There is something I don't understand about how eval works.
Suppose I have a function foo:
function foo() {
console.log("test");
}
And then I write
eval("foo()");
or
eval("foo" + "();");
The function foo is executed and I have "test" printed out.
However, if I write:
eval("function foo() { console.log(\"foo\"); }();");
or
eval("function foo() { console.log(\"foo\"); }" + "();");
I get "SyntaxError: Unexpected token )".
Why is that? I saw that when I pass the function name it is evaluated into the function's code, so I though it's supposed to be the same as "eval("foo" + "();");"
I'm using Chrome 27 if it makes any difference.
Because no answer is specific about why the last snippet fails, and nobody seems to be warning you for the dangers of eval:
eval, according to MDN:
A string representing a JavaScript expression, statement, or sequence of statements. The expression can include variables and properties of existing objects.
Your string, "function foo(){console.log('foo');}()" actually consists of 2 statements, that make no sense to the JS engine. Well, the second one doesn't:
function foo(){console.log('foo');}//function declaration, fine
;//<-- js adds implied statement terminator
()//no invocation, because no function reference, group nothing ==> does not compute, raise error
That's why you have to turn your function declaration statement into an expression, by adding an operator. Typically, this is the grouping operator: ()
(function foo(){ console.log('foo')});
The function declaration is now an expression, everything (the grouping () included) can be seen as a statement, unless the code that follows belongs to the code above. In this case, the invoking parentheses clearly do, so the JS engine sorts it out for you.
For clarity, some say the preferred notation is:
(function f(){}());//<-- invoking parentheses inside group
Which makes sense, because the invocation is part of the statement, after all.
If you don't like all these parentheses, any operator will do:
~function(){}();//bitwise not
+function(){}();//coerce to number
Are all equally valid. Note that they will change the possible return values of the function expression.
(function(){}());//resolves to undefined
+function(){}();//resolves to NaN, because undefined is NotANumber
How your eval could look, then, is any of the following:
eval("(function (){console.log('foo');}());");
eval("~function (){console.log('foo');}();");
eval("!function (){console.log('foo');}();");
And so on, and so forth...
Lastly, eval evaluates code in the global scope, so any code eval-ed containing functions can, and likely will, polute the global namespace. Mallicious code will also have access to everything, so be weary of XSS attacks and all other JS based techniques.
Bottom line, eval is evil, especially since all browsers now support JSON.parse natively, and for those that don't, there still is a tried and tested JSON2.js file out there.
Using eval in strict mode does make things slightly safer, but doesn't prevent XSS attacks at all, the code can still manipulate the DOM, for example, and reassign exposed DOM references or objects.
Google "why eval is evil" if you want to find out more.
Also check the ECMAScript specs
function x() {} is a statement, whereas everything else you're doing is an expression, and you can only evaluate expressions.
The other answers explain how you can turn this statement into an expression. Thought I'd let you know why you actually get an error, though. :)
eval("(function() { console.log(\"foo\"); })()");
This is called a self-executing anonymous function.
you would need to call
eval("function foo() { console.log(\"foo\"); };" + " foo();");
in order for this to work;
Put brackets around your function
eval("(function foo() { console.log(\"foo\"); })()");
It is like calling
(function foo() { console.log("foo"); })()
The main thing you should understand about eval is that it is evil. Don't use it... ever. There is always a way to achieve the same thing by dynamically creating a script document or by using a closure.
Read Douglas Crockford: JavaScript the good parts.
Related
This question already has answers here:
Explain the encapsulated anonymous function syntax
(10 answers)
Closed 7 years ago.
I'm reading up on JavaScript IIFE and so far the understand concept, but I am wondering about the outside parenthesis. Specifically, why are they required? For example,
(function() {var msg='I love JavaScript'; console.log(msg);}());
works great, but
function() {var msg='I love JavaScript'; console.log(msg);}();
generates a syntax error. Why? There are lots of discussions on IIFE, but I'm not seeing a clear explanation about why the parentheses are required.
