Javascript constructor copies array as empty when it should contain items - javascript

My code:
function BayesNet(vars) {
this.variables = {};
this.numVars = Object.keys(this.variables).length;
for (v in vars) {
this.variables[v] = new BayesNode(vars[v]);
this.variables[v].CPTorder = this.generateDomainRows(this.variables[v].parents);
this.variables[v].fullCPT = {}
for (var i = 0; i < this.variables[v].CPTorder.length; i++) {
this.variables[v].fullCPT[this.variables[v].CPTorder[i]] = this.variables[v].CPT[i];
}
this.variables[v].blocks = false;
}
}
function BayesNode(obj) {
this.parents = obj.parents;
this.children = obj.children;
if (typeof obj.domain == 'undefined')
this.domain = ['T','F'];
else
this.domain = obj.domain;
this.observation = obj.observation;
this.CPT = obj.CPT;
this.sampleDistribution = [];
for (var i = 0; i < this.CPT.length; i++) {
var s = [];
if(this.CPT[i].length == this.domain.length - 1)
this.CPT[i].push(1 - sumArray(this.CPT[i]));
s.push(this.CPT[i][0]);
for (var j = 1; j < this.domain.length - 1; j++) {
s.push(this.CPT[i][j]+s[j-1]);
}
s.push(1.0);
this.sampleDistribution.push(s);
}
//TODO: Check if CPT is valid
}
My problem is that BayesNode.parent is copied incorrectly.
BayesNode.parent should be an array containing items, and when I run the debugger through the constructor, this.parents is the correct value . However, once I go back to the BayesNet constructor, parents is an empty array. What could be causing this? All other variables in the object behave as expected.

Javascript executes function calls asynchronously. This is the root cause of your issue. You should use callbacks to execute code that is dependent on results of function calls.
Let me explain this using your code:
this.variables[v] = new BayesNode(vars[v]);
this.variables[v].CPTorder = this.generateDomainRows(this.variables[v].parents);
When you call the constructor, JS does NOT wait for the function to finish executing before moving onto the next line of code. When JS comes across "this.variables[v].parents", it is empty, because, the function call in the previous line is still executing asynchronously.
Javascript code design requires a different approach as compared to most other languages.

I don't see any issues in your code, its strange why its becoming empty. but to solve the problem there is a way. change the code as follow.
after this line
this.variables[v] = new BayesNode(vars[v]);
Add the follow
this.variables[v].parents = vars[v].parents;
I see you are not modifying the parents in the constructor, it will work for time being before you find out whats happening. you might have done this already :)

Related

angularjs - ng-repeat being called before scope variable is defined

I have a md-tabs setup, I have it binded to $scope.selectedIndex so that I can change it by code when I need to.
I use a button, call my function that updates data, I then change $scope.selectedIndex to the tab number needed (In this case 3) that will then change the selected tab.
All of that is fine, except it's calling $scope.selectedIndex before .forEach() call is finished, which results in a ng-repeat not working as it exits silently with no errors since its not defined.
Button Calls:
ng-click="initTVShow(show)"
Function:
$scope.initTVShow = function(show) {
var rng = "";
var possible = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";
for( var i=0; i < 10; i++ )
rng += possible.charAt(Math.floor(Math.random() * possible.length));
$scope.show = show;
$scope.show_name = show.title.split(' ').join('') + "-" + rng;
$scope.poster = show.images.poster;
$scope.backdrop = show.images.poster;
$scope.seasons = [{},{}];
getShow(show.imdb_id, function (err, show) {
for (let i = 0; i < 10; i++) {
$scope.seasons[i] = show.episodes.filter((episode) => episode.season === i);
}
});
$scope.selectedIndex=3;
};
As you can see, $scope.selectedIndex=3; basically changes to Tab #4 (0-based).
Then the following ng-repeat does not seem to work and most likely because .forEach hasn't finished yet. I tested with $scope.seasons with 2 empty index's to test and that works.
<span ng-repeat="season in seasons">{{season.number}}</span>
I think that this is because the function getShow retrieves data asynchronously. Show it to us.
Try this:
getShow(show.imdb_id, function (err, show) {
show.episodes.forEach(function(array){
$scope.seasons[array.season] = array;
});
$scope.selectedIndex=3;
});
It's because the data not ready and the DOM is.
you can solve it easily by adding ng-if on the element and check if the data ready.
Like:<span ng-if="season.number" ng-repeat="season in seasons">{{season.number}}</span>
I found the solution, it was actually just caused by the for() loop (Which pre-edit, was a .forEach), I simple added a callback to it, which changed the tab and it just WORKED!
Changed:
for (let i = 0; i < 11; i++) { }
To:
for (let i = 0; i < 11 || function(){ $scope.selectedIndex=3; }(); i++) { }

