2013-09-23 16:13:38 -05:00
<!doctype html>
< html lang = "en" >
< head >
< meta charset = "UTF-8" >
< meta name = "referrer" content = "never" >
< meta http-equiv = "X-UA-Compatible" content = "IE=edge" >
< base target = "_blank" >
< link rel = "alternate" type = "application/rss+xml" title = "RSS Feed" href = "/feed/archivebot.rss" >
< link rel = "alternate" type = "application/atom+xml" title = "Atom Feed" href = "/feed/archivebot.atom" >
< link rel = "icon" type = "image/png" href = "/assets/favicon.png" >
< title > ArchiveBot dashboard 2.0< / title >
< / head >
< body >
< style >
html, body {
background-color: #D0C0AE;
font-family: Tahoma, Arial, sans-serif;
font-size: 13px;
}
.padded-page {
padding: 20px 27px 20px 27px;
}
.header {
font-family: Arial, sans-serif;
font-weight: bold;
font-size: 18px;
margin: 0 0 20px 0;
display: flex;
align-items: flex-end;
justify-content: space-between;
flex-flow: row nowrap;
}
.job-header {
display: flex;
align-items: flex-end;
justify-content: space-between;
flex-flow: row nowrap;
}
.job-info {
white-space: nowrap;
overflow: hidden;
}
.job-url {
font-family: Arial, sans-serif;
font-size: 14px;
font-weight: bold;
text-decoration: none;
}
.job-ident {
margin: 0 1px 0 0;
border: 0;
background-color: #D0C0AE;
color: #786552;
font-weight: bold;
text-align: right;
}
.log-window {
background-color: #FFF7E1;
overflow-y: scroll;
height: 192px;
border: 1px solid #999;
margin: 0 0 1em 0;
border-radius: 3px;
}
.log-window-stopped {
border: 1px solid #222;
box-shadow: 2px 2px 4px #888;
}
.line-normal {
white-space: pre;
width: 100%;
padding: 0 0 0 5px;
box-sizing: border-box;
}
.line-error {
white-space: pre;
width: 100%;
background-color: #FFB9B9;
padding: 0 0 0 5px;
box-sizing: border-box;
}
.line-warning {
white-space: pre;
width: 100%;
background-color: #F7DB7D;
padding: 0 0 0 5px;
box-sizing: border-box;
}
.line-redirect {
white-space: pre;
width: 100%;
background-color: #E7CEEA;
padding: 0 0 0 5px;
box-sizing: border-box;
}
.line-ignore {
white-space: pre;
width: 100%;
color: #999;
padding: 0 0 0 5px;
box-sizing: border-box;
}
.line-stdout {
white-space: pre;
width: 100%;
background-color: #DCD8CB;
padding: 0 0 0 5px;
box-sizing: border-box;
}
a {
color: #000;
text-decoration: none;
}
a.ignore {
color: #999 !important;
}
.underlined-a {
text-decoration: underline;
}
.bold {
font-weight: bold;
}
#help {
background-color: #FFF7E1;
font-family: Arial, sans-serif;
font-size: 14px;
border-radius: 5px;
padding: 0.01em 1em 0.01em 1em;
margin-bottom: 1em;
}
#help p {
padding: 0.20em 0 0.20em 0;
}
#help p a {
text-decoration: underline;
}
.undisplayed {
display: none;
}
< / style >
< div class = "padded-page" >
< div class = "header" >
< div >
< a href = "http://archiveteam.org/index.php?title=ArchiveBot" class = "underlined-a" > ArchiveBot< / a >
tracking ~< span id = "num-crawls" > 0< / span > crawls.< br >
< / div >
< div >
< a href = "#" onclick = "ds.toggleHelp();return false;" class = "underlined-a" > Help!< / a >
< / div >
< / div >
< div id = "critical-info" >
< noscript >
Need JavaScript (ES5+) and WebSocket -> TCP:4567
< / noscript >
< div id = "help" class = "undisplayed" >
< p >
This page shows all of the crawls that < a href = "http://archiveteam.org/index.php?title=ArchiveBot" > ArchiveBot< / a > is currently running.
