// ==UserScript==
// @name Make Bookmarklets from Javascript URLs
// @namespace MBFU
// @description When it sees a link to a userscript or general Javascript URL, adds a Bookmarklet besides it, which you can drag to your toolbar to load the script when you next need it (running outside Greasemonkey of course).
// @include http://hwi.ath.cx/*/gm_scripts/*
// @include http://hwi.ath.cx/*/userscripts/*
// @include http://*userscripts.org/*
// @include https://*userscripts.org/*
// @include http://openuserjs.org/*
// @include https://openuserjs.org/*
// @include http://greasyfork.org/*
// @include https://greasyfork.org/*
// @include http://*/*
// @include https://*/*
// @exclude http://hwi.ath.cx/code/other/gm_scripts/joeys_userscripts_and_bookmarklets_overview.html
// @exclude http://neuralyte.org/~joey/gm_scripts/joeys_userscripts_and_bookmarklets_overview.html
// @grant none
// @version 1.2.7
// ==/UserScript==
// BUG: We had (i%32) in a userscript (DLT) but when this was turned into a bookmarklet and dragged into Chrome, the debugger showed it had become (i2), causing the script to error with "i2 is not defined". Changing the code to (i % 32) worked around the problem.
// TODO: All links with using dummy=random should change the value on mouseover/mousemove/click, so they can be re-used live without refreshing the page.
// Static bookmarklets could timeout and re-build after ... 10 seconds? ^^
// Most people will NOT want the NoCache version. It's only useful for script developers.
// Firefox/Greasemonkey does not like to install scripts ending ".user.js?dummy=123"
// But in Chrome it is a useful way to prevent the script from being cached when you are developing it.
var inGoogleChrome = window && window.navigator && window.navigator.vendor.match(/Google/);
// It doesn't make much sense for me to make it browser-dependent, because I share my bookmarks between Chrome and Firefox (using XMarks).
var preventBrowserFromCachingBookmarklets = inGoogleChrome;
//var preventBrowserFromCachingBookmarklets = true;
// Sometimes Chrome refuses to acknowledge that a script has been updated, and repeatedly installs an old version from its cache!
var preventCachingOfInstallScripts = inGoogleChrome;
// BUG: Chrome sometimes installs a new instance of an extension for each click, rather than overwriting (upgrading) the old. However disabling preventBrowserFromCachingBookmarklets does not fix that. It may have been the name "Wikimedia+"?
var addGreasemonkeyLibToBookmarklets = true;
// If enabled, as well as *.user.js files, it will also offer bookmarklets for *.js files
var activateOnAllJavascriptFiles = false;
// DONE: All bookmarklets optionally preload the Fallback GMAPI.
// DONE: All bookmarklets optionally load in non-caching fashion (for changing scripts).
// DONE: Use onload event to ensure prerequisite scripts are loaded before dependent scripts.
// DONE via FBGMAPI: We could provide neat catches for GM_ API commands so they won't fail entirely.
// TODO: Provide extra feature, which allows the Bookmarks to actually trigger
// the userscript properly-running inside Greasemonkey, if this userscript is
// present to handle the request, otherwise (with warning) load outside GM.
// TODO: Support @include/@excude and @require meta rules?
// This requires parsing the script's header using XHR before creating each bookmarklet.
// DONE: Optionally create static bookmarklet, with all code inline, rather than loaded from a URL.
// // comments will need to be removed or converted to /*...*/ comments
var addDateToStaticBookmarklets = true;
var defaultScripts = [];
var includeGMCompat = addGreasemonkeyLibToBookmarklets;
if (includeGMCompat) {
// defaultScripts.push("http://hwi.ath.cx/code/other/gm_scripts/fallbackgmapi/fallbackgmapi.user.js");
defaultScripts.push("//neuralyte.org/~joey/gm_scripts/fallbackgmapi/fallbackgmapi.user.js");
}
function buildLiveBookmarklet(url) {
var neverCache = preventBrowserFromCachingBookmarklets;
var scriptsToLoad = defaultScripts.slice(0);
// The bookmarklet should load scripts using the same protocol as the page. Otherwise browsers may give error messages about "mixed content", and refuse to load the script.
