subscriptions: Update live/upcoming/premier durations upon check

The durations were previously set to "LIVE", "UPCOMING", etc. and
would not be updated once the livestream was over or the upcoming
video was published.

1 files changed, 53 insertions, 24 deletions

diff --git a/youtube/subscriptions.py b/youtube/subscriptions.py
index b5f3c83..04d3c5a 100644
--- a/youtube/subscriptions.py
+++ b/youtube/subscriptions.py
@@ -108,8 +108,7 @@ def _subscribe(channels):
         with connection as cursor:
             channel_ids_to_check = [channel[0] for channel in channels if not _is_subscribed(cursor, channel[0])]
 
-            rows = ((channel_id, channel_name, 0, 0) for channel_id,
-                    channel_name in channels)
+            rows = ((channel_id, channel_name, 0, 0) for channel_id, channel_name in channels)
             cursor.executemany('''INSERT OR IGNORE INTO subscribed_channels (yt_channel_id, channel_name, time_last_checked, next_check_time)
                                   VALUES (?, ?, ?, ?)''', rows)
 
@@ -236,8 +235,7 @@ def _get_channel_names(cursor, channel_ids):
     return result
 
 
-def _channels_with_tag(cursor, tag, order=False, exclude_muted=False,
-                       include_muted_status=False):
+def _channels_with_tag(cursor, tag, order=False, exclude_muted=False, include_muted_status=False):
     ''' returns list of (channel_id, channel_name) '''
 
     statement = '''SELECT yt_channel_id, channel_name'''
@@ -434,8 +432,10 @@ def autocheck_setting_changed(old_value, new_value):
         stop_autocheck_system()
 
 
-settings.add_setting_changed_hook('autocheck_subscriptions',
-                                  autocheck_setting_changed)
+settings.add_setting_changed_hook(
+    'autocheck_subscriptions',
+    autocheck_setting_changed
+)
 if settings.autocheck_subscriptions:
     start_autocheck_system()
 # ----------------------------
@@ -574,7 +574,6 @@ def _get_upstream_videos(channel_id):
                 video_item['time_published'] = youtube_timestamp_to_posix(video_item['time_published']) - i  # subtract a few seconds off the videos so they will be in the right order
             except Exception:
                 print(video_item)
-
         else:
             video_item['is_time_published_exact'] = False
             video_item['time_published'] = None
@@ -601,7 +600,6 @@ def _get_upstream_videos(channel_id):
             # 1 month between videos
             videos[i]['time_published'] = now - i*3600*24*30
 
-        video_item['channel_id'] = channel_id
 
     if len(videos) == 0:
         average_upload_period = 4*7*24*3600  # assume 1 month for channel with no videos
@@ -620,26 +618,31 @@ def _get_upstream_videos(channel_id):
     with open_database() as connection:
         with connection as cursor:
 
-            # calculate how many new videos there are
-            existing_vids = set(row[0] for row in cursor.execute(
-                '''SELECT video_id
+            # Get video ids and duration of existing vids so we
+            # can see how many new ones there are and update
+            # livestreams/premiers
+            existing_vids = list(cursor.execute(
+                '''SELECT video_id, duration
                    FROM videos
                    INNER JOIN subscribed_channels
                        ON videos.sql_channel_id = subscribed_channels.id
                    WHERE yt_channel_id=?
                    ORDER BY time_published DESC
                    LIMIT 30''', [channel_id]).fetchall())
+            existing_vid_ids = set(row[0] for row in existing_vids)
+            existing_durs = dict(existing_vids)
 
             # new videos the channel has uploaded since last time we checked
             number_of_new_videos = 0
             for video in videos:
-                if video['id'] in existing_vids:
+                if video['id'] in existing_vid_ids:
                     break
                 number_of_new_videos += 1
 
             is_first_check = cursor.execute('''SELECT time_last_checked FROM subscribed_channels WHERE yt_channel_id=?''', [channel_id]).fetchone()[0] in (None, 0)
             time_videos_retrieved = int(time.time())
             rows = []
+            update_rows = []
             for i, video_item in enumerate(videos):
                 if (is_first_check
                         or number_of_new_videos > 6
@@ -655,16 +658,34 @@ def _get_upstream_videos(channel_id):
                     time_noticed = video_item['time_published']
                 else:
                     time_noticed = time_videos_retrieved
-                rows.append((
-                    video_item['channel_id'],
-                    video_item['id'],
-                    video_item['title'],
-                    video_item['duration'],
-                    video_item['time_published'],
-                    video_item['is_time_published_exact'],
-                    time_noticed,
-                    video_item['description'],
-                ))
+
+                # videos which need durations updated
+                non_durations = ('upcoming', 'none', 'live', '')
+                v_id = video_item['id']
+                if (existing_durs.get(v_id) is not None
+                    and existing_durs[v_id].lower() in non_durations
+                    and video_item['duration'] not in non_durations
+                ):
+                    update_rows.append((
+                        video_item['title'],
+                        video_item['duration'],
+                        video_item['time_published'],
+                        video_item['is_time_published_exact'],
+                        video_item['description'],
+                        video_item['id'],
+                    ))
+                # all other videos
+                else:
+                    rows.append((
+                        video_item['channel_id'],
+                        video_item['id'],
+                        video_item['title'],
+                        video_item['duration'],
+                        video_item['time_published'],
+                        video_item['is_time_published_exact'],
+                        time_noticed,
+                        video_item['description'],
+                    ))
 
             cursor.executemany('''INSERT OR IGNORE INTO videos (
                                       sql_channel_id,
@@ -677,6 +698,13 @@ def _get_upstream_videos(channel_id):
                                       description
                                   )
                                   VALUES ((SELECT id FROM subscribed_channels WHERE yt_channel_id=?), ?, ?, ?, ?, ?, ?, ?)''', rows)
+            cursor.executemany('''UPDATE videos SET
+                                      title=?,
+                                      duration=?,
+                                      time_published=?,
+                                      is_time_published_exact=?,
+                                      description=?
+                                  WHERE video_id=?''', update_rows)
             cursor.execute('''UPDATE subscribed_channels
                               SET time_last_checked = ?, next_check_time = ?
                               WHERE yt_channel_id=?''', [int(time.time()), next_check_time, channel_id])
@@ -809,7 +837,7 @@ def import_subscriptions():
         error = 'Unsupported file format: ' + mime_type
         error += (' . Only subscription.json, subscriptions.csv files'
                   ' (from Google Takeouts)'
-                  ' and XML OPML files exported from Youtube\'s'
+                  ' and XML OPML files exported from YouTube\'s'
                   ' subscription manager page are supported')
         return (flask.render_template('error.html', error_message=error),
                 400)
@@ -1004,7 +1032,8 @@ def get_subscriptions_page():
                     'muted': muted,
                 })
 
-    return flask.render_template('subscriptions.html',
+    return flask.render_template(
+        'subscriptions.html',
         header_playlist_names=local_playlist.get_playlist_names(),
         videos=videos,
         num_pages=math.ceil(number_of_videos_in_db/60),