From 6fd50f06813feb879c411c9a1b094f77676aa592 Mon Sep 17 00:00:00 2001
From: "J. King" <jking@jkingweb.ca>
Date: Fri, 27 Jul 2018 19:11:33 -0400
Subject: [PATCH] Ensure char and byte position never goes beyond the end of
 the string

---
 lib/UTF8String.php        |  8 +++++---
 tests/cases/TestCodec.php | 30 ++++++++++++++++++++++++++++++
 2 files changed, 35 insertions(+), 3 deletions(-)

diff --git a/lib/UTF8String.php b/lib/UTF8String.php
index 04faa81..4a2e2b6 100644
--- a/lib/UTF8String.php
+++ b/lib/UTF8String.php
@@ -31,7 +31,9 @@ class UTF8String {
     public function nextChr(): string {
         // get the byte at the current position
         $b = @$this->string[$this->posByte];
-        if (ord($b) < 0x80) {
+        if ($b === "") {
+            return "";
+        } elseif (ord($b) < 0x80) {
             // if the byte is an ASCII character or end of input, simply return it
             $this->posChar++;
             $this->posByte++;
@@ -50,16 +52,16 @@ class UTF8String {
         // this function effectively implements https://encoding.spec.whatwg.org/#utf-8-decoder
         // though it differs from a slavish implementation because it operates on only a single
         // character rather than a whole stream
-        $this->posChar++;
         // optimization for ASCII characters
         $b = @$this->string[$this->posByte];
         if ($b=="") {
-            $this->posByte++;
             return false;
         } elseif (($b = ord($b)) < 0x80) {
+            $this->posChar++;
             $this->posByte++;
             return $b;
         }
+        $this->posChar++;
         $point = 0;
         $seen = 0;
         $needed = 1;
diff --git a/tests/cases/TestCodec.php b/tests/cases/TestCodec.php
index fe74dfb..f78db68 100644
--- a/tests/cases/TestCodec.php
+++ b/tests/cases/TestCodec.php
@@ -107,6 +107,36 @@ class TestConf extends \PHPUnit\Framework\TestCase {
         $this->assertSame(0, $s->posChr());
         $this->assertSame(0, $s->posByte());
     }
+    
+    /** 
+     * @covers \MensBeam\UTF8\UTF8String::posChr
+     * @covers \MensBeam\UTF8\UTF8String::posByte
+    */
+    public function testTraversePastTheEndOfAString() {
+        $s = new UTF8String("a");
+        $this->assertSame(0, $s->posChr());
+        $this->assertSame(0, $s->posByte());
+
+        $this->assertSame("a", $s->nextChr());
+        $this->assertSame(1, $s->posChr());
+        $this->assertSame(1, $s->posByte());
+
+        $this->assertSame("", $s->nextChr());
+        $this->assertSame(1, $s->posChr());
+        $this->assertSame(1, $s->posByte());
+
+        $s = new UTF8String("a");
+        $this->assertSame(0, $s->posChr());
+        $this->assertSame(0, $s->posByte());
+
+        $this->assertSame(ord("a"), $s->nextOrd());
+        $this->assertSame(1, $s->posChr());
+        $this->assertSame(1, $s->posByte());
+
+        $this->assertSame(false, $s->nextOrd());
+        $this->assertSame(1, $s->posChr());
+        $this->assertSame(1, $s->posByte());
+    }
 
     public function provideStrings() {
         return [