< / p >
< p >
To pause scrolling, move your mouse inside a log window.
< / p >
< p >
To clear all finished jobs, reload the page.
< / p >
< p >
Mouse over the job start date or the response count for additional information.
< / p >
< p >
If your adblocker is enabled for this domain, you will see slower performance, and some URLs will not be displayed.
< / p >
< p >
To use ArchiveBot, drop by < a href = "http://chat.efnet.org:9090/?nick=&channels=%23archivebot&Login=Login" > #archivebot< / a > on EFNet. < a href = "https://raw2.github.com/ArchiveTeam/ArchiveBot/master/COMMANDS" > Issue commands< / a > by typing them into the channel. You will need channel operator (@) or voice (+) status to issue archiving jobs; just ask for help or leave a message with the website you want to archive.
< / p >
< p >
These < a href = "https://github.com/ArchiveTeam/ArchiveBot/tree/master/db/ignore_patterns" > ignore sets< / a > are available for crawls. The < a href = "https://github.com/ArchiveTeam/ArchiveBot/blob/master/db/ignore_patterns/global.json" > global< / a > ignore set automatically applies to all crawls.
< / p >
< p >
GitHub: < a href = "https://github.com/ArchiveTeam/ArchiveBot" > ArchiveBot< / a > and < a href = "https://github.com/ArchiveTeam/dashboard2" > dashboard2< / a > .
< / p >
< / div >
< / div >
< div id = "traffic" > < / div >
< div id = "logs" > < / div >
< / div >
< script >
"use strict";
var assert = function(condition, message) {
if(!condition) {
throw message || "Assertion failed";
}
};
var byId = function(id) {
return document.getElementById(id);
};
var text = function(s) {
return document.createTextNode(s);
};
/**
* appendChild but accepts strings and arrays of children|strings
*/
var appendAny = function(e, thing) {
if(Array.isArray(thing)) {
for(var i=0; i < thing.length ; i + + ) {
appendAny(e, thing[i]);
}
} else if(typeof thing == "string") {
e.appendChild(text(thing));
} else {
// TODO: Check that it's actually a DOM node first
e.appendChild(thing);
}
};
/**
* Create DOM element with attributes and children from Array< node | string > |node|string
*/
var h = function(elem, attrs, thing) {
var e = document.createElement(elem);
if(attrs != null) {
for(var attr in attrs) {
if(attr == "spellcheck" || attr == "readonly") {
e.setAttribute(attr, attrs[attr]);
} else {
e[attr] = attrs[attr];
}
}
}
if(thing != null) {
appendAny(e, thing);
}
return e;
};
var href = function(href, text) {
var a = h("a");
a.href = href;
a.textContent = text;
return a;
};
var prettyJson = function(obj) {
return JSON.stringify(obj, undefined, 2);
};
// Copied from Coreweb/js_coreweb/cw/string.js
/**
* Like Python's s.split(delim, num) and s.split(delim)
* This does *NOT* implement Python's no-argument s.split()
*
* @param {string} s The string to split.
* @param {string} sep The separator to split by.
* @param {number} maxsplit Maximum number of times to split.
*
* @return {!Array.< string > } The splitted string, as an array.
*/
var split = function(s, sep, maxsplit) {
assert(typeof sep == "string",
"arguments[1] of split must be a separator string");
if(maxsplit === undefined || maxsplit < 0 ) {
return s.split(sep);
}
var pieces = s.split(sep);
var head = pieces.splice(0, maxsplit);
// after the splice, pieces is shorter and no longer has the `head` elements.
if(pieces.length > 0) {
var tail = pieces.join(sep);
head.push(tail); // no longer just the head.
