media format detection

app/sync/migrations/0016_auto_20201208_0518.py

@@ -0,0 +1,18 @@
+# Generated by Django 3.1.4 on 2020-12-08 05:18
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('sync', '0015_auto_20201207_0744'),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name='source',
+            name='fallback',
+            field=models.CharField(choices=[('f', 'Fail, do not download any media'), ('n', 'Get next best resolution or codec instead'), ('h', 'Get next best resolution but at least HD')], db_index=True, default='h', help_text='What do do when media in your source resolution and codecs is not available', max_length=1, verbose_name='fallback'),
+        ),
+    ]

app/sync/models.py

@@ -72,13 +72,13 @@ class Source(models.Model):
     )
 
     FALLBACK_FAIL = 'f'
-    FALLBACK_NEXT_SD = 's'
+    FALLBACK_NEXT_BEST = 'n'
-    FALLBACK_NEXT_HD = 'h'
+    FALLBACK_NEXT_BEST_HD = 'h'
-    FALLBACKS = (FALLBACK_FAIL, FALLBACK_NEXT_SD, FALLBACK_NEXT_HD)
+    FALLBACKS = (FALLBACK_FAIL, FALLBACK_NEXT_BEST, FALLBACK_NEXT_BEST_HD)
     FALLBACK_CHOICES = (
         (FALLBACK_FAIL, _('Fail, do not download any media')),
-        (FALLBACK_NEXT_SD, _('Get next best SD media or codec instead')),
+        (FALLBACK_NEXT_BEST, _('Get next best resolution or codec instead')),
-        (FALLBACK_NEXT_HD, _('Get next best HD media or codec instead')),
+        (FALLBACK_NEXT_BEST_HD, _('Get next best resolution but at least HD'))
     )
 
     # Fontawesome icons used for the source on the front end
@@ -218,7 +218,7 @@ class Source(models.Model):
         max_length=1,
         db_index=True,
         choices=FALLBACK_CHOICES,
-        default=FALLBACK_NEXT_HD,
+        default=FALLBACK_NEXT_BEST_HD,
         help_text=_('What do do when media in your source resolution and codecs is not available')
     )
     has_failed = models.BooleanField(
@@ -300,6 +300,10 @@ class Source(models.Model):
     def source_resolution_height(self):
         return self.RESOLUTION_MAP.get(self.source_resolution, 0)
 
+    @property
+    def can_fallback(self):
+        return self.fallback != self.FALLBACK_FAIL
+
     def index_media(self):
         '''
             Index the media source returning a list of media metadata as dicts.
@@ -494,94 +498,237 @@ class Media(models.Model):
         fields = self.METADATA_FIELDS.get(field, {})
         return fields.get(self.source.source_type, '')
 
