X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fpornhub.py;h=75ed69cde3db4c4116be3089f77f8f43f3205271;hb=f41347260c2c2cf723bc2bb8a5c11f67a22175d5;hp=f5f3e6593a2bc788b05efd155cff2e851fab729f;hpb=278d061a0c5eae20963c0a6df4b9b13fd1537186;p=youtube-dl diff --git a/youtube_dl/extractor/pornhub.py b/youtube_dl/extractor/pornhub.py index f5f3e6593..75ed69cde 100644 --- a/youtube_dl/extractor/pornhub.py +++ b/youtube_dl/extractor/pornhub.py @@ -14,9 +14,9 @@ from ..compat import ( ) from .openload import PhantomJSwrapper from ..utils import ( + determine_ext, ExtractorError, int_or_none, - js_to_json, orderedSet, remove_quotes, str_to_int, @@ -24,7 +24,29 @@ from ..utils import ( ) -class PornHubIE(InfoExtractor): +class PornHubBaseIE(InfoExtractor): + def _download_webpage_handle(self, *args, **kwargs): + def dl(*args, **kwargs): + return super(PornHubBaseIE, self)._download_webpage_handle(*args, **kwargs) + + webpage, urlh = dl(*args, **kwargs) + + if any(re.search(p, webpage) for p in ( + r'
]+\bonload=["\']go\(\)', + r'document\.cookie\s*=\s*["\']RNKEY=', + r'document\.location\.reload\(true\)')): + url_or_request = args[0] + url = (url_or_request.get_full_url() + if isinstance(url_or_request, compat_urllib_request.Request) + else url_or_request) + phantom = PhantomJSwrapper(self, required_version='2.0') + phantom.get(url, html=webpage) + webpage, urlh = dl(*args, **kwargs) + + return webpage, urlh + + +class PornHubIE(PornHubBaseIE): IE_DESC = 'PornHub and Thumbzilla' _VALID_URL = r'''(?x) https?:// @@ -128,26 +150,6 @@ class PornHubIE(InfoExtractor): 'only_matching': True, }] - def _download_webpage_handle(self, *args, **kwargs): - def dl(*args, **kwargs): - return super(PornHubIE, self)._download_webpage_handle(*args, **kwargs) - - webpage, urlh = dl(*args, **kwargs) - - if any(re.search(p, webpage) for p in ( - r']+\bonload=["\']go\(\)', - r'document\.cookie\s*=\s*["\']RNKEY=', - r'document\.location\.reload\(true\)')): - url_or_request = args[0] - url = (url_or_request.get_full_url() - if isinstance(url_or_request, compat_urllib_request.Request) - else url_or_request) - phantom = PhantomJSwrapper(self, required_version='2.0') - phantom.get(url, html=webpage) - webpage, urlh = dl(*args, **kwargs) - - return webpage, urlh - @staticmethod def _extract_urls(webpage): return re.findall( @@ -168,7 +170,7 @@ class PornHubIE(InfoExtractor): def dl_webpage(platform): self._set_cookie(host, 'platform', platform) return self._download_webpage( - 'http://www.%s/view_video.php?viewkey=%s' % (host, video_id), + 'https://www.%s/view_video.php?viewkey=%s' % (host, video_id), video_id, 'Downloading %s webpage' % platform) webpage = dl_webpage('pc') @@ -225,12 +227,13 @@ class PornHubIE(InfoExtractor): else: thumbnail, duration = [None] * 2 - if not video_urls: - tv_webpage = dl_webpage('tv') - + def extract_js_vars(webpage, pattern, fatal=True): assignments = self._search_regex( - r'(var.+?mediastring.+?)', tv_webpage, - 'encoded url').split(';') + pattern, webpage, 'encoded url', fatal=fatal) + if not assignments: + return {} + + assignments = assignments.split(';') js_vars = {} @@ -252,11 +255,31 @@ class PornHubIE(InfoExtractor): assn = re.sub(r'var\s+', '', assn) vname, value = assn.split('=', 1) js_vars[vname] = parse_js_value(value) + return js_vars - video_url = js_vars['mediastring'] - if video_url not in video_urls_set: - video_urls.append((video_url, None)) - video_urls_set.add(video_url) + def add_video_url(video_url): + v_url = url_or_none(video_url) + if not v_url: + return + if v_url in video_urls_set: + return + video_urls.append((v_url, None)) + video_urls_set.add(v_url) + + if not video_urls: + FORMAT_PREFIXES = ('media', 'quality') + js_vars = extract_js_vars( + webpage, r'(var\s+(?:%s)_.+)' % '|'.join(FORMAT_PREFIXES), + fatal=False) + if js_vars: + for key, format_url in js_vars.items(): + if any(key.startswith(p) for p in FORMAT_PREFIXES): + add_video_url(format_url) + + if not video_urls: + js_vars = extract_js_vars( + dl_webpage('tv'), r'(var.+?mediastring.+?)') + add_video_url(js_vars['mediastring']) for mobj in re.finditer( r']+\bclass=["\']downloadBtn\b[^>]+\bhref=(["\'])(?P