url = url.replace(/^http[s]*:[/][/]/, "//");
scriptsToLoad.push(url);
var neverCacheStr = ( neverCache ? "+'?dummy='+new Date().getTime()" : "" );
/*
var toRun = "(function(){\n";
for (var i=0;i<scriptsToLoad.length;i++) {
var script = scriptsToLoad[i];
toRun += " var newScript = document.createElement('script');\n";
toRun += " newScript.src = '" + script + "'" + neverCacheStr + ";\n";
toRun += " document.body.appendChild(newScript);\n";
}
toRun += "})();";
*/
var toRun = "(function(){\n";
// Chrome has no .toSource() or uneval(), so we use JSON. :f
toRun += "var scriptsToLoad="+JSON.stringify(scriptsToLoad)+";\n";
toRun += "function loadNext() {\n";
toRun += " if (scriptsToLoad.length == 0) { return; }\n";
toRun += " var next = scriptsToLoad.shift();\n";
toRun += " var newScript = document.createElement('script');\n";
toRun += " newScript.src = next"+neverCacheStr+";\n";
toRun += " newScript.onload = loadNext;\n";
toRun += " newScript.onerror = function(e){ console.error('Problem loading script: '+next,e); };\n";
toRun += " document.body.appendChild(newScript);\n";
toRun += "}\n";
toRun += "loadNext();\n";
toRun += "})(); (void 0);";
var name = getNameFromFilename(url);
/*
if (neverCache) {
name = name + " (NoCache)";
}
if (includeGMCompat) {
name = name + " (FBAPI)";
}
*/
var newLink = document.createElement("A");
newLink.href = "javascript:" + toRun;
newLink.textContent = name;
newLink.title = newLink.href;
var newContainer = document.createElement("div");
// Act like a span, do not wrap onto separate lines
// Removed because it was problematic on some sites (which?)
//newContainer.style.whiteSpace = 'nowrap';
newContainer.appendChild(document.createTextNode("(Live Bookmarklet: "));
newContainer.appendChild(newLink);
var extraString = ( neverCache || includeGMCompat ? neverCache && includeGMCompat ? " (no-caching, with GM fallbacks)" : neverCache ? " (no-caching)" : " (with GM fallbacks)" : "" );
// DISABLED HERE:
extraString = "";
if (extraString) {
// newContainer.appendChild(document.createTextNode(extraString));
// var extraText = document.createElement("span");
// extraText.style.fontSize = '80%';
var extraText = document.createElement("small");
extraText.textContent = extraString;
newContainer.appendChild(extraText);
}
newContainer.appendChild(document.createTextNode(")"));
newContainer.style.paddingLeft = '8px';
// link.parentNode.insertBefore(newContainer,link.nextSibling);
return newContainer;
}
function reportMessage(msg) {
console.log(msg);
}
function reportWarning(msg) {
console.warn(msg);
}
function reportError(msg) {
console.error(msg);
}
function doesItCompile(code) {
try {
var f = new Function(code);
} catch (e) {
return false;
}
return true;
}
/* For more bugs try putting (function(){ ... })(); wrapper around WikiIndent! */
function fixComments(line) {
//// Clear // comment line
line = line.replace(/^[ \t]*\/\/.*/g,'');
//// Wrap // comment in /*...*/ (Dangerous! if comment contains a */ !)
// line = line.replace(/^([ \t]*)\/\/(.*)/g,'$1/*$2*/');
//// Clear trailing comment (after a few chars we deem sensible)
//// This still doesn't handle some valid cases.
var trailingComment = /([;{}()\[\],\. \t])\s*\/\/.*/g;
//// What might break: An odd number of "s or 's, any /*s or */s
var worrying = /(["']|\/\*|\*\/)/;
// Here is a breaking example:
// var resNodes = document.evaluate("//div[@id='res']//li", document, null, XPathResult.UNORDERED_NODE_SNAPSHOT_TYPE, null);
// var hasTrailingComment = line.match(trailingComment);
var hasTrailingComment = trailingComment.exec(line);
if (hasTrailingComment) {
/*
if (line.match(worrying)) {
reportWarning("Warning: trailingComment matches: "+hasTrailingComment);
}
*/
var compiledBefore = doesItCompile(line);
var newLine = line.replace(trailingComment,'$1');
var compilesAfter = doesItCompile(newLine);
if (compiledBefore && !compilesAfter) {
reportWarning("Aborted // stripping on: "+line);
} else {
// Accept changes
line = newLine;
}
}
return line;
}
function cleanupSource(source) {
// console.log("old length: "+source.length);
var lines = source.split('\n');
// console.log("lines: "+lines.length);
for (var i=0;i<lines.length;i++) {
lines[i] = fixComments(lines[i]);
}
// source = lines.join('\n');
source = lines.join(' ');
// console.log("new length: "+source.length);
//// For Bookmarklet Builder's reformatter:
source = source.replace("(function","( function",'g');
return source;
}
// We could cache urls, at least during this one page visit
// Specifically the ones we request repeatedly for statis bmls (fbgmapi).
