]>
Commit | Line | Data |
---|---|---|
e299f6d2 PH |
1 | from __future__ import unicode_literals |
2 | ||
8e05c870 MO |
3 | |
4 | from .common import InfoExtractor | |
842cca7d | 5 | from ..utils import ( |
48fb963b | 6 | determine_ext, |
667d9648 | 7 | ExtractorError, |
842cca7d S |
8 | int_or_none, |
9 | js_to_json, | |
bee6451f | 10 | merge_dicts, |
48fb963b | 11 | urljoin, |
842cca7d | 12 | ) |
8e05c870 | 13 | |
6f5dcd4e | 14 | |
8e05c870 | 15 | class PornHdIE(InfoExtractor): |
5886b38d | 16 | _VALID_URL = r'https?://(?:www\.)?pornhd\.com/(?:[a-z]{2,4}/)?videos/(?P<id>\d+)(?:/(?P<display_id>.+))?' |
e3944c26 S |
17 | _TESTS = [{ |
18 | 'url': 'http://www.pornhd.com/videos/9864/selfie-restroom-masturbation-fun-with-chubby-cutie-hd-porn-video', | |
48fb963b | 19 | 'md5': '87f1540746c1d32ec7a2305c12b96b25', |
e3944c26 S |
20 | 'info_dict': { |
21 | 'id': '9864', | |
22 | 'display_id': 'selfie-restroom-masturbation-fun-with-chubby-cutie-hd-porn-video', | |
23 | 'ext': 'mp4', | |
24 | 'title': 'Restroom selfie masturbation', | |
25 | 'description': 'md5:3748420395e03e31ac96857a8f125b2b', | |
ec85ded8 | 26 | 'thumbnail': r're:^https?://.*\.jpg', |
e3944c26 | 27 | 'view_count': int, |
70c3ee13 | 28 | 'like_count': int, |
e3944c26 | 29 | 'age_limit': 18, |
bee6451f S |
30 | }, |
31 | 'skip': 'HTTP Error 404: Not Found', | |
e3944c26 | 32 | }, { |
e299f6d2 | 33 | 'url': 'http://www.pornhd.com/videos/1962/sierra-day-gets-his-cum-all-over-herself-hd-porn-video', |
bee6451f | 34 | 'md5': '1b7b3a40b9d65a8e5b25f7ab9ee6d6de', |
e299f6d2 | 35 | 'info_dict': { |
4b9cced1 | 36 | 'id': '1962', |
842cca7d | 37 | 'display_id': 'sierra-day-gets-his-cum-all-over-herself-hd-porn-video', |
4b9cced1 | 38 | 'ext': 'mp4', |
bee6451f | 39 | 'title': 'md5:98c6f8b2d9c229d0f0fde47f61a1a759', |
4b9cced1 | 40 | 'description': 'md5:8ff0523848ac2b8f9b065ba781ccf294', |
ec85ded8 | 41 | 'thumbnail': r're:^https?://.*\.jpg', |
842cca7d | 42 | 'view_count': int, |
70c3ee13 | 43 | 'like_count': int, |
4b9cced1 | 44 | 'age_limit': 18, |
e3944c26 | 45 | }, |
e3944c26 | 46 | }] |
8e05c870 MO |
47 | |
48 | def _real_extract(self, url): | |
5ad28e7f | 49 | mobj = self._match_valid_url(url) |
4b9cced1 | 50 | video_id = mobj.group('id') |
842cca7d | 51 | display_id = mobj.group('display_id') |
8e05c870 | 52 | |
842cca7d | 53 | webpage = self._download_webpage(url, display_id or video_id) |
8e05c870 | 54 | |
ceff3fd8 | 55 | title = self._html_search_regex( |
1b381853 S |
56 | [r'<span[^>]+class=["\']video-name["\'][^>]*>([^<]+)', |
57 | r'<title>(.+?) - .*?[Pp]ornHD.*?</title>'], webpage, 'title') | |
4b9cced1 | 58 | |
667d9648 | 59 | sources = self._parse_json(js_to_json(self._search_regex( |
3902cdd0 | 60 | r"(?s)sources'?\s*[:=]\s*(\{.+?\})", |
667d9648 S |
61 | webpage, 'sources', default='{}')), video_id) |
62 | ||
bee6451f | 63 | info = {} |
667d9648 | 64 | if not sources: |
bee6451f S |
65 | entries = self._parse_html5_media_entries(url, webpage, video_id) |
66 | if entries: | |
67 | info = entries[0] | |
68 | ||
69 | if not sources and not info: | |
667d9648 S |
70 | message = self._html_search_regex( |
71 | r'(?s)<(div|p)[^>]+class="no-video"[^>]*>(?P<value>.+?)</\1', | |
72 | webpage, 'error message', group='value') | |
73 | raise ExtractorError('%s said: %s' % (self.IE_NAME, message), expected=True) | |
74 | ||
807962f4 | 75 | formats = [] |
e6fe993c | 76 | for format_id, video_url in sources.items(): |
48fb963b | 77 | video_url = urljoin(url, video_url) |
f7a211dc PH |
78 | if not video_url: |
79 | continue | |
e6fe993c S |
80 | height = int_or_none(self._search_regex( |
81 | r'^(\d+)[pP]', format_id, 'height', default=None)) | |
f7a211dc PH |
82 | formats.append({ |
83 | 'url': video_url, | |
48fb963b | 84 | 'ext': determine_ext(video_url, 'mp4'), |
e6fe993c S |
85 | 'format_id': format_id, |
86 | 'height': height, | |
f7a211dc | 87 | }) |
bee6451f S |
88 | if formats: |
89 | info['formats'] = formats | |
90 | self._sort_formats(info['formats']) | |
8e05c870 | 91 | |
667d9648 | 92 | description = self._html_search_regex( |
bee6451f S |
93 | (r'(?s)<section[^>]+class=["\']video-description[^>]+>(?P<value>.+?)</section>', |
94 | r'<(div|p)[^>]+class="description"[^>]*>(?P<value>[^<]+)</\1'), | |
95 | webpage, 'description', fatal=False, | |
96 | group='value') or self._html_search_meta( | |
97 | 'description', webpage, default=None) or self._og_search_description(webpage) | |
667d9648 S |
98 | view_count = int_or_none(self._html_search_regex( |
99 | r'(\d+) views\s*<', webpage, 'view count', fatal=False)) | |
100 | thumbnail = self._search_regex( | |
3902cdd0 | 101 | r"poster'?\s*:\s*([\"'])(?P<url>(?:(?!\1).)+)\1", webpage, |
bee6451f | 102 | 'thumbnail', default=None, group='url') |
667d9648 | 103 | |
70c3ee13 | 104 | like_count = int_or_none(self._search_regex( |
bee6451f S |
105 | (r'(\d+)</span>\s*likes', |
106 | r'(\d+)\s*</11[^>]+>(?: |\s)*\blikes', | |
70c3ee13 J |
107 | r'class=["\']save-count["\'][^>]*>\s*(\d+)'), |
108 | webpage, 'like count', fatal=False)) | |
109 | ||
bee6451f | 110 | return merge_dicts(info, { |
6f5dcd4e | 111 | 'id': video_id, |
842cca7d | 112 | 'display_id': display_id, |
4b9cced1 S |
113 | 'title': title, |
114 | 'description': description, | |
115 | 'thumbnail': thumbnail, | |
116 | 'view_count': view_count, | |
70c3ee13 | 117 | 'like_count': like_count, |
4b9cced1 S |
118 | 'formats': formats, |
119 | 'age_limit': 18, | |
bee6451f | 120 | }) |