Develop Reference

Antlr4 Javascript Visitor

I'm currently trying to develope a JavaScript Compiler with the help of an Antlr4 Visitor. I've got this already implemented with Java but cannot figure out how to do this in JavaScript. Probably somebody can answer me a few questions?
1: In Java there is a Visitor.visit function. If im right this isn't possibile with Javascript. Is there a work around for this?
2: My Javascript Visitor got all the generated visiting functions but when I use console.log(ctx) the context is undefined. Any idea why?
Extract from the SimpleVisitor.js:
// Visit a parse tree produced by SimpleParser#parse.
SimpleVisitor.prototype.visitParse = function(ctx) {
console.log(ctx);
};
Main js file:
var antlr4 = require('lib/antlr4/index');
var SimpleLexer = require('antlr4/SimpleLexer');
var SimpleParser = require('antlr4/SimpleParser');
var SimpleVisitor = require('antlr4/SimpleVisitor');
var input = "double hallo = 1;";
var chars = new antlr4.InputStream(input);
var lexer = new SimpleLexer.SimpleLexer(chars);
var tokens = new antlr4.CommonTokenStream(lexer);
var parser = new SimpleParser.SimpleParser(tokens);
var visitor = new SimpleVisitor.SimpleVisitor();
parser.buildParseTrees = true;
var tree = parser.parse();
visitor.visitParse();
This is probably enough to start with ...
Bruno
Edit:
Probably the context is undefined because I call the function without arguments but where do I get the "starting"-context?
Edit2:
So I think I get the idea how this should work out. One Question remaining how do I determine which rule to call next inside each visitor function?

The basic idea behind the visitor is that you have to handle all the logic by yourself. To do this I generated the visitor using antlr. My own visitor overrides all functions that I need to implement my logic.
create lexer, tokens, ...
var antlr4 = require('antlr4/index');
var SimpleJavaLexer = require('generated/GrammarLexer');
var SimpleJavaParser = require('generated/GrammarParser');
var SimpleJavaVisitor = require('generated/GrammarVisitor');
var Visitor = require('./Visitor');
var input = "TestInput";
var chars = new antlr4.InputStream(input);
var lexer = new GrammarLexer.GrammarLexer(chars);
var tokens = new antlr4.CommonTokenStream(lexer);
var parser = new GrammarParser.GrammarParser(tokens);
var visitor = new Visitor.Visitor();
parser.buildParseTrees = true;
var tree = parser.parse();
and call your entry function
visitor.visitTest(tree);
inside your new visitor you need to implement your new logic to determine which function to call next (the right context as argument is important)
var GrammarVisitor = require('generated/GrammarVisitor').GrammarVisitor;
function Visitor () {
SimpleJavaVisitor.call(this);
return this;
};
Visitor.prototype = Object.create(GrammarVisitor.prototype);
Visitor.prototype.constructor = Visitor;
Visitor.prototype.visitTest = function(ctx) {
// implement logic to determine which function to visit
// then call next function and with the right context
this.visitBlock(ctx.block());
};
I hope you can understand my basic idea. If anybody got any questions just comment.

