yt_dlp/cookies.py

   1 import contextlib
   2 import ctypes
   3 import json
   4 import os
   5 import shutil
   6 import struct
   7 import subprocess
   8 import sys
   9 import tempfile
  10 import time
  11 from datetime import datetime, timedelta, timezone
  12 from enum import Enum, auto
  13 from hashlib import pbkdf2_hmac
  14
  15 from .aes import (
  16     aes_cbc_decrypt_bytes,
  17     aes_gcm_decrypt_and_verify_bytes,
  18     unpad_pkcs7,
  19 )
  20 from .compat import compat_b64decode, compat_cookiejar_Cookie
  21 from .dependencies import (
  22     _SECRETSTORAGE_UNAVAILABLE_REASON,
  23     secretstorage,
  24     sqlite3,
  25 )
  26 from .minicurses import MultilinePrinter, QuietMultilinePrinter
  27 from .utils import Popen, YoutubeDLCookieJar, error_to_str, expand_path
  28
  29 CHROMIUM_BASED_BROWSERS = {'brave', 'chrome', 'chromium', 'edge', 'opera', 'vivaldi'}
  30 SUPPORTED_BROWSERS = CHROMIUM_BASED_BROWSERS | {'firefox', 'safari'}
  31
  32
  33 class YDLLogger:
  34     def __init__(self, ydl=None):
  35         self._ydl = ydl
  36
  37     def debug(self, message):
  38         if self._ydl:
  39             self._ydl.write_debug(message)
  40
  41     def info(self, message):
  42         if self._ydl:
  43             self._ydl.to_screen(f'[Cookies] {message}')
  44
  45     def warning(self, message, only_once=False):
  46         if self._ydl:
  47             self._ydl.report_warning(message, only_once)
  48
  49     def error(self, message):
  50         if self._ydl:
  51             self._ydl.report_error(message)
  52
  53     class ProgressBar(MultilinePrinter):
  54         _DELAY, _timer = 0.1, 0
  55
  56         def print(self, message):
  57             if time.time() - self._timer > self._DELAY:
  58                 self.print_at_line(f'[Cookies] {message}', 0)
  59                 self._timer = time.time()
  60
  61     def progress_bar(self):
  62         """Return a context manager with a print method. (Optional)"""
  63         # Do not print to files/pipes, loggers, or when --no-progress is used
  64         if not self._ydl or self._ydl.params.get('noprogress') or self._ydl.params.get('logger'):
  65             return
  66         file = self._ydl._out_files.error
  67         try:
  68             if not file.isatty():
  69                 return
  70         except BaseException:
  71             return
  72         return self.ProgressBar(file, preserve_output=False)
  73
  74
  75 def _create_progress_bar(logger):
  76     if hasattr(logger, 'progress_bar'):
  77         printer = logger.progress_bar()
  78         if printer:
  79             return printer
  80     printer = QuietMultilinePrinter()
  81     printer.print = lambda _: None
  82     return printer
  83
  84
  85 def load_cookies(cookie_file, browser_specification, ydl):
  86     cookie_jars = []
  87     if browser_specification is not None:
  88         browser_name, profile, keyring = _parse_browser_specification(*browser_specification)
  89         cookie_jars.append(extract_cookies_from_browser(browser_name, profile, YDLLogger(ydl), keyring=keyring))
  90
  91     if cookie_file is not None:
  92         is_filename = YoutubeDLCookieJar.is_path(cookie_file)
  93         if is_filename:
  94             cookie_file = expand_path(cookie_file)
  95
  96         jar = YoutubeDLCookieJar(cookie_file)
  97         if not is_filename or os.access(cookie_file, os.R_OK):
  98             jar.load(ignore_discard=True, ignore_expires=True)
  99         cookie_jars.append(jar)
 100
 101     return _merge_cookie_jars(cookie_jars)
 102
 103
 104 def extract_cookies_from_browser(browser_name, profile=None, logger=YDLLogger(), *, keyring=None):
 105     if browser_name == 'firefox':
 106         return _extract_firefox_cookies(profile, logger)
 107     elif browser_name == 'safari':
 108         return _extract_safari_cookies(profile, logger)
 109     elif browser_name in CHROMIUM_BASED_BROWSERS:
 110         return _extract_chrome_cookies(browser_name, profile, keyring, logger)
 111     else:
 112         raise ValueError(f'unknown browser: {browser_name}')
 113
 114
 115 def _extract_firefox_cookies(profile, logger):
 116     logger.info('Extracting cookies from firefox')
 117     if not sqlite3:
 118         logger.warning('Cannot extract cookies from firefox without sqlite3 support. '