}
return head;
};
/**
* [[1, 2], [3, 4]] -> {1: 2, 3: 4}
*/
var intoObject = function(arr) {
var obj = {};
arr.forEach(function(e) {
obj[e[0]] = e[1];
});
return obj;
};
var getQueryArgs = function() {
var pairs = location.search.replace("?", "").split("&");
if(pairs == "") {
return {};
}
return intoObject(pairs.map(function(e) { return split(e, "=", 1); }));
};
var isSafari = navigator.userAgent.indexOf("Safari") != -1;
/*** End of utility code ***/
var JobsTracker = function() {
this.known = {};
this.sorted = [];
this.finishedArray = [];
this.finishedSet = {};
};
JobsTracker.prototype.countActive = function() {
return this.sorted.length - this.finishedArray.length;
};
JobsTracker.prototype.resort = function() {
this.sorted.sort(function(a, b) { return a["started_at"] > b["started_at"] ? -1 : 1 });
};
/**
* Returns true if a new job was added
*/
JobsTracker.prototype.handleJobData = function(jobData) {
var ident = jobData["ident"];
var alreadyKnown = ident in this.known;
if(!alreadyKnown) {
this.known[ident] = true;
this.sorted.push(jobData);
this.resort();
}
if(jobData["finished"] & & !(ident in this.finishedSet)) {
this.finishedSet[ident] = true;
this.finishedArray.push(ident);
}
return !alreadyKnown;
};
var JobRenderInfo = function(logWindow, logSegment, statsElements, lineCountWindow, lineCountSegments) {
this.logWindow = logWindow;
this.logSegment = logSegment;
this.statsElements = statsElements;
this.lineCountWindow = lineCountWindow;
this.lineCountSegments = lineCountSegments;
};
var Reusable = {
obj_className_line_normal: {"className": "line-normal"},
obj_className_line_error: {"className": "line-error"},
obj_className_line_warning: {"className": "line-warning"},
obj_className_line_redirect: {"className": "line-redirect"},
obj_className_line_ignore: {"className": "line-ignore"},
obj_className_line_stdout: {"className": "line-stdout"},
obj_className_bold: {"className": "bold"}
};
// http://stackoverflow.com/questions/2901102/how-to-print-a-number-with-commas-as-thousands-separators-in-javascript
var numberWithCommas = function(s) {
assert(typeof s == "string", "pass string to numberWithCommas");
return s.replace(/\B(?=(\d{3})+(?!\d))/g, ",");
};
var toStringTenths = function(n) {
var s = "" + (Math.round(10 * n) / 10);
if(s.indexOf(".") == -1) {
s += ".0";
}
return s;
};
var getTotalResponses = function(jobData) {
return (
parseInt(jobData["r1xx"]) +
parseInt(jobData["r2xx"]) +
parseInt(jobData["r3xx"]) +
parseInt(jobData["r4xx"]) +
parseInt(jobData["r5xx"]) +
parseInt(jobData["runk"]));
};
var getSummaryResponses = function(jobData) {
return (
"1xx: " + numberWithCommas(String(jobData["r1xx"])) + "\n" +
"2xx: " + numberWithCommas(String(jobData["r2xx"])) + "\n" +
"3xx: " + numberWithCommas(String(jobData["r3xx"])) + "\n" +
"4xx: " + numberWithCommas(String(jobData["r4xx"])) + "\n" +
"5xx: " + numberWithCommas(String(jobData["r5xx"])) + "\n" +
"Unknown: " + numberWithCommas(String(jobData["runk"])));
};
var JobsRenderer = function(container, historyLines, showNicks) {
this.container = container;
this.historyLines = historyLines;
this.showNicks = showNicks;
this.linesPerSegment = Math.max(1, Math.round(this.historyLines / 10));
this.jobs = new JobsTracker();
// ident -> JobRenderInfo
this.renderInfo = {};
this.mouseInside = null;
this.numCrawls = byId('num-crawls');
};
JobsRenderer.prototype._getNextJobInSorted = function(ident) {
for(var i=0; i < this.jobs.sorted.length ; i + + ) {
var e = this.jobs.sorted[i];
if(e["ident"] == ident) {
return this.jobs.sorted[i+1];
}
}
return null;
};
JobsRenderer.prototype._createLogSegment = function() {
return h('div');
};
JobsRenderer.prototype._createLogContainer = function(jobData) {
var ident = jobData["ident"];
var beforeJob = this._getNextJobInSorted(ident);
var beforeElement = beforeJob == null ? null : byId("log-container-" + beforeJob["ident"]);
var logSegment = this._createLogSegment();
var logWindowAttrs = {
"className": "log-window",
"id": "log-window-" + ident,
"onmouseenter": function(ev) {
this.mouseInside = ident;
ev.target.classList.add('log-window-stopped');
}.bind(this),
"onmouseleave": function(ev) {
this.mouseInside = null;
ev.target.classList.remove('log-window-stopped');
}.bind(this)
}
// If you reach the end of a log window, the browser annoyingly
// starts to scroll the page instead. We prevent this behavior here.
