Merge pull request #2 from dennisinteractive/28103_handle_mutiple_fields

28103 handle mutiple fields

dennis_link_checker.module

@@ -9,7 +9,9 @@ use Dennis\Link\Checker\LinkLocalisation;
 use Dennis\Link\Checker\Processor;
 use Dennis\Link\Checker\EntityHandler;
 use Dennis\Link\Checker\Analyzer;
-use SystemQueue as Queue;
+use Dennis\Link\Checker\Throttler;
+use Dennis\Link\Checker\Database;
+use Dennis\Link\Checker\Queue;
 
 /**
  * PSR-4 autoloader
@@ -37,11 +39,20 @@ function dennis_link_checker_setup(array $nids) {
     ->setMaxRedirects(10)
     ->setInternalOnly(TRUE)
     ->setLocalisation(LinkLocalisation::ORIGINAL)
+    ->setFieldNames(variable_get('dennis_link_checker_fields', array('body')))
     ->setNodeList($nids);
 
   $queue = new Queue('dennis_link_checker');
   $entity_handler = new EntityHandler($config);
-  $analyzer = new Analyzer($config);
+
+  // Make sure we don't request more than one page per second.
+  $curl_throttler = new Throttler(1);
+
+  // Database object that allows interaction with the DB.
+  $database = new Database();
+
+  $analyzer = new Analyzer($config, $curl_throttler, $database);
+
   return new Processor($config, $queue, $entity_handler, $analyzer);
 }
 

src/Dennis/Link/Checker/Analyzer.php

@@ -10,18 +10,38 @@
  * @package Dennis\Link\Checker
  */
 class Analyzer implements AnalyzerInterface {
-
-  protected $host;
-
+  /**
+   * @var ConfigInterface
+   */
   protected $config;
 
+  /**
+   * @var number of redirects in current chain.
+   */
   protected $redirectCount;
 
+  /**
+   * @var int maximum number of seconds to spend resolving links.
+   */
+  protected $linkTimeLimit = 480;
+
+  /**
+   * @var Throttler
+   */
+  protected $curlThrottler;
+
+  /**
+   * @var Database
+   */
+  protected $database;
+
   /**
    * @inheritDoc
    */
-  public function __construct(ConfigInterface $config) {
+  public function __construct(ConfigInterface $config, Throttler $curl_throttler, Database $database) {
     $this->config = $config;
+    $this->curlThrottler = $curl_throttler;
+    $this->database = $database;
   }
 
   /**
@@ -45,8 +65,17 @@ public function getSiteHost() {
    * @inheritDoc
    */
   public function multipleLinks($links) {
-    foreach ($links as &$link) {
-      $link = $this->link($link);
+    $timeout = $this->linkTimeLimit + time();
+
+    foreach ($links as $link) {
+      if (time() >= $timeout) {
+        throw new TimeoutException(sprintf('Could not process %s links within %s seconds',
+          count($links),
+          $this->linkTimeLimit));
+      }
+      $this->link($link);
+      // Keep the DB connection alive whilst we are processing external links.
+      $this->database->keepConnectionAlive();
     }
 
     return $links;
@@ -56,6 +85,9 @@ public function multipleLinks($links) {
    * @inheritDoc
    */
   public function link(LinkInterface $link) {
+    // Make sure we only process one link per configured number of seconds.
+    $this->curlThrottler->throttle();
+
     // Only redirect 301's so cannot use CURLOPT_FOLLOWLOCATION
     $this->redirectCount = 0;
 
@@ -75,7 +107,8 @@ public function link(LinkInterface $link) {
         ->setNumberOfRedirects($this->redirectCount);
 
     } catch (ResourceFailException $e) {
-      $link->setError($e->getMessage(), $e->getCode());
+      $link->setNumberOfRedirects($this->redirectCount)
+        ->setError($e->getMessage(), $e->getCode());
     }
 
     return $link;
@@ -90,8 +123,13 @@ public function link(LinkInterface $link) {
    */
   protected function followRedirects($url) {
     $info = $this->getInfo($url);
+
     if (!empty($info['redirect_url'])) {
       if ($info['http_code'] == 301) {
+        // Throw exception if we have reached our redirect limit.
+        if ($this->redirectCount > $this->config->getMaxRedirects()) {
+          throw new ResourceFailException(sprintf('Maximum of %s redirects reached.', $this->config->getMaxRedirects()));
+        }
         // Do the redirect
         $this->redirectCount++;
         return $this->followRedirects($info['redirect_url']);
@@ -123,8 +161,10 @@ protected function getInfo($url) {
       return $info;
     }
     else {
+      $errno = curl_errno($ch);
+      $error = curl_error($ch);
       curl_close($ch);
-      throw new ResourceFailException(curl_error($ch), curl_errno($ch));
+      throw new ResourceFailException($error, $errno);
     }
 
