Allow filtering on min/max length of media

tubesync/sync/filtering.py

@@ -28,6 +28,10 @@ def filter_media(instance: Media):
     if filter_filter_text(instance):
         skip = True
 
+    # Check if the video is longer than the max, or shorter than the min
+    if filter_duration(instance):
+        skip = True
+
     # Check if skipping
     if instance.skip != skip:
         instance.skip = skip
@@ -96,3 +100,36 @@ def filter_source_cutoff(instance: Media):
             return True
 
     return False
+
+
+# Check if we skip based on duration (min/max)
+def filter_duration(instance: Media):
+    if not instance.source.filter_seconds:
+        return False
+
+    duration = instance.duration
+    if not duration:
+        # Attempt fallback to slower metadata field, this adds significant time, new media won't need this
+        # Tests show fetching instance.duration can take as long as the rest of the filtering
+        if instance.metadata_duration:
+            duration = instance.metadata_duration
+            instance.duration = duration
+            instance.save()
+        else:
+            log.info(f'Media: {instance.source} / {instance} has no duration stored, not skipping')
+            return False
+
+    duration_limit = instance.source.filter_seconds
+    if instance.source.filter_seconds_min and duration < duration_limit:
+        # Filter out videos that are shorter than the minimum
+        log.info(f'Media: {instance.source} / {instance} is shorter ({duration}) than '
+                 f'the minimum duration ({duration_limit}), skipping')
+        return True
+
+    if not instance.source.filter_seconds_min and duration > duration_limit:
+        # Filter out videos that are greater than the maximum
+        log.info(f'Media: {instance.source} / {instance} is longer ({duration}) than '
+                 f'the maximum duration ({duration_limit}), skipping')
+        return True
+
+    return False

tubesync/sync/migrations/0023_media_duration_filter.py

@@ -0,0 +1,41 @@
+# Generated by pac
+
+from django.db import migrations, models
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('sync', '0022_add_delete_files_on_disk'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='media',
+            name='duration',
+            field=models.PositiveIntegerField(
+                verbose_name='duration',
+                blank=True,
+                null=True,
+                help_text='Duration of media in seconds'),
+        ),
+        migrations.AddField(
+            model_name='source',
+            name='filter_seconds',
+            field=models.PositiveIntegerField(
+                verbose_name='filter seconds',
+                blank=True,
+                null=True,
+                help_text='Filter Media based on Min/Max duration. Leave blank or 0 to disable filtering'),
+        ),
+        migrations.AddField(
+            model_name='source',
+            name='filter_seconds_min',
+            field=models.BooleanField(
+                verbose_name='filter seconds min',
+                choices=[(True, 'Minimum Length'),(False, 'Maximum Length')],
+                default=True,
+                help_text='When Filter Seconds is > 0, do we skip on minimum (video shorter than limit) or maximum ('
+                          'video greater than maximum) video duration'
+            ),
+        ),
+    ]

tubesync/sync/models.py

@@ -104,6 +104,11 @@ class Source(models.Model):
         (FALLBACK_NEXT_BEST_HD, _('Get next best resolution but at least HD'))
     )
 
+    FILTER_SECONDS_CHOICES = (
+        (True, _('Minimum Length')),
+        (False, _('Maximum Length')),
+    )
+
     EXTENSION_M4A = 'm4a'
     EXTENSION_OGG = 'ogg'
     EXTENSION_MKV = 'mkv'
@@ -293,6 +298,19 @@ class Source(models.Model):
         blank=True,
         help_text=_('Regex compatible filter string for video titles')
     )
+    filter_seconds = models.PositiveIntegerField(
+                _('filter seconds'),
+                blank=True,
+                null=True,
+                help_text=_('Filter Media based on Min/Max duration. Leave blank or 0 to disable filtering')
+    )
+    filter_seconds_min = models.BooleanField(
+        _('filter seconds min/max'),
+        choices=FILTER_SECONDS_CHOICES,
+        default=True,
+        help_text=_('When Filter Seconds is > 0, do we skip on minimum (video shorter than limit) or maximum (video '
+                    'greater than maximum) video duration')
+    )
     delete_removed_media = models.BooleanField(
         _('delete removed media'),
         default=False,
@@ -785,7 +803,7 @@ class Media(models.Model):
         _('manual_skip'),
         db_index=True,
         default=False,
-        help_text=_('Media marked as "skipped", won\' be downloaded')
+        help_text=_('Media marked as "skipped", won\'t be downloaded')
     )
     downloaded = models.BooleanField(
         _('downloaded'),
@@ -859,6 +877,13 @@ class Media(models.Model):
         help_text=_('Size of the downloaded media in bytes')
     )
 
+    duration = models.PositiveIntegerField(
+        _('duration'),
+        blank=True,
+        null=True,
+        help_text=_('Duration of media in seconds')
+    )
+
     def __str__(self):
         return self.key
 
@@ -1115,7 +1140,7 @@ class Media(models.Model):
             return None
 
     @property
-    def duration(self):
+    def metadata_duration(self):
         field = self.get_metadata_field('duration')
         duration = self.loaded_metadata.get(field, 0)
         try:

tubesync/sync/tasks.py

@@ -294,6 +294,10 @@ def download_media_metadata(media_id):
     if upload_date:
         media.published = timezone.make_aware(upload_date)
 
+    # Store duration in DB so it's fast to access
+    if media.metadata_duration:
+        media.duration = media.metadata_duration
+
     filter_media(media)
     # Check we can download the media item
     if not media.skip:

tubesync/sync/views.py

@@ -296,8 +296,8 @@ class ValidateSourceView(FormView):
 
 class EditSourceMixin:
     model = Source
-    fields = ('source_type', 'key', 'name', 'directory', 'filter_text', 'media_format',
-              'index_schedule', 'download_media', 'download_cap', 'delete_old_media',
+    fields = ('source_type', 'key', 'name', 'directory', 'filter_text', 'filter_seconds', 'filter_seconds_min',
+              'media_format', 'index_schedule', 'download_media', 'download_cap', 'delete_old_media',
               'delete_removed_media', 'days_to_keep', 'source_resolution', 'source_vcodec',
               'source_acodec', 'prefer_60fps', 'prefer_hdr', 'fallback', 'copy_channel_images',
               'delete_removed_media', 'delete_files_on_disk', 'days_to_keep', 'source_resolution',