WIP: Issue/3 New broken and redirected link rule

src/Rule/BrokenLink.php

@@ -0,0 +1,64 @@
+<?php
+
+namespace CidiLabs\PhpAlly\Rule;
+
+use DOMElement;
+
+/**
+*  Links that are broken need to be removed or manually updated.
+*  Based on UDOIT 2.8.X https://github.com/ucfopen/UDOIT/blob/classic/lib/Udoit.php
+*  contributions by Emily Sachs
+*/
+class BrokenLink extends BaseRule
+{
+
+	public function id()
+	{
+		return self::class;
+	}
+
+	public function check()
+	{
+		$links = array();
+		foreach ($this->getAllElements('a') as $a) {
+			$href = $a->getAttribute('href');
+			if ($href) {
+				$links[$href] = $a;
+			}
+		}
+		$this->checkLink($links);
+
+		return count($this->issues);
+	}
+
+	private function checkLink($links) {
+		$curls = array();
+		$mcurl = curl_multi_init();
+		foreach (array_keys($links) as $i => $link) {
+			$curls[$i] = curl_init();
+			curl_setopt($curls[$i], CURLOPT_URL, $link);
+			curl_setopt($curls[$i], CURLOPT_HEADER, true);
+			curl_setopt($curls[$i], CURLOPT_NOBODY, true);
+			curl_setopt($curls[$i], CURLOPT_REFERER, true);
+			curl_setopt($curls[$i], CURLOPT_TIMEOUT, 2);
+			curl_setopt($curls[$i], CURLOPT_TIMEOUT, 2);
+			curl_setopt($curls[$i], CURLOPT_AUTOREFERER, true);
+			curl_setopt($curls[$i], CURLOPT_RETURNTRANSFER, true);
+			curl_setopt($curls[$i], CURLOPT_FOLLOWLOCATION, true);
+			curl_multi_add_handle($mcurl, $curls[$i]);
+		}
+		$running = null;
+		do {
+			curl_multi_exec($mcurl, $running);
+		} while ($running > 0);
+		foreach (array_keys($links) as $i => $link) {
+			$status = curl_getinfo($curls[$i], CURLINFO_RESPONSE_CODE);
+			// If the status is greater than or equal to 400 the link is broken.
+			if ($status >= 400) {
+				$this->setIssue($links[$link]);
+			}
+			curl_multi_remove_handle($mcurl, $curls[$i]);
+		}
+		curl_multi_close($mcurl);
+	}
+}

