From 95ee40fa8cdca9cff0aba033a4352fc0621a9583 Mon Sep 17 00:00:00 2001
From: Lars-Dominik Braun <lars@6xq.net>
Date: Mon, 5 Mar 2018 19:06:55 +0100
Subject: Add generic click behavior script

Configureable. Clicks elements matching one (or more) CSS selectors once
or multiple times.

Currently supported: Facebook, Twitter, Disqus (embedded iframe)
---
 crocoite/behavior.py              |  15 +++---
 crocoite/data/click.js            | 111 ++++++++++++++++++++++++++++++++++++++
 crocoite/data/per-site/twitter.js |  30 -----------
 3 files changed, 119 insertions(+), 37 deletions(-)
 create mode 100644 crocoite/data/click.js
 delete mode 100644 crocoite/data/per-site/twitter.js

(limited to 'crocoite')

diff --git a/crocoite/behavior.py b/crocoite/behavior.py
index e4ec93b..26841aa 100644
--- a/crocoite/behavior.py
+++ b/crocoite/behavior.py
@@ -222,17 +222,18 @@ class Screenshot (Behavior):
                 'X-Chrome-Viewport': viewport})
         writer.write_record (record)
 
-### Site-specific scripts ###
+class Click (JsOnload):
+    """ Generic link clicking """
+
+    name = 'click'
+    scriptPath = 'click.js'
 
-class Twitter (HostnameFilter, JsOnload):
-    name = 'twitter'
-    scriptPath = 'per-site/twitter.js'
-    hostname = ['com', 'twitter']
+### Site-specific scripts ###
 
 # available behavior scripts. Order matters, move those modifying the page
 # towards the end of available
-generic = [Scroll, EmulateScreenMetrics]
-perSite = [Twitter]
+generic = [Scroll, EmulateScreenMetrics, Click]
+perSite = []
 available = generic + perSite + [Screenshot, DomSnapshot]
 availableNames = set (map (lambda x: x.name, available))
 
diff --git a/crocoite/data/click.js b/crocoite/data/click.js
new file mode 100644
index 0000000..7013487
--- /dev/null
+++ b/crocoite/data/click.js
@@ -0,0 +1,111 @@
+/*	Generic clicking
+ *
+ *  We can’t just click every clickable object, since there may be side-effects
+ *  like navigating to a different location. Thus whitelist known elements.
+ */
+
+(function(){
+const selectorFlag = Object.freeze ({
+	none: 0,
+	multi: 1, /* click item multiple times */
+});
+const sites = Object.freeze ([
+	{
+		hostname: /^www\.facebook\.com$/i,
+		selector: [
+			/* show more comments */
+			{s: 'a.UFIPagerLink[role=button]', flags: selectorFlag.none},
+			/* show nested comments*/
+			{s: 'a.UFICommentLink[role=button]', flags: selectorFlag.none},
+			],
+	}, {
+		hostname: /^twitter\.com$/i,
+		selector: [
+			/* expand threads */
+			{s: 'a.ThreadedConversation-moreRepliesLink', flags: selectorFlag.none},
+			/* show hidden profiles */
+			{s: 'button.ProfileWarningTimeline-button', flags: selectorFlag.none},
+			/* show hidden/sensitive media */
+			{s: 'button.Tombstone-action.js-display-this-media', flags: selectorFlag.none},
+			],
+	}, {
+		hostname: /^disqus\.com$/i,
+		selector: [
+			/* load more comments */
+			{s: 'a.load-more__button', flags: selectorFlag.multi},
+			],
+	}
+	]);
+
+/* pick selectors matching current location */
+let hostname = document.location.hostname;
+let selector = [];
+for (let s of sites) {
+	if (s.hostname.test (hostname)) {
+		selector = selector.concat (s.selector);
+	}
+}
+
+function makeClickEvent () {
+	return new MouseEvent('click', {
+				view: window,
+				bubbles: true,
+				cancelable: true
+				});
+}
+
+/* throttle clicking */
+let queue = [];
+function click () {
+	let o = queue.shift ();
+	if (o !== undefined) {
+		o.dispatchEvent (makeClickEvent ());
+	}
+	return queue.length > 0;
+}
+
+/*	Element is visible if itself and all of its parents are
+ */
+function isVisible (o) {
+	if (o === null || !(o instanceof Element)) {
+		return true;
+	}
+	let style = window.getComputedStyle (o);
+	if ('parentNode' in o) {
+		return style.display !== 'none' && isVisible (o.parentNode);
+	} else {
+		return style.display !== 'none';
+	}
+}
+
+/*	Elements are considered clickable if they are a) visible and b) not
+ *	disabled
+ */
+function isClickable (o) {
+	return !o.hasAttribute ('disabled') && isVisible (o);
+}
+
+/* some sites don’t remove/replace the element immediately, so keep track of
+ * which ones we already clicked */
+let have = new Set ();
+function discover () {
+	for (let s of selector) {
+		let obj = document.querySelectorAll (s.s);
+		for (let o of obj) {
+			if (!have.has (o) && isClickable (o)) {
+				queue.push (o);
+				if (!(s.flags & selectorFlag.multi)) {
+					have.add (o);
+				}
+			}
+		}
+	}
+	if (queue.length > 0) {
+		window.setInterval (click, 50);
+	}
+	return true;
+}
+
+/* XXX: can we use a mutation observer instead? */
+window.setInterval (discover, 1000);
+}());
diff --git a/crocoite/data/per-site/twitter.js b/crocoite/data/per-site/twitter.js
deleted file mode 100644
index 2773f64..0000000
--- a/crocoite/data/per-site/twitter.js
+++ /dev/null
@@ -1,30 +0,0 @@
-/* Fixups for twitter:
- * - Some accounts are hidden behind a “suspicious activity” message, click
- *   that.
- * - Click “more replies” buttons periodically (as they popup when scrolling)
- */
-(function(){
-function makeClickEvent () {
-	return new MouseEvent('click', {
-				view: window,
-				bubbles: true,
-				cancelable: true
-				});
-}
-function expandThread () {
-	let links = document.querySelectorAll('a.ThreadedConversation-moreRepliesLink');
-	for (let i = 0; i < links.length; i++) {
-		links[i].dispatchEvent (makeClickEvent ());
-	}
-	return true;
-}
-function showProfile () {
-	var show = document.querySelector ("button.ProfileWarningTimeline-button");
-	if (show) {
-		show.dispatchEvent (makeClickEvent ());
-	}
-}
-window.addEventListener("load", showProfile);
-/* XXX: can we use a mutation observer instead? */
-window.setInterval (expandThread, 1000);
-}());
-- 
cgit v1.2.3