 119                        'Please use a python interpreter compiled with sqlite3 support')
 120         return YoutubeDLCookieJar()
 121
 122     if profile is None:
 123         search_root = _firefox_browser_dir()
 124     elif _is_path(profile):
 125         search_root = profile
 126     else:
 127         search_root = os.path.join(_firefox_browser_dir(), profile)
 128
 129     cookie_database_path = _find_most_recently_used_file(search_root, 'cookies.sqlite', logger)
 130     if cookie_database_path is None:
 131         raise FileNotFoundError(f'could not find firefox cookies database in {search_root}')
 132     logger.debug(f'Extracting cookies from: "{cookie_database_path}"')
 133
 134     with tempfile.TemporaryDirectory(prefix='yt_dlp') as tmpdir:
 135         cursor = None
 136         try:
 137             cursor = _open_database_copy(cookie_database_path, tmpdir)
 138             cursor.execute('SELECT host, name, value, path, expiry, isSecure FROM moz_cookies')
 139             jar = YoutubeDLCookieJar()
 140             with _create_progress_bar(logger) as progress_bar:
 141                 table = cursor.fetchall()
 142                 total_cookie_count = len(table)
 143                 for i, (host, name, value, path, expiry, is_secure) in enumerate(table):
 144                     progress_bar.print(f'Loading cookie {i: 6d}/{total_cookie_count: 6d}')
 145                     cookie = compat_cookiejar_Cookie(
 146                         version=0, name=name, value=value, port=None, port_specified=False,
 147                         domain=host, domain_specified=bool(host), domain_initial_dot=host.startswith('.'),
 148                         path=path, path_specified=bool(path), secure=is_secure, expires=expiry, discard=False,
 149                         comment=None, comment_url=None, rest={})
 150                     jar.set_cookie(cookie)
 151             logger.info(f'Extracted {len(jar)} cookies from firefox')
 152             return jar
 153         finally:
 154             if cursor is not None:
 155                 cursor.connection.close()
 156
 157
 158 def _firefox_browser_dir():
 159     if sys.platform in ('linux', 'linux2'):
 160         return os.path.expanduser('~/.mozilla/firefox')
 161     elif sys.platform == 'win32':
 162         return os.path.expandvars(R'%APPDATA%\Mozilla\Firefox\Profiles')
 163     elif sys.platform == 'darwin':
 164         return os.path.expanduser('~/Library/Application Support/Firefox')
 165     else:
 166         raise ValueError(f'unsupported platform: {sys.platform}')
 167
 168
 169 def _get_chromium_based_browser_settings(browser_name):
 170     # https://chromium.googlesource.com/chromium/src/+/HEAD/docs/user_data_dir.md
 171     if sys.platform in ('linux', 'linux2'):
 172         config = _config_home()
 173         browser_dir = {
 174             'brave': os.path.join(config, 'BraveSoftware/Brave-Browser'),
 175             'chrome': os.path.join(config, 'google-chrome'),
 176             'chromium': os.path.join(config, 'chromium'),
 177             'edge': os.path.join(config, 'microsoft-edge'),
 178             'opera': os.path.join(config, 'opera'),
 179             'vivaldi': os.path.join(config, 'vivaldi'),
 180         }[browser_name]
 181
 182     elif sys.platform == 'win32':
 183         appdata_local = os.path.expandvars('%LOCALAPPDATA%')
 184         appdata_roaming = os.path.expandvars('%APPDATA%')
 185         browser_dir = {
 186             'brave': os.path.join(appdata_local, R'BraveSoftware\Brave-Browser\User Data'),
 187             'chrome': os.path.join(appdata_local, R'Google\Chrome\User Data'),
 188             'chromium': os.path.join(appdata_local, R'Chromium\User Data'),
 189             'edge': os.path.join(appdata_local, R'Microsoft\Edge\User Data'),
 190             'opera': os.path.join(appdata_roaming, R'Opera Software\Opera Stable'),
 191             'vivaldi': os.path.join(appdata_local, R'Vivaldi\User Data'),
 192         }[browser_name]
 193
 194     elif sys.platform == 'darwin':
 195         appdata = os.path.expanduser('~/Library/Application Support')
 196         browser_dir = {
 197             'brave': os.path.join(appdata, 'BraveSoftware/Brave-Browser'),
 198             'chrome': os.path.join(appdata, 'Google/Chrome'),
 199             'chromium': os.path.join(appdata, 'Chromium'),