Best practice: functions within loops

I got the following code which works perfectly. What it does is: in a table it highlights the corresponding table header cell and table first column cell when you hover over any table cell.
// Row & Column Highlight
(function() {
var gridCellRow = null,
gridCellCol = null,
tableElement = document.getElementsByClassName('inner_table');
for (var i = 0, len_i = tableElement.length; i < len_i; i++) {
if (tableElement[i].getElementsByClassName('row_label_cell').length > 0) {
var gridCell = tableElement[i].getElementsByClassName('input_cell');
for (var j = 0, len_j = gridCell.length; j < len_j; j++) {
function gridCellParents(currentCell) {
return gridCellRow = currentCell.parentNode.firstElementChild,
gridCellCol = currentCell.parentNode.parentNode.rows[0].cells[currentCell.cellIndex];
}
gridCell[j].addEventListener('mouseover', (function() {
gridCellParents(this);
gridCellRow.classList.add('highlight');
gridCellCol.classList.add('highlight');
}));
gridCell[j].addEventListener('mouseout', (function() {
gridCellRow.classList.remove('highlight');
gridCellCol.classList.remove('highlight');
}));
}
}
}
}());
However, JSHint tells me, that
for (var j = 0, len_j = gridCell.length; j < len_j; j++) {
function gridCellParents(currentCell) {
return gridCellRow = currentCell.parentNode.firstElementChild,
gridCellCol = currentCell.parentNode.parentNode.rows[0].cells[currentCell.cellIndex];
}
is not best practice "Function declarations should not be placed in blocks. Use a function expression or move the statement to the top of the outer function."
as well as
gridCell[j].addEventListener('mouseover', (function() {
gridCellParents(this);
gridCellRow.classList.add('highlight');
gridCellCol.classList.add('highlight');
}));
gridCell[j].addEventListener('mouseout', (function() {
gridCellRow.classList.remove('highlight');
gridCellCol.classList.remove('highlight');
}));
}
is not best practice "Don't make functions within a loop."
So how am I correctly and according to best practice building this whole function?
Function deceleration shouldn't be within loops because it makes no
sense to re-create the same function over and over again, in a
"continuous flow" (unlike other situation where the same function
might be created again, in a more complex code). The main reason is
because of hoisting and it strongly goes against javascript
principles to write functions declarations inside loops.
A good starting point, with a more ordered code:
// Row & Column Highlight
(function() {
var gridCellRow,
gridCellCol,
gridCell,
tableElement = document.getElementsByClassName('inner_table');
function gridCellParents(currentCell) {
gridCellRow = currentCell.parentNode.firstElementChild,
gridCellCol = currentCell.parentNode.parentNode.rows[0].cells[currentCell.cellIndex];
}
function onMouseEnter() {
gridCellParents(this);
gridCellRow.classList.add('highlight');
gridCellCol.classList.add('highlight');
}
function onMuoseLeave() {
gridCellRow.classList.remove('highlight');
gridCellCol.classList.remove('highlight');
}
for (var i = 0, len_i = tableElement.length; i < len_i; i++) {
if (tableElement[i].getElementsByClassName('row_label_cell').length > 0) {
gridCell = tableElement[i].getElementsByClassName('input_cell');
for (var j = 0, len_j = gridCell.length; j < len_j; j++) {
gridCell[j].addEventListener('mouseenter', onMouseEnter);
gridCell[j].addEventListener('mouseleave', onMuoseLeave);
}
}
}}());
As you can see, I've modified your events to mousenter and mouseleave which might better suit your needs and be better for overall performance.
Update - delegated version:
// Row & Column Highlight
(function() {
var gridCell,
tableElement = document.querySelectorAll('.inner_table');
function getCellParents(cell){
return {
row : cell.parentNode.firstElementChild, // row
col : cell.parentNode.parentNode.rows[0].cells[cell.cellIndex] // col
};
}
function updateGridCellParents(cell, state) {
state = state ? 'add' : 'remove';
var parents = getCellParents(cell);
parents.row.classList[state]('highlight');
parents.col.classList[state]('highlight');
}
funciton checkTarget(target){
// make sure the element is what we expected it to be
return target.className.indexOf('input_cell') != 0;
}
function onMouseEvents(e){
checkTarget(e.target) && updateGridCellParents(e.target, e.type == "mouseover");
}
document.body.addEventListener('mouseover', onMouseEvents);
document.body.addEventListener('mouseout', onMouseEvents);
})();
In addition to the previous answer, I think it is important to also state why it is a bad practice.
The issue when creating functions inside loops is that they often use values that depends on the loop's iteration. Let's have an example.
// Create three function, that writes their number
var funcs = [];
for(var i=0; i<3; i++){
funcs.push(function(){
document.write(i);
});
}
// Call them.
funcs.forEach(function(f){
f();
});
One may expect the above code to write 1 then 2 then 3. However, because variables in JS are not block-scoped but function-scoped (except for the new let and const), the closure of all three of these functions will actually use the exact same i: 3, the last value it had been given (and thus the value it still has).
Because of this behaviour, this is very easy to make mistakes. Hence, it is not recommended.
If you need to create a function that depends of the value of a loop, you can use a factory.
// Create a factory function that returns a
// function that writes the argument.
function writerFactory(msg){
return function(){
document.write(msg);
}
}
// Create three functions, that write their number.
var funcs = [];
for(var i=0; i<3; i++){
funcs.push(writerFactory(i));
}
// Call them.
funcs.forEach(function(f){
f();
});
This time, each function has a different closure: the one that is created by each call of the factory. They all have access to a different msg.