+    def iter_formats(self):
+        for fmt in self.formats:
+            yield parse_media_format(fmt)
+
     def get_best_combined_format(self):
         '''
             Attempts to see if there is a single, combined audio and video format that
             exactly matches the source requirements. This is used over separate audio
-            and video formats if possible.
+            and video formats if possible. Combined formats are the easiest to check
-
+            for as they must exactly match the source profile be be valid.
-            single format structure = {
-                'format_id': '22',
-                'url': '... long url ...',
-                'player_url': None,
-                'ext': 'mp4',
-                'width': 1280,
-                'height': 720,
-                'acodec': 'mp4a.40.2',
-                'abr': 192,
-                'vcodec': 'avc1.64001F',
-                'asr': 44100,
-                'filesize': None,
-                'format_note': '720p',
-                'fps': 30,
-                'tbr': 1571.695,
-                'format': '22 - 1280x720 (720p)',
-                'protocol': 'https',
-                'http_headers': {... dict of headers ...}
-            }
-
-            or for hdr = {
-                'format_id': '336',
-                'url': '... long url ...',
-                'player_url': None,
-                'asr': None,
-                'filesize': 312014985,
-                'format_note': '1440p60 HDR',
-                'fps': 60,
-                'height': 1440,
-                'tbr': 16900.587,
-                'width': 2560,
-                'ext': 'webm',
-                'vcodec': 'vp9.2',
-                'acodec': 'none',
-                'downloader_options': {'http_chunk_size': 10485760},
-                'format': '336 - 2560x1440 (1440p60 HDR)',
-                'protocol': 'https',
-                'http_headers': {... dict of headers ...}
-            }
-
         '''
-        candidates = []
+        for fmt in self.iter_formats():
-        for fmt in self.formats:
-            parsed_fmt = parse_media_format(fmt)
             # Check height matches
-            print(self.source.source_resolution_height, parsed_fmt['height'])
+            if self.source.source_resolution.strip().upper() != fmt['format']:
-            if self.source.source_resolution_height != parsed_fmt['height']:
-                print()
                 continue
-            print('height OK')
             # Check the video codec matches
-            print(self.source.source_vcodec, parsed_fmt['vcodec'])
+            if self.source.source_vcodec != fmt['vcodec']:
-            if self.source.source_vcodec != parsed_fmt['vcodec']:
-                print()
                 continue
-            print('vcodec OK')
             # Check the audio codec matches
-            print(self.source.source_acodec, parsed_fmt['acodec'])
+            if self.source.source_acodec != fmt['acodec']:
-            if self.source.source_acodec != parsed_fmt['acodec']:
-                print()
                 continue
-            print('acodec OK')
+            # if the source prefers 60fps, check for it
-            # All OK so far...
+            if self.source.prefer_60fps:
-            candidates.append(parsed_fmt)
+                if not fmt['is_60fps']:
-            print()
+                    continue
-        for c in candidates:
+            # If the source prefers HDR, check for it
-            print(c)
+            if self.source.prefer_hdr:
-        return 'combined'
+                if not fmt['is_hdr']:
+                    continue
+            # If we reach here, we have a combined match!
+            return True, fmt['id']
+        return False, False
 
     def get_best_audio_format(self):
         '''
             Finds the best match for the source required audio format. If the source
             has a 'fallback' of fail this can return no match.
         '''
-        return 'audio'
+        # Order all audio-only formats by bitrate
+        audio_formats = []
+        for fmt in self.iter_formats():
+            # If the format has a video stream, skip it
+            if fmt['vcodec']:
+                continue
+            audio_formats.append(fmt)
+        audio_formats = list(reversed(sorted(audio_formats, key=lambda k: k['abr'])))
+        if not audio_formats:
+            # Media has no audio formats at all
+            return False, False
+        # Find the highest bitrate audio format with a matching codec
+        for fmt in audio_formats:
+            if self.source.source_acodec == fmt['acodec']:
+                # Matched!
+                return True, fmt['id']
+        # No codecs matched
+        if self.source.can_fallback:
+            # Can fallback, find the next highest bitrate non-matching codec
+            return False, audio_formats[0]
+        else:
+            # Can't fallback
+            return False, False
+
 
