PHP dealing with huge string - php

I have to replace xmlns with ns in my incomming xml in order to fix SimpleXMLElements xpath() function. Most functions do not have a performance problem. But there allways seems to be an overhead as the string grows.
E.g. preg_replace on a 2 MB string takes 50ms to process, even if I limit the replaces to 1 and the replace is done at the very beginning.
If I substr the first few characters and just replace that part it is slightly faster. But not really that what I want.
Is there any PHP method that would perform better in my problem? And if there is no option, could a simple php extension help, that just does Replace => SimpleXMLElement in C?

If you know exactly where the offending "x", "m" and "l" are, you can just use something like $xml[$x_pos] = ' '; $xml[$m_pos] = ' '; $xml[$l_pos] = ' ' to transform them into spaces. Or transform them into ns___ (where _ = space).

You're always going to get an overhead when trying to do this - you're dealing with a char array and trying to do replace multiple matching elements of the array (i.e. words).
50ms is not much of an overhead, unless (as I suspect) you're trying to do this in a loop?

50ms sounds pretty reasonable to me, for something like this. The requirement itself smells of something being wrong.
Is there any particular reason that you're using regular expressions? Why do people keep jumping to the overkill regex solution?
There is a bog-standard string replace function called str_replace that may do what you want in a fraction of the time (though whether this is right for you depends on how complex your search/replace is).

From the PHP source, as we can see, for example here:
http://svn.php.net/repository/php/php-src/branches/PHP_5_2/ext/standard/string.c
I don`t see, any copies, but I'm not expert in C. From the other hand we can see there many convert to string calls, which at 1st sight could copy values. If they copy values, then we in trouble here.
Only if we in trouble
Try to invent some str_replace wheel here with the help of string-by-char processing. For example we have string $somestring = "somevalue". In PHP we could work with it's chars by indexes as echo $somestring{0}, which will give us "s" or echo $somestring{2} which will give us "m". I'm not sure in this way, but it's possible, if official implimentations don't use references, as they should use.

Related

PHP Regex string returns two identical arrays

I've got a Regex query here to pull out all of the tags in a page. It looks like this:
preg_match_all('%<tr[^>]++>(.*?)</tr>%s', $pageText, $rows);
Problem is that while it does find all of the tags on the page in the return array it actually returns a multidimensional array, where each entry of the first array contains an array of all of the matches. In other words, it hands me multiple identical copies of the first array, IE the one I actually want.
Help please?
EDIT: Also relevant: I'm not allowed to use DOM for this application despite it being a significantly easier (and better) way of going about things.
What you're actually asking about is the $row[0] list, which redundantly contains the <tr>...</tr> blob again. If you just care about the (.*?) inner data, then use \K to reset the full match.
preg_match_all('=<tr\b[^>]*+>(.*?)</tr>\K=s', $pageText, $rows);
It's not possible to get rid of $row[0] completely. You'll have to ignore it, and use $row[1] alone.
Try this one:
preg_match_all('~<tr(?:\\s+[^>]*)?>(.*?)</tr>~si', $pageText, $rows);
var_dump($rows[1]);
Don't use % to wrap RegExps. It's a character somehow reserved for printf() like functions and with %s or %i at the end of your Pattern, it can be quite confusing.

I want to add one hash to all the hashes in a string (PHP)

So in my string, I have certain sections with hashes. For example, consider the string "#Hello, this is a sample string. This is another example of ###hashes".
I want to replace that with:
"##Hello, this is a sample string, This is another example of ####hashes".
(note that the number of hashes in each instance increased by one)
However, I'm not too sure how. I'd imagine it involved regular expressions, and I've searched a bit, but I'm not too sure what to do.
Can anyone help/lead me on the right path?
Cheers
preg_replace('/(#[^#])/', '#\1', $string);
This works too:
preg_replace('/#+/', '#$0', $string);