src/Rule/RedirectedLink.php

@@ -0,0 +1,114 @@
+<?php
+
+namespace CidiLabs\PhpAlly\Rule;
+
+use DOMElement;
+
+/**
+*  Links that are permanently redirected should be updated with the new link.
+*  Based on UDOIT 2.8.X https://github.com/ucfopen/UDOIT/blob/classic/lib/Udoit.php
+*  contributions by Emily Sachs
+*/
+class RedirectedLink extends BaseRule
+{
+
+	public function id()
+	{
+		return self::class;
+	}
+
+	public function check()
+	{
+		$links = array();
+		foreach ($this->getAllElements('a') as $a) {
+			$href = $a->getAttribute('href');
+			if ($href) {
+				$links[$href] = $a;
+			}
+		}
+		$this->checkLink($links);
+
+		return count($this->issues);
+	}
+
+	private function checkLink($links) {
+		$curls = array();
+		$mcurl = curl_multi_init();
+		foreach (array_keys($links) as $i => $link) {
+			$curls[$i] = curl_init();
+			curl_setopt($curls[$i], CURLOPT_URL, $link);
+			curl_setopt($curls[$i], CURLOPT_HEADER, true);
+			curl_setopt($curls[$i], CURLOPT_NOBODY, true);
+			curl_setopt($curls[$i], CURLOPT_REFERER, true);
+			curl_setopt($curls[$i], CURLOPT_TIMEOUT, 2);
+			curl_setopt($curls[$i], CURLOPT_TIMEOUT, 2);
+			curl_setopt($curls[$i], CURLOPT_AUTOREFERER, true);
+			curl_setopt($curls[$i], CURLOPT_RETURNTRANSFER, true);
+			curl_setopt($curls[$i], CURLOPT_FOLLOWLOCATION, true);
+			curl_multi_add_handle($mcurl, $curls[$i]);
+		}
+		$running = null;
+		do {
+			curl_multi_exec($mcurl, $running);
+		} while ($running > 0);
+		foreach (array_keys($links) as $i => $link) {
+			$status = curl_getinfo($curls[$i], CURLINFO_RESPONSE_CODE);
+			// If the status is 400 or greater the link is broken so dont bother checking.
+			if ($status < 400) {
+				$this->checkRedirect($links[$link]);
+			}
+			curl_multi_remove_handle($mcurl, $curls[$i]);
+		}
+		curl_multi_close($mcurl);
+	}
+
+	private function checkRedirect($original) {
+		$link = $original->getAttribute('href');
+		$curl = curl_init();
+		curl_setopt($curl, CURLOPT_URL, $link);
+		curl_setopt($curl, CURLOPT_HEADER, true);
+		curl_setopt($curl, CURLOPT_NOBODY, true);
+		curl_setopt($curl, CURLOPT_REFERER, true);
+		curl_setopt($curl, CURLOPT_TIMEOUT, 2);
+		curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
+
+		curl_exec($curl);
+		$redirect = curl_getinfo($curl, CURLINFO_REDIRECT_URL);
+		$status = curl_getinfo($curl, CURLINFO_RESPONSE_CODE);
+		curl_close($curl);
+
+		// Only permanent redirections are a problem
+		if ($status === 301 || $status === 308) {
+			$this->followPermanentRedirects($original, $redirect);
+		}
+	}
+
+	private function followPermanentRedirects($original, $link, $maxRedirects = 20) {
+		// Avoid infinite calls. 20 is chrome and firefox redirect limit.
+		if ($maxRedirects < 1) {
+			$this->setIssue($original, null, json_encode(array('redirect_url' => $link)));
+			return;
+		}
+
+		$curl = curl_init();
+		curl_setopt($curl, CURLOPT_URL, $link);
+		curl_setopt($curl, CURLOPT_HEADER, true);
+		curl_setopt($curl, CURLOPT_NOBODY, true);
+		curl_setopt($curl, CURLOPT_REFERER, true);
+		curl_setopt($curl, CURLOPT_TIMEOUT, 2);
+		curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
+
+		curl_exec($curl);
+		$redirect = curl_getinfo($curl, CURLINFO_REDIRECT_URL);
+		$status = curl_getinfo($curl, CURLINFO_RESPONSE_CODE);
+		curl_close($curl);
+
+		// Continue until we run out of permanent redirects
+		if ($status === 301 || $status === 308) {
+			$this->followPermanentRedirects($original, $redirect, $maxRedirects - 1);
+		} else {
+			$this->setIssue($original, null, json_encode(array('redirect_url' => $link)));
+		}
+	}
+}
+

src/rules.json

@@ -5,6 +5,7 @@
   "CidiLabs\\PhpAlly\\Rule\\AnchorSuspiciousLinkText",
   "CidiLabs\\PhpAlly\\Rule\\BaseFontIsNotUsed",
   "CidiLabs\\PhpAlly\\Rule\\BlinkIsNotUsed",
+  "CidiLabs\\PhpAlly\\Rule\\BrokenLink",
   "CidiLabs\\PhpAlly\\Rule\\ContentTooLong",
   "CidiLabs\\PhpAlly\\Rule\\CssTextHasContrast",
   "CidiLabs\\PhpAlly\\Rule\\CssTextStyleEmphasize",
@@ -29,6 +30,7 @@
   "CidiLabs\\PhpAlly\\Rule\\ObjectTagDetected",
   "CidiLabs\\PhpAlly\\Rule\\ParagraphNotUsedAsHeader",
   "CidiLabs\\PhpAlly\\Rule\\PreShouldNotBeUsedForTabularValues",
+  "CidiLabs\\PhpAlly\\Rule\\RedirectedLink",
   "CidiLabs\\PhpAlly\\Rule\\TableDataShouldHaveTableHeader",
   "CidiLabs\\PhpAlly\\Rule\\TableHeaderShouldHaveScope",
   "CidiLabs\\PhpAlly\\Rule\\VideoCaptionsMatchCourseLanguage",