There are two ways to create functions in JavaScript (well, 3, but let's ignore new Function()). You can either write a function declaration or write a function expression.
A function declaration in itself is a statement and statements by themselves don't return values (let's also ignore how the debugging console or Node.js REPL print return values of statements). A function expression however is a proper expression and expressions in JavaScript returns values that can be immediately used.
Now, you may have seen people saying that the following is a function expression:
var x = function () {};
It may be tempting to conclude that the syntax:
function () {};
is what makes it an expression. But that's wrong. The syntax above is what makes it an anonymous function. And anonymous functions can either be a declaration or an expression. What makes it an expression is this syntax:
var x = ...
That is, everything to the right of an = sign is an expression. Expressions make it easier to write math formulas in programming languages. So in general everywhere that math is expected to be processed is an expression.
Some of the forms of expressions in JavaScript include:
everything to the right of an = operator
things in braces () that are not function call braces
everything to the right of a math operator (+,-,*,/)
all the arguments to the ternary operator .. ? .. : ..
When you write:
function () {}
it is a declaration and does not return a value (the declared function). Therefore trying to call the non-result is an error.
But when you write:
(function () {})
it is an expression and returns a value (the declared function) which may be used immediately (for example, may be called or may be assigned).
Note the rules for what counts as expressions above. From that it follows that braces are not the only things that you can use to construct an IIFE. Below are valid ways for constructing IIFEs (because we write function expressions):
tmp=function(){}()
+function(){}()
-function(){}()
0/function(){}()
0*function(){}()
0?0:function(){}()
(function(){}())
(function(){})()
You may actually see one of the above non-standard forms (particularly the + version) in third-party libraries, because they want to save one byte. But I strongly advise you to only use the brace forms (either are fine), because they are widely recognized as IIFEs by other programmers.
The version of IIFE that is wrapped in parenthesis works, because this marks the declaration of the internal function declaration as an expression.
http://benalman.com/news/2010/11/immediately-invoked-function-expression/
For more detailed explanation please see:
Advanced JavaScript: Why is this function wrapped in parentheses?
HINT:
The invocation operator (()) only works with expressions, not declarations.
This will be a long-winded answer, but will give you the necessary background. In JavaScript there are two ways functions can be defined:
A function definition (the classical kind)
function foo() {
//why do we always use
}
and then the more obscure type, a function expression
var bar = function() {
//foo and bar
};
In essence the same thing is going on at execution. A function object is created, memory is allocated, and an identifier is bound to the function. The difference is in the syntax. The former is itself a statement which declares a new function, the latter is an expression.
The function expression gives us the ability to insert a function any place where a normal expression would be expected. This lends its way to anonymous functions and callbacks. Take for instance
setTimeout(500, function() {
//for examples
});
Here, the anonymous function will execute whenever setTimeout says so. If we want to execute a function expression immediately, however, we need to ensure the syntax is recognizable as an expression, otherwise we have ambiguity as to whether of not we mean a function expression or statement.
var fourteen = function sumOfSquares() {
var value = 0;
for (var i = 0; i < 4; i++)
value += i * i;
return value;
}();
Here sumOfSquares is immediately invoked because it can be recognized as an expression. fourteen becomes 14 and sumOfSquares is garbage-collected. In your example, the grouping operator () coerces its content into an expression, therefore the function is an expression and can be called immediately as such.
One important thing to note about the difference between my first foo and bar example though is hoisting. If you don't know what that it is, a quick Google search or two should tell you, but the quick and dirty definition is that hoisting is JavaScript's behavior to bring declarations (variables and functions) to the top of a scope. These declarations usually only hoist the identifier but not its initialized value, so the entire scope will be able to see the variable/function before it is assigned a value.
