From e572a1010b81f5864e610808c86848db4d6ed8e4 Mon Sep 17 00:00:00 2001 From: Erik Date: Sat, 17 Oct 2015 19:22:47 +0200 Subject: [PATCH] [youporn] Fix extraction [youporn] Added description and thumbnail [youporn] Added uploader and date [youporn] Removed Try and Except lines [youporn] Fixed date, fatal, formats and /s* [youporn] Undid removing comment about video url components & Undid and fixed removal of encrypted URL detection [youporn] Fix: Add encrypted link to links array only if not already in it [youporn] Fix: Add encrypted link to links array only if not already in it [youporn] Fix: cleanup --- youtube_dl/extractor/youporn.py | 55 ++++++++++++++------------------- 1 file changed, 23 insertions(+), 32 deletions(-) diff --git a/youtube_dl/extractor/youporn.py b/youtube_dl/extractor/youporn.py index 4ba7c36db7..546985f3af 100644 --- a/youtube_dl/extractor/youporn.py +++ b/youtube_dl/extractor/youporn.py @@ -4,6 +4,7 @@ from __future__ import unicode_literals import json import re import sys +import datetime from .common import InfoExtractor from ..compat import ( @@ -27,10 +28,11 @@ class YouPornIE(InfoExtractor): 'info_dict': { 'id': '505835', 'ext': 'mp4', - 'upload_date': '20101221', - 'description': 'Love & Sex Answers: http://bit.ly/DanAndJenn -- Is It Unhealthy To Masturbate Daily?', - 'uploader': 'Ask Dan And Jennifer', 'title': 'Sex Ed: Is It Safe To Masturbate Daily?', + 'description': 'Watch Sex Ed: Is It Safe To Masturbate Daily? at YouPorn.com - YouPorn is the biggest free porn tube site on the net!', + 'uploader': 'Ask Dan And Jennifer', + 'thumbnail': 'http://cdn5.image.youporn.phncdn.com/201012/17/505835/640x480/8/sex-ed-is-it-safe-to-masturbate-daily-8.jpg', + 'date': '20101221', 'age_limit': 18, } } @@ -45,45 +47,34 @@ class YouPornIE(InfoExtractor): webpage = self._download_webpage(req, video_id) age_limit = self._rta_search(webpage) - # Get JSON parameters - json_params = self._search_regex( - [r'videoJa?son\s*=\s*({.+})', - r'var\s+currentVideo\s*=\s*new\s+Video\((.+?)\)[,;]'], - webpage, 'JSON parameters') - try: - params = json.loads(json_params) - except ValueError: - raise ExtractorError('Invalid JSON') - self.report_extraction(video_id) - try: - video_title = params['title'] - upload_date = unified_strdate(params['release_date_f']) - video_description = params['description'] - video_uploader = params['submitted_by'] - thumbnail = params['thumbnails'][0]['image'] - except KeyError: - raise ExtractorError('Missing JSON parameter: ' + sys.exc_info()[1]) + video_title = self._html_search_regex(r'page_params.video_title = \'(.+?)\';', webpage, 'video URL', fatal=False) + video_description = self._html_search_meta('description', webpage, 'video DESC', fatal=False) + video_thumbnail = self._html_search_regex(r'page_params.imageurl\t=\t"(.+?)";', webpage, 'video THUMB', fatal=False) + video_uploader = self._html_search_regex(r"
By:
\n]+\">(.+?)", webpage, 'video UPLOADER', fatal=False) + video_date = self._html_search_regex(r"
\n (.+?)\n
", webpage, 'video DATE', fatal=False) + video_date = datetime.datetime.strptime(video_date, '%B %d, %Y').strftime('%Y%m%d') # Get all of the links from the page - DOWNLOAD_LIST_RE = r'(?s)' - download_list_html = self._search_regex(DOWNLOAD_LIST_RE, - webpage, 'download list').strip() - LINK_RE = r'' + DOWNLOAD_LIST_RE = r'(?s)sources: {\n(?P.*?)}' + download_list_html = self._search_regex(DOWNLOAD_LIST_RE, webpage, 'download list').strip() + LINK_RE = r': \'(.+?)\',' links = re.findall(LINK_RE, download_list_html) # Get all encrypted links - encrypted_links = re.findall(r'var encryptedQuality[0-9]{3}URL = \'([a-zA-Z0-9+/]+={0,2})\';', webpage) + encrypted_links = re.findall(r'page_params.encryptedQuality[0-9]{3,4}URL\s=\s\'([a-zA-Z0-9+/]+={0,2})\';', webpage) for encrypted_link in encrypted_links: link = aes_decrypt_text(encrypted_link, video_title, 32).decode('utf-8') - links.append(link) + # it's unclear if encryted links still differ from normal ones, so only include in links array if it's unique + if link not in links: + links.append(link) formats = [] for link in links: # A link looks like this: - # http://cdn1.download.youporn.phncdn.com/201210/31/8004515/480p_370k_8004515/YouPorn%20-%20Nubile%20Films%20The%20Pillow%20Fight.mp4?nvb=20121113051249&nva=20121114051249&ir=1200&sr=1200&hash=014b882080310e95fb6a0 + # http://cdn2b.public.youporn.phncdn.com/201012/17/505835/720p_1500k_505835/YouPorn%20-%20Sex%20Ed%20Is%20It%20Safe%20To%20Masturbate%20Daily.mp4?rs=200&ri=2500&s=1445599900&e=1445773500&h=5345d19ce9944ec52eb167abf24af248 # A path looks like this: - # /201210/31/8004515/480p_370k_8004515/YouPorn%20-%20Nubile%20Films%20The%20Pillow%20Fight.mp4 + # 201012/17/505835/720p_1500k_505835/YouPorn%20-%20Sex%20Ed%20Is%20It%20Safe%20To%20Masturbate%20Daily.mp4 video_url = unescapeHTML(link) path = compat_urllib_parse_urlparse(video_url).path format_parts = path.split('/')[4].split('_')[:2] @@ -111,11 +102,11 @@ class YouPornIE(InfoExtractor): return { 'id': video_id, - 'uploader': video_uploader, - 'upload_date': upload_date, 'title': video_title, - 'thumbnail': thumbnail, 'description': video_description, + 'thumbnail': video_thumbnail, + 'uploader': video_uploader, + 'date': video_date, 'age_limit': age_limit, 'formats': formats, }