 200             'edge': os.path.join(appdata, 'Microsoft Edge'),
 201             'opera': os.path.join(appdata, 'com.operasoftware.Opera'),
 202             'vivaldi': os.path.join(appdata, 'Vivaldi'),
 203         }[browser_name]
 204
 205     else:
 206         raise ValueError(f'unsupported platform: {sys.platform}')
 207
 208     # Linux keyring names can be determined by snooping on dbus while opening the browser in KDE:
 209     # dbus-monitor "interface='org.kde.KWallet'" "type=method_return"
 210     keyring_name = {
 211         'brave': 'Brave',
 212         'chrome': 'Chrome',
 213         'chromium': 'Chromium',
 214         'edge': 'Microsoft Edge' if sys.platform == 'darwin' else 'Chromium',
 215         'opera': 'Opera' if sys.platform == 'darwin' else 'Chromium',
 216         'vivaldi': 'Vivaldi' if sys.platform == 'darwin' else 'Chrome',
 217     }[browser_name]
 218
 219     browsers_without_profiles = {'opera'}
 220
 221     return {
 222         'browser_dir': browser_dir,
 223         'keyring_name': keyring_name,
 224         'supports_profiles': browser_name not in browsers_without_profiles
 225     }
 226
 227
 228 def _extract_chrome_cookies(browser_name, profile, keyring, logger):
 229     logger.info(f'Extracting cookies from {browser_name}')
 230
 231     if not sqlite3:
 232         logger.warning(f'Cannot extract cookies from {browser_name} without sqlite3 support. '
 233                        'Please use a python interpreter compiled with sqlite3 support')
 234         return YoutubeDLCookieJar()
 235
 236     config = _get_chromium_based_browser_settings(browser_name)
 237
 238     if profile is None:
 239         search_root = config['browser_dir']
 240     elif _is_path(profile):
 241         search_root = profile
 242         config['browser_dir'] = os.path.dirname(profile) if config['supports_profiles'] else profile
 243     else:
 244         if config['supports_profiles']:
 245             search_root = os.path.join(config['browser_dir'], profile)
 246         else:
 247             logger.error(f'{browser_name} does not support profiles')
 248             search_root = config['browser_dir']
 249
 250     cookie_database_path = _find_most_recently_used_file(search_root, 'Cookies', logger)
 251     if cookie_database_path is None:
 252         raise FileNotFoundError(f'could not find {browser_name} cookies database in "{search_root}"')
 253     logger.debug(f'Extracting cookies from: "{cookie_database_path}"')
 254
 255     decryptor = get_cookie_decryptor(config['browser_dir'], config['keyring_name'], logger, keyring=keyring)
 256
 257     with tempfile.TemporaryDirectory(prefix='yt_dlp') as tmpdir:
 258         cursor = None
 259         try:
 260             cursor = _open_database_copy(cookie_database_path, tmpdir)
 261             cursor.connection.text_factory = bytes
 262             column_names = _get_column_names(cursor, 'cookies')
 263             secure_column = 'is_secure' if 'is_secure' in column_names else 'secure'
 264             cursor.execute(f'SELECT host_key, name, value, encrypted_value, path, expires_utc, {secure_column} FROM cookies')
 265             jar = YoutubeDLCookieJar()
 266             failed_cookies = 0
 267             unencrypted_cookies = 0
 268             with _create_progress_bar(logger) as progress_bar:
 269                 table = cursor.fetchall()
 270                 total_cookie_count = len(table)
 271                 for i, line in enumerate(table):
 272                     progress_bar.print(f'Loading cookie {i: 6d}/{total_cookie_count: 6d}')
 273                     is_encrypted, cookie = _process_chrome_cookie(decryptor, *line)
 274                     if not cookie:
 275                         failed_cookies += 1
 276                         continue
 277                     elif not is_encrypted:
 278                         unencrypted_cookies += 1
 279                     jar.set_cookie(cookie)
 280             if failed_cookies > 0:
 281                 failed_message = f' ({failed_cookies} could not be decrypted)'
 282             else:
 283                 failed_message = ''
 284             logger.info(f'Extracted {len(jar)} cookies from {browser_name}{failed_message}')
 285             counts = decryptor._cookie_counts.copy()
 286             counts['unencrypted'] = unencrypted_cookies
