Added initial support for 'strict' parsing option

Author: paquettg

src/PHPHtmlParser/Content.php

@@ -41,6 +41,11 @@ public function __construct($content)
 		$this->pos     = 0;
 	}
 
+	public function getPosition()
+	{
+		return $this->pos;
+	}
+
 	/**
 	 * Gets the current character we are at.
 	 *

src/PHPHtmlParser/Dom.php

@@ -4,6 +4,7 @@
 use PHPHtmlParser\Dom\HtmlNode;
 use PHPHtmlParser\Dom\TextNode;
 use PHPHtmlParser\Exceptions\NotLoadedException;
+use PHPHtmlParser\Exceptions\StrictException;
 use stringEncode\Encode;
 
 class Dom {
@@ -539,6 +540,12 @@ protected function parseTag()
 			else
 			{
 				// no value attribute
+				if ($this->options->strict)
+				{
+					// can't have this in strict html
+					$character = $this->content->getPosition();
+					throw new StrictException("Tag '$tag' has an attribute '$name' with out a value! (character #$character)");
+				}
 				$node->getTag()->$name = [
 					'value'       => null,
 					'doubleQuote' => true,
@@ -556,6 +563,14 @@ protected function parseTag()
 		}
 		elseif (in_array($tag, $this->selfClosing))
 		{
+			
+			// Should be a self closing tag, check if we are strict
+			if ( $this->options->strict)
+			{
+				$character = $this->content->getPosition();
+				throw new StrictException("Tag '$tag' is not self clossing! (character #$character)");
+			}
+
 			// We force self closing on this tag.
 			$node->getTag()->selfClosing();
 		}

src/PHPHtmlParser/Exceptions/StrictException.php

@@ -0,0 +1,4 @@
+<?php
+namespace PHPHtmlParser\Exceptions;
+
+final class StrictException extends \Exception {}

src/PHPHtmlParser/Options.php

@@ -10,6 +10,7 @@ class Options {
 	 */
 	protected $defaults = [
 		'whitespaceTextNode' => true,
+		'strict'             => false,
 	];
 
 	/**

tests/Options/StrictTest.php

@@ -0,0 +1,55 @@
+<?php
+
+use PHPHtmlParser\Dom;
+use PHPHtmlParser\Exceptions\StrictException;
+
+class StrictTest extends PHPUnit_Framework_TestCase {
+
+	public function testConfigStrict()
+	{
+		$dom = new Dom;
+		$dom->setOptions([
+			'strict' => true,
+		]);
+		$dom->load('<div><p id="hey">Hey you</p> <p id="ya">Ya you!</p></div>');
+		$this->assertEquals(' ', $dom->getElementById('hey')->nextSibling()->text);
+	}
+
+	public function testConfigStrictMissingSelfClosing()
+	{
+		$dom = new Dom;
+		$dom->setOptions([
+			'strict' => true,
+		]);
+		try
+		{
+			// should throw an exception
+			$dom->load('<div><p id="hey">Hey you</p><br><p id="ya">Ya you!</p></div>');
+			// we should not get here
+			$this->assertTrue(false);
+		}
+		catch (StrictException $e)
+		{
+			$this->assertEquals("Tag 'br' is not self clossing! (character #31)", $e->getMessage());
+		}
+	}
+
+	public function testConfigStrictMissingAttribute()
+	{
+		$dom = new Dom;
+		$dom->setOptions([
+			'strict' => true,
+		]);
+		try
+		{
+			// should throw an exception
+			$dom->load('<div><p id="hey" block>Hey you</p> <p id="ya">Ya you!</p></div>');
+			// we should not get here
+			$this->assertTrue(false);
+		}
+		catch (StrictException $e)
+		{
+			$this->assertEquals("Tag 'p' has an attribute 'block' with out a value! (character #22)", $e->getMessage());
+		}
+	}
+}