     def get_best_video_format(self):
         '''
             Finds the best match for the source required video format. If the source
-            has a 'fallback' of fail this can return no match.
+            has a 'fallback' of fail this can return no match. Resolution is treated
+            as the most important factor to match.
         '''
-        return 'video'
+        min_height = getattr(settings, 'VIDEO_HEIGHT_CUTOFF', 360)
+        fallback_hd_cutoff = getattr(settings, 'VIDEO_HEIGHT_IS_HD', 500)
+        # Filter video-only formats by resolution that matches the source
+        video_formats = []
+        for fmt in self.iter_formats():
+            # If the format has an audio stream, skip it
+            if fmt['acodec']:
+                continue
+            if self.source.source_resolution.strip().upper() == fmt['format']:
+                video_formats.append(fmt)
+        # Check we matched some streams
+        if not video_formats:
+            # No streams match the requested resolution, see if we can fallback
+            if self.source.can_fallback:
+                # Find the next-best format matches by height
+                for fmt in self.iter_formats():
+                    # If the format has an audio stream, skip it
+                    if fmt['acodec']:
+                        continue
+                    if (fmt['height'] <= self.source.source_resolution_height and 
+                        fmt['height'] >= min_height):
+                        video_formats.append(fmt)
+            else:
+                # Can't fallback
+                return False, False
+        video_formats = list(reversed(sorted(video_formats, key=lambda k: k['height'])))
+        if not video_formats:
+            # Still no matches
+            return False, False
+        exact_match, best_match = None, None
+        # Of our filtered video formats, check for resolution + codec + hdr + fps match
+        if self.source.prefer_60fps and self.source.prefer_hdr:
+            for fmt in video_formats:
+                # Check for an exact match
+                if (self.source.source_resolution.strip().upper() == fmt['format'] and
+                    self.source.source_vcodec == fmt['vcodec'] and 
+                    fmt['is_hdr'] and
+                    fmt['is_60fps']):
+                    # Exact match
+                    exact_match, best_match = True, fmt
+                    break
+            if self.source.can_fallback:
+                if not best_match:
+                    for fmt in video_formats:
+                        # Check for a codec, hdr and fps match but drop the resolution
+                        if (self.source.source_vcodec == fmt['vcodec'] and 
+                            fmt['is_hdr'] and fmt['is_60fps']):
+                            # Close match
+                            exact_match, best_match = False, fmt
+                            break
+                if not best_match:
+                    for fmt in video_formats:
+                        # Check for hdr and fps match but drop the resolution and codec
+                        if fmt['is_hdr'] and fmt['is_60fps']:
+                            exact_match, best_match = False, fmt
+                            break
+                if not best_match:
+                    for fmt in video_formats:
+                        # Check for fps match but drop the resolution and codec and hdr
+                        if fmt['is_hdr'] and fmt['is_60fps']:
+                            exact_match, best_match = False, fmt
+                            break
+                if not best_match:
+                    # Match the highest resolution
+                    exact_match, best_match = False, video_formats[0]
+        # Check for resolution + codec + fps match
+        if self.source.prefer_60fps and not self.source.prefer_hdr:
+            for fmt in video_formats:
+                # Check for an exact match
+                if (self.source.source_resolution.strip().upper() == fmt['format'] and
+                    self.source.source_vcodec == fmt['vcodec'] and 
+                    fmt['is_60fps']):
+                    # Exact match
+                    exact_match, best_match = True, fmt
+                    break
+            if self.source.can_fallback:
+                if not best_match:
+                    for fmt in video_formats:
+                        # Check for a codec and fps match but drop the resolution
+                        if (self.source.source_vcodec == fmt['vcodec'] and 
+                            fmt['is_60fps']):
+                            exact_match, best_match = False, fmt
+                            break
+                if not best_match:
+                    for fmt in video_formats:
+                        # Check for an fps match but drop the resolution and codec
+                        if fmt['is_60fps']:
+                            exact_match, best_match = False, fmt
+                            break
+                if not best_match:
+                    # Match the highest resolution
+                    exact_match, best_match = False, video_formats[0]
+        # Check for resolution + codec + hdr
+        if self.source.prefer_hdr and not self.source.prefer_60fps:
+            for fmt in video_formats:
+                # Check for an exact match
+                if (self.source.source_resolution.strip().upper() == fmt['format'] and
+                    self.source.source_vcodec == fmt['vcodec'] and 
+                    fmt['is_hdr']):
+                    # Exact match
+                    exact_match, best_match = True, fmt
+                    break
+            if self.source.can_fallback:
+                if not best_match:
+                    for fmt in video_formats:
+                        # Check for a codec and hdr match but drop the resolution
+                        if (self.source.source_vcodec == fmt['vcodec'] and 
+                            fmt['is_hdr']):
+                            exact_match, best_match = True, fmt
+                            break
+                if not best_match:
+                    for fmt in video_formats:
+                        # Check for an hdr match but drop the resolution and codec
+                        if fmt['is_hdr']:
+                            exact_match, best_match = False, fmt
+                            break
+                if not best_match:
+                    # Match the highest resolution
+                    exact_match, best_match = False, video_formats[0]
+        # check for resolution + codec
+        if not self.source.prefer_hdr and not self.source.prefer_60fps:
+            for fmt in video_formats:
+                # Check for an exact match
+                if (self.source.source_resolution.strip().upper() == fmt['format'] and
+                    self.source.source_vcodec == fmt['vcodec'] and
+                    not fmt['is_60fps']):
+                    # Exact match
+                    exact_match, best_match = True, fmt
+                    break
+            if self.source.can_fallback:
+                if not best_match:
+                    for fmt in video_formats:
+                        # Check for a codec match without 60fps and drop the resolution
+                        if (self.source.source_vcodec == fmt['vcodec'] and
+                            not fmt['is_60fps']):
+                            exact_match, best_match = False, fmt
+                            break
+                if not best_match:
+                    for fmt in video_formats:
+                        # Check for a codec match but drop the resolution
+                        if self.source.source_vcodec == fmt['vcodec']:
+                            # Close match
+                            exact_match, best_match = False, fmt
+                            break
+                if not best_match:
+                    # Match the highest resolution
+                    exact_match, best_match = False, video_formats[0]
+        # See if we found a match
+        if best_match:
+            # Final check to see if the match we found was good enough
+            if exact_match:
+                return True, best_match['id']
+            elif self.source.can_fallback:
+                # Allow the fallback if it meets requirements
+                if (self.source.fallback == self.source.FALLBACK_NEXT_BEST_HD and
+                    best_match['height'] >= fallback_hd_cutoff):
+                    return False, best_match['id']
+                elif self.source.fallback == self.source.FALLBACK_NEXT_BEST:
+                    return False, best_match['id']
+        # Nope, failed to find match
+        return False, False
+
 