recursive function completes before recursive calls finish

I'm new to javascript and am not real capable with the asynchronous aspects, closures, etc. I have done a few days research on this and lots of trial & error but can't seem to get past my issue, which is:
I am trying to walk a tree structure, gathering all the bottom level nodes (those without child nodes). This node data gets loaded into global arrays (not optimal but needed). The walk function I'm using is recursive. But the asynchronous nature cause the first call to the function to return before the recursive calls return, so the complete tree doesn't get interrogated. I tried putting it in an anonymous function which seems to get the entire tree traversed, but then the global arrays are not being loaded (inaccessible?).
BTW, the real code is on a separate, isolated network so a direct cut & paste to here is not possible. Below is functional equivalent of the relevant parts (unless I've made a typo). Apologies for that. Any help would be appreciated.
var nodeList = new Array(); // global variable
function someFunction(rootNode) {
// unrelated processing here
walkTree(rootNode); // gather the childless nodes
return;
}
function walkTree(node) {
return function() { // required in order traverse the entire tree
// but with it, nodeList does not get populated
var num = node.numChildren();
var childNodes = node.getChildNodes();
for (var i=0; i<num; i++) {
var currentNode = childNodes.item(i);
if (currentNode.numChildren() > 0) {
walkTree(currentNode);
}
else {
var obj = new Object();
/// extract certain attributes of current node here
/// and make a variant
nodeList[nodeList.length] = obj;
}
} // END for
} // close anonymous function
} // END FUNCTION
If you don't need asynchronous execution, your code can be simplified:
var nodeList = [];
function someFunction(rootNode) {
// unrelated processing here
walkTree(rootNode); // gather the childless nodes
return;
}
function walkTree(node) {
var num = node.numChildren(),
childNodes = node.getChildNodes();
for (var i=0; i<num; i++) {
var currentNode = childNodes.item(i);
if (currentNode.numChildren() > 0) {
walkTree(currentNode);
}
else {
var obj = new Object();
/// extract certain attributes of current node here
/// and make a variant
nodeList.push(obj);
}
}
}
If you do need asynchronous execution, the implementation would depend on what asynchronous mechanism you used (web workers, simulation using setTimeout, a framework like Clumpy, etc.).
With Clumpy, for instance, you might code it like this (untested):
var nodeList = [],
clumpy = new Clumpy();
function someFunction(rootNode) {
// unrelated processing here
walkTree(rootNode); // gather the childless nodes
return;
}
function walkTree(node) {
var num = node.numChildren(),
childNodes = node.getChildNodes(),
i;
clumpy.for_loop(
function() { i = 0; },
function() { return i < num; },
function() { i++; },
function() {
var currentNode = childNodes.item(i);
if (currentNode.numChildren() > 0) {
walkTree(currentNode);
}
else {
var obj = new Object();
/// extract certain attributes of current node here
/// and make a variant
nodeList.push(obj);
}
}
);
}

