htmlentities in PHP but preserving html tags

You can get the list of correspondances character => entity used by htmlentities, with the function get_html_translation_table ; consider this code :

$list = get_html_translation_table(HTML_ENTITIES);
var_dump($list);

(You might want to check the second parameter to that function in the manual -- maybe you'll need to set it to a value different than the default one)

It will get you something like this :

array
  ' ' => string ' ' (length=6)
  '¡' => string '¡' (length=7)
  '¢' => string '¢' (length=6)
  '£' => string '£' (length=7)
  '¤' => string '¤' (length=8)
  ....
  ....
  ....
  'ÿ' => string 'ÿ' (length=6)
  '"' => string '"' (length=6)
  '<' => string '&lt;' (length=4)
  '>' => string '&gt;' (length=4)
  '&' => string '&amp;' (length=5)

Now, remove the correspondances you don't want :

unset($list['"']);
unset($list['<']);
unset($list['>']);
unset($list['&']);

Your list, now, has all the correspondances character => entity used by htmlentites, except the few characters you don't want to encode.

And now, you just have to extract the list of keys and values :

$search = array_keys($list);
$values = array_values($list);

And, finally, you can use str_replace to do the replacement :

$str_in = '<p><font style="color:#FF0000">Camión español</font></p>';
$str_out = str_replace($search, $values, $str_in);
var_dump($str_out);

And you get :

string '<p><font style="color:#FF0000">Cami&Atilde;&sup3;n espa&Atilde;&plusmn;ol</font></p>' (length=84)

Which looks like what you wanted ;-)

Edit : well, except for the encoding problem (damn UTF-8, I suppose -- I'm trying to find a solution for that, and will edit again)

Second edit couple of minutes after : it seem you'll have to use utf8_encode on the $search list, before calling str_replace :-(

Which means using something like this :

$search = array_map('utf8_encode', $search);

Between the call to array_keys and the call to str_replace.

And, this time, you should really get what you wanted :

string '<p><font style="color:#FF0000">Cami&oacute;n espa&ntilde;ol</font></p>' (length=70)

And here is the full portion of code :

$list = get_html_translation_table(HTML_ENTITIES);
unset($list['"']);
unset($list['<']);
unset($list['>']);
unset($list['&']);

$search = array_keys($list);
$values = array_values($list);
$search = array_map('utf8_encode', $search);

$str_in = '<p><font style="color:#FF0000">Camión español</font></p>';
$str_out = str_replace($search, $values, $str_in);
var_dump($str_in, $str_out);

And the full output :

string '<p><font style="color:#FF0000">Camión español</font></p>' (length=58)
string '<p><font style="color:#FF0000">Cami&oacute;n espa&ntilde;ol</font></p>' (length=70)

This time, it should be ok ^^
It doesn't really fit in one line, is might not be the most optimized solution ; but it should work fine, and has the advantage of allowing you to add/remove any correspondance character => entity you need or not.

Have fun !

Might not be terribly efficient, but it works

$sample = '<p><font style="color:#FF0000">Camión español</font></p>';

echo htmlspecialchars_decode(
    htmlentities($sample, ENT_NOQUOTES, 'UTF-8', false)
  , ENT_NOQUOTES
);

This is optimized version of the accepted answer.

$list = get_html_translation_table(HTML_ENTITIES);
unset($list['"']);
unset($list['<']);
unset($list['>']);
unset($list['&']);

$string = strtr($string, $list);

No solution short of a parser is going to be correct for all cases. Yours is a good case:

<p><font style="color:#FF0000">Camión español</font></p>

but do you also want to support:

<p><font>true if 5 < a && name == "joe"</font></p>

where you want it to come out as:

<p><font>true if 5 &lt; a &amp;&amp; name == &quot;joe&quot;</font></p>

Question: Can you do the encoding BEFORE you build the HTML. In other words can do something like:

"<p><font>" + htmlentities(inner) + "</font></p>"

You'll save yourself lots of grief if you can do that. If you can't, you'll need some way to skip encoding <, >, and " (as described above), or simply encode it all, and then undo it (eg. replace('&lt;', '<'))