// If the user wants to scroll the page, they need to move their
// mouse outside a log window first.
if(!isSafari) {
logWindowAttrs["onwheel"] = function(ev) {
// Note: offsetHeight is "wrong" by 2px but it doesn't matter
//console.log(ev, logWindow.scrollTop, (logWindow.scrollHeight - logWindow.offsetHeight));
if(ev.deltaY < 0 & & logWindow . scrollTop = = 0 ) {
ev.preventDefault();
} else if(ev.deltaY > 0 & & logWindow.scrollTop >= (logWindow.scrollHeight - logWindow.offsetHeight)) {
ev.preventDefault();
}
}
} else {
// Safari 7.0.5 can't preventDefault or stopPropagation an onwheel event,
// so use onmousewheel instead.
logWindowAttrs["onmousewheel"] = function(ev) {
//console.log(ev, logWindow.scrollTop, (logWindow.scrollHeight - logWindow.offsetHeight));
if(ev.wheelDeltaY > 0 & & logWindow.scrollTop == 0) {
ev.preventDefault();
} else if(ev.wheelDeltaY < 0 & & logWindow . scrollTop > = (logWindow.scrollHeight - logWindow.offsetHeight)) {
ev.preventDefault();
}
}
}
var statsElements = {
mb: h("span", null, "?"),
responses: h("span", null, "?"),
responsesPerSecond: h("span", null, "?"),
connections: h("span", null, "?"),
delay: h("span", null, "?"),
ignores: h("span", null, "?")
};
var logWindow = h('div', logWindowAttrs, logSegment);
var startedISOString = new Date(parseFloat(jobData["started_at"]) * 1000).toISOString();
var div = h(
'div',
{"id": "log-container-" + ident}, [
h("div", {"className": "job-header"}, [
h("span", {"className": "job-info"}, [
h("a", {"className": "job-url", "href": jobData["url"]}, jobData["url"]),
" on ",
h("span", {"title": startedISOString}, startedISOString.split("T")[0]),
(this.showNicks ? " by " + jobData["started_by"] : "") + "; ",
statsElements.mb,
h("span", null, " MB in "),
statsElements.responses,
" at ",
statsElements.responsesPerSecond,
"/s; ",
statsElements.connections,
" con. w/ ",
statsElements.delay,
" ms delay; ",
statsElements.ignores
]),
h("input", {
"className": "job-ident",
"type": "text",
"value": ident,
"size": "28",
"spellcheck": "false",
"readonly": "",
"onclick": function() { this.select(); }
})
]),
logWindow
]
);
this.renderInfo[ident] = new JobRenderInfo(logWindow, logSegment, statsElements, 0, [0]);
this.container.insertBefore(div, beforeElement);
}
JobsRenderer.prototype._renderDownloadLine = function(data, logSegment) {
if(data["is_warning"]) {
var attrs = Reusable.obj_className_line_warning;
} else if(data["is_error"]) {
var attrs = Reusable.obj_className_line_error;
} else if(data["response_code"] & & data["response_code"] >= 300 & & data["response_code"] < 400 ) {
var attrs = Reusable.obj_className_line_redirect;
} else {
var attrs = Reusable.obj_className_line_normal;
}
logSegment.appendChild(h("div", attrs, [
data["response_code"] + " " + data["wget_code"] + " ",
href(data["url"], data["url"])
]));
return 1;
};
JobsRenderer.prototype._renderIgnoreLine = function(data, logSegment) {
var attrs = Reusable.obj_className_line_ignore;
logSegment.appendChild(h("div", attrs, [
h('span', null, " IGNOR "),
h('a', {"href": data["url"], "className": "ignore"}, data["url"]),
h('span', Reusable.obj_className_bold, " by "),
data["pattern"]
]));
return 1;
};
JobsRenderer.prototype._renderStdoutLine = function(data, logSegment) {
var cleanedMessage = data["message"].replace(/[\r\n]+$/, "");
var renderedLines = 0;
if(!cleanedMessage) {
return renderedLines;
}
var lines = cleanedMessage.split("\n");
for(var i=0; i < lines.length ; i + + ) {
var line = lines[i];
if(!line) {
continue;
}
logSegment.appendChild(h("div", Reusable.obj_className_line_stdout, line));
renderedLines += 1;
}
return renderedLines;
};
JobsRenderer.prototype.handleData = function(data) {
var jobData = data["job_data"];
