[heuristics for paren linking (thanks @brion!)
Evan Prodromou <evan@controlyourself.ca>**20081102090241] {
hunk ./lib/util.php 714
-	$r = preg_replace('@https?://[^)\]>\s]+@', '<a href="\0" class="extlink">\0</a>', $r);
+	$r = preg_replace_callback('@https?://[^\]>\s]+@', 'common_render_uri_thingy', $r);
hunk ./lib/util.php 720
+function common_render_uri_thingy($matches) {
+	$uri = $matches[0];
+	$trailer = '';
+
+	# Some heuristics for extracting URIs from surrounding punctuation
+	# Strip from trailing text...
+	if (preg_match('/^(.*)([,.:"\']+)$/', $uri, $matches)) {
+		$uri = $matches[1];
+		$trailer = $matches[2];
+	}
+
+	$pairs = array(
+		']' => '[', # technically disallowed in URIs, but used in Java docs
+		')' => '(', # far too frequent in Wikipedia and MSDN
+	);
+	$final = substr($uri, -1, 1);
+	if (isset($pairs[$final])) {
+		$openers = substr_count($uri, $pairs[$final]);
+		$closers = substr_count($uri, $final);
+		if ($closers > $openers) {
+			// Assume the paren was opened outside the URI
+			$uri = substr($uri, 0, -1);
+			$trailer = $final . $trailer;
+		}
+	}
+	return '<a href="' . $uri . '" class="extlink">' . $uri . '</a>' . $trailer;
+}
+
}