   }

src/Dennis/Link/Checker/AnalyzerInterface.php

@@ -13,9 +13,12 @@ interface AnalyzerInterface {
 
   /**
    * AnalyzerInterface constructor.
+   *
    * @param ConfigInterface $config
+   * @param Throttler $curl_throttler
+   * @param Database $database
    */
-  public function __construct(ConfigInterface $config);
+  public function __construct(ConfigInterface $config, Throttler $curl_throttler, Database $database);
 
   /**
    * Checks the link.

src/Dennis/Link/Checker/Config.php

@@ -25,6 +25,8 @@ class Config implements ConfigInterface {
 
   protected $nids = FALSE;
 
+  protected $fieldNames = [];
+
   /**
    * @inheritDoc
    */
@@ -136,4 +138,21 @@ public function setNodeList(array $nids) {
   public function getNodeList() {
     return $this->nids;
   }
+
+  /**
+   * @inheritDoc
+   */
+  public function setFieldNames(array $field_names) {
+    $this->fieldNames = $field_names;
+
+    return $this;
+  }
+
+  /**
+   * @inheritDoc
+   */
+  public function getFieldNames() {
+    return $this->fieldNames;
+  }
+
 }

src/Dennis/Link/Checker/ConfigInterface.php

@@ -117,4 +117,19 @@ public function setNodeList(array $nids);
    * @return array
    */
   public function getNodeList();
+
+  /**
+   * The names of fields to check for links in.
+   *
+   * @param array $field_names
+   * @return self
+   */
+  public function setFieldNames(array $field_names);
+
+  /**
+   * The names of fields to check for links in.
+   *
+   * @return array
+   */
+  public function getFieldNames();
 }

src/Dennis/Link/Checker/Database.php

@@ -0,0 +1,35 @@
+<?php
+/**
+ * @file
+ * Database
+ */
+namespace Dennis\Link\Checker;
+
+/**
+ * Class Database
+ * @package Dennis\Link\Database
+ */
+class Database implements DatabaseInterface {
+  /**
+   * How often the DB should be pinged in seconds.
+   */
+  const interval = 15;
+
+  /**
+   * @var int last time the DB was pinged.
+   */
+  protected $pingTime = 0;
+
+  /**
+   * @inheritDoc
+   */
+  public function keepConnectionAlive() {
+    $now = time();
+    // If it's been more than 15 seconds... Ping!
+    if (($now - $this->pingTime) > self::interval) {
+      db_query('SELECT CURTIME()');
+      // Set the ping time.
+      $this->pingTime = $now;
+    }
+  }
+}

src/Dennis/Link/Checker/DatabaseInterface.php

@@ -0,0 +1,17 @@
+<?php
+/**
+ * @file
+ * DatabaseInterface
+ */
+namespace Dennis\Link\Checker;
+
+/**
+ * Interface DatabaseInterface
+ * @package Dennis\Link\Checker
+ */
+interface DatabaseInterface {
+  /**
+   * Keep database connection alive.
+   */
+  public function keepConnectionAlive();
+}

src/Dennis/Link/Checker/Entity.php

@@ -0,0 +1,64 @@
+<?php
+/**
+ * @file
+ * Entity
+ */
+namespace Dennis\Link\Checker;
+
+/**
+ * Class Entity
+ * @package Dennis\Link\Checker
+ */
+class Entity implements EntityInterface {
+  /**
+   * @var string
+   */
+  protected $entityType;
+
+  /**
+   * @var int
+   */
+  protected $entityId;
+
+  /**
+   * @var ConfigInterface
+   */
+  protected $config;
+
+  /**
+   * @inheritDoc
+   */
+  public function __construct($config, $entity_type, $entity_id) {
+    $this->config = $config;
+    $this->entityType = $entity_type;
+    $this->entityId = $entity_id;
+  }
+
+  /**
+   * @inheritDoc
+   */
+  public function getConfig() {
+    return $this->config;
+  }
+
+  /**
+   * @inheritDoc
+   */
+  public function entityId() {
+    return $this->entityId;
+  }
+
+  /**
+   * @inheritDoc
+   */
+  public function entityType() {
+    return $this->entityType;
+  }
+
+  /**
+   * @inheritDoc
+   */
+  public function getField($field_name) {
+    return new Field($this, $field_name);
+  }
+}