document generation only works the first time

I'm using openxml in my HTML5 mobile app to generate word documents on the mobile device.
In general openxml works fine and straight forward, but I'm struggling with an annyoing problem.
The document generation only works the first time after I've started the app. This time I can open and view the document. Restart the app means:
- Redeploy from development machine
- Removing the app from the task pane (pushing aside; I assume the app is removed then?)
The second time I get the message the document is corrupted and I'm unable to view the file
UPDATE:
I can't reproduce this behaviour when I'm running the app connected to the remote debugger without having a breakpoint set. Doing it this way I always get a working document.
I doesn't make a difference wether I do any changes on the document or not. Simply open and saving reproduce this error.
After doing some research I've found that structure of the docx.zip file of the working and the corrupt file is the same. They also have the same file length. But in the corrupt docx there are some files I've found some files having a wrong/invalid CRC. See here an example when trying to get a corrupt file out of the zip. Other files are working as expected.
The properties for this file are->
(CRC in a working version is: 44D3906C)
Code for processing the doc-template:
/*
* Process the template
*/
function processTemplate(doc64, callback)
{
"use strict";
console.log("PROCESS TEMPLATE");
var XAttribute = Ltxml.XAttribute;
var XCData = Ltxml.XCData;
var XComment = Ltxml.XComment;
var XContainer = Ltxml.XContainer;
var XDeclaration = Ltxml.XDeclaration;
var XDocument = Ltxml.XDocument;
var XElement = Ltxml.XElement;
var XName = Ltxml.XName;
var XNamespace = Ltxml.XNamespace;
var XNode = Ltxml.XNode;
var XObject = Ltxml.XObject;
var XProcessingInstruction = Ltxml.XProcessingInstruction;
var XText = Ltxml.XText;
var XEntity = Ltxml.XEntity;
var cast = Ltxml.cast;
var castInt = Ltxml.castInt;
var W = openXml.W;
var NN = openXml.NoNamespace;
var wNs = openXml.wNs;
var doc = new openXml.OpenXmlPackage(doc64);
// add a paragraph to the beginning of the document.
var body = doc.mainDocumentPart().getXDocument().root.element(W.body);
var tpl_row = ((doc.mainDocumentPart().getXDocument().descendants(W.tbl)).elementAt(1).descendants(W.tr)).elementAt(2);
var newrow = new XElement(tpl_row);
doc.mainDocumentPart().getXDocument().descendants(W.tbl).elementAt(1).add(newrow);
// callback(doc);
var mod_file = null;
var newfile;
var path;
if (doc != null && doc != undefined ) {
mod_file = doc.saveToBlob();
// Start writing document
path = "Templates";
newfile = "Templates/Bau.docx";
console.log("WRITE TEMPLATE DOCUMENT");
fs.root.getFile("Templates/" + "MyGenerated.docx", {create: true, exclusive: false},
function(fileEntry)
{
fileEntry.createWriter(
function(fileWriter)
{
fileWriter.onwriteend = function(e) {
console.log("TEMPLATE DOCUMENT WRITTEN:"+e.target.length);
};
fileWriter.onerror = function(e) {
console.log("ERROR writing DOCUMENT:" + e.code + ";" + e.message);
};
var blobreader = new FileReader();
blobreader.onloadend = function()
{
fileWriter.write(blobreader.result); // reader.result contains the contents of blob as a typed array
};
blobreader.readAsArrayBuffer(mod_file);
},
null);
}, null);
};
Any ideas what I'm doing wrong?

Thanks for posting about the error. There were some issues with jszip.js that I encountered when I was developing the Open XML SDK for JavaScript.
At the following link, there is a sample javascript app that demonstrates generating a document.
Open XML SDK for JavaScript Demo
In that app you can save multiple DOCXs, one after another, and they are not corrupted.
In order to work on this issue, I need to be able to re-produce locally. Maybe you can take that little working web app and replace parts with your parts until it is generating invalid files?
Cheers, Eric
P.S. I am traveling and have intermittent access to internet. If you can continue the thread on OpenXmlDeveloper.org, then it will help me to answer quicker. :-)

What made it work for me, was changing the way of adding images (Parts) to the document. I was using the type "binary" for adding images to document. I changed this to "base64"
So I changed the source from:
mydoc.addPart( "/word/"+reltarget, openXml.contentTypes.png, "binary", fotodata ); // add Image Part to doc
to:
mydoc.addPart( "/word/"+reltarget, openXml.contentTypes.png, "base64", window.btoa(fotodata) ); // add Image Part to doc

How to Get args from SP.UI.ModalDialog?

