Merge branch 'release/0.0.1'

README.md

@@ -1,4 +1,4 @@
-bee4/robots.txt v0.0.0
+bee4/robots.txt v0.0.1
 ======================
 
 [![Build Status](https://travis-ci.org/bee4/robots.txt.svg?branch=develop)](https://travis-ci.org/bee4/robots.txt)

src/Exception/DuplicateRuleException.php

@@ -0,0 +1,65 @@
+<?php
+/**
+ * This file is part of the beebot package.
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ *
+ * @copyright Bee4 2015
+ * @author	Stephane HULARD <[email protected]>
+ * @package Bee4\RobotsTxt\Exception
+ */
+
+namespace Bee4\RobotsTxt\Exception;
+
+use Exception;
+use Bee4\RobotsTxt\Rule;
+
+/**
+ * Class DuplicateRuleException
+ * Error thrown when the parser try to add 2 rules for the same UA
+ * @package Bee4\RobotsTxt\Exception
+ */
+class DuplicateRuleException extends Exception
+{
+	/**
+	 * @var Rule
+	 */
+	protected $rule;
+
+	/**
+	 * @var string
+	 */
+	protected $ua;
+
+	/**
+	 * Rule setter
+	 * @param Rule $rule
+	 * @return DuplicateRuleException
+	 */
+	public function setRule(Rule $rule) {
+		$this->rule = $rule;
+		return $this;
+	}
+	/**
+	 * @return Rule
+	 */
+	public function getRule() {
+		return $this->rule;
+	}
+
+	/**
+	 * User Agent setter
+	 * @param string $ua
+	 * @return DuplicateRuleException
+	 */
+	public function setUserAgent($ua) {
+		$this->ua = $ua;
+		return $this;
+	}
+	/**
+	 * @return string
+	 */
+	public function getUserAgent() {
+		return $this->ua;
+	}
+}

src/Parser.php

@@ -25,6 +25,7 @@ class Parser
 	protected $content;
 
 	public function __construct($content) {
+		//Remove the UTF8 BOM
 		$this->content = trim($content, "\xEF\xBB\xBF");
 	}
 

src/Rules.php

@@ -11,6 +11,8 @@
 
 namespace Bee4\RobotsTxt;
 
+use Bee4\RobotsTxt\Exception\DuplicateRuleException;
+
 /**
  * Class Rules
  * Represent a collection of Rules
@@ -26,6 +28,13 @@ class Rules
 	 */
 	protected $collection = [];
 
+	private $defaultRule;
+
+	public function __construct() {
+		$this->defaultRule = new Rule();
+		$this->add(self::DEFAULT_UA, $this->defaultRule);
+	}
+
 	/**
 	 * Add a new rule to the collection
 	 * @param string $ua
@@ -34,14 +43,22 @@ class Rules
 	 */
 	public function add($ua, Rule $rule) {
 		$ua = $this->handleUa($ua);
-		if( isset($this->collection[$ua]) ) {
-			throw new \RuntimeException('You can\'t add 2 rules for the same UserAgent');
+		if( isset($this->collection[$ua]) && $this->collection[$ua] !== $this->defaultRule ) {
+			throw (new DuplicateRuleException('You can\'t add 2 rules for the same UserAgent'))
+				->setRule($rule)
+				->setUserAgent($ua);
 		}
 		$this->collection[$ua] = $rule;
 
 		return $this;
 	}
 
+	/**
+	 * Check if the URL match for the given UA or not
+	 * @param string $ua
+	 * @param string $url
+	 * @return boolean
+	 */
 	public function match($ua, $url) {
 		if( ($rule = $this->get($ua)) === null ) {
 			return false;

test/units/ParserTest.php

@@ -12,6 +12,7 @@
 namespace Test\Bee4\RobotsTxt;
 
 use Bee4\RobotsTxt\Parser;
+use Bee4\RobotsTxt\Rules;
 use Bee4\RobotsTxt\ParserFactory;
 
 /**
@@ -27,6 +28,13 @@ class ParserTest extends \PHPUnit_Framework_TestCase
 Allow: /truite.php
 disallow: /";
 
+	protected $duplicateRuleContent = "User-agent: *
+Disallow: /mentions-legales/
+
+User-agent: *
+Allow: /truite.php";
+
+
 	public function testParse() {
 		$object = new Parser($this->content);
 		$rules = $object->parse();
@@ -41,9 +49,26 @@ public function testParse() {
 		$this->assertTrue($rules->match('Google-Bot v01', '/truite.php'));
 	}
 
+	public function testEmptyContentParse() {
+		$object = new Parser("");
+		$rules = $object->parse();
+
+		$rule = $rules->get(Rules::DEFAULT_UA);
+		$this->assertInstanceOf('\Bee4\RobotsTxt\Rule', $rule);
+		$this->assertTrue($rule->match('/another-page.html'));
+	}
+
+
+	/**
+	 * @expectedException Bee4\RobotsTxt\Exception\DuplicateRuleException
+	 */
+	public function testDuplicateRuleParse() {
+		$object = new Parser($this->duplicateRuleContent);
+		$rules = $object->parse();
+	}
+
 	public function testParserFactory() {
 		$rules = ParserFactory::build("http://www.bee4.fr");
-
 		$this->assertInstanceOf('\Bee4\RobotsTxt\Rule', $rules->get('*'));
 	}
 }