var added = this.jobs.handleJobData(jobData);
this.numCrawls.textContent = this.jobs.countActive();
if(added) {
this._createLogContainer(jobData);
}
var type = data["type"];
var ident = jobData["ident"];
var info = this.renderInfo[ident];
if(!info) {
console.warn("No render info for " + ident);
return;
}
var totalResponses = parseInt(getTotalResponses(jobData));
if(type == "download") {
var linesRendered = this._renderDownloadLine(data, info.logSegment);
} else if(type == "stdout") {
var linesRendered = this._renderStdoutLine(data, info.logSegment);
} else if(type == "ignore") {
var linesRendered = this._renderIgnoreLine(data, info.logSegment);
} else {
assert(false, "Unexpected message type " + type);
}
// Update stats
info.statsElements.mb.textContent =
numberWithCommas(
toStringTenths(
(parseInt(jobData["bytes_downloaded"]) / (1024 * 1024)).toString()));
info.statsElements.responses.textContent =
numberWithCommas(totalResponses.toString()) + " resp.";
info.statsElements.responses.title = getSummaryResponses(jobData);
var duration = Date.now()/1000 - parseFloat(jobData["started_at"]);
info.statsElements.responsesPerSecond.textContent =
toStringTenths(totalResponses/duration);
info.statsElements.connections.textContent = jobData["concurrency"];
var delayMin = parseInt(jobData["delay_min"]);
var delayMax = parseInt(jobData["delay_max"]);
info.statsElements.delay.textContent =
delayMin == delayMax ?
delayMin :
delayMin + "-" + delayMax;
info.statsElements.ignores.textContent =
jobData["suppress_ignore_reports"] ?
'hiding ignores' :
'showing ignores';
info.lineCountWindow += linesRendered;
info.lineCountSegments[info.lineCountSegments.length - 1] += linesRendered;
if(info.lineCountSegments[info.lineCountSegments.length - 1] >= this.linesPerSegment) {
//console.log("Created new segment", info);
var newSegment = this._createLogSegment();
info.logWindow.appendChild(newSegment);
info.logSegment = newSegment;
info.lineCountSegments.push(0);
}
if(this.mouseInside != ident) {
// We may have to remove more than one segment, if the user
// has paused the log window for a while.
while(info.lineCountWindow >= this.historyLines + this.linesPerSegment) {
var firstLogSegment = info.logWindow.firstChild;
assert(firstLogSegment != null, "info.logWindow.firstChild is null; " +
JSON.stringify({
"lineCountWindow": info.lineCountWindow,
"lineCountSegments": info.lineCountSegments}));
info.logWindow.removeChild(firstLogSegment);
info.lineCountWindow -= info.lineCountSegments[0];
info.lineCountSegments.shift();
}
// Scroll to the bottom
info.logWindow.scrollTop = info.logWindow.scrollHeight;
}
};
var BatchingQueue = function(callable, minInterval) {
this.callable = callable;
this._minInterval = minInterval;
this.queue = [];
this._timeout = null;
this._boundRunCallable = this._runCallable.bind(this);
};
BatchingQueue.prototype.setMinInterval = function(minInterval) {
this._minInterval = minInterval;
};
BatchingQueue.prototype._runCallable = function() {
this._timeout = null;
var queue = this.queue;
this.queue = [];
this.callable(queue);
};
BatchingQueue.prototype.callNow = function() {
if(this._timeout !== null) {
clearTimeout(this._timeout);
this._timeout = null;
}
this._runCallable();
};
BatchingQueue.prototype.push = function(v) {
this.queue.push(v);
if(this._timeout === null) {
this._timeout = setTimeout(this._boundRunCallable, this._minInterval);
}
};
var Decayer = function(initial, multiplier, max) {
this.initial = initial;
this.multiplier = multiplier;
this.max = max;
this.reset();
};
Decayer.prototype.reset = function() {
// First call to .decay() will multiply, but we want to get the `intitial`
// value on the first call to .decay(), so divide.