var sourcesLoaded = {};
// My first "promise":
function getSourceFor(url) {
return {
then: function(handleResponse){
// Prevent caching of this resource. I found this was needed one time on Chrome!
// It probably doesn't need to be linked to preventCachingOfInstallScripts or preventBrowserFromCachingBookmarklets.
url += "?dummy="+Math.random();
console.log("Loading "+url);
getURLThen(url, handlerFn, onErrorFn);
function handlerFn(res) {
var source = res.responseText;
handleResponse(source);
}
function onErrorFn(res) {
reportError("Failed to load "+url+": HTTP "+res.status);
}
// CONSIDER: We should return the next promise?
}
};
}
/* To avoid a multitude of premature network requests, the bookmarklet is not actually "compiled" until mouseover. */
function buildStaticBookmarklet(url) {
var newLink = document.createElement("a");
newLink.textContent = getNameFromFilename(url);
var newContainer = document.createElement("div");
// newContainer.style.whiteSpace = 'nowrap';
// Experimental:
newContainer.appendChild(document.createTextNode("(Static Bookmarklet: "));
newContainer.appendChild(newLink);
newContainer.appendChild(document.createTextNode(")"));
newContainer.style.paddingLeft = '8px';
newLink.style.textDecoration = 'underline';
// newLink.style.color = '#000080';
newLink.style.color = '#333366';
// link.parentNode.insertBefore(newContainer,link.nextSibling);
// The href may change before we fire (e.g. if dummy is appended) so we make a copy.
var href = url;
// setTimeout(function(){
// ,2000 * staticsRequested);
newLink.onmouseover = function(){
getStaticBookmarkletFromUserscript(href,whenGot);
newLink.style.color = '#ff7700';
newLink.onmouseover = null; // once
};
function whenGot(staticSrc, report) {
var extraText = "";
// Does it parse?
try {
var testFn = new Function(staticSrc);
newLink.style.color = ''; // Success! Color like a normal link
} catch (e) {
var msg = "PARSE FAILED";
// Firefox has this:
if (e.lineNumber) {
msg += " on line "+e.lineNumber;
}
msg += ":";
console.log("["+href+"] "+msg,e);
newLink.title = msg+" "+e;
newLink.style.color = 'red'; // color as error
window.lastParseError = e;
// return;
}
newLink.href = "javascript:" + staticSrc;
if (addDateToStaticBookmarklets) {
var d = new Date();
//var dateStr = d.getFullYear()+"."+(d.getMonth()+1)+"."+d.getDate();
var dateStr = d.toISOString().substring(0,10);
newLink.textContent = newLink.textContent + " ("+dateStr+")";
}
if (report.needsGMFallbacks) {
extraText += " uses GM";
} else {
extraText += " GM free";
}
if (extraText) {
newLink.parentNode.insertBefore( document.createTextNode(extraText), newLink.nextSibling);
}
// Just in case the browser is dumb, force it to re-analyse the link.
