From bcbd8a8050ce16e4cd6ea9ed86b00b567c4fb3d0 Mon Sep 17 00:00:00 2001
From: Duncan Sommerville <duncan.sommerville@gmail.com>
Date: Mon, 19 Oct 2015 15:36:34 -0400
Subject: [PATCH] Fix bug in auto ingest, immediately retry on failure

---
 .../application/services/PodcastEpisodeService.php  | 10 +++++++---
 python_apps/airtime-celery/airtime-celery/tasks.py  | 13 +++++++++----
 2 files changed, 16 insertions(+), 7 deletions(-)

diff --git a/airtime_mvc/application/services/PodcastEpisodeService.php b/airtime_mvc/application/services/PodcastEpisodeService.php
index 45ce09fe3..6e8276ed5 100644
--- a/airtime_mvc/application/services/PodcastEpisodeService.php
+++ b/airtime_mvc/application/services/PodcastEpisodeService.php
@@ -117,17 +117,21 @@ class Application_Service_PodcastEpisodeService extends Application_Service_Thir
         $ref = parent::updateTrackReference($task, $episodeId, $episodes, $status);
 
         if ($status == CELERY_SUCCESS_STATUS) {
-            foreach($episodes as $episode) {
+            foreach ($episodes as $episode) {
                 // Since we process episode downloads as a batch, individual downloads can fail
                 // even if the task itself succeeds
+                $dbEpisode = PodcastEpisodesQuery::create()
+                    ->findOneByDbId($episode->episodeid);
                 if ($episode->status) {
-                    $dbEpisode = PodcastEpisodesQuery::create()
-                        ->findOneByDbId($episode->episodeid);
                     $dbEpisode->setDbFileId($episode->fileid)
                         ->save();
+                } else {
+                    Logging::warn("Celery task $task episode $episode->episodeid unsuccessful with status $episode->status");
+                    $dbEpisode->delete();
                 }
             }
         }
+        // TODO: do we need a broader fail condition here?
 
         return $ref;
     }
diff --git a/python_apps/airtime-celery/airtime-celery/tasks.py b/python_apps/airtime-celery/airtime-celery/tasks.py
index 83a7faefc..a3904c95f 100644
--- a/python_apps/airtime-celery/airtime-celery/tasks.py
+++ b/python_apps/airtime-celery/airtime-celery/tasks.py
@@ -110,10 +110,10 @@ def podcast_download(episodes, callback_url, api_key):
             f = json.loads(re.content)  # Read the response from the media API to get the file id
             obj['fileid'] = f['id']
             obj['status'] = 1
-            response.append(obj)
         except Exception as e:
             logger.info('Error during file download: {0}'.format(e.message))
             obj['status'] = 0
+        response.append(obj)
     return json.dumps(response)
 
 
@@ -127,10 +127,15 @@ def get_filename(r):
     """
     # Try to get the filename from the content disposition
     d = r.headers.get('Content-Disposition')
+    filename = ''
     if d:
-        _, params = cgi.parse_header(d)
-        filename = params['filename']
-    else:
+        try:
+            _, params = cgi.parse_header(d)
+            filename = params['filename']
+        except Exception as e:
+            # We end up here if we get a Content-Disposition header with no filename
+            logger.warn("Couldn't find file name in Content-Disposition header, using url")
+    if not filename:
         # Since we don't necessarily get the filename back in the response headers,
         # parse the URL and get the filename and extension
         path = urlparse.urlsplit(r.url).path