this.current = this.initial / this.multiplier;
return this.current;
};
Decayer.prototype.decay = function() {
this.current = Math.min(this.current * this.multiplier, this.max);
return this.current;
};
var Dashboard = function() {
this.messageCount = 0;
var args = getQueryArgs();
var historyLines =
args["historyLines"] ?
Number(args["historyLines"]) :
navigator.userAgent.match(/Mobi/) ? 250 : 1000;
var batchTimeWhenVisible =
args["batchTimeWhenVisible"] ?
Number(args["batchTimeWhenVisible"]) :
125;
var showNicks =
args["showNicks"] ?
Number(args["showNicks"]) :
false;
this.dumpTraffic = args["dumpMax"] & & Number(args["dumpMax"]) > 0;
if(this.dumpTraffic) {
this.dumpMax = Number(args["dumpMax"]);
}
this.jobsRenderer = new JobsRenderer(byId('logs'), historyLines, showNicks);
var batchTimeWhenHidden = 5000;
var xhr = new XMLHttpRequest();
xhr.onload = function() {
var recentLines = JSON.parse(xhr.responseText);
for(var i=0; i < recentLines.length ; i + + ) {
this.handleData(recentLines[i]);
}
this.queue = new BatchingQueue(function(queue) {
//console.log("Queue has ", queue.length, "items");
for(var i=0; i < queue.length ; i + + ) {
this.handleData(JSON.parse(queue[i]));
}
}.bind(this), batchTimeWhenVisible);
this.decayer = new Decayer(1000, 1.5, 60000);
this.connectWebSocket();
document.addEventListener("visibilitychange", function() {
if(document.hidden) {
//console.log("Page has become hidden");
this.queue.setMinInterval(batchTimeWhenHidden);
} else {
//console.log("Page has become visible");
this.queue.setMinInterval(batchTimeWhenVisible);
this.queue.callNow();
}
}.bind(this), false);
}.bind(this);
2014-09-04 23:07:11 +00:00
xhr.open("GET", "/logs/recent?cb=" + Date.now() + Math.random());
2013-09-23 16:13:38 -05:00
xhr.send("");
};
Dashboard.prototype.handleData = function(data) {
this.messageCount += 1;
if(this.dumpTraffic & & this.messageCount < = this.dumpMax) {
byId('traffic').appendChild(h("pre", null, prettyJson(data)));
}
this.jobsRenderer.handleData(data);
};
Dashboard.prototype.connectWebSocket = function() {
2014-09-04 23:07:11 +00:00
this.ws = new WebSocket("ws://" + location.host + "/stream");
2013-09-23 16:13:38 -05:00
this.ws.onmessage = function(ev) {
this.queue.push(ev["data"]);
}.bind(this);
this.ws.onopen = function(ev) {
console.log("WebSocket opened:", ev);
this.decayer.reset();
}.bind(this);
this.ws.onclose = function(ev) {
console.log("WebSocket closed:", ev);
var delay = this.decayer.decay();
console.log("Reconnecting in", delay, "ms");
setTimeout(this.connectWebSocket.bind(this), delay);
}.bind(this);
};
Dashboard.prototype.toggleHelp = function() {
var help = byId('help');
if(help.classList.contains('undisplayed')) {
help.classList.remove('undisplayed');
} else {
help.classList.add('undisplayed');
}
};
var ds = new Dashboard();
< / script >
< / body >
< / html >
