X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fopenload.py;h=0c20d0177e0f2abbc5c0503babf51ba66715f374;hb=HEAD;hp=292476ef86cf8680fc59e58488f84cae44da7947;hpb=a88d461dff67205fcec684426afbcbeb4b0e7cf5;p=youtube-dl diff --git a/youtube_dl/extractor/openload.py b/youtube_dl/extractor/openload.py index 292476ef8..0c20d0177 100644 --- a/youtube_dl/extractor/openload.py +++ b/youtube_dl/extractor/openload.py @@ -1,108 +1,238 @@ # coding: utf-8 from __future__ import unicode_literals -import re - -from .common import InfoExtractor +import json +import os +import subprocess +import tempfile + +from ..compat import ( + compat_urlparse, + compat_kwargs, +) from ..utils import ( - determine_ext, + check_executable, + encodeArgument, ExtractorError, - get_element_by_id, - PhantomJSwrapper, + get_exe_version, + is_outdated_version, + std_headers, ) -class OpenloadIE(InfoExtractor): - _VALID_URL = r'https?://(?:openload\.(?:co|io)|oload\.tv)/(?:f|embed)/(?P[a-zA-Z0-9-_]+)' - - _TESTS = [{ - 'url': 'https://openload.co/f/kUEfGclsU9o', - 'md5': 'bf1c059b004ebc7a256f89408e65c36e', - 'info_dict': { - 'id': 'kUEfGclsU9o', - 'ext': 'mp4', - 'title': 'skyrim_no-audio_1080.mp4', - 'thumbnail': r're:^https?://.*\.jpg$', - }, - }, { - 'url': 'https://openload.co/embed/rjC09fkPLYs', - 'info_dict': { - 'id': 'rjC09fkPLYs', - 'ext': 'mp4', - 'title': 'movie.mp4', - 'thumbnail': r're:^https?://.*\.jpg$', - 'subtitles': { - 'en': [{ - 'ext': 'vtt', - }], - }, - }, - 'params': { - 'skip_download': True, # test subtitles only - }, - }, { - 'url': 'https://openload.co/embed/kUEfGclsU9o/skyrim_no-audio_1080.mp4', - 'only_matching': True, - }, { - 'url': 'https://openload.io/f/ZAn6oz-VZGE/', - 'only_matching': True, - }, { - 'url': 'https://openload.co/f/_-ztPaZtMhM/', - 'only_matching': True, - }, { - # unavailable via https://openload.co/f/Sxz5sADo82g/, different layout - # for title and ext - 'url': 'https://openload.co/embed/Sxz5sADo82g/', - 'only_matching': True, - }, { - 'url': 'https://oload.tv/embed/KnG-kKZdcfY/', - 'only_matching': True, - }] - - _USER_AGENT = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/56.0.2924.87 Safari/537.36' +def cookie_to_dict(cookie): + cookie_dict = { + 'name': cookie.name, + 'value': cookie.value, + } + if cookie.port_specified: + cookie_dict['port'] = cookie.port + if cookie.domain_specified: + cookie_dict['domain'] = cookie.domain + if cookie.path_specified: + cookie_dict['path'] = cookie.path + if cookie.expires is not None: + cookie_dict['expires'] = cookie.expires + if cookie.secure is not None: + cookie_dict['secure'] = cookie.secure + if cookie.discard is not None: + cookie_dict['discard'] = cookie.discard + try: + if (cookie.has_nonstandard_attr('httpOnly') + or cookie.has_nonstandard_attr('httponly') + or cookie.has_nonstandard_attr('HttpOnly')): + cookie_dict['httponly'] = True + except TypeError: + pass + return cookie_dict + + +def cookie_jar_to_list(cookie_jar): + return [cookie_to_dict(cookie) for cookie in cookie_jar] + + +class PhantomJSwrapper(object): + """PhantomJS wrapper class + + This class is experimental. + """ + + _TEMPLATE = r''' + phantom.onError = function(msg, trace) {{ + var msgStack = ['PHANTOM ERROR: ' + msg]; + if(trace && trace.length) {{ + msgStack.push('TRACE:'); + trace.forEach(function(t) {{ + msgStack.push(' -> ' + (t.file || t.sourceURL) + ': ' + t.line + + (t.function ? ' (in function ' + t.function +')' : '')); + }}); + }} + console.error(msgStack.join('\n')); + phantom.exit(1); + }}; + var page = require('webpage').create(); + var fs = require('fs'); + var read = {{ mode: 'r', charset: 'utf-8' }}; + var write = {{ mode: 'w', charset: 'utf-8' }}; + JSON.parse(fs.read("{cookies}", read)).forEach(function(x) {{ + phantom.addCookie(x); + }}); + page.settings.resourceTimeout = {timeout}; + page.settings.userAgent = "{ua}"; + page.onLoadStarted = function() {{ + page.evaluate(function() {{ + delete window._phantom; + delete window.callPhantom; + }}); + }}; + var saveAndExit = function() {{ + fs.write("{html}", page.content, write); + fs.write("{cookies}", JSON.stringify(phantom.cookies), write); + phantom.exit(); + }}; + page.onLoadFinished = function(status) {{ + if(page.url === "") {{ + page.setContent(fs.read("{html}", read), "{url}"); + }} + else {{ + {jscode} + }} + }}; + page.open(""); + ''' + + _TMP_FILE_NAMES = ['script', 'html', 'cookies'] @staticmethod - def _extract_urls(webpage): - return re.findall( - r']+src=["\']((?:https?://)?(?:openload\.(?:co|io)|oload\.tv)/embed/[a-zA-Z0-9-_]+)', - webpage) - - def _real_extract(self, url): - video_id = self._match_id(url) - url = 'https://openload.co/embed/%s/' % video_id - headers = { - 'User-Agent': self._USER_AGENT, - } - - webpage = self._download_webpage(url, video_id, headers=headers) - - if 'File not found' in webpage or 'deleted by the owner' in webpage: - raise ExtractorError('File not found', expected=True, video_id=video_id) - - phantom = PhantomJSwrapper(self, required_version='2.0') - webpage, _ = phantom.get(url, html=webpage, video_id=video_id, headers=headers) + def _version(): + return get_exe_version('phantomjs', version_re=r'([0-9.]