I have tried other online suggestions without success.
So...
My function opening a SharePoint dialog passes agrs into the prescribed option object, like so:
SETTING UP THE DIALOG:
Nothing magical here...
function openEmailDialog() {
var options = SP.UI.$create_DialogOptions(),
url = '../Pages/EmailDocument.aspx';
options.title = "Email Documents";
options.width = 1024;
options.height = 400;
options.allowMaximize = false;
options.url = url;
options.args = { DidYouGetThis: true };
SP.UI.ModalDialog.showModalDialog(options);
};
Next...
Upon opening the target URL, most online examples recommend the following JavaScript to extract the args BACK from the dialog, like so:
GETTING THE ARGS:
Remember, this is JavaScript in a new page which was just opened as a dialog...
$(document).ready(function () {
// This fails because "get_childDialog" doesn't exist
var args = SP.UI.ModalDialog.get_childDialog().get_args();
});
This fails because the SP.UI.ModalDialog object has no get_childDialog function.

Use var args = window.frameElement.dialogArgs;
The article I used for reference.
Live Article.

Here is a tutorial discussing this very concept. - http://www.sharepointdevelopment.me/2011/06/passing-data-to-and-from-sharepoint-modal-dialogs/

Browser crashes after 10-15 mins

In my app I'm displaying 10 charts (charts are from dygraphs.) to monitor data. For displaying charts I'm getting data from my sever by sending ajax request to 4 servlets on every 5 seconds. After 10-15 mins (don't know exact time.) my browser crashes saying "aw!! snap." What could be the reason? Is it javascript that is causing it? or is it because I'm sending request every 5 seconds?
Browser tested: Firefox and Chorme.
Note:- When I refresh the browser after crash it again works fine for 10-15 mins.
JS code:
var i=0;
var loc = new String();
var conn = new String();
var heapUsage = new String();
var cpuUsage = new String();
var thrdCnt = new String();
var heapUsageConsole = new String();
var cpuUsageConsole = new String();
var thrdCntConsole = new String();
var user = new String();
var MemTotal = new String();
function jubking(){
var xmlhttp;
if (window.XMLHttpRequest) {
xmlhttp = new XMLHttpRequest();
} else {
xmlhttp = new ActiveXObject("Microsoft.XMLHTTP");
}
var url = "MonitorDBServlet";
xmlhttp.open("POST", url, false);
xmlhttp.send(null);
var str = xmlhttp.responseText;
var strArr = str.split(",");
url = "MonitorTomcatServlet";
xmlhttp.open("POST", url, false);
xmlhttp.send(null);
var appstr = xmlhttp.responseText;
var appArr = appstr.split(",");
url = "MonitorConsoleTomcatServlet";
xmlhttp.open("POST", url, false);
xmlhttp.send(null);
var appstrConsole = xmlhttp.responseText;
var appArrConsole = appstrConsole.split(",");
url = "CpuMemoryServlet";
xmlhttp.open("POST", url, false);
xmlhttp.send(null);
var statesStr = xmlhttp.responseText;
var states = statesStr.split(",");
if(i>30){
loc = loc.substring(loc.indexOf("\n")+1);
loc += i+","+strArr[0]+","+strArr[1]+"\n";
//--- Do same thing all other var
} else {
loc += i+","+strArr[0]+","+strArr[1]+"\n";
//--- Do same thing all other var
}
document.getElementById("dbSize").innerHTML = strArr[3];
document.getElementById("HeapMemoryUsageMax").innerHTML = appArr[1];
document.getElementById("HeapMemoryUsageMaxConsole").innerHTML = appArrConsole[1];
g = new Dygraph(document.getElementById("dbLocks"),
",locksheld,lockswait\n"+loc+"");
g = new Dygraph(document.getElementById("activeConnection"),
",Connections\n"+conn+"");
g = new Dygraph(document.getElementById("example2"),
",heapUsage\n"+heapUsage+"");
g = new Dygraph(document.getElementById("example3"),
",cpuUsage\n"+cpuUsage+"");
g = new Dygraph(document.getElementById("example4"),
",thread,peakThread\n"+thrdCnt+"");
g = new Dygraph(document.getElementById("example6"),
",heapUsage\n"+heapUsageConsole+"");
g = new Dygraph(document.getElementById("example7"),
",\n"+cpuUsageConsole+"");
g = new Dygraph(document.getElementById("example8"),
",thread,peakThread\n"+thrdCntConsole+"");
g = new Dygraph(document.getElementById("cpuStates"),
",user,system,nice,idle\n"+user+"");
g = new Dygraph(document.getElementById("memStates"),
",MT,MF,B,C,ST,SF\n"+MemTotal+"");
i = i + 1;
setTimeout("jubking()", 5000);
}

