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     compat_kwargs,
  12 )
  13 from ..utils import (
  14     check_executable,
  15     encodeArgument,
  16     ExtractorError,
  17     get_exe_version,
  18     is_outdated_version,
  19     std_headers,
  20     process_communicate_or_kill,
  21 )
  22
  23
  24 def cookie_to_dict(cookie):
  25     cookie_dict = {
  26         'name': cookie.name,
  27         'value': cookie.value,
  28     }
  29     if cookie.port_specified:
  30         cookie_dict['port'] = cookie.port
  31     if cookie.domain_specified:
  32         cookie_dict['domain'] = cookie.domain
  33     if cookie.path_specified:
  34         cookie_dict['path'] = cookie.path
  35     if cookie.expires is not None:
  36         cookie_dict['expires'] = cookie.expires
  37     if cookie.secure is not None:
  38         cookie_dict['secure'] = cookie.secure
  39     if cookie.discard is not None:
  40         cookie_dict['discard'] = cookie.discard
  41     try:
  42         if (cookie.has_nonstandard_attr('httpOnly')
  43                 or cookie.has_nonstandard_attr('httponly')
  44                 or cookie.has_nonstandard_attr('HttpOnly')):
  45             cookie_dict['httponly'] = True
  46     except TypeError:
  47         pass
  48     return cookie_dict
  49
  50
  51 def cookie_jar_to_list(cookie_jar):
  52     return [cookie_to_dict(cookie) for cookie in cookie_jar]
  53
  54
  55 class PhantomJSwrapper(object):
  56     """PhantomJS wrapper class
  57
  58     This class is experimental.
  59     """
  60
  61     _TEMPLATE = r'''
  62         phantom.onError = function(msg, trace) {{
  63           var msgStack = ['PHANTOM ERROR: ' + msg];
  64           if(trace && trace.length) {{
  65             msgStack.push('TRACE:');
  66             trace.forEach(function(t) {{
  67               msgStack.push(' -> ' + (t.file || t.sourceURL) + ': ' + t.line
  68                 + (t.function ? ' (in function ' + t.function +')' : ''));
  69             }});
  70           }}
  71           console.error(msgStack.join('\n'));
  72           phantom.exit(1);
  73         }};
  74         var page = require('webpage').create();
  75         var fs = require('fs');
  76         var read = {{ mode: 'r', charset: 'utf-8' }};
  77         var write = {{ mode: 'w', charset: 'utf-8' }};
  78         JSON.parse(fs.read("{cookies}", read)).forEach(function(x) {{
  79           phantom.addCookie(x);
  80         }});
  81         page.settings.resourceTimeout = {timeout};
  82         page.settings.userAgent = "{ua}";
  83         page.onLoadStarted = function() {{
  84           page.evaluate(function() {{
  85             delete window._phantom;
  86             delete window.callPhantom;
  87           }});
  88         }};
  89         var saveAndExit = function() {{
  90           fs.write("{html}", page.content, write);
  91           fs.write("{cookies}", JSON.stringify(phantom.cookies), write);
  92           phantom.exit();
  93         }};
  94         page.onLoadFinished = function(status) {{
  95           if(page.url === "") {{
  96             page.setContent(fs.read("{html}", read), "{url}");
  97           }}
  98           else {{
  99             {jscode}
 100           }}
 101         }};
 102         page.open("");
 103     '''
 104
 105     _TMP_FILE_NAMES = ['script', 'html', 'cookies']
 106
 107     @staticmethod
 108     def _version():
 109         return get_exe_version('phantomjs', version_re=r'([0-9.]+)')
 110
 111     def __init__(self, extractor, required_version=None, timeout=10000):
 112         self._TMP_FILES = {}
 113
 114         self.exe = check_executable('phantomjs', ['-v'])
 115         if not self.exe:
 116             raise ExtractorError('PhantomJS executable not found in PATH, '
 117                                  'download it from http://phantomjs.org',
 118                                  expected=True)
 119
 120         self.extractor = extractor
 121
 122         if required_version:
 123             version = self._version()
 124             if is_outdated_version(version, required_version):
 125                 self.extractor._downloader.report_warning(
 126                     'Your copy of PhantomJS is outdated, update it to version '
 127                     '%s or newer if you encounter any errors.' % required_version)
 128
 129         self.options = {
 130             'timeout': timeout,
 131         }
 132         for name in self._TMP_FILE_NAMES:
 133             tmp = tempfile.NamedTemporaryFile(delete=False)
 134             tmp.close()
