I have code taken from Google to harvest URLs from Google. I saved it as filename.js. When I run the file it showed "'document' is undefined". The part of the code which is showing problem is
// ==UserScript==
// @name Google URL Harvester
// @namespace http://userscripts.org/scripts/show/42998
// @description Harvests URLs from a Google Search
// @include http://www.google.co.uk/
// @include http://www.google.com/
// ==/UserScript==
var btn_container;
var inputs = document.getElementsByTagName("input");
for (var i = 0; i < inputs.length; i++) {
if (inputs[i].name == "btnG")
btn_container = inputs[i].parentNode;
}
function find_next_link(html) {
var url = html.match(/(<a href="[^"]+">)\s*<span[^>]+style="[^"]*background-position:\s?-76px\s/);
if (url == null)
return false;
var div = document.createElement("div");
div.innerHTML = url[1];
return div.firstChild.href;
}
function harvest(query_url, callback) {
ajax(query_url, function(e){
var als = e.match(/<a[^>]+class=l[^>]*>/g);
for (var i = 0; i < als.length; i++) {
urls.push(als[i].match(/href="([^"]+)"/)[1]);
}
var next_url = find_next_link(e);
if (next_url)
harvest(next_url, callback);
else
callback();
});
}
function ajax(url, callback) {
var req = new XMLHttpRequest();
req.onreadystatechange = function() {
if (req.readyState == 4 && req.status == 200) {
callback(req.responseText);
}
}
开发者_StackOverflow req.open("GET", url, true);
req.send("");
}
var new_button = document.createElement("input");
new_button.type = "button";
new_button.value = "Harvest URLs";
new_button.setAttribute("onsubmit", "return false;");
btn_container.appendChild(new_button);
var urls = [];
new_button.addEventListener("click", function(){
var query_url = unsafeWindow.document.forms[0].action + "?num=100&q="+escape(unsafeWindow.document.forms[0].q.value);
document.body.innerHTML = "<img src='http://oneworldwebsites.com/images/wheel%20throbber.gif' />";
harvest(query_url, function() {
document.body.innerHTML = urls.join("<br/>");
});
},false);
Here I have not defined document(if it is necessary). Can anybody please rectify the error in this code. Operating system is Windows 7.
I am saving this document to my desktop as harv.js and running it. Am I doing anything wrong?
Google like most websites updates their structure over time.
for (var i = 0; i < inputs.length; i++) {
if (inputs[i].name == "btnG")
btn_container = inputs[i].parentNode;
}
Needs to become
for (var i = 0; i < inputs.length; i++) {
if (inputs[i].name == "btnK") //<<------G to K
btn_container = inputs[i].parentNode;
}
I can just guess that the error is thrown because you try to access the "input" element, before creating it later. Moreover you shouldn't use a html element as a identifier in JavaScript. You could also use Firebug to pinpoint the error.
精彩评论