     def get_format_str(self):
         '''

app/sync/templates/sync/media-item.html

@@ -33,6 +33,10 @@
         <td class="hide-on-small-only">Desired format</td>
         <td><span class="hide-on-med-and-up">Desired format<br></span><strong>{{ media.source.format_summary }}</strong></td>
       </tr>
+      <tr title="Fallback setting on the source">
+        <td class="hide-on-small-only">Fallback</td>
+        <td><span class="hide-on-med-and-up">Fallback<br></span><strong>{{ media.source.get_fallback_display }}</strong></td>
+      </tr>
       <tr title="Has the media been downloaded">
         <td class="hide-on-small-only">Downloaded</td>
         <td><span class="hide-on-med-and-up">Downloaded<br></span><strong>{% if media.downloaded %}<i class="fas fa-check"></i>{% else %}<i class="fas fa-times"></i>{% endif %}</strong></td>
@@ -48,11 +52,11 @@
         </td>
       </tr>
       <tr title="Best available format for source requirements">
-        <td class="hide-on-small-only">Best match</td>
+        <td class="hide-on-small-only">Matched formats</td>
-        <td><span class="hide-on-med-and-up">Best match<br></span><strong>
+        <td><span class="hide-on-med-and-up">Matched formats<br></span>
-          audio: {{ media.get_best_audio_format }}<br>
+          Combined: <strong>{% if combined_format %}{{ combined_format }} {% if combined_exact %}(exact match){% else %}(fallback){% endif %}{% else %}No match{% endif %}</strong><br>
-          video: {{ media.get_best_video_format }}<br>
+          Audio: <strong>{% if audio_format %}{{ audio_format }} {% if audio_exact %}(exact match){% else %}(fallback){% endif %}{% else %}No match{% endif %}</strong><br>
-          combo: {{ media.get_best_combined_format }}
+          Video: <strong>{% if video_format %}{{ video_format }} {% if video_exact %}(exact match){% else %}(fallback){% endif %}{% else %}No match{% endif %}
         </strong></td>
       </tr>
     </table>

app/sync/templates/sync/source.html

@@ -18,6 +18,7 @@
     <a href="{% url 'sync:tasks-completed' %}?filter={{ source.pk }}" class="btn">View tasks<span class="hide-on-small-only"> linked to this source</span> <i class="far fa-fw fa-clock"></i></a>
   </div>
 </div>
+{% include 'infobox.html' with message=message %}
 {% if source.has_failed %}{% include 'errorbox.html' with message='This source has encountered permanent failures listed at the bottom of this page, check its settings' %}{% endif %}
 <div class="row">
   <div class="col s12">

