PREC Modifers

[bluewalrus]

So I've been using the modifers (http://php.net/manual/en/reference.pcre.pattern.modifiers.php) I thought correctly but have run into a few problems recently.

I had this


$code_is = preg_replace('/<li><p.*?>(.*?)<\/p>.*?<\/li>/s', "<li>$1</li>", $code_is);

I then tried


$code_is = preg_replace('/<li><p.*?>(.*?)<\/p>.*?<\/li>/m', "<li>$1</li>", $code_is);

and


$code_is = preg_replace('/<li><p.*?>(.*?)<\/p>.*?<\/li>/sm', "<li>$1</li>", $code_is);

None of which worked as I intended. I'm trying to replace any line item with a paragraph in it there are new lines and tabs in there. None of the above solutions work, the below solution somewhat works but not for all occurances so I figure there must be a better way or modifer.


$code_is = preg_replace('/<li>[.|\n|\r|\t]*?<p.*?>(.*?)<\/p>[.|\n|\r|\t]*?<\/li>/s', "<li>$1</li>", $code_is);

[james438]

sometimes newlines are expressed as \r\n as well. When you say that it is not working I am assuming that you mean it is not matching. I have not worked with the multiline modifier much, but it applies to the use of ^ and $ commands, which you are not using, so I wouldn't worry about it with your situation.

/s tells the dot to recognize newlines, so I would keep using it in your situation. Using the dot in [.|\n|\r|\t] makes the \n|\r|\t parts meaningless since the dot will match anything for one character. You may as well replace [.|\n|\r|\t] with ".".

Not promising anything, but could you post an example of what you want matched and what it should look like after it has been processed?

Just a typo, but your title says PREC when it should be PCRE ;)

later

[bluewalrus]

Yes, it's not matching I think because of the new lines/tabs I'd like it to just completely ignore any white spaces (tabs, returns, enters, spaces, [&nbsp;, &#160;(if possible) ]).


<li>
<p>blah blah blah random text
</p></li>

Replaced as


<li>blah blah blah random text</li>

[djr33]

You could pre-process the text by removing all of those with a simple str_replace(). Then you wouldn't need to worry about it. I'm not sure if that would apply to your project though.

[bluewalrus]

How would I do it with str_replace? That examples is 3 lines of a larger string. I'd want to keep tabs, new lines, etc. if they aren't inside an li.

I guess a better sample would be:


<p>Some stuff up here</p>

<ol>
<li>1</li>
<li>2</li>
<li>
<p>blah blah blah random text
</p></li>
</ol>

<img src="a.jpg" alt="a" />

<a href="link.html">link text</a>


The formatting of this would be


<p>Some stuff up here</p>

<ol>
<li>1</li>
<li>2</li>
<li>blah blah blah random text</li>
</ol>

<img src="a.jpg" alt="a" />

<a href="link.html">link text</a>

[djr33]

Won't work well then. My idea was more for a full HTML (or XML) parser: first strip all white space (since it is irrelevant) then process the tags. But if you are going to need the formatted source code, there's no way around that.

[james438]

Why not just use this?


<?php
$code_is=" <p>Some stuff up here</p>

<ol>
<li>1</li>
<li>2</li>
<li>
<p>blah blah blah random text
</p></li>
</ol>

<img src=\"a.jpg\" alt=\"a\" />

<a href=\"link.html\">link text</a>";
$code_is=preg_replace('/<li>\s{1,}(<p>|\b)/s','<li>',$code_is);
$code_is=preg_replace('/(\s)+(<\/p>)?<\/li>/s',"</li>",$code_is);
?>
<textarea cols="100" rows="50"><?php print htmlentities($code_is);?></textarea>

"+" means one or more, which is the same as "{1,}".
"?" means zero or one, which is another way of saying that it may be present at this location and it might not.

[james438]

Actually, this is better:


$code_is=preg_replace('/<li>\s+(<p>|\b)/s','<li>',$code_is);
$code_is=preg_replace('/\s*?<\/p>?<\/li>/s',"</li>",$code_is);

EDIT: even better:


$code_is=preg_replace('/<li>\s+(<p>|\b)/s','<li>',$code_is);
$code_is=preg_replace('/\s*?<\/p>?\s*?<\/li>/s','</li>',$code_is);

[james438]

If this answers your question, please mark this thread as resolved.

[bluewalrus]

Been busy haven't had a chance to try it quite yet. What does the \b do?

[james438]

good question. \b is a bit harder to understand; at least for me it was. \b is the boundary between a word character and a non word character. For example

$test='####tttt';
$test=preg_replace('/\b/','correct',$test);
echo "$test";

// this produces: ####correcttttt

I wasn't sure how to do what you wanted without \b, because it would have failed under certain circumstances.

[bluewalrus]

Thanks, yes that could did work I finally got around to finishing my word/open office/ckeditor convertor.

I don't have a mark as resolved option anymore though I guess maybe an admin will mark it for me.

[bluewalrus]

I've returned with a varied question (this one is to remove duplicating tags). I've got this code now


$code_is = preg_replace('/<([a-z,A-Z].*?)>\s+<\1>/s', " <$1>", $code_is);
//initially had $code_is = preg_replace('/<(.*?)>\s+<\1>/s', " <$1>", $code_is);
$code_is = preg_replace('/<(\/.*?)><\1>/s', "<$1> ", $code_is);


which is suppose to filter something like


<em> <em>et al.</em></em>

to


<em>et al.</em>

Well it works there it is deleting <td> tags entered like this


<td>&nbsp;</td>
<td>&nbsp;</td>


which becomes


<td>&nbsp;</td>

[james438]

We should probably try and get a moderator to split this thread.

Anyway


<?php
$code_is="<em> <em>et al.</em></em>
<td>&nbsp;</td>
<td>&nbsp;</td>";
$code_is = preg_replace('/<([a-zA-Z].*?)>\s*<\1>/s', " <$1>", $code_is);
$code_is = preg_replace('/<(\/.*?)><\1>/s', "<$1> ", $code_is);
echo $code_is;
?>

I didn't look too closely into why it wasn't working. All I did was fix a few of the more obvious errors and then tried it out again and it seems to be working.

EDIT: This is a little better though:


$code_is = preg_replace('/<([a-zA-Z].*?)>\s*<\1>/si', " <$1>", $code_is);
$code_is = preg_replace('/<(\/.*?)><\1>/si', "<$1> ", $code_is);