Fixed paquettg#187 and added tests

Author: paquettg

src/PHPHtmlParser/Content.php

@@ -4,6 +4,7 @@
 
 namespace PHPHtmlParser;
 
+use PHPHtmlParser\Exceptions\ContentLengthException;
 use PHPHtmlParser\Exceptions\LogicalException;
 
 /**
@@ -74,14 +75,27 @@ public function char(?int $char = null): string
      * Moves the current position forward.
      *
      * @chainable
+     * @throws ContentLengthException
      */
     public function fastForward(int $count): Content
     {
+        if (!$this->canFastForward()) {
+            // trying to go over the content length, throw exception
+            throw new ContentLengthException('Attempt to fastForward pass the length of the content.');
+        }
         $this->pos += $count;
 
         return $this;
     }
 
+    /**
+     * Checks if we can move the position forward.
+     */
+    public function canFastForward(): bool
+    {
+        return \strlen($this->content) > $this->pos;
+    }
+
     /**
      * Moves the current position backward.
      *
@@ -197,14 +211,15 @@ public function copyByToken(string $token, bool $char = false, bool $escape = fa
     /**
      * Skip a given set of characters.
      *
-     * @return Content|string
+     * @throws LogicalException
      */
-    public function skip(string $string, bool $copy = false)
+    public function skip(string $string, bool $copy = false): string
     {
         $len = \strspn($this->content, $string, $this->pos);
-
-        // make it chainable if they don't want a copy
-        $return = $this;
+        if ($len === false) {
+            throw new LogicalException('Strspn returned false with position ' . $this->pos . '.');
+        }
+        $return = '';
         if ($copy) {
             $return = \substr($this->content, $this->pos, $len);
             if ($return === false) {

src/PHPHtmlParser/Dom.php

@@ -12,6 +12,7 @@
 use PHPHtmlParser\Dom\TextNode;
 use PHPHtmlParser\Exceptions\ChildNotFoundException;
 use PHPHtmlParser\Exceptions\CircularException;
+use PHPHtmlParser\Exceptions\ContentLengthException;
 use PHPHtmlParser\Exceptions\CurlException;
 use PHPHtmlParser\Exceptions\LogicalException;
 use PHPHtmlParser\Exceptions\NotLoadedException;
@@ -646,7 +647,13 @@ private function parseTag(): array
         }
 
         // check if this is a closing tag
-        if ($this->content->fastForward(1)->char() == '/') {
+        try {
+            $this->content->fastForward(1);
+        } catch (ContentLengthException $exception) {
+            // we are at the end of the file
+            return $return;
+        }
+        if ($this->content->char() == '/') {
             // end tag
             $tag = $this->content->fastForward(1)
                 ->copyByToken('slash', true);
@@ -683,7 +690,12 @@ private function parseTag(): array
         ) {
             $space = $this->content->skipByToken('blank', true);
             if (empty($space)) {
-                $this->content->fastForward(1);
+                try {
+                    $this->content->fastForward(1);
+                } catch (ContentLengthException $exception) {
+                    // reached the end of the content
+                    break;
+                }
                 continue;
             }
 
@@ -764,7 +776,9 @@ private function parseTag(): array
             }
         }
 
-        $this->content->fastForward(1);
+        if ($this->content->canFastForward()) {
+            $this->content->fastForward(1);
+        }
 
         $return['status'] = true;
         $return['node'] = $node;

src/PHPHtmlParser/Exceptions/ContentLengthException.php

@@ -0,0 +1,14 @@
+<?php
+
+declare(strict_types=1);
+
+namespace PHPHtmlParser\Exceptions;
+
+use Exception;
+
+/**
+ * Class EmptyCollectionException.
+ */
+final class ContentLengthException extends Exception
+{
+}

tests/DomTest.php

@@ -649,4 +649,12 @@ public function testCompatibleWithWordPressShortcode()
         $this->assertEquals(' [wprs_alert type="success" content="this is a short code" /] ', $node->innerHtml);
 
     }
+
+    public function testBrokenHtml()
+    {
+        $dom = new Dom();
+        $dom->loadStr('<the thing broke itV');
+
+        $this->assertEquals('<the thing broke itv></the>', $dom->outerHtml);
+    }
 }