]>
Commit | Line | Data |
---|---|---|
add96eb9 | 1 | import base64 |
2 | import urllib.parse | |
3 | ||
e897bd82 | 4 | from .bokecc import BokeCCBaseIE |
e650659b | 5 | from ..utils import ( |
b3602f68 | 6 | ExtractorError, |
e650659b | 7 | determine_ext, |
b3602f68 | 8 | traverse_obj, |
e897bd82 | 9 | update_url_query, |
e650659b | 10 | ) |
fda7d31a PH |
11 | |
12 | ||
5633b4d3 | 13 | class InfoQIE(BokeCCBaseIE): |
533f67d3 | 14 | _VALID_URL = r'https?://(?:www\.)?infoq\.com/(?:[^/]+/)+(?P<id>[^/]+)' |
282cb9c7 | 15 | |
533f67d3 | 16 | _TESTS = [{ |
c0a7c608 | 17 | 'url': 'http://www.infoq.com/presentations/A-Few-of-My-Favorite-Python-Things', |
edec83a0 | 18 | 'md5': 'b5ca0e0a8c1fed93b0e65e48e462f9a2', |
c0a7c608 | 19 | 'info_dict': { |
02a63fad | 20 | 'id': 'A-Few-of-My-Favorite-Python-Things', |
c0a7c608 PH |
21 | 'ext': 'mp4', |
22 | 'description': 'Mike Pirnat presents some tips and tricks, standard libraries and third party packages that make programming in Python a richer experience.', | |
23 | 'title': 'A Few of My Favorite [Python] Things', | |
9d069c47 | 24 | }, |
533f67d3 S |
25 | }, { |
26 | 'url': 'http://www.infoq.com/fr/presentations/changez-avis-sur-javascript', | |
27 | 'only_matching': True, | |
02a63fad YCH |
28 | }, { |
29 | 'url': 'http://www.infoq.com/cn/presentations/openstack-continued-delivery', | |
30 | 'md5': '4918d0cca1497f2244572caf626687ef', | |
31 | 'info_dict': { | |
32 | 'id': 'openstack-continued-delivery', | |
33 | 'title': 'OpenStack持续交付之路', | |
34 | 'ext': 'flv', | |
35 | 'description': 'md5:308d981fb28fa42f49f9568322c683ff', | |
36 | }, | |
b3602f68 | 37 | 'skip': 'Sorry, the page you visited does not exist', |
45024183 MW |
38 | }, { |
39 | 'url': 'https://www.infoq.com/presentations/Simple-Made-Easy', | |
40 | 'md5': '0e34642d4d9ef44bf86f66f6399672db', | |
41 | 'info_dict': { | |
42 | 'id': 'Simple-Made-Easy', | |
43 | 'title': 'Simple Made Easy', | |
44 | 'ext': 'mp3', | |
45 | 'description': 'md5:3e0e213a8bbd074796ef89ea35ada25b', | |
46 | }, | |
47 | 'params': { | |
48 | 'format': 'bestaudio', | |
49 | }, | |
533f67d3 | 50 | }] |
fda7d31a | 51 | |
45024183 | 52 | def _extract_rtmp_video(self, webpage): |
7560096d | 53 | # The server URL is hardcoded |
a0566bbf | 54 | video_url = 'rtmpe://videof.infoq.com/cfx/st/' |
7560096d KW |
55 | |
56 | # Extract video URL | |
edec83a0 | 57 | encoded_id = self._search_regex( |
02a63fad YCH |
58 | r"jsclassref\s*=\s*'([^']*)'", webpage, 'encoded id', default=None) |
59 | ||
add96eb9 | 60 | real_id = urllib.parse.unquote(base64.b64decode(encoded_id).decode('utf-8')) |
7560096d | 61 | playpath = 'mp4:' + real_id |
fda7d31a | 62 | |
02a63fad | 63 | return [{ |
45024183 | 64 | 'format_id': 'rtmp_video', |
02a63fad YCH |
65 | 'url': video_url, |
66 | 'ext': determine_ext(playpath), | |
67 | 'play_path': playpath, | |
68 | }] | |
fda7d31a | 69 | |
e650659b | 70 | def _extract_cf_auth(self, webpage): |
197224b7 S |
71 | policy = self._search_regex(r'InfoQConstants\.scp\s*=\s*\'([^\']+)\'', webpage, 'policy') |
