Request for help: regexp

[Pekka]

In EE 2 stylesheet.php there is some code to remove all style start and end tags plus comment codes before style is sent out. The code that now does it is very simple and does not tolerate any other than exact <style> tag used in default templates.

$stylesheet_html = str_replace("<style type=\"text/css\" media=\"screen\">","",$stylesheet_html); // very restricted. someone who knows regexp could make these idiot proof!
$stylesheet_html = str_replace("</style>","",$stylesheet_html);
$stylesheet_html = str_replace("<!--","",$stylesheet_html);
$stylesheet_html = str_replace("-->","",$stylesheet_html);

Does anyone know regexp to make one which removes all possible variations inside <style>? All help is appreciated! Thanks!

[DavidW]

Clearly the regex needs to be case insensitive for the tags (even though anything other than lower case breaks XHTML compliance), but you can get case insensitivity using eregi_replace().

Though, in some ways, Perl compatible regular expressions are more powerful, faster and binary safe, not everyone necessarily has PCRE turned on in PHP.


If you want to match any <style...> tag caselessly - which is, I think, the only latitude you need, then:

$stylesheet_html = eregi_replace("<style[^>]*>","",$stylesheet_html);
should do the job (I've checked this regex in a UNIX shell using grep, but not in PHP).

[^>] means every character except >. Put a * after it for 0 or more repeats. The > afterwards requires the match to finish with a >.


I don't see any need to change the other three lines, as you hinted in your post.

Does this do what you want?



David

[DavidW]

Thinking further, defensive coding suggests using a case insensitive match for the second line to make sure that </style> is stripped. str_ireplace() should do the job there.

We now have the following snippet:
$stylesheet_html = eregi_replace("<style[^>]*>","",$stylesheet_html);
$stylesheet_html = str_ireplace("</style>","",$stylesheet_html);
$stylesheet_html = str_replace("<!--","",$stylesheet_html);
$stylesheet_html = str_replace("-->","",$stylesheet_html);

(I've lost the indenting - throw it back in at will).



David

[Pekka]

Can't use str_ireplace as it is only in PHP 5 or later.

My RISC way of coding would suggest

$stylesheet_html = eregi_replace("<style[^>]*>","",$stylesheet_html);
$stylesheet_html = str_replace("</style>","",$stylesheet_html);
$stylesheet_html = str_replace("</STYLE>","",$stylesheet_html);
$stylesheet_html = str_replace("</Style>","",$stylesheet_html);
$stylesheet_html = str_replace("<!--","",$stylesheet_html);
$stylesheet_html = str_replace("-->","",$stylesheet_html);

but isn't there a way to make it with regexp (I don't know what is the case insensitive area switch is in regexp), just a guess :) :

$stylesheet_html = eregi_replace("</[s|S][t|T][y|Y][l|L][e|E]>","",$stylesheet_html);

[DavidW]

In enhanced regexes (as opposed to Perl regexes), it's the function used that determines case sensitivity. eregi_replace() is case insensitive, ereg_replace() is the regular, case sensitive function.

Just use $stylesheet_html = eregi_replace("</style>","",$stylesheet_html); for the second line - no need to mess around with anything else. The first line I gave above is already case insensitive because of the use of eregi_replace().

In other words, we have:
$stylesheet_html = eregi_replace("<style[^>]*>","",$stylesheet_html);
$stylesheet_html = eregi_replace("</style>","",$stylesheet_html);
$stylesheet_html = str_replace("<!--","",$stylesheet_html);
$stylesheet_html = str_replace("-->","",$stylesheet_html);

As a final teaching point, [S|s] is incorrect - it should be [Ss]. You're getting mixed up with the syntax for alternatives - (bar|foo). Don't do it anyway - it's messy - simply use a case insensitive regex.



David

[Pekka]

That looks good and works. Thanks!! :)

[segal3]

Should we be updating our code somewhere?

[DavidW]

This is included in 2.02 - it makes the code more robust.



David