newLink.parentNode.insertBefore(newLink, newLink.nextSibling);
}
return newContainer;
}
function getStaticBookmarkletFromUserscript(href, callbackGotStatic) {
getSourceFor(href).then(function (userscriptSource) {
var hasGrantNone = userscriptSource.match('\\n//\\s*@grant\\s+none\\s*\\n');
var needsGMFallbacks = ! hasGrantNone;
var scriptsToLoad = needsGMFallbacks ? defaultScripts.slice(0) : [];
loadScripts(scriptsToLoad, allSourcesLoaded);
function allSourcesLoaded(scriptSources) {
scriptSources.push(userscriptSource);
var toRun = "";
for (var i=0; i<scriptSources.length; i++) {
if (!scriptSources[i]) {
reportError("Expected contents of "+scriptsToLoad[i]+" but got: "+scriptSources[i]);
}
var cleaned = cleanupSource(scriptSources[i]);
toRun += "(function(){\n";
toRun += cleaned;
toRun += "})();\n";
}
toRun += "(void 0);";
var report = {
needsGMFallbacks: needsGMFallbacks,
};
callbackGotStatic(toRun, report);
}
});
function loadScripts(scriptsToLoad, callbackScriptsLoaded) {
if (scriptsToLoad.length === 0) {
return callbackScriptsLoaded([]);
}
var scriptSources = [];
var numLoaded = 0;
for (var i=0; i<scriptsToLoad.length; i++) {
loadSourceIntoArray(i);
}
function loadSourceIntoArray(i) {
getSourceFor(scriptsToLoad[i]).then(function(source){
if (!source) {
reportError("Failed to acquire source for: "+href);
}
scriptSources[i] = source;
numLoaded++;
if (numLoaded == scriptsToLoad.length) {
callbackScriptsLoaded(scriptSources);
}
});
}
}
}
// In this case, link points to a folder containing a userscript.
// We guess the userscript's name from the folder's name.
function addQuickInstall(link) {
if (link.parentNode.tagName == 'TD') {
link.parentNode.style.width = '60%';
}
var br2 = document.createElement("br");
link.parentNode.insertBefore(br2,link.nextSibling);
var br = document.createElement("br");
link.parentNode.insertBefore(br,link.nextSibling.nextSibling);
var name = link.href.match(/([^\/]*)\/$/)[1];
var newLink = document.createElement("A");
newLink.href = link.href + name+".user.js";
newLink.textContent = "Install Userscript"; // name+".user.js";
var newContainer = document.createElement("span");
newContainer.appendChild(document.createTextNode(" ["));
newContainer.appendChild(newLink);
newContainer.appendChild(document.createTextNode("]"));
newContainer.style.paddingLeft = '8px';
link.parentNode.insertBefore(newContainer,br);
link.style.color = 'black';
link.style.fontWeight = 'bold';
newContainer.appendChild(buildLiveBookmarklet(newLink.href));
newContainer.appendChild(buildStaticBookmarklet(newLink.href));
newContainer.appendChild(buildLiveUserscript(newLink));
popupSourceOnHover(newLink);
// Do this after the other two builders have used the .href
if (preventCachingOfInstallScripts) {
newLink.href = newLink.href + '?dummy='+new Date().getTime();
}
}
function getURLThen(url,handlerFn,onErrorFn) {
var req = new XMLHttpRequest();
req.open("get", url, true);
req.onreadystatechange = function (aEvt) {
if (req.readyState == 4) {
if(req.status == 200) {
// Got it
handlerFn(req);
} else {
var msg = ("XHR failed with status "+req.status+"\n");
window.status = msg;
onErrorFn(req);
console.warn(msg);
}
}
};
req.send(null);
}
var frame = null;
function loadSourceViewer(url, newLink, evt) {
// window.lastEVT = evt;
if (frame && frame.parentNode) {
frame.parentNode.removeChild(frame);
}
frame = document.createElement("div");
function reportToFrame(msg) {
frame.appendChild( document.createTextNode(msg) );
}
// This doesn't work. Loading it directly into the iframe triggers Greasemonkey to install it!
//frame.src = url;
// What we need to do is get the script with an XHR, then place it into the div.
// This seems to fire immediately in Firefox!
var cleanup = function(evt) {
frame.parentNode.removeChild(frame);
document.body.removeEventListener("click",cleanup,false);
// frame.removeEventListener("mouseout",cleanup,false);
};
document.body.addEventListener("click",cleanup,false);
getURLThen(url, function(res){
// We were using pre instead of div to get monospace like <tt> or <code>
// However since we are commonly reading the description, sans seems better.
var displayDiv = document.createElement("div");
displayDiv.style.fontSize = '0.8em';
displayDiv.style.whiteSpace = "pre-wrap";
var displayCode = document.createElement("pre");
displayCode.textContent = res.responseText;
displayCode.style.maxHeight = "100%";
displayCode.style.overflow = "auto";
displayDiv.appendChild(displayCode);
while (frame.firstChild) {
frame.removeChild(frame.firstChild);
}
frame.appendChild(displayDiv);
if (typeof Rainbow != null) {
/*
// Works fine in Chrome, but in Firefox it causes Rainbow to fail with "too much recursion".
