interesting preg_replace problem

right off the bat, here's my preg_replace call: preg_replace('/\{([^}]+)}/e', "'{<a href=view_entry.php?word='. urlencode(trim('\\1')) .'>\\1</a>}'", $string); what I'm doing is toying with the jargon file (www.jargon.org), just dumping into a mysql database and pulling it out and parsing it for the web. Nothing new, just playing with it for the hell of it. my problem is that I have data that looks like this in $string: 1. vi. [orig. from BASIC's CHAIN' statement] To hand off execution to a child or successor without going through the {OS} command interpreter that invoked it. The state of the parent program is lost and there is no returning to it. Though this facility used to be common on memory-limited micros and is still widely supported for backward compatibility, the jargon usage is semi-obsolescent; in particular, most Unix programmers will think of this as an {exec}. Oppose the more modern subshell'. 2. n. A series of linked data areas within an operating system or application. `Chain rattling' is the process of repeatedly running through the linked data areas searching for one which is of interest to the executing program. The implication is that there is a very large number of links on the chain. and, this preg_replace call is turning it into this: 1. vi. [orig. from BASIC's CHAIN' statement] To hand off execution to a child or successor without going through the {<a href=view_entry.php?word=OS>OS</a>} command interpreter that invoked it. The state of the parent program is lost and there is no returning to it. Though this facility used to be common on memory-limited micros and is still widely supported for backward compatibility, the jargon usage is semi-obsolescent; in particular, most Unix programmers will think of this as an {<a href=view_entry.php?word=%60Ch> Ch</a>}. Oppose the more modern subshell'. 2. n. A series of linked data areas within an operating system or application. Chain rattling' is the process of repeatedly running through the linked data areas searching for one which is of interest to the executing program. The implication is that there is a very large number of links on the chain. the problem here is that it's replacing the {exec} with a hyperlinked {`Ch}, which is from later in the string. This is what it does with every one of these. the first {} pattern matches and replaces fine, but any {} pattern after that gets corrupted with data from later in the string. If its at the end of the string, it gets replaces with a long string of Zs, which is very odd. I'm trying to just link like the first {OS} gets linked. I'm sure it's just something silly in my regex, anyone spot it? I'm stumped. -Rex