From f190f46d482268c47337fdc7e4174f0197a9c41d Mon Sep 17 00:00:00 2001
From: girst <tobi@isticktoit.net>
Date: Tue, 4 Aug 2020 18:06:27 +0200
Subject: [PATCH] split metadata in essential and extended

extended is only used by /watch, but essential also for
store_video_metadata(). this allows is to move parse_metadata() to
the youtube blueprint.
---
 app/common/common.py | 35 ++++++++++++++++++++++-------------
 1 file changed, 22 insertions(+), 13 deletions(-)

diff --git a/app/common/common.py b/app/common/common.py
index fbb6f2c..4508124 100644
--- a/app/common/common.py
+++ b/app/common/common.py
@@ -140,7 +140,7 @@ def update_channel(db, xmldata, from_webhook=False):
                 if from_webhook: current_app.logger.warning(f"  is webhook and new") # XXX: remove
                 _, _, meta, _, _ = get_video_info(video['video_id'])
                 if meta:
-                    meta = prepare_metadata(meta)
+                    meta = video_metadata(meta)
                     published = dateutil.parser.parse(meta['published'])
                     if from_webhook: current_app.logger.warning(f"  uploaded {published}") # XXX: remove
                     if (now - published).days < 7:
@@ -261,6 +261,25 @@ def unscramble(cipher, algo):  # test video id: UxxajLWwzqY
     sig = cipher.get('sig', [''.join(signature)])[0]
     return f"{cipher['url'][0]}&{sp}={sig}"
 
+def video_metadata(metadata):
+    if not metadata:
+        return {}
+
+    meta1 = metadata['videoDetails']
+    meta2 = metadata['microformat']['playerMicroformatRenderer']
+
+    published_at = meta2.get('liveBroadcastDetails',{}) \
+        .get('startTimestamp', f"{meta2['publishDate']}T00:00:00Z")
+
+    return {
+        'title': meta1['title'],
+        'author': meta1['author'],
+        'channel_id': meta1['channelId'],
+        'published': published_at,
+        'views': int(meta1['viewCount']),
+        'length': int(meta1['lengthSeconds']),
+    }
+
 def prepare_metadata(metadata):
     meta1 = metadata['videoDetails']
     meta2 = metadata['microformat']['playerMicroformatRenderer']
@@ -338,19 +357,9 @@ def prepare_metadata(metadata):
     whitelisted = sorted(meta2.get('availableCountries',[]))
     blacklisted = sorted(set(all_countries) - set(whitelisted))
 
-    published_at = f"{meta2['publishDate']}T00:00:00Z" # yyyy-mm-dd
-    # 'premiere' videos (and livestreams?) have a ISO8601 date available:
-    if 'liveBroadcastDetails' in meta2 and 'startTimestamp' in meta2['liveBroadcastDetails']: # TODO: tighten up
-        published_at = meta2['liveBroadcastDetails']['startTimestamp']
-
     return {
-        'title': meta1['title'],
-        'author': meta1['author'],
-        'channel_id': meta1['channelId'],
+        **video_metadata(metadata),
         'description': meta1['shortDescription'],
-        'published': published_at,
-        'views': meta1['viewCount'],
-        'length': int(meta1['lengthSeconds']),
         'rating': meta1['averageRating'],
         'category': meta2['category'],
         'aspectr': aspect_ratio,
@@ -373,7 +382,7 @@ def store_video_metadata(video_id):
         if new_video:
             _, _, meta, _, _ = get_video_info(video_id)
             if meta:
-                meta = prepare_metadata(meta)
+                meta = video_metadata(meta)
                 c.execute("""
                     INSERT OR IGNORE INTO videos (id, channel_id, title, published, crawled)
                                    VALUES (?, ?, ?, datetime(?), datetime(?))
-- 
2.39.3