Rainbow.extend('javascript', [
{
'name': 'importantcomment',
'pattern': /(\/\/|\#) @(name|description|include) [\s\S]*?$/gm
},
], false);
*/
setTimeout(function(){
displayCode.setAttribute('data-language', "javascript");
displayCode.style.fontSize = '100%';
Rainbow.color(displayCode.parentNode, function(){
console.log("Rainbow finished.");
});
},50);
}
// frame.addEventListener("mouseout",cleanup,false);
// newLink.title = res.responseText;
var lines = res.responseText.split("\n");
for (var i=0;i<lines.length;i++) {
var line = lines[i];
if (line.match(/@description\s/)) {
var descr = line.replace(/.*@description\s*/,'');
newLink.title = descr;
break;
}
}
}, function(res){
reportToFrame("Failed to load "+url+": HTTP "+res.status);
});
/*
frame.style.position = 'fixed';
frame.style.top = evt.clientY+4+'px';
frame.style.left = evt.clientX+4+'px';
*/
// frame.style.position = 'absolute';
// frame.style.top = evt.layerY+12+'px';
// frame.style.left = evt.layerX+12+'px';
// frame.style.top = evt.layerY - window.innerHeight*35/100 + 'px';
// frame.style.left = evt.layerX + 64 + 'px';
// frame.style.width = "70%";
// frame.style.height = "70%";
frame.style.position = 'fixed';
frame.style.right = '2%';
frame.style.width = '50%';
frame.style.top = '10%';
frame.style.height = '80%';
frame.style.backgroundColor = 'white';
frame.style.color = 'black';
frame.style.padding = '8px';
frame.style.border = '2px solid #555555';
document.body.appendChild(frame);
reportToFrame("Loading...");
}
function buildSourceViewer(url) {
var newLink = document.createElement("A");
// newLink.href = '#';
newLink.textContent = "Source";
newLink.addEventListener('click',function(e) {
loadSourceViewer(url, newLink, e);
},false);
// TODO: Problem with .user.js files and Chrome:
// In Chrome, opens an empty iframe then the statusbar says it wants to install an extension.
// For Chrome we could try: frame.src = "view-source:"+...;
var extra = document.createElement("span");
extra.appendChild(document.createTextNode("["));
extra.appendChild(newLink);
extra.appendChild(document.createTextNode("]"));
extra.style.paddingLeft = '8px';
// link.parentNode.insertBefore(extra,link.nextSibling);
return extra;
}
function popupSourceOnHover(link) {
var hoverTimer = null;
function startHover(evt) {
stopHover(evt);
hoverTimer = setTimeout(function(){
loadSourceViewer(link.href, link, evt);
stopHover(evt);
// link.removeEventListener("mouseover",startHover,false);
// link.removeEventListener("mouseout",stopHover,false);
},1500);
}
function stopHover(evt) {
clearTimeout(hoverTimer);
hoverTimer = null;
}
link.addEventListener("mouseover",startHover,false);
link.addEventListener("mouseout",stopHover,false);
// If they click on it before waiting to hover, they probably don't want the popup:
link.addEventListener("click",stopHover,false);
}
function buildLiveUserscript(link) {
//// This isn't working any more. data:// lost its power circa 2006 due to abuse.
//// Create a clickable link that returns a sort-of file to the browser using the "data:" protocol.
//// That file would be a new userscript for installation.
//// We can generate the contents of this new userscript at run-time.
//// The current one we generate runs (no @includes), and loads the latest userscript from its website via script injection.
