Merge pull request #44 from nlevitt/AITFIVE-451-3

followup to https://github.com/internetarchive/umbra/pull/43

umbra/behaviors.d/psu24.js

@@ -0,0 +1,110 @@
+
+var umbraBehavior = {
+	IDLE_TIMEOUT_SEC : 10,
+	idleSince : null,
+	alreadyClicked : {},
+
+	intervalFunc: function() {
+		var clickedSomething = false;
+		var somethingLeftBelow = false;
+		var somethingLeftAbove = false;
+
+		var iframes = document.querySelectorAll("iframe");
+		var documents = Array(iframes.length + 1);
+		documents[0] = document;
+		
+		for (var i = 0; i < iframes.length; i++) {
+			documents[i+1] = iframes[i].contentWindow.document;
+		}
+
+		for (var j = 0; j < documents.length; j++) {
+			var clickTargets = documents[j].querySelectorAll("a[id='load-more']");
+			for (var i = 0; i < clickTargets.length; i++) {
+				if (clickTargets[i].className === "disabled") {
+					continue;
+				}
+
+				var where = this.aboveBelowOrOnScreen(clickTargets[i]);
+				if (where == 0) {
+					console.log("clicking on " + clickTargets[i].outerHTML);
+					// do mouse over event on click target
+					// since some urls are requsted only on
+					// this event - see
+					// https://webarchive.jira.com/browse/AITFIVE-451
+					var mouseOverEvent = document.createEvent('Events');
+					mouseOverEvent.initEvent("mouseover",true, false);
+					clickTargets[i].dispatchEvent(mouseOverEvent);
+					clickTargets[i].click();
+					clickedSomething = true;
+					this.idleSince = null;
+
+					break; //break from clickTargets loop, but not from iframe loop
+				} else if (where > 0) {
+					somethingLeftBelow = true;
+				} else if (where < 0) {
+					somethingLeftAbove = true;
+				}
+			}
+		}
+
+		if (!clickedSomething) {
+			if (somethingLeftAbove) {
+				console.log("scrolling UP because everything on this screen has been clicked but we missed something above");
+				window.scrollBy(0, -500);
+				this.idleSince = null;
+			} else if (somethingLeftBelow) {
+				console.log("scrolling because everything on this screen has been clicked but there's more below document.body.clientHeight="
+								+ document.body.clientHeight);
+				window.scrollBy(0, 200);
+				this.idleSince = null;
+			} else if (window.scrollY + window.innerHeight < document.documentElement.scrollHeight) {
+				console.log("scrolling because we're not to the bottom yet document.body.clientHeight="
+								+ document.body.clientHeight);
+				window.scrollBy(0, 200);
+				this.idleSince = null;
+			} else if (this.idleSince == null) {
+				this.idleSince = Date.now();
+			}
+		}
+
+		if (!this.idleSince) {
+			this.idleSince = Date.now();
+		}
+	},
+
+	aboveBelowOrOnScreen: function(e) {
+		var eTop = e.getBoundingClientRect().top;
+		if (eTop < window.scrollY) {
+			return -1; // above
+		} else if (eTop > window.scrollY + window.innerHeight) {
+			return 1; // below
+		} else {
+			return 0; // on screen
+		}
+	},
+
+	start: function() {
+		var that = this;
+		this.intervalId = setInterval(function() {
+			that.intervalFunc()
+		}, 250);
+	},
+
+	isFinished: function() {
+		if (this.idleSince != null) {
+			var idleTimeMs = Date.now() - this.idleSince;
+			if (idleTimeMs / 1000 > this.IDLE_TIMEOUT_SEC) {
+				return true;
+			}
+		}
+		return false;
+	},
+};
+
+// Called from outside of this script.
+var umbraBehaviorFinished = function() {
+	return umbraBehavior.isFinished()
+};
+
+umbraBehavior.start();
+

umbra/behaviors.d/simpleclicks.js.in

@@ -1,44 +1,113 @@
-var umbraSimpleClicksBehavior = {
+var umbraBehavior = {
-        IDLE_TIMEOUT_SEC: 10,
+	IDLE_TIMEOUT_SEC : 10,
-        idleSince: null,
+	idleSince : null,
-        alreadyClicked: {},
+	alreadyClicked : {},
 
-        intervalFunc: function() {
+	intervalFunc : function() {
-                var clickTargets = document.querySelectorAll("${click_css_selector}");
+		var clickedSomething = false;
+		var somethingLeftBelow = false;
+		var somethingLeftAbove = false;
+		var cssSelector = "${click_css_selector}";
 
-                for (var i = 0; i < clickTargets.length; i++) {
+		var iframes = document.querySelectorAll("iframe");
-                        var key = clickTargets[i].outerHTML;
+		var documents = Array(iframes.length + 1);
-                        if (!this.alreadyClicked[key]) {
+		documents[0] = document;
-                                console.log("clicking on " + key);
+		
-                                clickTargets[i].click();
+		for (var i = 0; i < iframes.length; i++) {
-                                this.alreadyClicked[key] = true;
+			documents[i+1] = iframes[i].contentWindow.document;
-                                this.idleSince = null;
+		}
-                                return;
-                        }
-                }
 
