# This is a combination of 2 commits.

# The first commit's message is:

# This is a combination of 4 commits.
# The first commit's message is:
Transcript bucket settings.

# The 2nd commit message will be skipped:

#	Transcript file field to use django storage.

# The 3rd commit message will be skipped:

#	example method to get_or_create the transcript url.

# The 4th commit message will be skipped:

#	Adding file_name to the transcript method.

# The 2nd commit message will be skipped:

#	Transcript bucket settings.

.gitignore

@@ -69,3 +69,5 @@ venv/
 venvs/
 
 video-images/
+
+video-transcripts/

edxval/api.py

@@ -198,7 +198,7 @@ def create_video_transcript(video_id, language, transcript_url, transcript_forma
     transcript = Transcript.objects.create(
         video_id=video_id,
         language=language,
-        transcript_url=transcript_url,
+        transcript=transcript_url,
         fmt=transcript_format,
         provider=provider,
     )
@@ -222,7 +222,7 @@ def update_video_transcript(video_id, language, transcript_url, transcript_forma
     """
     transcript = Transcript.objects.get(video_id=video_id, language=language)
     transcript.language = language
-    transcript.transcript_url = transcript_url
+    transcript.transcript = transcript_url
     transcript.fmt = transcript_format
     transcript.provider = provider
     transcript.save()
@@ -230,6 +230,11 @@ def update_video_transcript(video_id, language, transcript_url, transcript_forma
     return TranscriptSerializer(transcript).data
 
 
+def create_update_video_transcript(video_id, language, file_data, file_name=None):
+    video_transcript, _ = Transcript.create_or_update(video_id=video_id,language=language, file_data=file_data,file_name=file_name)
+    return video_transcript.transcript_url()
+
+
 def get_course_video_image_url(course_id, edx_video_id):
     """
     Returns course video image url or None if no image found

edxval/migrations/0006_auto_20170816_0325.py

+# -*- coding: utf-8 -*-
+from __future__ import unicode_literals
+
+from django.db import migrations, models
+import django.utils.timezone
+import model_utils.fields
+import edxval.models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('edxval', '0005_videoimage'),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='Transcript',
+            fields=[
+                ('id', models.AutoField(verbose_name='ID', serialize=False, auto_created=True, primary_key=True)),
+                ('created', model_utils.fields.AutoCreatedField(default=django.utils.timezone.now, verbose_name='created', editable=False)),
+                ('modified', model_utils.fields.AutoLastModifiedField(default=django.utils.timezone.now, verbose_name='modified', editable=False)),
+                ('video_id', models.CharField(max_length=255)),
+                ('transcript', edxval.models.CustomizableFileField(null=True, blank=True)),
+                ('language', models.CharField(max_length=8, db_index=True)),
+                ('provider', models.CharField(default=b'Custom', max_length=30, choices=[(b'Custom', b'Custom'), (b'3PlayMedia', b'3PlayMedia'), (b'Cielo24', b'Cielo24')])),
+                ('fmt', models.CharField(db_index=True, max_length=20, choices=[(b'srt', b'SubRip'), (b'sjson', b'SRT JSON')])),
+            ],
+        ),
+        migrations.RemoveField(
+            model_name='subtitle',
+            name='video',
+        ),
+        migrations.DeleteModel(
+            name='Subtitle',
+        ),
+        migrations.AlterUniqueTogether(
+            name='transcript',
+            unique_together=set([('video_id', 'language')]),
+        ),
+    ]

edxval/models.py

@@ -26,7 +26,7 @@ from django.utils.six import python_2_unicode_compatible
 
 from model_utils.models import TimeStampedModel
 
-from edxval.utils import video_image_path, get_video_image_storage
+from edxval.utils import video_image_path, get_video_image_storage, video_transcript_path, get_video_transcript_storage
 
 logger = logging.getLogger(__name__)  # pylint: disable=C0103
 
@@ -357,6 +357,35 @@ class TranscriptFormat(object):
     )
 
