status }, gotStatus)
+	}
+	function gotStatus (response) {
+		console.log('got status', response)
+		if (response && response.status === "on") {
+			console.log('its on', loaded)
+			if (! loaded) {
+				console.log('started')
+				setTimeout(() => { start() }, 5000)
+				// send(document.body.innerText)
+				// setInterval(function(){
+				// 	send(document.body.innerText)
+				// }, 10000)
+			}
+			loaded = true
+		}
+	}
+	function onMessage (request, sender, sendResponse) { 
+		switch (request.method) {
+			case 'start':
+				start()
+				break
+			case 'stop':
+				listening = false
+				break
+		}
+	}
+	function start(){
+		console.log(window.location.href, listening)
+		if (window.location.href.indexOf('schol' + 'ar' + '.go' + 'og' + 'le') === -1) return
+		if (listening) return
+		listening = true
+		energize()
+	}
+
+	function energize(){
+		console.log('energize')
+		const records = toArray($$(".gs_r")).map((el, i) => {
+		  let data = {}
+		  let link = el.querySelector("h3 a")
+		  if (link) {
+		    data.link = link.href
+		  }
+		  let pdfLink = el.querySelector(".gs_or_ggsm a")
+		  if (pdfLink) {
+		    data.pdfLink = pdfLink.href
+		  }
+		  let attribution = el.querySelector('.gs_a')
+		  if (attribution) {
+		    data.attribution = attribution.innerText
+		    data.attributionLinks = toArray(attribution.querySelectorAll('a')).map(a => ({
+		      href: a.href,
+		      name: a.innerText,
+		    }))
+		  }
+		  let snippet = el.querySelector('.gs_a')
+		  if (snippet) {
+		    data.snippet = snippet.innerText
+		  }
+		  let citationLink = el.querySelector('.gs_fl a:nth-of-type(3)')
+		  if (citationLink && citationLink.innerText.match('Cited by')) {
+		    data.citationLink = citationLink.href
+		    data.citationCount = parseInt(citationLink.innerText.replace(/^\s*Cited by /, ''), 10) || -1
+		  }
+		  return data
+		})
+		let record = {
+			title: document.querySelector('title').innerText,
+			url: window.location.href,
+			records: records,
+		}
+	  send(JSON.stringify(record))
+		let nextLink = $("#gs_n td:last-child a")
+		if (nextLink) {
+			setTimeout(() => { nextLink.click() }, 19000 + (Math.random() * 21000))
+		}
+		send("done")
+	}
+	function send (text) {
+		chrome.extension.sendMessage({ type: TYPES.Text, data: text }, function(){})
+	}
+
+	init()
+
+})()