72 | signature = self._search_regex(r'InfoQConstants\.scs\s*=\s*\'([^\']+)\'', webpage, 'signature') | |
73 | key_pair_id = self._search_regex(r'InfoQConstants\.sck\s*=\s*\'([^\']+)\'', webpage, 'key-pair-id') | |
e650659b RA |
74 | return { |
75 | 'Policy': policy, | |
76 | 'Signature': signature, | |
77 | 'Key-Pair-Id': key_pair_id, | |
78 | } | |
edec83a0 | 79 | |
45024183 MW |
80 | def _extract_http_video(self, webpage): |
81 | http_video_url = self._search_regex(r'P\.s\s*=\s*\'([^\']+)\'', webpage, 'video URL') | |
e650659b | 82 | http_video_url = update_url_query(http_video_url, self._extract_cf_auth(webpage)) |
02a63fad | 83 | return [{ |
45024183 | 84 | 'format_id': 'http_video', |
22d07ba4 | 85 | 'url': http_video_url, |
a0566bbf | 86 | 'http_headers': {'Referer': 'https://www.infoq.com/'}, |
edec83a0 | 87 | }] |
02a63fad | 88 | |
45024183 | 89 | def _extract_http_audio(self, webpage, video_id): |
b3602f68 ES |
90 | try: |
91 | http_audio_url = traverse_obj(self._form_hidden_inputs('mp3Form', webpage), 'filename') | |
92 | except ExtractorError: | |
93 | http_audio_url = None | |
74da8565 | 94 | if not http_audio_url: |
45024183 MW |
95 | return [] |
96 | ||
45024183 MW |
97 | # base URL is found in the Location header in the response returned by |
98 | # GET https://www.infoq.com/mp3download.action?filename=... when logged in. | |
add96eb9 | 99 | http_audio_url = urllib.parse.urljoin('http://ress.infoq.com/downloads/mp3downloads/', http_audio_url) |
e650659b | 100 | http_audio_url = update_url_query(http_audio_url, self._extract_cf_auth(webpage)) |
45024183 MW |
101 | |
102 | # audio file seem to be missing some times even if there is a download link | |
103 | # so probe URL to make sure | |
e650659b | 104 | if not self._is_valid_url(http_audio_url, video_id): |
45024183 MW |
105 | return [] |
106 | ||
107 | return [{ | |
108 | 'format_id': 'http_audio', | |
109 | 'url': http_audio_url, | |
110 | 'vcodec': 'none', | |
45024183 MW |
111 | }] |
112 | ||
02a63fad YCH |
113 | def _real_extract(self, url): |
114 | video_id = self._match_id(url) | |
115 | webpage = self._download_webpage(url, video_id) | |
116 | ||
04f3fd2c | 117 | video_title = self._html_extract_title(webpage) |
02a63fad YCH |
118 | video_description = self._html_search_meta('description', webpage, 'description') |
119 | ||
120 | if '/cn/' in url: | |
121 | # for China videos, HTTP video URL exists but always fails with 403 | |
5633b4d3 | 122 | formats = self._extract_bokecc_formats(webpage, video_id) |
02a63fad | 123 | else: |
45024183 | 124 | formats = ( |
3089bc74 S |
125 | self._extract_rtmp_video(webpage) |
126 | + self._extract_http_video(webpage) | |
127 | + self._extract_http_audio(webpage, video_id)) | |
02a63fad | 128 | |
c0a7c608 | 129 | return { |
fda7d31a | 130 | 'id': video_id, |
fda7d31a | 131 | 'title': video_title, |
fda7d31a | 132 | 'description': video_description, |
edec83a0 | 133 | 'formats': formats, |
c0a7c608 | 134 | } |