Added adapter-specific exception. Added second attempt to obtain PSL ignoring SSL peer. Protect PHP cache to be rewrited with an empty array in case of error.

Throwing an HTTP adapter's exception if content couldn't be fetched
Added handling of exceptions during getting PSL
Prevent writing an empty data into a cache file
(by  Alex Shu)

Author: Alex Shu

data/public-suffix-list.php

@@ -11,6 +11,9 @@
 
 namespace Pdp\HttpAdapter;
 
+
+use Pdp\HttpAdapter\Exception;
+
 /**
  * cURL http adapter.
  *
@@ -31,8 +34,20 @@ public function getContent($url)
     {
         $ch = curl_init();
         curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
+        curl_setopt($ch, CURLOPT_FAILONERROR, true);
+        curl_setopt($ch, CURLOPT_TIMEOUT, 5);
         curl_setopt($ch, CURLOPT_URL, $url);
+
         $content = curl_exec($ch);
+
+        if ($errNo = curl_errno($ch)) {
+            throw new Exception\CurlHttpAdapterException("CURL error [{$errNo}]: " . curl_error($ch), $errNo);
+        }
+
+        $responseCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
+        if ($responseCode !== 200) {
+            throw new Exception\CurlHttpAdapterException('Wrong HTTP response code: ' . $responseCode, $responseCode);
+        }
         curl_close($ch);
 
         return $content;

src/Pdp/HttpAdapter/CurlHttpAdapter.php

@@ -0,0 +1,9 @@
+<?php
+
+namespace Pdp\HttpAdapter\Exception;
+
+
+class CurlHttpAdapterException extends HttpAdapterException
+{
+
+}

src/Pdp/HttpAdapter/Exception/CurlHttpAdapterException.php

@@ -0,0 +1,9 @@
+<?php
+
+namespace Pdp\HttpAdapter\Exception;
+
+
+class HttpAdapterException extends \Exception
+{
+
+}

src/Pdp/HttpAdapter/Exception/HttpAdapterException.php

@@ -30,6 +30,8 @@ interface HttpAdapterInterface
      * @param string $url
      *
      * @return string Retrieved content
+     *
+     * @throws Exception\HttpAdapterException
      */
     public function getContent($url);
 }

src/Pdp/HttpAdapter/HttpAdapterInterface.php

@@ -11,6 +11,7 @@
 
 namespace Pdp;
 
+use Pdp\HttpAdapter\Exception\HttpAdapterException;
 use Pdp\HttpAdapter\HttpAdapterInterface;
 
 /**
@@ -69,13 +70,17 @@ public function refreshPublicSuffixList()
         $publicSuffixListArray = $this->parseListToArray(
             $this->cacheDir . '/' . self::PDP_PSL_TEXT_FILE
         );
-        $this->writePhpCache($publicSuffixListArray);
+
+        // do not empty existing PHP cache file if source TXT is empty
+        if (is_array($publicSuffixListArray) && !empty($publicSuffixListArray)) {
+            $this->writePhpCache($publicSuffixListArray);
+        }
     }
 
     /**
      * Obtain Public Suffix List from its online source and write to cache dir.
      *
-     * @return int Number of bytes that were written to the file
+     * @return int|bool Number of bytes that were written to the file OR false in case of error
      */
     public function fetchListFromSource()
     {
@@ -185,13 +190,20 @@ public function writePhpCache(array $publicSuffixList)
      */
     public function getList()
     {
-        if (!file_exists($this->cacheDir . '/' . self::PDP_PSL_PHP_FILE)) {
-            $this->refreshPublicSuffixList();
+        $filename = $this->cacheDir . '/' . self::PDP_PSL_PHP_FILE;
+
+        if (file_exists($filename)) {
+            $data = include $filename;
+        } else {
+            try {
+                $this->refreshPublicSuffixList();
+                $data = include $filename;
+            } catch (\Exception $e) {
+                $data = array();
+            }
         }
 
