Replacing only parts of the text

Anon

I want to make some replacements in a string taht contains HTML code using eregi_replace().

The problem is that I want eregi_replace() to search only outside HTML tags.

Ex.: Replace dot character "." with asterix char "*" but leave the "." char that is inside tags.

$string = "test...<a href=www.phpbuilder.com>test.";

after replacing...

$string = "test**<a href=www.phpbuilder.com>test";

vsego

$newStr = ereg_replace("^{(([^{<](<[^>]>))).",}} "\1", $oldStr);
There are probably other eregs as well, but I think this one should work (wrote it by heart, so I'm not sure, but it's definitely a good start).
Be carefull - it might be slow on big strings!

Anon

the traditional way to do this (ie, the C way) is

for($i=0;$i<strlen($the_file);$i++)
if($the_file[$i]=='.')
$the_file[$i]='*';

this is easily modified to only replace stuff outside of carets (pointy brackets) by using a toggle. when you hit an open caret you turn replacement off, when you hit a close caret you turn it back on... viz.

$replace = true;
for($i=0;$i<strlen($the_file);$i++)
{
if($the_file[$i]=='<')
$replace = false;
if($the_file[$i]=='>')
$replace = true;
if($replace && ($the_file[$i]='.')
$the_file[$i]='*';
}

note that i haven't tested this at all, it's just off the top of my head, but it should work.

also note that you cannot put carets in your body copy that are not part of html tags... you'll need to use the lt/gt codes! you can make the above a bit more robust by adding some counters to see if the number of open and close carets is equal, ie

$replace = true;
for($i=0;$i<strlen($the_file);$i++)
{
if($the_file[$i]=='<')
{$replace = false;$ltcount++;}
if($the_file[$i]=='>')
{$replace = true;$gtcount++}
if($replace && ($the_file[$i]='.')
$the_file[$i]='*';
}

if($gtcount != $ltcount)
echo "you have unmatched carets!!!";

really, all regular expressions are is a nice syntax to control this byte-by-byte march through a string. often it's faster and easier to do the marching yourself!

vsego

Regular expressions are faster
We both neglected quots
<tag <tag>text> will be treated as
<tag <tag>>text&gt;
so you don't need to count < and >

It does make sence to assume that input HTML is correct. Providing that is another problem, depending on source and purpose of the parsed text(s).

Anon

Thank you!

But it doesn't work. There must be some other way...

This pattern returns the same string without replacing the dot character with asterix.

Anon

Sorry 'bout that. I've missed one "" and forgot that you need while-loop here because one replace, of course, does only one "type" of change:
i.e. it will change all "<><>." to "<><>", but not "<>." or "blah.". Here is the code I've tested:

<?php
$oldStr = "a.a<b>a.a</b>a.a<a href=\"http://www.borland.com/\">Borland.com</a>is cool.a<b>a.a</b>";
$reg_exp = "^{(([^{<](<[^>]>))).";}}
$newStr = $oldStr;
while (ereg($reg_exp, $newStr, $reOut))
$newStr = ereg_replace($reg_exp, "\1*", $newStr);
echo htmlentities($newStr);
?>

Good luck,

Vedran