From ad32b52756a3ffecc49c218f26c9602cd36658f0 Mon Sep 17 00:00:00 2001 From: Unrud Date: Fri, 6 Dec 2019 05:49:03 +0100 Subject: [PATCH] Replace PhantomJS with Node.js sandbox PhantomJS is unmaintained and depends on the deprecated QtWebKit. --- youtube_dl/YoutubeDL.py | 4 +- youtube_dl/extractor/openload.py | 238 ++++--------------------------- youtube_dl/extractor/pornhub.py | 25 +++- 3 files changed, 45 insertions(+), 222 deletions(-) diff --git a/youtube_dl/YoutubeDL.py b/youtube_dl/YoutubeDL.py index f5cb46308..01796b98c 100755 --- a/youtube_dl/YoutubeDL.py +++ b/youtube_dl/YoutubeDL.py @@ -95,7 +95,7 @@ from .utils import ( ) from .cache import Cache from .extractor import get_info_extractor, gen_extractor_classes, _LAZY_LOADER -from .extractor.openload import PhantomJSwrapper +from .extractor.openload import NodejsWrapper from .downloader import get_suitable_downloader from .downloader.rtmp import rtmpdump_version from .postprocessor import ( @@ -2285,7 +2285,7 @@ class YoutubeDL(object): exe_versions = FFmpegPostProcessor.get_versions(self) exe_versions['rtmpdump'] = rtmpdump_version() - exe_versions['phantomjs'] = PhantomJSwrapper._version() + exe_versions['nodejs'] = NodejsWrapper.version() exe_str = ', '.join( '%s %s' % (exe, v) for exe, v in sorted(exe_versions.items()) diff --git a/youtube_dl/extractor/openload.py b/youtube_dl/extractor/openload.py index 0c20d0177..9920ae1e0 100644 --- a/youtube_dl/extractor/openload.py +++ b/youtube_dl/extractor/openload.py @@ -1,238 +1,50 @@ # coding: utf-8 from __future__ import unicode_literals -import json -import os +import base64 import subprocess -import tempfile -from ..compat import ( - compat_urlparse, - compat_kwargs, -) from ..utils import ( check_executable, encodeArgument, ExtractorError, get_exe_version, - is_outdated_version, - std_headers, ) -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'] +class NodejsWrapper(object): + """Node.js wrapper class""" @staticmethod - def _version(): - return get_exe_version('phantomjs', version_re=r'([0-9.]+)') + def version(): + return get_exe_version('nodejs', version_re=r'([0-9.]+)') - def __init__(self, extractor, required_version=None, timeout=10000): - self._TMP_FILES = {} - - self.exe = check_executable('phantomjs', ['-v']) + def __init__(self): + self.exe = check_executable('nodejs', ['-v']) if not self.exe: - raise ExtractorError('PhantomJS executable not found in PATH, ' - 'download it from http://phantomjs.org', + raise ExtractorError('Node.js executable not found in PATH, ' + 'download it from http://nodejs.org', expected=True) - self.extractor = extractor - - 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) - - self.options = { - 'timeout': timeout, - } - 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();'): + def run_in_sandbox(self, jscode, timeout=10000): """ - Downloads webpage (if needed) and executes JS + Executes JS in a sandbox 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 + jscode: code to be executed + timeout: number of milliseconds before execution is cancelled - 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() + jscode_wrapper = ( + 'const vm = require("vm");' + 'const jscode = Buffer.from("{jscode_b64}", "base64").toString("binary");' + 'const options = {{timeout: {timeout:d}}};' + 'process.stdout.write(String(vm.runInNewContext(jscode, {{}}, options)));').format( + jscode_b64=base64.b64encode(jscode.encode('utf-8')).decode(), + timeout=timeout) + args = [encodeArgument(a) for a in [self.exe, '-e', jscode_wrapper]] + p = subprocess.Popen(args, stdout=subprocess.PIPE, stderr=subprocess.PIPE) + stdout, stderr = 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)) + raise ExtractorError(stderr.decode('utf-8', 'replace')) + return stdout.decode('utf-8') diff --git a/youtube_dl/extractor/pornhub.py b/youtube_dl/extractor/pornhub.py index ba0ad7da2..ca7fb8e07 100644 --- a/youtube_dl/extractor/pornhub.py +++ b/youtube_dl/extractor/pornhub.py @@ -11,8 +11,9 @@ from ..compat import ( compat_HTTPError, compat_str, compat_urllib_request, + compat_urlparse, ) -from .openload import PhantomJSwrapper +from .openload import NodejsWrapper from ..utils import ( determine_ext, ExtractorError, @@ -31,16 +32,26 @@ class PornHubBaseIE(InfoExtractor): 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\(true\)')): + if re.search(r']+\bonload=["\']go\(\)', webpage): 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) + domain = compat_urlparse.urlparse(url).netloc + jscode = self._search_regex( + r']*>(?P', + webpage, 'jscode', flags=re.DOTALL, group='script') + self.to_screen('Executing JS on webpage') + cookie = NodejsWrapper().run_in_sandbox(( + 'let document = {{location:{{reload: function() {{}}}}}};' + '{jscode};//*/\n' + 'go();' + 'document.cookie').format(jscode=jscode)) + cookie_name = cookie.split("=")[0].strip() + if cookie_name != 'RNKEY': + raise ExtractorError('Unexpected cookie name %r' % cookie_name) + cookie_value = cookie[len(cookie_name) + 1:].split(";")[0].strip() + self._set_cookie(domain, cookie_name, cookie_value) webpage, urlh = dl(*args, **kwargs) return webpage, urlh