yt_dlp/extractor/openload.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import json
   5 import os
   6 import subprocess
   7 import tempfile
   8
   9 from ..compat import (
  10     compat_urlparse,
  11 )
  12 from ..utils import (
  13     check_executable,
  14     encodeArgument,
  15     ExtractorError,
  16     get_exe_version,
  17     is_outdated_version,
  18     Popen,
  19 )
  20
  21
  22 def cookie_to_dict(cookie):
  23     cookie_dict = {
  24         'name': cookie.name,
  25         'value': cookie.value,
  26     }
  27     if cookie.port_specified:
  28         cookie_dict['port'] = cookie.port
  29     if cookie.domain_specified:
  30         cookie_dict['domain'] = cookie.domain
  31     if cookie.path_specified:
  32         cookie_dict['path'] = cookie.path
  33     if cookie.expires is not None:
  34         cookie_dict['expires'] = cookie.expires
  35     if cookie.secure is not None:
  36         cookie_dict['secure'] = cookie.secure
  37     if cookie.discard is not None:
  38         cookie_dict['discard'] = cookie.discard
  39     try:
  40         if (cookie.has_nonstandard_attr('httpOnly')
  41                 or cookie.has_nonstandard_attr('httponly')
  42                 or cookie.has_nonstandard_attr('HttpOnly')):
  43             cookie_dict['httponly'] = True
  44     except TypeError:
  45         pass
  46     return cookie_dict
  47
  48
  49 def cookie_jar_to_list(cookie_jar):
  50     return [cookie_to_dict(cookie) for cookie in cookie_jar]
  51
  52
  53 class PhantomJSwrapper(object):
  54     """PhantomJS wrapper class
  55
  56     This class is experimental.
  57     """
  58
  59     _TEMPLATE = r'''
  60         phantom.onError = function(msg, trace) {{
  61           var msgStack = ['PHANTOM ERROR: ' + msg];
  62           if(trace && trace.length) {{
  63             msgStack.push('TRACE:');
  64             trace.forEach(function(t) {{
  65               msgStack.push(' -> ' + (t.file || t.sourceURL) + ': ' + t.line
  66                 + (t.function ? ' (in function ' + t.function +')' : ''));
  67             }});
  68           }}
  69           console.error(msgStack.join('\n'));
  70           phantom.exit(1);
  71         }};
  72         var page = require('webpage').create();
  73         var fs = require('fs');
  74         var read = {{ mode: 'r', charset: 'utf-8' }};
  75         var write = {{ mode: 'w', charset: 'utf-8' }};
  76         JSON.parse(fs.read("{cookies}", read)).forEach(function(x) {{
  77           phantom.addCookie(x);
  78         }});
  79         page.settings.resourceTimeout = {timeout};
  80         page.settings.userAgent = "{ua}";
  81         page.onLoadStarted = function() {{
  82           page.evaluate(function() {{
  83             delete window._phantom;
  84             delete window.callPhantom;
  85           }});
  86         }};
  87         var saveAndExit = function() {{
  88           fs.write("{html}", page.content, write);
  89           fs.write("{cookies}", JSON.stringify(phantom.cookies), write);
  90           phantom.exit();
  91         }};
  92         page.onLoadFinished = function(status) {{
  93           if(page.url === "") {{
  94             page.setContent(fs.read("{html}", read), "{url}");
  95           }}
  96           else {{
  97             {jscode}
  98           }}
  99         }};
 100         page.open("");
 101     '''
 102
 103     _TMP_FILE_NAMES = ['script', 'html', 'cookies']
 104
 105     @staticmethod
 106     def _version():
 107         return get_exe_version('phantomjs', version_re=r'([0-9.]+)')
 108
 109     def __init__(self, extractor, required_version=None, timeout=10000):
 110         self._TMP_FILES = {}
 111
 112         self.exe = check_executable('phantomjs', ['-v'])
 113         if not self.exe:
 114             raise ExtractorError('PhantomJS executable not found in PATH, '
 115                                  'download it from http://phantomjs.org',
 116                                  expected=True)
 117
 118         self.extractor = extractor
 119
 120         if required_version:
 121             version = self._version()
 122             if is_outdated_version(version, required_version):
 123                 self.extractor._downloader.report_warning(
 124                     'Your copy of PhantomJS is outdated, update it to version '
 125                     '%s or newer if you encounter any errors.' % required_version)
 126
 127         self.options = {
 128             'timeout': timeout,
 129         }
 130         for name in self._TMP_FILE_NAMES:
 131             tmp = tempfile.NamedTemporaryFile(delete=False)
