Git-Mirrors/brozzler
1
0
Fork 0
mirror of https://github.com/internetarchive/brozzler.git synced 2025-04-20 23:56:34 -04:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

More changes for handling psu24 site

Browse Source
This commit is contained in:
Hunter Stern 2015-09-15 18:03:08 -07:00
parent ea41653c44
commit 3467670900
3 changed files with 55 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

65
umbra/behaviors.d/simpleclicks.js.in
View File

@ -1,3 +1,5 @@
var umbraEndConditionTarget;
var umbraAboveBelowOrOnScreen = function(e) {
var eTop = e.getBoundingClientRect().top;
if (eTop < window.scrollY) {
@ -19,7 +21,14 @@ var umbraSimpleScrollsAndClicksBehavior = {
var somethingLeftBelow = false;
var somethingLeftAbove = false;
var cssSelector = "${click_css_selector}";
var cssSelectorClickEndCondition = "${click_css_selector_end_condition}";
var cssSelectorClickComputedStyleEndCondition = "${click_css_selector_computed_style_end_condition}";
//fixup cssSelectorClickEndCondition value if this argument has not been set in behaviors.yaml
if (cssSelectorClickEndCondition == "") {
cssSelectorClickEndCondition = "#uniq-selector-for-nothing";
}
var iframes = document.querySelectorAll("iframe");
var documents = Array(iframes.length + 1);
documents[0] = document;
@ -30,34 +39,46 @@ var umbraSimpleScrollsAndClicksBehavior = {
for (var j = 0; j < documents.length; j++) {
var clickTargets = documents[j].querySelectorAll("button.sc-button-play, button.playButton");
var clickTargets = documents[j].querySelectorAll(cssSelector);
umbraEndConditionTarget = documents[j].querySelector(cssSelectorClickEndCondition);
if (umbraEndConditionTarget) {
if (cssSelectorClickComputedStyleEndCondition != "") {
var dynamicCode = umbraCreateDynamicCodeToCheckSelectorCondition(cssSelectorClickComputedStyleEndCondition);
if (eval(dynamicCode)) {
return;
}
}
else if (cssSelectorClickComputedStyleEndCondition == "") {
return;
}
}
for ( var i = 0; i < clickTargets.length; i++) {
if (clickTargets[i].umbraClicked) {
//has already been clicked so no need to check again. if we did check
// and element was above screen, we would go back up and never reach the bottom
if (clickTargets[i].umbraClicked && cssSelectorClickEndCondition == "#uniq-selector-for-nothing") {
//has already been clicked so no need to check again unless there is a click end condition value set
//for this url in behaviors.yaml. then we keep clicking until the end condition is met
continue;
}
var where = umbraAboveBelowOrOnScreen(clickTargets[i]);
if (where == 0) {
if (!clickTargets[i].umbraClicked) {
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;
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; //break from clickTargets loop, but not from iframe loop
} else if (where > 0) {
somethingLeftBelow = true;
} else if (where < 0) {
@ -112,6 +133,12 @@ var umbraSimpleScrollsAndClicksBehavior = {
},
};
var umbraCreateDynamicCodeToCheckSelectorCondition = function(condition) {
if (!condition || condition == "") return;
return eval("var dynamicFunction = new Function('return function testCondition(target){ if (!target) return false; var computedStyle=window.getComputedStyle(target); return computedStyle." + condition + " }' )(); dynamicFunction(umbraEndConditionTarget);");
}
// Called from outside of this script.
var umbraBehaviorFinished = function() {
return umbraSimpleScrollsAndClicksBehavior.isFinished()

8
umbra/behaviors.py
View File

@ -32,7 +32,13 @@ class Behavior:
behavior_js = os.path.sep.join(__file__.split(os.path.sep)[:-1] + ["behaviors.d"] + [behavior["behavior_js"]])
behavior["script"] = open(behavior_js, encoding="utf-8").read()
elif "click_css_selector" in behavior:
behavior["script"] = simpleclicks_js_template.substitute(click_css_selector=behavior["click_css_selector"])
if "click_css_selector_end_condition" not in behavior:
behavior["click_css_selector_end_condition"] = "";
if "click_css_selector_computed_style_end_condition" not in behavior:
behavior["click_css_selector_computed_style_end_condition"] = "";
behavior["script"] = simpleclicks_js_template.substitute(click_css_selector=behavior["click_css_selector"], click_css_selector_end_condition=behavior["click_css_selector_end_condition"], click_css_selector_computed_style_end_condition=behavior["click_css_selector_computed_style_end_condition"])
return Behavior._behaviors

2
umbra/behaviors.yaml
View File

@ -31,6 +31,8 @@ behaviors:
# https://webarchive.jira.com/browse/ARI-4128
url_regex: '^https?://(?:www\.)?psu24.psu.edu/.*$'
click_css_selector: a[id='load-more']
click_css_selector_end_condition: a[id='load-more'][class='disabled']
click_css_selector_computed_style_end_condition: visibility=='hidden'
- # https://webarchive.jira.com/browse/AITFIVE-451
url_regex: '^https?://(?:www\.)?soundcloud.com/.*$'
click_css_selector: button.sc-button-play, button.playButton