/* DISABLED
// BUG: data:{...}.user.js does not interest my Chromium
var name = getNameFromFilename(link.href)+" Live!";
var name = "Install LiveLoader";
var whatToRun = '(function(){\n'
+ ' var ns = document.createElement("script");\n'
+ ' ns.src = "' + encodeURI(getCanonicalUrl(link.href)) + '";\n'
+ ' document.getElementsByTagName("head")[0].appendChild(ns);\n'
+ '})();\n';
var newLink = document.createElement("a");
newLink.textContent = name;
newLink.href = "data:text/javascript;charset=utf-8,"
+ "// ==UserScript==%0A"
+ "// @namespace LiveLoader%0A"
+ "// @name " + name + " LIVE%0A"
+ "// @description Loads " +name+ " userscript live from " + link.href + "%0A"
+ "// ==/UserScript==%0A"
+ "%0A"
+ encodeURIComponent(whatToRun) + "%0A"
+ "//.user.js";
var extra = document.createElement("span");
extra.appendChild(document.createTextNode("["));
extra.appendChild(newLink);
extra.appendChild(document.createTextNode("]"));
extra.style.paddingLeft = '8px';
link.parentNode.insertBefore(extra,link.nextSibling);
*/
return document.createTextNode("");
}
function getCanonicalUrl(url) {
if (url.substring(0,1)=="/") {
url = document.location.protocol + "://" + document.location.domain + "/" + url;
}
if (!url.match("://")) {
url = document.location.href.match("^[^?]*/") + url;
}
return url;
}
function getNameFromFilename(href) {
var isUserscript = href.match(/\.user\.js$/);
// Remove any leading folders and trailing ".user.js"
var name = href.match(/[^\/]*$/)[0].replace(/\.user\.js$/,'');
name = decodeURIComponent(name);
// The scripts on userscripts.org do not have their name in the filename,
// but we can get the name from the page title!
if (document.location.host=="userscripts.org" && document.location.pathname=="/scripts/show/"+name) {
var scriptID = name;
name = document.title.replace(/ for Greasemonkey/,'');
// Optionally, include id in name:
name += " ("+scriptID+")";
}
if (isUserscript) {
var words = name.split("_");
for (var i=0;i<words.length;i++) {
if (words[i].length) {
var c = words[i].charCodeAt(0);
if (c>=97 && c<=122) {
c = 65 + (c - 97);
words[i] = String.fromCharCode(c) + words[i].substring(1);
}
}
}
name = words.join(" ");
} else {
// It's just a Javascript file
name = "Load "+name;
}
return name;
}
var links = document.getElementsByTagName("A");
//// We used to process backwards (for less height recalculation).
//// But this was messing up maxStaticsToRequest.
//// But now backwards processing is restored, to avoid producing multiple bookmarks!
for (var i=links.length;i--;) {
//for (var i=0;i<links.length;i++) {
var link = links[i];
if (link.getAttribute('data-make-bookmarklet') === 'false') {
continue;
}
var regexp = activateOnAllJavascriptFiles ? /\.js$/ : /\.user\.js/;
// If we see a direct link to a user script, create buttons for it.
if (link.href.match(regexp)) {
var url = link.href;
/*
if (link.hostname === 'github.com' && !link.href.match(/[/]raw[/]/)) {
// Ignore links to github.com/foo/bar/blob/master/script.js
// Accept links to github.com/foo/bar/raw/master/script.js
continue;
}
*/
// Turn github /blob/ links into /raw/ links
if (link.hostname === 'github.com' && !link.href.match(/[/]raw[/]/)) {
url = url.replace(/[/]blob[/]/, '/raw/');
}
var where = link;
function insert(newElem) {
where.parentNode.insertBefore(newElem,where.nextSibling);
where = newElem;
}
insert(buildLiveBookmarklet(url));
insert(buildStaticBookmarklet(url));
insert(buildLiveUserscript(link));
insert(buildSourceViewer(url));
}
// If the current page looks like a Greasemonkey Userscript Folder, then
// create an installer for every subfolder (assuming a script is inside it).
if (document.location.pathname.match(/\/(gm_scripts|userscripts)\//)) {
if (link.href.match(/\/$/) && link.textContent!=="Parent Directory") {
addQuickInstall(link);
}
}
}
/*
var promise(getURLThen,url) {
var handler;
getURLThen(url,handlerFn,handlerFn);
function handlerFn(res) {
var source = res.responseText;
if (handler) {
handler(source);
} else {
reportError("No handler set for: "+
}
}
return {
then: function(handleResponse){
handler = handleResponse;
}
};
}
*/