-                if (!this.idleSince) {
+		for (var j = 0; j < documents.length; j++) {
-                        this.idleSince = Date.now();
-                }
-        },
 
-        start: function() {
+			var clickTargets = documents[j].querySelectorAll(cssSelector);
-                var that = this;
-                this.intervalId = setInterval(function(){ that.intervalFunc() }, 250);
-        },
 
-        isFinished: function() {
+			for ( var i = 0; i < clickTargets.length; i++) {
-                if (this.idleSince != null) {
+				if (clickTargets[i].umbraClicked) {
-                        var idleTimeMs = Date.now() - this.idleSince;
+					continue;
-                        if (idleTimeMs / 1000 > this.IDLE_TIMEOUT_SEC) {
+				}
-                                return true;
+
-                        }
+				var where = this.aboveBelowOrOnScreen(clickTargets[i]);
-                }
+
-                return false;
+				if (where == 0) {
-        },
+					console.log("clicking on " + clickTargets[i].outerHTML);
+					// do mouse over event on click target
+					// since some urls are requsted only on
+					// this event - see
+					// https://webarchive.jira.com/browse/AITFIVE-451
+					var mouseOverEvent = document.createEvent('Events');
+					mouseOverEvent.initEvent("mouseover",true, false);
+					clickTargets[i].dispatchEvent(mouseOverEvent);
+					clickTargets[i].click();
+					clickedSomething = true;
+					this.idleSince = null;
+					clickTargets[i].umbraClicked = true;
+
+					break; //break from clickTargets loop, but not from iframe loop
+				} else if (where > 0) {
+					somethingLeftBelow = true;
+				} else if (where < 0) {
+					somethingLeftAbove = true;
+				}
+			}
+		}
+
+		if (!clickedSomething) {
+			if (somethingLeftAbove) {
+				// console.log("scrolling UP because everything on this screen has been clicked but we missed something above");
+				window.scrollBy(0, -500);
+				this.idleSince = null;
+			} else if (somethingLeftBelow) {
+				// console.log("scrolling because everything on this screen has been clicked but there's more below document.body.clientHeight="
+				// 				+ document.body.clientHeight);
+				window.scrollBy(0, 200);
+				this.idleSince = null;
+			} else if (window.scrollY + window.innerHeight < document.documentElement.scrollHeight) {
+				// console.log("scrolling because we're not to the bottom yet document.body.clientHeight="
+				// 				+ document.body.clientHeight);
+				window.scrollBy(0, 200);
+				this.idleSince = null;
+			} else if (this.idleSince == null) {
+				this.idleSince = Date.now();
+			}
+		}
+
+		if (!this.idleSince) {
+			this.idleSince = Date.now();
+		}
+	},
+
+	start : function() {
+		var that = this;
+		this.intervalId = setInterval(function() {
+			that.intervalFunc()
+		}, 250);
+	},
+
+	isFinished : function() {
+		if (this.idleSince != null) {
+			var idleTimeMs = Date.now() - this.idleSince;
+			if (idleTimeMs / 1000 > this.IDLE_TIMEOUT_SEC) {
+				return true;
+			}
+		}
+		return false;
+	},
+
+	aboveBelowOrOnScreen : function(e) {
+		var eTop = e.getBoundingClientRect().top;
+		if (eTop < window.scrollY) {
+			return -1; // above
+		} else if (eTop > window.scrollY + window.innerHeight) {
+			return 1; // below
+		} else {
+			return 0; // on screen
+		}
+	},
 };
 
 // Called from outside of this script.
-var umbraBehaviorFinished = function() { return umbraSimpleClicksBehavior.isFinished() };
+var umbraBehaviorFinished = function() {
+	return umbraBehavior.isFinished()
+};
 
-umbraSimpleClicksBehavior.start();
+umbraBehavior.start();

umbra/behaviors.yaml

@@ -16,6 +16,10 @@ behaviors:
    url_regex: '^https?://(?:www\.)?vimeo\.com/.*$'
    behavior_js: vimeo.js
    request_idle_timeout_sec: 10
+ -
+   url_regex: '^https?://(?:www\.)?psu24.psu.edu/.*$'
+   behavior_js: psu24.js
+   request_idle_timeout_sec: 10
  -
    url_regex: '^https?://(?:www\.)?instagram\.com/.*$'
    behavior_js: instagram.js
@@ -28,6 +32,14 @@ behaviors:
    url_regex: '^https?://(?:www\.)?usask.ca/.*$'
    click_css_selector: a[id='feature-next']
    request_idle_timeout_sec: 10
+ - # https://webarchive.jira.com/browse/AITFIVE-451
+   url_regex: '^https?://(?:www\.)?soundcloud.com/.*$'
+   click_css_selector: button.sc-button-play, button.playButton
+   request_idle_timeout_sec: 10
+ - # https://webarchive.jira.com/browse/AITFIVE-463
+   url_regex: '^https?://(?:www\.)?christophercerrone.com/.*$'
+   click_css_selector: button.playButton.medium
+   request_idle_timeout_sec: 10
  - # default fallback brhavior
    url_regex: '^.*$'
    request_idle_timeout_sec: 10