You can use about:crashes in FF to view the specific reason for your crash. As mentioned by others, you could be leaking memory if you're caching off data (assigning it to a variable) returned by your AJAX call and not clearing it when the next call is made.
Edit:
Just saw your comment - 1,923,481 K is definitely too much - you're leaking data somewhere. What OS are you running? If you run FF from console in *nix, you usually get some form of a dump into console when something's going wrong (not sure about Windows).
You could possibly try decreasing your poll intervals to once every few seconds and step through the script using Firebug or Chrome's debugger to see what's happening. Worst case, start commenting things out until you figure out exactly what is making your app crash. And then, figure out a way to fix it :)

I suspect that your dygraphs usage is, as you note in your comments, the source of your trouble. It looks like you're binding new graphs over and over again when you only want to update the data, using a moving window for the data would also help. Try reworking your updater to work like this pseudo-JavaScript:
var graphs = {
dbLocks: {
graph: new DyGraph(/* ... */),
data: [ ]
},
activeConnection: {
graph: new DyGraph(/* ... */),
data: [ ]
},
// etc.
};
var DATA_WINDOW_SIZE = 1000; // Or whatever works for you.
function update(which, new_data) {
var g = graphs[which];
g.data.push(new_data);
if(g.data.length > DATA_WINDOW_SIZE)
g.data.shift();
g.graph.updateOptions({ file: g.data });
}
function jubking() {
// Launch all your AJAX calls and bind a callback to each
// one. The success callback would call the update() function
// above to update the graph and manage the data window.
// Wait for all the above asynchronous AJAX calls to finish and
// then restart the timer for the next round.
setTimeout(jubking, 5000);
}
The basic idea is to use window on your data with a reasonable maximum width so that the data doesn't grow to chew up all your memory. As you add a new data point at the end of your data cache, you drop old ones off the other end once you hit your maximum comfortable size.
You can find some techniques for waiting for several asynchronous AJAX calls to finish over here: How to confirm when more than one AJAX call has completed? (disclosure: yes, that's one of my other answers).

The answer above advocates re-using your Dygraph object and calling g.updateOptions({file:...}) to reduce memory usage. This is a great way to do it.
The other way is to call g.destroy() before you redefine the Dygraph object. This will make dygraphs clear out all of its internal arrays and DOM references. Example:
g = new Dygraph(...);
g.destroy();
g = new Dygraph(...);
Read more here: http://blog.dygraphs.com/2012/01/preventing-dygraphs-memory-leaks.html

web page loading time

var startTime = new Date().getTime();
var myWin = window.open("http://www.hizlial.com/hediyelik/hediyelik-urunler/zippo-jack-daniels-ltr-flask-hediye-seti_16.004.4126.1233.htm","_blank")
window.onload = function() { opener.window.endtime = new Date().getTime(); }
var endTime = opener.window.endtime;
var timeTaken = endTime-startTime;
document.write(timeTaken);
i edited the new code but this time i couldnt write the timeTaken value? so whats wrong here?

That's because javascript isn't a threaded language, meaning that it doesn't wait for things to happen, it will open the page and while the page is loading will go on and do other stuff. What you would need to do is to add this into the target page:
window.onload = function() { window.opener.endtime = new Date().getTime(); }
Or as ramazan murat wrote:
var startTime = new Date().getTime(); var myWin = window.open("www.mozilla.com");
window.onload = function() { opener.window.endtime = new Date().getTime(); };
var timeTaken = endTime-startTime;
Another way to go about this would be to do this:
<script>start = new Date().getTime();</script>
<iframe src="http://www.mozilla.com/" onload="end=new Date().getTime();"></iframe>

I would use Net in Firebug to test a page's load time.

Use Google Chrome's developer tool's timeline function.

If you don't need to do this programmatically, then Firebug's Net tab will give you all the information you need to know about how long given resources on a page take to request.