try to more quickly check the state of a li task job before preparing…

… from crawl
webignition · Mar 20, 2015 · 141537b · 141537b
1 parent 6de9eb7
commit 141537b
Show file tree

Hide file tree

Showing 2 changed files with 33 additions and 12 deletions.
diff --git a/src/SimplyTestable/ApiBundle/Controller/TaskController.php b/src/SimplyTestable/ApiBundle/Controller/TaskController.php
@@ -45,14 +45,14 @@ public function completeAction($canonical_url, $task_type, $parameter_hash) {
             return $this->sendFailureResponse();
         }        
 
-        $taskType = $this->getTaskTypeService()->getByName($task_type);        
-        
+        $taskType = $this->getTaskTypeService()->getByName($task_type);
+
         $tasks = $this->getTaskService()->getEquivalentTasks($canonical_url, $taskType, $parameter_hash, $this->getTaskService()->getIncompleteStates());
 
         if (count($tasks) === 0) {
             return $this->sendGoneResponse();
         }
-        
+
         $endDateTime = new \DateTime($this->getArguments('completeByUrlAndTaskTypeAction')->get('end_date_time'));
         $rawOutput = $this->getArguments('completeByUrlAndTaskTypeAction')->get('output');
 
@@ -89,20 +89,22 @@ public function completeAction($canonical_url, $task_type, $parameter_hash) {
                 $this->getJobService()->complete($task->getJob());
             }
 
-            if ($task->getType()->equals($this->getTaskTypeService()->getByName('URL discovery'))) {
+            if ($task->getType()->equals($urlDiscoveryTaskType)) {
                 if ($this->getJobService()->isCompleted($task->getJob())) {
-                    $crawlJobContainer = $this->getCrawlJobContainerService()->getForJob($task->getJob());
+                    if ($this->getCrawlJobContainerService()->getEntityRepository()->doesCrawlTaskParentStateMatchState($task, $this->getJobService()->getFailedNoSitemapState())) {
+                        $crawlJobContainer = $this->getCrawlJobContainerService()->getForJob($task->getJob());
 
-                    foreach ($crawlJobContainer->getParentJob()->getRequestedTaskTypes() as $taskType) {
-                        /* @var $taskType TaskType */
-                        $taskTypeParameterDomainsToIgnoreKey = strtolower(str_replace(' ', '-', $taskType->getName())) . '-domains-to-ignore';
+                        foreach ($crawlJobContainer->getParentJob()->getRequestedTaskTypes() as $taskType) {
+                            /* @var $taskType TaskType */
+                            $taskTypeParameterDomainsToIgnoreKey = strtolower(str_replace(' ', '-', $taskType->getName())) . '-domains-to-ignore';
 
-                        if ($this->container->hasParameter($taskTypeParameterDomainsToIgnoreKey)) {
-                            $this->getJobPreparationService()->setPredefinedDomainsToIgnore($taskType, $this->container->getParameter($taskTypeParameterDomainsToIgnoreKey));
+                            if ($this->container->hasParameter($taskTypeParameterDomainsToIgnoreKey)) {
+                                $this->getJobPreparationService()->setPredefinedDomainsToIgnore($taskType, $this->container->getParameter($taskTypeParameterDomainsToIgnoreKey));
+                            }
                         }
-                    }
 
-                    $this->getJobPreparationService()->prepareFromCrawl($crawlJobContainer);
+                        $this->getJobPreparationService()->prepareFromCrawl($crawlJobContainer);
+                    }
                 }
 
                 $this->getResqueQueueService()->enqueue(

diff --git a/src/SimplyTestable/ApiBundle/Repository/CrawlJobContainerRepository.php b/src/SimplyTestable/ApiBundle/Repository/CrawlJobContainerRepository.php
@@ -3,10 +3,29 @@
 
 use Doctrine\ORM\EntityRepository;
 use SimplyTestable\ApiBundle\Entity\Job\Job;
+use SimplyTestable\ApiBundle\Entity\State;
+use SimplyTestable\ApiBundle\Entity\Task\Task;
 use SimplyTestable\ApiBundle\Entity\User;
 
 class CrawlJobContainerRepository extends EntityRepository
 {
+
+    public function doesCrawlTaskParentStateMatchState(Task $task, State $state) {
+        $queryBuilder = $this->createQueryBuilder('CrawlJobContainer');
+        $queryBuilder->join('CrawlJobContainer.parentJob', 'ParentJob');
+        $queryBuilder->join('CrawlJobContainer.crawlJob', 'CrawlJob');
+        $queryBuilder->join('ParentJob.state', 'State');
+
+        $queryBuilder->select('State.name');
+
+        $queryBuilder->where('CrawlJob = :CrawlJob');
+        $queryBuilder->setParameter('CrawlJob', $task->getJob());
+        $queryBuilder->setMaxResults(1);
+
+        $result = $queryBuilder->getQuery()->getResult();
+        return (count($result) === 0) ? false : $result[0]['name'] == $state->getName();
+    }
+
 
     /**
      *