 135             self._TMP_FILES[name] = tmp
 136
 137     def __del__(self):
 138         for name in self._TMP_FILE_NAMES:
 139             try:
 140                 os.remove(self._TMP_FILES[name].name)
 141             except (IOError, OSError, KeyError):
 142                 pass
 143
 144     def _save_cookies(self, url):
 145         cookies = cookie_jar_to_list(self.extractor._downloader.cookiejar)
 146         for cookie in cookies:
 147             if 'path' not in cookie:
 148                 cookie['path'] = '/'
 149             if 'domain' not in cookie:
 150                 cookie['domain'] = compat_urlparse.urlparse(url).netloc
 151         with open(self._TMP_FILES['cookies'].name, 'wb') as f:
 152             f.write(json.dumps(cookies).encode('utf-8'))
 153
 154     def _load_cookies(self):
 155         with open(self._TMP_FILES['cookies'].name, 'rb') as f:
 156             cookies = json.loads(f.read().decode('utf-8'))
 157         for cookie in cookies:
 158             if cookie['httponly'] is True:
 159                 cookie['rest'] = {'httpOnly': None}
 160             if 'expiry' in cookie:
 161                 cookie['expire_time'] = cookie['expiry']
 162             self.extractor._set_cookie(**compat_kwargs(cookie))
 163
 164     def get(self, url, html=None, video_id=None, note=None, note2='Executing JS on webpage', headers={}, jscode='saveAndExit();'):
 165         """
 166         Downloads webpage (if needed) and executes JS
 167
 168         Params:
 169             url: website url
 170             html: optional, html code of website
 171             video_id: video id
 172             note: optional, displayed when downloading webpage
 173             note2: optional, displayed when executing JS
 174             headers: custom http headers
 175             jscode: code to be executed when page is loaded
 176
 177         Returns tuple with:
 178             * downloaded website (after JS execution)
 179             * anything you print with `console.log` (but not inside `page.execute`!)
 180
 181         In most cases you don't need to add any `jscode`.
 182         It is executed in `page.onLoadFinished`.
 183         `saveAndExit();` is mandatory, use it instead of `phantom.exit()`
 184         It is possible to wait for some element on the webpage, for example:
 185             var check = function() {
 186               var elementFound = page.evaluate(function() {
 187                 return document.querySelector('#b.done') !== null;
 188               });
 189               if(elementFound)
 190                 saveAndExit();
 191               else
 192                 window.setTimeout(check, 500);
 193             }
 194
 195             page.evaluate(function(){
 196               document.querySelector('#a').click();
 197             });
 198             check();
 199         """
 200         if 'saveAndExit();' not in jscode:
 201             raise ExtractorError('`saveAndExit();` not found in `jscode`')
 202         if not html:
 203             html = self.extractor._download_webpage(url, video_id, note=note, headers=headers)
 204         with open(self._TMP_FILES['html'].name, 'wb') as f:
 205             f.write(html.encode('utf-8'))
 206
 207         self._save_cookies(url)
 208
 209         replaces = self.options
 210         replaces['url'] = url
 211         user_agent = headers.get('User-Agent') or std_headers['User-Agent']
 212         replaces['ua'] = user_agent.replace('"', '\\"')
 213         replaces['jscode'] = jscode
 214
 215         for x in self._TMP_FILE_NAMES:
 216             replaces[x] = self._TMP_FILES[x].name.replace('\\', '\\\\').replace('"', '\\"')
 217
 218         with open(self._TMP_FILES['script'].name, 'wb') as f:
 219             f.write(self._TEMPLATE.format(**replaces).encode('utf-8'))
 220
 221         if video_id is None:
 222             self.extractor.to_screen('%s' % (note2,))
 223         else:
 224             self.extractor.to_screen('%s: %s' % (video_id, note2))
 225
 226         p = subprocess.Popen([
 227             self.exe, '--ssl-protocol=any',
 228             self._TMP_FILES['script'].name
 229         ], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 230         out, err = process_communicate_or_kill(p)
 231         if p.returncode != 0:
 232             raise ExtractorError(
 233                 'Executing JS failed\n:' + encodeArgument(err))
 234         with open(self._TMP_FILES['html'].name, 'rb') as f:
 235             html = f.read().decode('utf-8')
 236
 237         self._load_cookies()
 238
 239         return (html, encodeArgument(out))