for..in
does not loop through the indexes of an array, it loops through the enumerable property names of an object. It happens that the only enumerable properties array instances have, by default, are array indexes, and so it mostly works to think it does array indexes in limited situations. But that's not what for..in
does, and misunderstanding this will bite you. :-) It breaks as soon as you add any further properties to the array (a perfectly valid thing to do) or any library you're using decides to extend the array prototype (also a valid thing to do).
In any case, what you get back from document.getElementsByTagName
isn't an array. It's a NodeList
. Your best bet for iterating through NodeList
s is to use an explicit index a'la Pointy's answer -- e.g., a straight counting loop:
var i;
for (i = 0; i < list.length; ++i) {
// ... do your thing ...
}
Somewhat off-topic because it doesn't relate to your NodeList
, but: When you are actually working with a real array, because arrays in JavaScript are sparse, there is applicability for for..in
, you just have to be clear about what you're doing (looping through property names, not indexes). You might want to loop only as many times as the array has actual entries, rather than looping through all the indexes in the gaps in the sparse array. Here's how you do that:
var a, name;
a = [];
a[0] = "zero";
a[10000] = "ten thousand";
for (name in a) {
// Only process this property name if it's a property of the
// instance itself (not its prototype), and if the name survives
// transition to and from a string unchanged -- e.g., it's numeric
if (a.hasOwnProperty(name) && parseInt(name) == name) {
alert(a[name]);
}
}
The above only alerts twice, "zero" and "ten thousand"; whereas a straight counting loop without the checks would alert 10,001 times (mostly saying "undefined" because it's looping through the gap).