def pull_cmd(cfg, parser, parsed, args):
if not parsed.quiet:
parser.print_byline()
username = get_from_config(cfg, 'username')
api_url = get_from_config(cfg, 'api_url')
cat_title = get_from_config(cfg, 'category')
# Command line api_key overrides config-set api_key
api_key = parsed.apikey
if not api_key:
try:
api_key = cfg.get('project', 'api_key')
except ConfigParser.NoOptionError:
pass
if not api_key:
err('Specify an api key either in steve.ini, on command line, '
'or in API_KEY file.')
return 1
if not username or not api_url or not cat_title or not api_key:
return 1
api = steve.restapi.API(api_url)
all_categories = steve.restapi.get_content(
api.category.get(username=username, api_key=api_key,
limit=0))
cat = [cat_item for cat_item in all_categories['objects']
if cat_item['title'] == cat_title]
if not cat:
err('Category "{0}" does not exist.'.format(cat_title))
return 1
# Get the category from the list of 1.
cat = cat[0]
out('Retrieved category.')
data = []
for counter, video_url in enumerate(cat['videos']):
# Lame, but good enough for now.
video_id = video_url.split('/')[-2]
video_data = steve.restapi.get_content(
api.video(video_id).get(username=username,
api_key=api_key))
out('Working on "{0}"'.format(video_data['slug']))
# Nix some tastypie bits from the data.
for bad_key in ('resource_uri',):
if bad_key in video_data:
del video_data[bad_key]
# Add id.
video_data['id'] = video_id
fn = 'json/{0:4d}_{1}.json'.format(counter, video_data['slug'])
data.append((fn, video_data))
out('Saving files....')
save_json_files(cfg, data)
return 0
}
return item
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument("--maxresults", help="Max results", default=50)
parser.add_argument("-c", "--channel", action='store_true', help="YouTube channel id")
parser.add_argument("-p", "--playlist", action='store_true', help="YouTube playlist id")
args = parser.parse_args()
cfg = get_project_config()
scraper = YouTubeScraper(cfg, max_results=args.maxresults)
if args.channel:
channel_id = get_from_config(cfg, 'channel_id', 'youtube')
print("scraping channel {}".format(channel_id))
data = scraper.scrape_channel(channel_id)
save_json_files(cfg, data)
elif args.playlist:
playlist_id = get_from_config(cfg, 'playlist_id', 'youtube')
print("scraping playlist {}".format(playlist_id))
data = scraper.scrape_playlist(playlist_id)
save_json_files(cfg, data)
else:
print("nothing to do. no channel or playlist requested")
def pull_cmd(cfg, parser, parsed, args):
if not parsed.quiet:
parser.print_byline()
username = get_from_config(cfg, 'username')
api_url = get_from_config(cfg, 'api_url')
cat_title = get_from_config(cfg, 'category')
# Command line api_key overrides config-set api_key
api_key = parsed.apikey
if not api_key:
try:
api_key = cfg.get('project', 'api_key')
except ConfigParser.NoOptionError:
pass
if not api_key:
err('Specify an api key either in steve.ini, on command line, '
'or in API_KEY file.')
return 1
if not username or not api_url or not cat_title or not api_key:
return 1
api = steve.restapi.API(api_url)
all_categories = steve.restapi.get_content(
api.category.get(username=username, api_key=api_key, limit=0))
cat = [
cat_item for cat_item in all_categories['objects']
if cat_item['title'] == cat_title
]
if not cat:
err('Category "{0}" does not exist.'.format(cat_title))
return 1
# Get the category from the list of 1.
cat = cat[0]
out('Retrieved category.')
data = []
for counter, video_url in enumerate(cat['videos']):
# Lame, but good enough for now.
video_id = video_url.split('/')[-2]
video_data = steve.restapi.get_content(
api.video(video_id).get(username=username, api_key=api_key))
out('Working on "{0}"'.format(video_data['slug']))
# Nix some tastypie bits from the data.
for bad_key in ('resource_uri', ):
if bad_key in video_data:
del video_data[bad_key]
# Add id.
video_data['id'] = video_id
fn = 'json/{0:4d}_{1}.json'.format(counter, video_data['slug'])
data.append((fn, video_data))
out('Saving files....')