 132             tmp.close()
 133             self._TMP_FILES[name] = tmp
 134
 135     def __del__(self):
 136         for name in self._TMP_FILE_NAMES:
 137             try:
 138                 os.remove(self._TMP_FILES[name].name)
 139             except (IOError, OSError, KeyError):
 140                 pass
 141
 142     def _save_cookies(self, url):
 143         cookies = cookie_jar_to_list(self.extractor._downloader.cookiejar)
 144         for cookie in cookies:
 145             if 'path' not in cookie:
 146                 cookie['path'] = '/'
 147             if 'domain' not in cookie:
 148                 cookie['domain'] = compat_urlparse.urlparse(url).netloc
 149         with open(self._TMP_FILES['cookies'].name, 'wb') as f:
 150             f.write(json.dumps(cookies).encode('utf-8'))
 151
 152     def _load_cookies(self):
 153         with open(self._TMP_FILES['cookies'].name, 'rb') as f:
 154             cookies = json.loads(f.read().decode('utf-8'))
 155         for cookie in cookies:
 156             if cookie['httponly'] is True:
 157                 cookie['rest'] = {'httpOnly': None}
 158             if 'expiry' in cookie:
 159                 cookie['expire_time'] = cookie['expiry']
 160             self.extractor._set_cookie(**cookie)
 161
 162     def get(self, url, html=None, video_id=None, note=None, note2='Executing JS on webpage', headers={}, jscode='saveAndExit();'):
 163         """
 164         Downloads webpage (if needed) and executes JS
 165
 166         Params:
 167             url: website url
 168             html: optional, html code of website
 169             video_id: video id
 170             note: optional, displayed when downloading webpage
 171             note2: optional, displayed when executing JS
 172             headers: custom http headers
 173             jscode: code to be executed when page is loaded
 174
 175         Returns tuple with:
 176             * downloaded website (after JS execution)
 177             * anything you print with `console.log` (but not inside `page.execute`!)
 178
 179         In most cases you don't need to add any `jscode`.
 180         It is executed in `page.onLoadFinished`.
 181         `saveAndExit();` is mandatory, use it instead of `phantom.exit()`
 182         It is possible to wait for some element on the webpage, for example:
 183             var check = function() {
 184               var elementFound = page.evaluate(function() {
 185                 return document.querySelector('#b.done') !== null;
 186               });
 187               if(elementFound)
 188                 saveAndExit();
 189               else
 190                 window.setTimeout(check, 500);
 191             }
 192
 193             page.evaluate(function(){
 194               document.querySelector('#a').click();
 195             });
 196             check();
 197         """
 198         if 'saveAndExit();' not in jscode:
 199             raise ExtractorError('`saveAndExit();` not found in `jscode`')
 200         if not html:
 201             html = self.extractor._download_webpage(url, video_id, note=note, headers=headers)
 202         with open(self._TMP_FILES['html'].name, 'wb') as f:
 203             f.write(html.encode('utf-8'))
 204
 205         self._save_cookies(url)
 206
 207         replaces = self.options
 208         replaces['url'] = url
 209         user_agent = headers.get('User-Agent') or self.extractor.get_param('http_headers')['User-Agent']
 210         replaces['ua'] = user_agent.replace('"', '\\"')
 211         replaces['jscode'] = jscode
 212
 213         for x in self._TMP_FILE_NAMES:
 214             replaces[x] = self._TMP_FILES[x].name.replace('\\', '\\\\').replace('"', '\\"')
 215
 216         with open(self._TMP_FILES['script'].name, 'wb') as f:
 217             f.write(self._TEMPLATE.format(**replaces).encode('utf-8'))
 218
 219         if video_id is None:
 220             self.extractor.to_screen('%s' % (note2,))
 221         else:
 222             self.extractor.to_screen('%s: %s' % (video_id, note2))
 223
 224         p = Popen(
 225             [self.exe, '--ssl-protocol=any', self._TMP_FILES['script'].name],
 226             stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 227         out, err = p.communicate_or_kill()
 228         if p.returncode != 0:
 229             raise ExtractorError(
 230                 'Executing JS failed\n:' + encodeArgument(err))
 231         with open(self._TMP_FILES['html'].name, 'rb') as f:
 232             html = f.read().decode('utf-8')
 233
 234         self._load_cookies()
 235
 236         return (html, encodeArgument(out))