With function definitions this is not the case, here the entire declaration is hoisted and will be visible throughout the containing scope.
console.log("lose your " + function() {
fiz(); //will execute fiz
buzz(); //throws TypeError
function fiz() {
console.log("lose your scoping,");
}
var buzz = function() {
console.log("and win forever");
};
return "sanity";
}()); //prints "lose your scoping, lose your sanity"
It seems like a stupid question but I really can't wrap my head around it. I began to wonder this when I was thinking of why there is a semicolon after variable declaration but not function declaration. So does that mean a function declaration is an expression?
And what is the official meaning for statement? I've been finding the answer saying that a statement is a "command" that tells computer what to do, but isn't a function declaration telling the computer what to do? Does it have something to do with when it's get loaded when executing the code?
I began to wonder this when I was thinking of why there is a semicolon after variable declaration but not function declaration
Not every statement ends with a semicolon.
Examples of statements with trailing semicolon:
Variable declaration: var foo = <expression>;
Expression statement: <expression>;
Do while loop: do <statement> while (<expression>);
Examples without trailing semicolons:
If statement: if (<expression>) <statement>
Try catch statement: try <block> catch (<identifier>) <block>
For loop: for (...) <statement>
Block statement: { <statement list> }
So does that mean a function declaration is an expression?
No, it's more complicated than that. Function declarations are not statements. They are not expressions either, they are source elements. Statements and function declarations are both source elements. **
And what is the official meaning for statement?
I can't tell you the official definition, but in the context of JS I would say something like "it's an instruction that does not produce an assignable result/value". Expressions on the other hand produce a result that can be used in other expressions.
** Good to know, but a bit off-topic: Since function declarations are not statements, they are technically not allowed to be used inside blocks. This becomes even more apparent if we also consider hoisting. This example should throw a syntax error in every browser, but unfortunately it doesn't.
if (true) {
function foo() { alert('foo'); }
} else {
function foo() { alert('bar'); }
}
foo();
This leads to different behaviors in different browser. While Chrome will show bar, Firefox will show foo. Chrome just hoists both function declarations, and the second overrides the first one. Firefox interprets both declarations as something like a function expression.
Try it yourself in different browsers.
A variable declaration can have side-effects (eg, var x = alert(42);).
A function declaration cannot.
There is a JSLint option, one of The Good Parts in fact, that "[requires] parens around immediate invocations," meaning that the construction
(function () {
// ...
})();
would instead need to be written as
(function () {
// ...
}());
My question is this -- can anyone explain why this second form might be considered better? Is it more resilient? Less error-prone? What advantage does it have over the first form?
Since asking this question, I have come to understand the importance of having a clear visual distinction between function values and the values of functions. Consider the case where the result of immediate invocation is the right-hand side of an assignment expression:
var someVar = (function () {
// ...
}());
Though the outermost parentheses are syntactically unnecessary, the opening parenthesis gives an up-front indication that the value being assigned is not the function itself but rather the result of the function being invoked.
This is similar to Crockford's advice regarding capitalization of constructor functions -- it is meant to serve as a visual cue to anyone looking at the source code.
From Douglass Crockford's style convention guide: (search for "invoked immediately")
When a function is to be invoked immediately, the entire invocation expression should be wrapped in parens so that it is clear that the value being produced is the result of the function and not the function itself.
So, basically, he feels it makes more clear the distinction between function values, and the values of functions. So, it's an stylistic matter, not really a substantive difference in the code itself.
updated reference, old PPT no longer exists
Immediately Called Anonymous Functions get wrapped it in parens because:
They are function expressions and leaving parens out would cause it to be interpreted as a function declaration which is a syntax error.
Function expressions cannot start with the word function.
When assigning the function expression to a variable, the function itself is not returned, the return value of the function is returned, hence the parens evaluate what's inside them and produce a value. when the function is executed, and the trailing parens ..}() cause the function to execute immediately.
Or, use:
void function () {
...
} ()
This question already has answers here:
Closed 10 years ago.
Possible Duplicate:
What does the exclamation mark do before the function?
