Re: [PHP-I18N] Difficulties using preg_replace with Latin9 and Unicode characters (Resolved)

You don't seem to be using any delimiters in your regexes. No wonder it 
doesn't work.

As for UTF-8, you can pass those strings to preg_replace(), but you need 
to use /u modifier. Like this:

  preg_replace("/[Ff]útbol/u", "Soccer", $text);

-Andrei
http://10fathoms.org/vu - daily photoblog

Erik Norvelle wrote:
> Greetings,
> 
> This was originally a question, but since I ended up solving the issue 
> on my own, I thought I would post my solution.
> 
> I have written a program for aiding in translating documents from 
> Spanish to English, which relies heavily on regular expressions.  Mostly 
> it works, but there are a few characters which cause problems for the 
> regular expression engine.  For instance, the following regular 
> expressions do not match correctly:
> 
> preg_replace("Dña\. ", "Ms\. ", $text); [Matches as /D.*/]
> preg_replace("[Ff]útbol", "Soccer", $text); [Matches as /[Ff].*/]
> preg_replace("1º", "1st", $text); [Matches as /1.*/]
> 
> Plus a few others.  It appears that upon hitting one of these 
> troublesome characters, the preg engine stops parsing and uses whatever 
> "legal" characters it has found up to that point as the "real" regex, 
> ignoring whatever comes after.
> 
> I have tried saving the files in various encodings, in particular, 
> UTF-8, as well as the native Latin9 encoding, to see if PHP would pick 
> up the encoding and respond correctly. No luck, alas.
> The regexes are stored in a MySQL database, with encoding 
> "utf8_unicode_ci", so in theory the function iconv should work to change 
> the encoding.  I have tried the following:
> 
> $regex = iconv("UTF-8", "ISO-8859-1", $trans['patron']);
> 
> This should, in theory, change the pattern (stored in UTF-8 in the DB) 
> into a nice Latin1 pattern.  However, it truncates the pattern, much as 
> PHP does automatically.  For instance, "Sociedad Española de 
> Cardiología" becomes "Sociedad Espa", and "Dña." becomes "D", etc.
> 
> The solution was to tell MySQL to perform the conversion to Latin1 prior 
> to executing the SELECT query to retrieve the Regexes.  MySQL does a 
> better job than PHP in translating between character sets, it would appear:
> 
> mysql_query("SET character_set_results=latin1");
> 
> This has fixed the problems that I had.
> 
> HTH,
> Erik Norvelle
>