save_json_files(cfg, data)
return 0
def pull(cfg, ctx, quiet, apikey):
"""Pulls data from a richard instance."""
if not quiet:
click.echo(VERSION)
username = get_from_config(cfg, 'username')
api_url = get_from_config(cfg, 'api_url')
cat_title = get_from_config(cfg, 'category')
# Command line api_key overrides config-set api_key
if not apikey:
try:
apikey = cfg.get('project', 'api_key')
except NoOptionError:
pass
if not apikey:
raise click.ClickException(
u'Specify an api key either in {0}, on command line, '
u'or in API_KEY file.'.format(get_project_config_file_name())
)
if not username or not api_url or not cat_title or not apikey:
raise click.ClickException(u'Missing username, api_url or api_key.')
api = steve.restapi.API(api_url)
all_categories = steve.restapi.get_content(
api.category.get(username=username, api_key=apikey,
limit=0))
cat = [cat_item for cat_item in all_categories['objects']
if cat_item['title'] == cat_title]
if not cat:
raise click.ClickException(u'Category "{0}" does not exist.'.format(cat_title))
# Get the category from the list of 1.
cat = cat[0]
click.echo('Retrieved category.')
data = []
for counter, video_url in enumerate(cat['videos']):
# Lame, but good enough for now.
video_id = video_url.split('/')[-2]
video_data = steve.restapi.get_content(
api.video(video_id).get(username=username,
api_key=apikey))
click.echo('Working on "{0}"'.format(video_data['slug']))
# Nix some tastypie bits from the data.
for bad_key in ('resource_uri',):
if bad_key in video_data:
del video_data[bad_key]
# Add id.
video_data['id'] = video_id
fn = 'json/{0:4d}_{1}.json'.format(counter, video_data['slug'])
data.append((fn, video_data))
click.echo('Saving files....')
save_json_files(cfg, data)
videos = category['videos']
# In[25]:
stevedata = []
for v in videos:
r = requests.get(v)
j = r.json()
stevedata.append(('%s.json' % j['id'], j))
# In[27]:
save_json_files(cfg, stevedata)
# In[30]:
soup = BeautifulSoup(open('talks.html'))
# In[38]:
proposals_soup = soup.find_all('div', class_='proposal_list_summary')
# In[41]:
psoup = proposals_soup[0]
def pull(cfg, ctx, quiet, apikey):
"""Pulls data from a richard instance."""
if not quiet:
click.echo(VERSION)
username = get_from_config(cfg, 'username')
api_url = get_from_config(cfg, 'api_url')
cat_title = get_from_config(cfg, 'category')
# Command line api_key overrides config-set api_key
if not apikey:
try:
apikey = cfg.get('project', 'api_key')
except ConfigParser.NoOptionError:
pass
if not apikey:
raise click.ClickException(
u'Specify an api key either in {0}, on command line, '
u'or in API_KEY file.'.format(get_project_config_file_name())
)
if not username or not api_url or not cat_title or not apikey:
raise click.ClickException(u'Missing username, api_url or api_key.')
api = steve.restapi.API(api_url)
all_categories = steve.restapi.get_content(
api.category.get(username=username, api_key=apikey,
limit=0))
cat = [cat_item for cat_item in all_categories['objects']
if cat_item['title'] == cat_title]
if not cat:
raise click.ClickException(u'Category "{0}" does not exist.'.format(cat_title))
# Get the category from the list of 1.
cat = cat[0]
click.echo('Retrieved category.')
data = []
for counter, video_url in enumerate(cat['videos']):
video_id = get_video_id(video_url)
video_data = steve.restapi.get_content(
api.video(video_id).get(username=username,
api_key=apikey))
click.echo('Working on "{0}"'.format(video_data['slug']))
# Nix some tastypie bits from the data.
for bad_key in ('resource_uri',):
if bad_key in video_data:
del video_data[bad_key]
# Add id.
video_data['id'] = video_id
fn = 'json/{0:4d}_{1}.json'.format(counter, video_data['slug'])
data.append((fn, video_data))
click.echo('Saving files....')
save_json_files(cfg, data)