Regex for a Function Call with Multiple Optional Parameters

I'm looking for a regex that will scan a document to match a function call, and return the value of the first parameter (a string literal) only.
The function call could look like any of the following:
MyFunction("MyStringArg");
MyFunction("MyStringArg", true);
MyFunction("MyStringArg", true, true);
I'm currently using:
$pattern = '/Use\s*\(\s*"(.*?)\"\s*\)\s*;/';
This pattern will only match the first form, however.
Thanks in advance for your help!
Update
I was able to solve my problem with:
$pattern = '/Use\s*\(\s*"(.*?)\"/';
Thanks Justin!
~Scott
If you only care about the value of the first parameter, you can just chop off the end of the regex:
$pattern = '/Use\s*\(\s*"(.*?)\"/';
However, you should understand that this (or any pure-regex solution for this problem) will not be perfect, and there will be some possible cases it handles incorrectly. In this case, you'll get false positives, and escaped quotes (\") will break it.
You can ignore escaped quotes by complicating it a bit:
$pattern = '/Use\s*\(\s*"(.*?)(?!<(?:\\\\)*\\)\"/';
This ignores " characters inside the quoted string if they have an odd number of backslashes in front of them.
However, the false-postives issue can't be helped without introducing false-negatives, and vice versa. This is because PHP is an irregular language, so it can't be parsed with "pure" regex, and even modern regex engines that allow recursion are going to need some pretty complex code to do a really thorough job at this.
All I'm saying is, if you're planning a one-off job to quickly scrape through some PHP you wrote yourself, regex is probably fine. If you're looking for something robust and open-ended that will do this on arbitrary PHP code, you need some kind of reflection or PHP parser.
This might be slightly simpler, though will only work if you have double quotes and not single quotes:
$pattern = /Use\s*[^\"]*\"([^\"]*)\"/

How to match anything except a pattern between two tags

I am attempting to match a string which is composed of HTML. Basically it is an image gallery so there is a lot of similarity in the string. There are a lot of <dl> tags in the string, but I am looking to match the last <dl>(.?)+</dl> combo that comes before a </div>.
The way I've devised to do this is to make sure that there aren't any <dl's inside the <dl></dl> combo I'm matching. I don't care what else is there, including other tags and line breaks.
I decided I had to do it with regular expressions because I can't predict how long this substring will be or anything that's inside it.
Here is my current regex that only returns me an array with two NULL indicies:
preg_match_all('/<dl((?!<dl).)+<\/dl>(?=<\/div>)/', $foo, $bar)
As you can see I use negative lookahead to try and see if there is another <dl> within this one. I've also tried negative lookbehind here with the same results. I've also tried using +? instead of just + to no avail. Keep in mind that there's no pattern <dl><dl></dl> or anything, but that my regex is either matching the first <dl> and the last </dl> or nothing at all.
Now I realize . won't match line breaks but I've tried anything I could imagine there and it still either provides me with the NULL indicies or nearly the whole string (from the very first occurance of <dl to </dl></div>, which includes several other occurances of <dl>, exactly what I didn't want). I honestly don't know what I'm doing incorrectly.
Thanks for your help! I've spent over an hour just trying to straighten out this one problem and it's about driven me to pulling my hair out.
Don't use regular expressions for irregular languages like HTML. Use a parser instead. It will save you a lot of time and pain.
I would suggest to use tidy instead. You can easily extra all the desired tags with their contents, even for broken HTML.
In general I would not recommend to write a parser using regex.
See http://www.php.net/tidy
As crazy as it is, about 2 minutes after I posted this question, I found a way that worked.
preg_match_all('/<dl([^\z](?!<dl))+?<\/dl>(?=<\/div>)/', $foo, $bar);
The [^\z] craziness is just a way I used to say "match all characters, even line breaks"

Whitelist in php

I have an input for users where they are supposed to enter their phone number. The problem is that some people write their phone number with hyphens and spaces in them. I want to put the input trough a filter to remove such things and store only digits in my database.
I figured that I could do some str_replace() for the whitespaces and special chars.
However I think that a better approach would be to pick out just the digits instead of removing everything else. I think that I have heard the term "whitelisting" about this.
Could you please point me in the direction of solving this in PHP?
Example: I want the input "0333 452-123-4" to result in "03334521234"
Thanks!
This is a non-trivial problem because there are lots of colloquialisms and regional differences. Please refer to What is the best way for converting phone numbers into international format (E.164) using Java? It's Java but the same rules apply.
I would say that unless you need something more fully-featured, keep it simple. Create a list of valid regular expressions and check the input against each until you find a match.
If you want it really simple, simply remove non-digits:
$phone = preg_replace('![^\d]+!', '', $phone);
By the way, just picking out the digits is, by definition, the same as removing everything else. If you mean something different you may want to rephrase that.
$number = filter_var(str_replace(array("+","-"), '', $number), FILTER_SANITIZE_NUMBER_INT);
Filter_Var removes everything but pluses and minuses, and str_replace gets rid of those.
or you could use preg_replace
$number = preg_replace('/[^0-9]/', '', $number);
You could do it two ways. Iterate through each index in the string, and run is_numeric() on it, or you could use a regular expression on the string.
On the client side I do recommand using some formating that you design when creating a form. This is good for zip or telephone fields. Take a look at this jquery plugin for a reference. It will much easy later on the server side.

Categories