+)') - decoded_id = get_element_by_id('streamurl', webpage) + def __init__(self, extractor, required_version=None, timeout=10000): + self._TMP_FILES = {} - video_url = 'https://openload.co/stream/%s?mime=true' % decoded_id + self.exe = check_executable('phantomjs', ['-v']) + if not self.exe: + raise ExtractorError('PhantomJS executable not found in PATH, ' + 'download it from http://phantomjs.org', + expected=True) - title = self._og_search_title(webpage, default=None) or self._search_regex( - r']+class=["\']title["\'][^>]*>([^<]+)', webpage, - 'title', default=None) or self._html_search_meta( - 'description', webpage, 'title', fatal=True) + self.extractor = extractor - entries = self._parse_html5_media_entries(url, webpage, video_id) - entry = entries[0] if entries else {} - subtitles = entry.get('subtitles') + if required_version: + version = self._version() + if is_outdated_version(version, required_version): + self.extractor._downloader.report_warning( + 'Your copy of PhantomJS is outdated, update it to version ' + '%s or newer if you encounter any errors.' % required_version) - info_dict = { - 'id': video_id, - 'title': title, - 'thumbnail': entry.get('thumbnail') or self._og_search_thumbnail(webpage, default=None), - 'url': video_url, - # Seems all videos have extensions in their titles - 'ext': determine_ext(title, 'mp4'), - 'subtitles': subtitles, - 'http_headers': headers, + self.options = { + 'timeout': timeout, } - return info_dict + for name in self._TMP_FILE_NAMES: + tmp = tempfile.NamedTemporaryFile(delete=False) + tmp.close() + self._TMP_FILES[name] = tmp + + def __del__(self): + for name in self._TMP_FILE_NAMES: + try: + os.remove(self._TMP_FILES[name].name) + except (IOError, OSError, KeyError): + pass + + def _save_cookies(self, url): + cookies = cookie_jar_to_list(self.extractor._downloader.cookiejar) + for cookie in cookies: + if 'path' not in cookie: + cookie['path'] = '/' + if 'domain' not in cookie: + cookie['domain'] = compat_urlparse.urlparse(url).netloc + with open(self._TMP_FILES['cookies'].name, 'wb') as f: + f.write(json.dumps(cookies).encode('utf-8')) + + def _load_cookies(self): + with open(self._TMP_FILES['cookies'].name, 'rb') as f: + cookies = json.loads(f.read().decode('utf-8')) + for cookie in cookies: + if cookie['httponly'] is True: + cookie['rest'] = {'httpOnly': None} + if 'expiry' in cookie: + cookie['expire_time'] = cookie['expiry'] + self.extractor._set_cookie(**compat_kwargs(cookie)) + + def get(self, url, html=None, video_id=None, note=None, note2='Executing JS on webpage', headers={}, jscode='saveAndExit();'): + """ + Downloads webpage (if needed) and executes JS + + Params: + url: website url + html: optional, html code of website + video_id: video id + note: optional, displayed when downloading webpage + note2: optional, displayed when executing JS + headers: custom http headers + jscode: code to be executed when page is loaded + + Returns tuple with: + * downloaded website (after JS execution) + * anything you print with `console.log` (but not inside `page.execute`!) + + In most cases you don't need to add any `jscode`. + It is executed in `page.onLoadFinished`. + `saveAndExit();` is mandatory, use it instead of `phantom.exit()` + It is possible to wait for some element on the webpage, for example: + var check = function() { + var elementFound = page.evaluate(function() { + return document.querySelector('#b.done') !== null; + }); + if(elementFound) + saveAndExit(); + else + window.setTimeout(check, 500); + } + + page.evaluate(function(){ + document.querySelector('#a').click(); + }); + check(); + """ + if 'saveAndExit();' not in jscode: + raise ExtractorError('`saveAndExit();` not found in `jscode`') + if not html: + html = self.extractor._download_webpage(url, video_id, note=note, headers=headers) + with open(self._TMP_FILES['html'].name, 'wb') as f: + f.write(html.encode('utf-8')) + + self._save_cookies(url) + + replaces = self.options + replaces['url'] = url + user_agent = headers.get('User-Agent') or std_headers['User-Agent'] + replaces['ua'] = user_agent.replace('"', '\\"') + replaces['jscode'] = jscode + + for x in self._TMP_FILE_NAMES: + replaces[x] = self._TMP_FILES[x].name.replace('\\', '\\\\').replace('"', '\\"') + + with open(self._TMP_FILES['script'].name, 'wb') as f: + f.write(self._TEMPLATE.format(**replaces).encode('utf-8')) + + if video_id is None: + self.extractor.to_screen('%s' % (note2,)) + else: + self.extractor.to_screen('%s: %s' % (video_id, note2)) + + p = subprocess.Popen([ + self.exe, '--ssl-protocol=any', + self._TMP_FILES['script'].name + ], stdout=subprocess.PIPE, stderr=subprocess.PIPE) + out, err = p.communicate() + if p.returncode != 0: + raise ExtractorError( + 'Executing JS failed\n:' + encodeArgument(err)) + with open(self._TMP_FILES['html'].name, 'rb') as f: + html = f.read().decode('utf-8') + + self._load_cookies() + + return (html, encodeArgument(out))