[youtube] Fix playlist and feed extraction (closes #25675)
[youtube-dl] / devscripts / make_lazy_extractors.py
1 from __future__ import unicode_literals, print_function
2
3 from inspect import getsource
4 import io
5 import os
6 from os.path import dirname as dirn
7 import sys
8
9 print('WARNING: Lazy loading extractors is an experimental feature that may not always work', file=sys.stderr)
10
11 sys.path.insert(0, dirn(dirn((os.path.abspath(__file__)))))
12
13 lazy_extractors_filename = sys.argv[1]
14 if os.path.exists(lazy_extractors_filename):
15     os.remove(lazy_extractors_filename)
16
17 from youtube_dl.extractor import _ALL_CLASSES
18 from youtube_dl.extractor.common import InfoExtractor, SearchInfoExtractor
19
20 with open('devscripts/lazy_load_template.py', 'rt') as f:
21     module_template = f.read()
22
23 module_contents = [
24     module_template + '\n' + getsource(InfoExtractor.suitable) + '\n',
25     'class LazyLoadSearchExtractor(LazyLoadExtractor):\n    pass\n']
26
27 ie_template = '''
28 class {name}({bases}):
29     _VALID_URL = {valid_url!r}
30     _module = '{module}'
31 '''
32
33 make_valid_template = '''
34     @classmethod
35     def _make_valid_url(cls):
36         return {valid_url!r}
37 '''
38
39
40 def get_base_name(base):
41     if base is InfoExtractor:
42         return 'LazyLoadExtractor'
43     elif base is SearchInfoExtractor:
44         return 'LazyLoadSearchExtractor'
45     else:
46         return base.__name__
47
48
49 def build_lazy_ie(ie, name):
50     valid_url = getattr(ie, '_VALID_URL', None)
51     s = ie_template.format(
52         name=name,
53         bases=', '.join(map(get_base_name, ie.__bases__)),
54         valid_url=valid_url,
55         module=ie.__module__)
56     if ie.suitable.__func__ is not InfoExtractor.suitable.__func__:
57         s += '\n' + getsource(ie.suitable)
58     if hasattr(ie, '_make_valid_url'):
59         # search extractors
60         s += make_valid_template.format(valid_url=ie._make_valid_url())
61     return s
62
63
64 # find the correct sorting and add the required base classes so that sublcasses
65 # can be correctly created
66 classes = _ALL_CLASSES[:-1]
67 ordered_cls = []
68 while classes:
69     for c in classes[:]:
70         bases = set(c.__bases__) - set((object, InfoExtractor, SearchInfoExtractor))
71         stop = False
72         for b in bases:
73             if b not in classes and b not in ordered_cls:
74                 if b.__name__ == 'GenericIE':
75                     exit()
76                 classes.insert(0, b)
77                 stop = True
78         if stop:
79             break
80         if all(b in ordered_cls for b in bases):
81             ordered_cls.append(c)
82             classes.remove(c)
83             break
84 ordered_cls.append(_ALL_CLASSES[-1])
85
86 names = []
87 for ie in ordered_cls:
88     name = ie.__name__
89     src = build_lazy_ie(ie, name)
90     module_contents.append(src)
91     if ie in _ALL_CLASSES:
92         names.append(name)
93
94 module_contents.append(
95     '_ALL_CLASSES = [{0}]'.format(', '.join(names)))
96
97 module_src = '\n'.join(module_contents) + '\n'
98
99 with io.open(lazy_extractors_filename, 'wt', encoding='utf-8') as f:
100     f.write(module_src)