For loop inside function (newbie)

I would like to have some variables that my for loop uses inside a function scope (not global).
I tried to wrap the for loop inside a function like this but it results in console error:
function() {
var data = livingroomTableData;
for(var i = data[0]; i < data[1]; i++) {
var elemvalue = data[2] + format(i) + ".png";
livingroomTableArray[i] = elemvalue;
}
}
I would like the data variable to have the values of livingroomTableData only inside this for loop (not globally). In other loops I will input a different variable into the data variable.
Oh yes, and as you can probably tell, I'm a total newbie. :S
There is only function scope in javascript, block scope does not exist, so you can't let the variable only inside the for loop. What you could do is to create a function scope.
Code example:
(function(livingroomTableData) {
var data = livingroomTableData;
//... the rest code
})(livingroomTableData);
The big problem is this line:
for(var i = data[0]; i < data[1]; i++) {
That means, starting with i as the first element of the array, do the code in the loop, incrementing i by one at the end of each run until i is not less than the second element of data.
I'd rewrite it to show you a working version, but its not clear what you actually want to do.
function() {
for(var i = 0; i < livingroomTableData.length; i++) {
var data = livingroomTableData[i];
//your code here...
}
}

Speed test: while() cycle vs. jQuery's each() function

I have an array of objects called targets and I want to execute a function on each of those objects. The first method:
targets.each(function() {
if (needScrollbars($(this))) {
wrap($(this), id);
id = id + 1;
}
});
This method gives execution speed of ~125ms. The second method is:
var i=0;
while (targets[i] != undefined) {
if (needScrollbars($(this))) {
wrap($(this), id);
id = id + 1;
}
i = i+1;
}
This second method takes whopping 1385ms to execute and I get my head around that. Does anyone have any idea why a bare bones cycle runs slower than a function which I'm only guessing that's doing (just guessing) a whole lot more than a simple cycle?
Thank you.
They are totally different. The this in the first example is the current target, in the second example this is the "external" this. You should change the second example as:
var i=0;
while (targets[i] != undefined) {
var cur = $(targets[i]);
if (needScrollbars(cur)) {
wrap(cur, id);
id = id + 1;
}
i = i+1;
}
The relevant quote
More importantly, the callback is fired in the context of the current DOM element, so the keyword this refers to the element.
But I don't know why you haven't written as:
for (var i = 0; i < targets.length; i++)
{
var cur = $(targets[i]);
if (needScrollbars(cur)) {
wrap(cur, id);
id = id + 1;
}
}
And in the end the each "method" is easier to comprehend (for me).
Your second method is not functionally equivalent to the first one.
Why? Because it uses this, making it a closure on the global scope. Of course the second method is slower: it continuously shells out jQuery objects made out of global scope. Try that benchmark again with:
var i=0;
while (targets[i] !== undefined) {
var o = $(targets[i]);
if (needScrollbars(o)) {
wrap(o, id);
id++;
}
i++;
}

Categories