2014-11-26 14:01:20 -05:00
|
|
|
from __future__ import unicode_literals
|
|
|
|
|
2021-01-24 08:40:02 -05:00
|
|
|
from ..utils import load_plugins
|
|
|
|
|
2016-02-10 08:01:31 -05:00
|
|
|
try:
|
|
|
|
from .lazy_extractors import *
|
|
|
|
from .lazy_extractors import _ALL_CLASSES
|
2016-02-21 06:28:58 -05:00
|
|
|
_LAZY_LOADER = True
|
2021-01-24 08:40:02 -05:00
|
|
|
_PLUGIN_CLASSES = []
|
2016-02-10 08:01:31 -05:00
|
|
|
except ImportError:
|
2016-02-21 06:28:58 -05:00
|
|
|
_LAZY_LOADER = False
|
2021-01-28 00:52:13 -05:00
|
|
|
|
|
|
|
if not _LAZY_LOADER:
|
2016-02-10 08:01:31 -05:00
|
|
|
from .extractors import *
|
|
|
|
_ALL_CLASSES = [
|
|
|
|
klass
|
|
|
|
for name, klass in globals().items()
|
|
|
|
if name.endswith('IE') and name != 'GenericIE'
|
|
|
|
]
|
|
|
|
_ALL_CLASSES.append(GenericIE)
|
2013-06-23 16:36:24 -04:00
|
|
|
|
2021-05-08 11:15:14 -04:00
|
|
|
_PLUGIN_CLASSES = load_plugins('extractor', 'IE', globals())
|
|
|
|
_ALL_CLASSES = _PLUGIN_CLASSES + _ALL_CLASSES
|
|
|
|
|
2013-08-24 15:10:03 -04:00
|
|
|
|
2016-02-10 07:16:18 -05:00
|
|
|
def gen_extractor_classes():
|
|
|
|
""" Return a list of supported extractors.
|
|
|
|
The order does matter; the first extractor matched is the one handling the URL.
|
|
|
|
"""
|
|
|
|
return _ALL_CLASSES
|
|
|
|
|
|
|
|
|
2013-06-23 16:36:24 -04:00
|
|
|
def gen_extractors():
|
|
|
|
""" Return a list of an instance of every supported extractor.
|
|
|
|
The order does matter; the first extractor matched is the one handling the URL.
|
|
|
|
"""
|
2016-02-10 07:16:18 -05:00
|
|
|
return [klass() for klass in gen_extractor_classes()]
|
2013-06-23 16:36:24 -04:00
|
|
|
|
2013-08-24 15:10:03 -04:00
|
|
|
|
2015-01-07 01:20:20 -05:00
|
|
|
def list_extractors(age_limit):
|
|
|
|
"""
|
|
|
|
Return a list of extractors that are suitable for the given age,
|
|
|
|
sorted by extractor ID.
|
|
|
|
"""
|
|
|
|
|
|
|
|
return sorted(
|
|
|
|
filter(lambda ie: ie.is_suitable(age_limit), gen_extractors()),
|
|
|
|
key=lambda ie: ie.IE_NAME.lower())
|
|
|
|
|
|
|
|
|
2013-06-23 16:36:24 -04:00
|
|
|
def get_info_extractor(ie_name):
|
|
|
|
"""Returns the info extractor class with the given ie_name"""
|
2014-11-23 15:20:46 -05:00
|
|
|
return globals()[ie_name + 'IE']
|