-        $this->list = new PublicSuffixList(
-            include $this->cacheDir . '/' . self::PDP_PSL_PHP_FILE
-        );
+        $this->list = new PublicSuffixList((array) $data);
 
         return $this->list;
     }
@@ -208,10 +220,16 @@ public function getList()
      */
     protected function write($filename, $data)
     {
-        $result = @file_put_contents($this->cacheDir . '/' . $filename, $data);
+        $data = trim($data);
+        $filename = $this->cacheDir . '/' . $filename;
+
+        if (empty($data)) {
+            throw new \Exception("No data to write into '{$filename}'");
+        }
 
+        $result = @file_put_contents($filename, $data . PHP_EOL);
         if ($result === false) {
-            throw new \Exception("Cannot write '" . $this->cacheDir . '/' . "$filename'");
+            throw new \Exception("Cannot write '{$filename}'");
         }
 
         return $result;

src/Pdp/PublicSuffixListManager.php

@@ -65,9 +65,9 @@ public function testPublicSuffixSpec()
         $this->checkPublicSuffix('test.ac', 'test.ac');
         // TLD with only 1 (wildcard) rule.
         $this->checkPublicSuffix('cy', null);
-        $this->checkPublicSuffix('c.cy', null);
-        $this->checkPublicSuffix('b.c.cy', 'b.c.cy');
-        $this->checkPublicSuffix('a.b.c.cy', 'b.c.cy');
+        $this->checkPublicSuffix('c.cy', 'c.cy');
+        $this->checkPublicSuffix('b.c.cy', 'c.cy');
+        $this->checkPublicSuffix('a.b.c.cy', 'c.cy');
         // More complex TLD.
         $this->checkPublicSuffix('jp', null);
         $this->checkPublicSuffix('test.jp', 'test.jp');

tests/src/Pdp/CheckPublicSuffixTest.php

@@ -2,6 +2,9 @@
 
 namespace Pdp\HttpAdapter;
 
+
+use Pdp\HttpAdapter\Exception;
+
 /**
  * @group internet
  */
@@ -32,4 +35,16 @@ public function testGetContent()
         $this->assertNotNull($content);
         $this->assertContains('google', $content);
     }
+
+    public function testExceptionBadUrl()
+    {
+        $this->setExpectedException('Pdp\HttpAdapter\Exception\HttpAdapterException', '', CURLE_COULDNT_RESOLVE_HOST);
+        $content = $this->adapter->getContent('https://aaaa.aaaa');
+    }
+
+    public function testExceptionBadHttpsCertificate()
+    {
+        $this->setExpectedException('Pdp\HttpAdapter\Exception\HttpAdapterException', '', CURLE_SSL_PEER_CERTIFICATE);
+        $content = $this->adapter->getContent('https://tv.eurosport.com/');
+    }
 }

tests/src/Pdp/HttpAdapter/CurlHttpAdapterTest.php

@@ -189,7 +189,7 @@ public function parseDataProvider()
             array('cea-law.co.il', 'co.il', 'cea-law.co.il', null, 'cea-law.co.il'),
             array('http://edition.cnn.com/WORLD/', 'com', 'cnn.com', 'edition', 'edition.cnn.com'),
             array('http://en.wikipedia.org/', 'org', 'wikipedia.org', 'en', 'en.wikipedia.org'),
-            array('a.b.c.cy', 'c.cy', 'b.c.cy', 'a', 'a.b.c.cy'),
+            array('a.b.c.cy', 'cy', 'c.cy', 'a.b', 'a.b.c.cy'),
             array('https://test.k12.ak.us', 'k12.ak.us', 'test.k12.ak.us', null, 'test.k12.ak.us'),
             array('www.scottwills.co.uk', 'co.uk', 'scottwills.co.uk', 'www', 'www.scottwills.co.uk'),
             array('b.ide.kyoto.jp', 'ide.kyoto.jp', 'b.ide.kyoto.jp', null, 'b.ide.kyoto.jp'),