So I was going back and looking over some of my own code as well as some other javascript code and I realized that a while back when I started writing javascript libraries I was using closures that looked something like this:
(function( window, document, undefined ) {
// code
})( window, document );
But then I saw some bootstrap code and I changed to this syntax:
! function (window, document, undefined) {
// code
}(window, document);
Now, if I'm not wrong (and please correct me if I am), placing the '!' in front of my anonymous function just causes it to be treated as '()' then returns (into nowhere?) a boolean value of whether the value returned by the function was not undefined, null, or empty.
What I'm wondering is, is there really a difference between using the '!()' syntax over '()()'? Are there certain browsers that will complain?
Any thoughts are appreciated, thanks! XD
What you're asking about is self-calling functions, otherwise known as IIFE (immediately invoked function expression). Which is a different thing from a closure. Although it does create a closure (indeed, all functions in javascript create closures, not just IIFE).
It's understandable that you may confuse the two issues though, since IIFE are usually introduced in the context of explaining closures. But be aware that they are different things. Closures are private, shared "global-like" variables. IIFE are functions that gets called immediately upon their definitions.
Now, how does IIFE work? The clue is in the name. It's the "FE" IIFE - function expression.
In javascript, as you know, there are two ways of creating functions - using function declarations:
function foo () {}
and using function expressions:
foo = function () {}
A function expression is simply a function declared in the context* of an expression. What are expressions in javascript? Simply any statement that evaluates something.
Traditionally, people recognize expressions as:
anything on the right side of the = sign
a = /* expression */
anything in braces
(/* expression */)
So traditionally those are the two "standard" ways for declaring function expressions:
foo = function(){}
and
(function(){})
And the second syntax makes it easy to then execute the function object returned by the expression. But, really, an expression is anywhere js does math (or logic, which is math anyway). So adding an operator to a function declaration also turns it into an expression. The following works because they are unary operators (meaning, they are legal without anything on the left hand side):
!function(){}()
+function(){}()
-function(){}() // I especially like this one because it
// looks like a command line switch
typeof function(){}()
But you can also use binary operators if you use some throw-away value or variable with it. The following also work:
x=function(){}()
0==function(){}()
1*function(){}()
2/function(){}()
Heck, you can even abuse the ternary operator:
0?0:function(){}() // valid and works!
There's nothing magical about it. It's not a specific syntax baked into javascript. Just like (function(){}()) is not a specific syntax for IIFE. It's just that when declared in an expression, functions return themselves as objects which can be called immediately.
But I'd advise against using any of the non-standard forms above though. For the same reason you asked this question - most javascript programmers are not used to seeing them and it can cause confusion. I myself didn't realize that you can do this until you asked the question. Still, it's a useful thing to know for when you need to write things like minifiers, code generators etc.
* I'm using "context" in it's traditional definition here not the javascript specific meaning of "context" as defined in the spec.
Nothing, they are just two different ways of achieving the same thing. The () is slightly more readable though.
Why I can write
var foo = function(){}();
But can not
function(){}();
Are there are any design reasons?
The first example is an assignment: the right-hand side is an expression, and the immediate execution of an anonymous function makes sense.
The second example is a declaration: once the closing "}" is hit the declaration has ended. Parens on their own don't make sense--they must contain an expression. The trailing ")" is an error.
Standalone declarations must be turned into expressions:
(function() {})(); // Or...
(function() {}());
The first makes the declaration an expression, then executes the result. The second turns both declaration and execution into an expression.
See also When do I use parenthesis and when do I not?
You can (function(){})();, and you aren't naming the function in: var foo = function(){}();
You are setting foo to the return value of the function which in your case is undefined, because all functions return something in JavaScript.
The first use of function
var foo = function(){}()
is in expression position, not statement position. The second one, on the other hand, is at the top-level, and works as a function statement. That is, 'function' can be used in two different contexts, and it can mean subtly different things.
If you want to make anonymous functions without names, you do that with a function expression. And, just because of the way JavaScript language grammar works, you'll need parens in certain context, as Dave Newton mentions, because the place where you're putting the word 'function' can be confused with its statement version (which does require a name).