Extraction: Proper error handling for terminated or non-existant channels

This commit is contained in:
James Taylor 2019-09-08 17:48:02 -07:00
parent bd343ed71f
commit 216231f9a6
3 changed files with 18 additions and 7 deletions

View File

@ -186,6 +186,8 @@ def get_channel_page(channel_id, tab='videos'):
info = yt_data_extract.extract_channel_info(json.loads(polymer_json), tab)
if info['errors']:
return flask.render_template('error.html', error_message = '\n'.join(info['errors']))
post_process_channel_info(info)
if tab in ('videos', 'search'):
info['number_of_videos'] = number_of_videos
@ -226,6 +228,9 @@ def get_channel_page_general_url(base_url, tab, request):
info = yt_data_extract.extract_channel_info(json.loads(polymer_json), tab)
if info['errors']:
return flask.render_template('error.html', error_message = '\n'.join(info['errors']))
post_process_channel_info(info)
if tab in ('videos', 'search'):
info['number_of_videos'] = 1000

View File

@ -455,7 +455,12 @@ def _get_upstream_videos(channel_id):
print('Failed to read atoma feed for ' + channel_status_name)
traceback.print_exc()
videos = yt_data_extract.extract_channel_info(json.loads(channel_tab), 'videos')['items']
channel_info = yt_data_extract.extract_channel_info(json.loads(channel_tab), 'videos')
if channel_info['errors']:
print('Error checking channel ' + channel_status_name + ': ' + ', '.join(channel_info['errors']))
return
videos = channel_info['items']
for i, video_item in enumerate(videos):
if 'description' not in video_item:
video_item['description'] = ''

View File

@ -281,6 +281,7 @@ def parse_info_prepare_for_html(renderer, additional_info={}):
def extract_channel_info(polymer_json, tab):
info = {'errors': []}
response = polymer_json[1]['response']
try:
microformat = response['microformat']['microformatDataRenderer']
@ -289,18 +290,18 @@ def extract_channel_info(polymer_json, tab):
# example terminated channel: https://www.youtube.com/channel/UCnKJeK_r90jDdIuzHXC0Org
except KeyError:
if 'alerts' in response and len(response['alerts']) > 0:
result = ''
for alert in response['alerts']:
result += alert['alertRenderer']['text']['simpleText'] + '\n'
flask.abort(200, result)
info['errors'].append(alert['alertRenderer']['text']['simpleText'])
return info
elif 'errors' in response['responseContext']:
for error in response['responseContext']['errors']['error']:
if error['code'] == 'INVALID_VALUE' and error['location'] == 'browse_id':
flask.abort(404, 'This channel does not exist')
raise
info['errors'].append('This channel does not exist')
return info
info['errors'].append('Failure getting microformat')
return info
info = {}
info['current_tab'] = tab