More performance improvements, and a regression fix

lib/UTF8.php

@@ -24,7 +24,7 @@ abstract class UTF8 {
         start:
         // get the byte at the specified position
         $b = @$string[$pos];
-        if ($b < "\x80" || $b=="") {
+        if ($b < "\x80") {
             // if the byte is an ASCII character or end of input, simply return it
             $next = $pos + 1;
             return $b;
@@ -97,7 +97,7 @@ abstract class UTF8 {
         do {
             $pos--;
             $t++;
-            $b = ($pos < strlen($string)) ? $string[$pos] : "";
+            $b = @$string[$pos];
         } while (
             $b >= "\x80" && $b <= "\xBF" && // continuation bytes
             ($t < 4 || $errMode==self::M_SKIP) && // stop after four bytes, unless we're skipping invalid sequences
@@ -186,7 +186,10 @@ abstract class UTF8 {
         start:
         // optimization for ASCII characters
         $b = @$string[$pos];
-        if ($b < "\x80") {
+        if ($b=="") {
+            $next = $pos + 1;
+            return null;
+        } elseif ($b < "\x80") {
             $next = $pos + 1;
             return ord($b);
         }

perf/perf.php

@@ -14,22 +14,12 @@ $files = [
 ];
 
 $tests = [
-    'Intl code points' => ["intl", function(string $text): int {
-        $t = 0;
-        $i = \IntlBreakIterator::createCodePointInstance();
-        $i->setText($text);
-        foreach ($i as $o) {
-            $p = $i->getLastCodePoint();
-            $t++;
-        }
-        return $t;
-    }],
-    'Native code points' => ["", function(string $text): int {
+    'Native characters' => ["", function(string $text): int {
         $t = 0;
         $pos = 0;
         $eof = strlen($text);
         while ($pos <= $eof) {
-            $p = UTF8::ord($text, $pos, $pos);
+            $p = UTF8::get($text, $pos, $pos);
             $t++;
         }
         return $t;
@@ -44,20 +34,12 @@ $tests = [
         }
         return $t;
     }],
-    'Native characters' => ["", function(string $text): int {
+    'Native code points' => ["", function(string $text): int {
         $t = 0;
         $pos = 0;
         $eof = strlen($text);
         while ($pos <= $eof) {
-            $p = UTF8::get($text, $pos, $pos);
-            $t++;
-        }
-        return $t;
-    }],
-    'PCRE split characters' => ["pcre", function(string $text): int {
-        $t = 0;
-        foreach (preg_split('//u',  $text) as $c) {
-            $p = $c;
+            $p = UTF8::ord($text, $pos, $pos);
             $t++;
         }
         return $t;