app/sync/utils.py

@@ -142,11 +142,21 @@ def parse_media_format(format_dict):
         acodec = None
     if acodec == 'NONE':
         acodec = None
+    try:
+        fps = int(format_dict.get('fps', 0))
+    except (ValueError, TypeError):
+        fps = 0
+    format_full = format_dict.get('format_note', '').strip().upper()
+    format_str = format_full[:-2] if format_full.endswith('60') else format_full
     return {
         'id': format_dict.get('format_id', ''),
+        'format': format_str,
+        'format_verbose': format_dict.get('format', ''),
         'height': format_dict.get('height', 0),
-        'is_60fps': format_dict.get('fps', 0) == 60,
-        'is_hdr': 'HDR' in format_dict.get('format', '').upper(),
         'vcodec': vcodec,
+        'vbr': format_dict.get('tbr', 0),
         'acodec': acodec,
+        'abr': format_dict.get('abr', 0),
+        'is_60fps': fps > 50,
+        'is_hdr': 'HDR' in format_dict.get('format', '').upper(),
     }

app/sync/views.py

@@ -41,9 +41,7 @@ class SourcesView(ListView):
     context_object_name = 'sources'
     paginate_by = settings.SOURCES_PER_PAGE
     messages = {
-        'source-created': _('Your new source has been added'),
         'source-deleted': _('Your selected source has been deleted.'),
-        'source-updated': _('Your selected source has been updated.'),
     }
 
     def __init__(self, *args, **kwargs):
@@ -235,7 +233,7 @@ class AddSourceView(CreateView):
         return initial
 
     def get_success_url(self):
-        url = reverse_lazy('sync:sources')
+        url = reverse_lazy('sync:source', kwargs={'pk': self.object.pk})
         return append_uri_params(url, {'message': 'source-created'})
 
 
@@ -243,9 +241,23 @@ class SourceView(DetailView):
 
     template_name = 'sync/source.html'
     model = Source
+    messages = {
+        'source-created': _('Your new source has been created'),
+        'source-updated': _('Your source has been updated.'),
+    }
+
+    def __init__(self, *args, **kwargs):
+        self.message = None
+        super().__init__(*args, **kwargs)
+
+    def dispatch(self, request, *args, **kwargs):
+        message_key = request.GET.get('message', '')
+        self.message = self.messages.get(message_key, '')
+        return super().dispatch(request, *args, **kwargs)
 
     def get_context_data(self, *args, **kwargs):
         data = super().get_context_data(*args, **kwargs)
+        data['message'] = self.message
         data['errors'] = []
         for error in get_source_completed_tasks(self.object.pk, only_errors=True):
             error_message = get_error_message(error)
@@ -264,7 +276,7 @@ class UpdateSourceView(UpdateView):
               'source_acodec', 'prefer_60fps', 'prefer_hdr', 'fallback')
 
     def get_success_url(self):
-        url = reverse_lazy('sync:sources')
+        url = reverse_lazy('sync:source', kwargs={'pk': self.object.pk})
         return append_uri_params(url, {'message': 'source-updated'})
 
 
@@ -366,6 +378,19 @@ class MediaItemView(DetailView):
     template_name = 'sync/media-item.html'
     model = Media
 
+    def get_context_data(self, *args, **kwargs):
+        data = super().get_context_data(*args, **kwargs)
+        combined_exact, combined_format = self.object.get_best_combined_format()
+        audio_exact, audio_format = self.object.get_best_audio_format()
+        video_exact, video_format = self.object.get_best_video_format()
+        data['combined_exact'] = combined_exact
+        data['combined_format'] = combined_format
+        data['audio_exact'] = audio_exact
+        data['audio_format'] = audio_format
+        data['video_exact'] = video_exact
+        data['video_format'] = video_format
+        return data
+
 
 class TasksView(ListView):
     '''

app/tubesync/settings.py

@@ -130,6 +130,10 @@ MEDIA_THUMBNAIL_WIDTH = 430                 # Width in pixels to resize thumbnai
 MEDIA_THUMBNAIL_HEIGHT = 240                # Height in pixels to resize thumbnails to
 
 
+VIDEO_HEIGHT_CUTOFF = 360       # Smallest resolution in pixels permitted to download
+VIDEO_HEIGHT_IS_HD = 500        # Height in pixels to count as 'HD'
+
+
 YOUTUBE_DEFAULTS = {
     'no_color': True,       # Do not use colours in output
     'age_limit': 99,        # 'Age in years' to spoof