使用PHP，如何可靠地识别（查找）源字符串中的已知Javascript变量名称？

Question

Let's say I have a PHP string that contains the following text, or some other text that is valid Javascript code if put into an interpreter:

$js_script PHP variable

var loaded = [];
var loading = [];
var downloading = [];

function load(name)
{
    loading.push(name);
    downloading.push(name);

    // do some asynchronous stuff, with the following callback:
    (function () {
        loaded.push(name);
        downloading = [];
    });
}

I want to replace the variable loading (as in replace the variable name itself in the entire code) to l1 , without using a full javascript parser, and thus producing the following output:

$js_script PHP variable

var loaded = [];
var l1 = [];
var downloading = [];

function load(name)
{
    l1.push(name);
    downloading.push(name);

    // do some asynchronous stuff, with the following callback:
    (function () {
        loaded.push(name);
        downloading = [];
    });
}

The reasons I assume avoiding using a parser is a good choice:

• they seem to be extreme overkill for this single task,
• this is the only modification I wish to perform on the input code,
• I explicitly know the variable name I'm going to replace.

I'm not concerned with performance, since the resulting output is to be cached. I thought about using regex for this, that is, finding all occurrences of "loading" in the code-string using:

\bloading\b(?=[^"]*(?:"[^"]*"[^"]*)*$)

... which obviously doesn't support apostrophes yet, but other than that, seems to handle the task right so far.

Is there something I'm missing? I know that JavaScript in wholesale is not a regular language, but how about this part? Should I forget about using regex, and utilize a parser instead?

Answer 1

If you use preg_replace , it's pretty simple:

$out = preg_replace("/(\W)(loading)(\W)/", "$1l1$3", $in);

An example can be found here .

EDIT: I did a quick benchmark, and running one replace like this on the file you specified took my computer ~2 μs on average (I put the benchmarking code here )