tests/BrokenLinkTest.php

@@ -0,0 +1,35 @@
+<?php
+
+use CidiLabs\PhpAlly\Rule\BrokenLink;
+
+class BrokenLinkTest extends PhpAllyTestCase {
+    public function testCheckValid()
+    {
+        $html = '<div><a href="www.google.com">I am a link.</a><div>';
+        $dom = new \DOMDocument('1.0', 'utf-8');
+        $dom->loadHTML($html);
+        $rule = new BrokenLink($dom);
+
+        $this->assertEquals(0, $rule->check(), 'BrokenLink should have no issue.');
+    }
+
+    public function testCheckBroken400()
+    {
+        $html = '<div><a href="http://www.deadlinkcity.com/error-page.asp?e=400">I am a link.</a><div>';
+        $dom = new \DOMDocument('1.0', 'utf-8');
+        $dom->loadHTML($html);
+        $rule = new BrokenLink($dom);
+
+        $this->assertEquals(1, $rule->check(), 'BrokenLink should have one issue.');
+    }
+
+    public function testCheckBroken404()
+    {
+        $html = '<div><a href="https://webaim.org/brokenlink">I am a link.</a><div>';
+        $dom = new \DOMDocument('1.0', 'utf-8');
+        $dom->loadHTML($html);
+        $rule = new BrokenLink($dom);
+
+        $this->assertEquals(1, $rule->check(), 'BrokenLink should have one issue.');
+    }
+}

tests/PhpAllyTest.php

@@ -29,7 +29,7 @@ public function testCheckOne()
         $this->phpAllyReportTest($report);
     }
 
-    public function testCheckMany() 
+    public function testCheckMany()
     {
         $ally = new PhpAlly();
         $options = [
@@ -41,17 +41,17 @@ public function testCheckMany()
         $report = $ally->checkMany($this->getManyHtml(), $ally->getRuleIds(), $options);
         $issues = $report->getIssues();
         $issue = reset($issues);
-        
-        $this->assertCount(6, $issues, 'Total report should have 5 issues.');
+
+        $this->assertCount(6, $issues, 'Total report should have 6 issues.');
         $this->phpAllyIssueTest($issue);
         $this->phpAllyReportTest($report);
     }
 
-    
+
     protected function phpAllyReportTest($report)
     {
         $issues = $report->getIssues();
-        
+
         foreach($issues as $issue) {
             $this->phpAllyIssueTest($issue);
         }
@@ -64,4 +64,4 @@ protected function phpAllyIssueTest(PhpAllyIssue $issue)
         $this->assertEquals(DOMElement::class, get_class($issue->getPreviewElement()), 'Issue return DomElement for getPreviewElement()');
     }
 
-}
+}

tests/PhpAllyTestCase.php

@@ -18,10 +18,10 @@ protected function getManyHtml()
     {
         return '<div>
             <p style="color: #000;">Paragraph text is here.</p>
-            <a href="https://google.com"> </a>
-            <a href="https://google.com">Click Here</a>
-            <a href="https://google.com"> </a>
-            
+            <a href="https://www.google.com/"> </a>
+            <a href="https://www.google.com/">Click Here</a>
+            <a href="https://www.google.com/"> </a>
+
             <p style="color: #0000FF";"><strong>Paragraph text does have enough contrast.</strong></p>
 
             <p>Paragraph text <span style="color: #000">has</span> enough contrast.</p>
@@ -129,4 +129,4 @@ protected function getImageHtml()
     {
         return '';
     }
-}
+}

tests/RedirectedLinkTest.php

@@ -0,0 +1,42 @@
+<?php
+
+use CidiLabs\PhpAlly\Rule\RedirectedLink;
+
+class RedirectedLinkTest extends PhpAllyTestCase {
+    public function testCheckNotRedirected()
+    {
+        $html = '<div><a href="www.google.com">I am a link.</a><div>';
+        $dom = new \DOMDocument('1.0', 'utf-8');
+        $dom->loadHTML($html);
+        $rule = new RedirectedLink($dom);
+
+        $this->assertEquals(0, $rule->check(), 'RedirectedLink should have no issue.');
+    }
+
+    public function testCheckRedirected()
+    {
+        $html = '<div><a href="https://online.ucf.edu/udoit">I am a link.</a><div>';
+        $dom = new \DOMDocument('1.0', 'utf-8');
+        $dom->loadHTML($html);
+        $rule = new RedirectedLink($dom);
+
+        $this->assertEquals(1, $rule->check(), 'RedirectedLink should have one issue.');
+    }
+
+    public function testCheckRedirectedAndMetadata()
+    {
+        $html = '<div><a href="https://online.ucf.edu/udoit">I am a link.</a><div>';
+        $dom = new \DOMDocument('1.0', 'utf-8');
+        $dom->loadHTML($html);
+        $rule = new RedirectedLink($dom);
+
+        // Check if metadata is present with a new link
+        $result = $rule->check();
+        if ($rule->getIssues() && count($rule->getIssues()) == 1) {
+            $meta = $rule->getIssues()[0]->getMetadata();
+            $result = 1 + $result;
+        }
+
+        $this->assertEquals(2, $result, 'RedirectedLink should have one issue.');
+    }
+}