 
+class CustomizableFileField(models.FileField):
+    """
+    Subclass of FileField that allows custom settings to not
+    be serialized (hard-coded) in migrations. Otherwise,
+    migrations include optional settings for storage (such as
+    the storage class and bucket name); we don't want to
+    create new migration files for each configuration change.
+    """
+    def __init__(self, *args, **kwargs):
+        kwargs.update(dict(
+            upload_to=video_transcript_path,
+            storage=get_video_transcript_storage(),
+            max_length=500,  # allocate enough for filepath
+            blank=True,
+            null=True
+        ))
+        super(CustomizableFileField, self).__init__(*args, **kwargs)
+
+    def deconstruct(self):
+        """
+        Override base class method.
+        """
+        name, path, args, kwargs = super(CustomizableFileField, self).deconstruct()
+        del kwargs['upload_to']
+        del kwargs['storage']
+        del kwargs['max_length']
+        return name, path, args, kwargs
+
+
 class Transcript(TimeStampedModel):
     """
     Transcript for a video
@@ -371,7 +400,7 @@ class Transcript(TimeStampedModel):
     """
     # It can be an edx_video_id or an external video id (e.g. in case of external URLs - YT/MP4/WEBM etc.)
     video_id = models.CharField(max_length=255)
-    transcript_url = models.TextField(null=True, blank=True)
+    transcript = CustomizableFileField()
     language = models.CharField(max_length=8, db_index=True)
     provider = models.CharField(
         max_length=30,
@@ -389,6 +418,31 @@ class Transcript(TimeStampedModel):
     def __str__(self):
         return '{lang} Transcript for {video}'.format(lang=self.language, video=self.video_id)
 
+    def transcript_url(self):
+        """
+        Return transcript url for a course video transcript.
+        """
+        storage = get_video_transcript_storage()
+        return storage.url(self.transcript.name)
+
+    @classmethod
+    def create_or_update(cls, video_id, language, file_name=None, file_data=None):
+        """
+        Create Transcript object.
+        """
+        video_transcript, created = cls.objects.get_or_create(video_id=video_id, language=language, fmt=TranscriptFormat.SJSON)
+
+        with closing(file_data) as transcript_file_data:
+            file_name = '{uuid}{ext}'.format(uuid=uuid4().hex, ext=os.path.splitext(file_name)[1])
+            try:
+                video_transcript.transcript.save(file_name, transcript_file_data)
+            except Exception:  # pylint: disable=broad-except
+                logger.exception('VAL: Video Transcript save failed to storage for video_id [%s]', video_id)
+                raise
+
+        return video_transcript, created
+
+
 
 @receiver(models.signals.post_save, sender=Video)
 def video_status_update_callback(sender, **kwargs):  # pylint: disable=unused-argument

edxval/serializers.py

@@ -54,10 +54,18 @@ class TranscriptSerializer(serializers.ModelSerializer):
     """
     Serializer for Transcript objects
     """
+    transcript_url = serializers.SerializerMethodField()
+
     class Meta:  # pylint: disable=C1001, C0111
         model = Transcript
         lookup_field = 'video_id'
-        fields = ('video_id', 'transcript_url', 'language', 'provider', 'fmt')
+        fields = ('video_id', 'language', 'provider', 'fmt', 'transcript_url')
+
+    def get_transcript_url(self, obj):
+        """
+        Return relative url for the object
+        """
+        return obj.transcript_url
 
 
 class CourseSerializer(serializers.RelatedField):

edxval/settings.py

@@ -190,3 +190,13 @@ VIDEO_IMAGE_SETTINGS = dict(
     VIDEO_IMAGE_MIN_BYTES=100,
     DIRECTORY_PREFIX='video-images/',
 )
+
+VIDEO_TRANSCRIPTS_SETTINGS = dict(
+    # Backend storage
+    # STORAGE_CLASS='storages.backends.s3boto.S3BotoStorage',
+    # STORAGE_KWARGS=dict(bucket='video-transcripts-bucket'),
+    # If you are changing prefix value then update the .gitignore accordingly
+    # so that transcripts created during tests due to upload should be ignored
+    VIDEO_TRANSCRIPTS_MAX_BYTES=3145728,
+    DIRECTORY_PREFIX='video-transcripts/',
+)

edxval/utils.py

@@ -29,3 +29,28 @@ def get_video_image_storage():
         # during edx-platform loading this method gets called but settings are not ready yet
         # so in that case we will return default(FileSystemStorage) storage class instance
         return get_storage_class()()
+
+
+def video_transcript_path(video_transcript_instance, filename):  # pylint:disable=unused-argument
+    """
+    Returns video transcript path.
+
+    Arguments:
+        video_transcript_instance (VideoImage): This is passed automatically by models.CustomizableImageField
+        filename (str): name of image file
+    """
+    return u'{}{}'.format(settings.VIDEO_TRANSCRIPTS_SETTINGS.get('DIRECTORY_PREFIX', ''), filename)
+
+
+def get_video_transcript_storage():
+    """
+    Return the configured django storage backend.
+    """
+    if hasattr(settings, 'VIDEO_TRANSCRIPTS_SETTINGS'):
+        return get_storage_class(
+            settings.VIDEO_TRANSCRIPTS_SETTINGS.get('STORAGE_CLASS'),
+        )(**settings.VIDEO_TRANSCRIPTS_SETTINGS.get('STORAGE_KWARGS', {}))
+    else:
+        # during edx-platform loading this method gets called but settings are not ready yet
+        # so in that case we will return default(FileSystemStorage) storage class instance
+        return get_storage_class()()