heckyel/yt-local
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge short and video parsing even further

Browse Source 
Use multi_get and multi_deep_get for tag differences
Replace the duration check with conservative_update
This commit is contained in:
Jesus E 2023-06-17 16:14:02 -04:00
parent a4299dc917
commit d7f934b7b2
No known key found for this signature in database
GPG Key ID: 159C8F8BC9AED8B6
1 changed files with 24 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
youtube/yt_data_extract/common.py
View File

@ -298,10 +298,11 @@ def extract_item_info(item, additional_info={}):
info['time_published'] = timestamp.group(1) info['time_published'] = timestamp.group(1)
if primary_type == 'video': if primary_type == 'video':
info['id'] = item.get('videoId') info['id'] = multi_deep_get(item,
if not info['id']: ['videoId'],
info['id'] = deep_get(item,'navigationEndpoint', 'watchEndpoint', ['navigationEndpoint', 'watchEndpoint', 'videoId'],
'videoId') ['navigationEndpoint', 'reelWatchEndpoint', 'videoId'], # shorts
)
info['view_count'] = extract_int(item.get('viewCountText')) info['view_count'] = extract_int(item.get('viewCountText'))
# dig into accessibility data to get view_count for videos marked as recommended, and to get time_published # dig into accessibility data to get view_count for videos marked as recommended, and to get time_published
@ -319,40 +320,34 @@ def extract_item_info(item, additional_info={}):
if info['view_count']: if info['view_count']:
info['approx_view_count'] = '{:,}'.format(info['view_count']) info['approx_view_count'] = '{:,}'.format(info['view_count'])
else: else:
info['approx_view_count'] = extract_approx_int(item.get('shortViewCountText')) info['approx_view_count'] = extract_approx_int(multi_get(item,
'shortViewCountText',
'viewCountText') # shorts
)
# handle case where it is "No views" # handle case where it is "No views"
if not info['approx_view_count']: if not info['approx_view_count']:
if ('No views' in item.get('shortViewCountText', '') if ('No views' in item.get('shortViewCountText', '')
or 'no views' in accessibility_label.lower()): or 'no views' in accessibility_label.lower()
or 'No views' in extract_str(item.get('viewCountText', '')) # shorts
):
info['view_count'] = 0 info['view_count'] = 0
info['approx_view_count'] = '0' info['approx_view_count'] = '0'
info['duration'] = extract_str(item.get('lengthText')) info['duration'] = extract_str(item.get('lengthText'))
if info['duration'] is None: # shorts
if not info['id']:
info['id'] = deep_get(item,'navigationEndpoint',
'reelWatchEndpoint', 'videoId')
info['approx_view_count'] = extract_approx_int(item.get('viewCountText'))
# handle case where it is "No views"
if not info['approx_view_count']:
if ('No views' in extract_str(item.get('viewCountText', ''))):
info['view_count'] = 0
info['approx_view_count'] = '0'
# dig into accessibility data to get duration for shorts # dig into accessibility data to get duration for shorts
accessibility_label = multi_deep_get(item, accessibility_label = deep_get(item,
['accessibility', 'accessibilityData', 'label'], 'accessibility', 'accessibilityData', 'label',
default='') default='')
duration = re.search(r'(\d+) (second|seconds|minute) - play video$', duration = re.search(r'(\d+) (second|seconds|minute) - play video$',
accessibility_label) accessibility_label)
if duration: if duration:
if duration.group(2) == 'minute': if duration.group(2) == 'minute':
info['duration'] = '1:00' conservative_update(info, 'duration', '1:00')
else: else:
info['duration'] = '0:' + duration.group(1).zfill(2) conservative_update(info,
'duration', '0:' + duration.group(1).zfill(2))
# if it's an item in a playlist, get its index # if it's an item in a playlist, get its index
if 'index' in item: # url has wrong index on playlist page if 'index' in item: # url has wrong index on playlist page