 287             logger.debug(f'cookie version breakdown: {counts}')
 288             return jar
 289         finally:
 290             if cursor is not None:
 291                 cursor.connection.close()
 292
 293
 294 def _process_chrome_cookie(decryptor, host_key, name, value, encrypted_value, path, expires_utc, is_secure):
 295     host_key = host_key.decode()
 296     name = name.decode()
 297     value = value.decode()
 298     path = path.decode()
 299     is_encrypted = not value and encrypted_value
 300
 301     if is_encrypted:
 302         value = decryptor.decrypt(encrypted_value)
 303         if value is None:
 304             return is_encrypted, None
 305
 306     return is_encrypted, compat_cookiejar_Cookie(
 307         version=0, name=name, value=value, port=None, port_specified=False,
 308         domain=host_key, domain_specified=bool(host_key), domain_initial_dot=host_key.startswith('.'),
 309         path=path, path_specified=bool(path), secure=is_secure, expires=expires_utc, discard=False,
 310         comment=None, comment_url=None, rest={})
 311
 312
 313 class ChromeCookieDecryptor:
 314     """
 315     Overview:
 316
 317         Linux:
 318         - cookies are either v10 or v11
 319             - v10: AES-CBC encrypted with a fixed key
 320             - v11: AES-CBC encrypted with an OS protected key (keyring)
 321             - v11 keys can be stored in various places depending on the activate desktop environment [2]
 322
 323         Mac:
 324         - cookies are either v10 or not v10
 325             - v10: AES-CBC encrypted with an OS protected key (keyring) and more key derivation iterations than linux
 326             - not v10: 'old data' stored as plaintext
 327
 328         Windows:
 329         - cookies are either v10 or not v10
 330             - v10: AES-GCM encrypted with a key which is encrypted with DPAPI
 331             - not v10: encrypted with DPAPI
 332
 333     Sources:
 334     - [1] https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/
 335     - [2] https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/key_storage_linux.cc
 336         - KeyStorageLinux::CreateService
 337     """
 338
 339     _cookie_counts = {}
 340
 341     def decrypt(self, encrypted_value):
 342         raise NotImplementedError('Must be implemented by sub classes')
 343
 344
 345 def get_cookie_decryptor(browser_root, browser_keyring_name, logger, *, keyring=None):
 346     if sys.platform in ('linux', 'linux2'):
 347         return LinuxChromeCookieDecryptor(browser_keyring_name, logger, keyring=keyring)
 348     elif sys.platform == 'darwin':
 349         return MacChromeCookieDecryptor(browser_keyring_name, logger)
 350     elif sys.platform == 'win32':
 351         return WindowsChromeCookieDecryptor(browser_root, logger)
 352     else:
 353         raise NotImplementedError(f'Chrome cookie decryption is not supported on this platform: {sys.platform}')
 354
 355
 356 class LinuxChromeCookieDecryptor(ChromeCookieDecryptor):
 357     def __init__(self, browser_keyring_name, logger, *, keyring=None):
 358         self._logger = logger
 359         self._v10_key = self.derive_key(b'peanuts')
 360         password = _get_linux_keyring_password(browser_keyring_name, keyring, logger)
 361         self._v11_key = None if password is None else self.derive_key(password)
 362         self._cookie_counts = {'v10': 0, 'v11': 0, 'other': 0}
 363
 364     @staticmethod
 365     def derive_key(password):
 366         # values from
 367         # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_linux.cc
 368         return pbkdf2_sha1(password, salt=b'saltysalt', iterations=1, key_length=16)
 369
 370     def decrypt(self, encrypted_value):
 371         version = encrypted_value[:3]
 372         ciphertext = encrypted_value[3:]
 373
 374         if version == b'v10':
 375             self._cookie_counts['v10'] += 1
 376             return _decrypt_aes_cbc(ciphertext, self._v10_key, self._logger)
 377
 378         elif version == b'v11':
 379             self._cookie_counts['v11'] += 1
 380             if self._v11_key is None:
 381                 self._logger.warning('cannot decrypt v11 cookies: no key found', only_once=True)
 382                 return None
 383             return _decrypt_aes_cbc(ciphertext, self._v11_key, self._logger)
 384
 385         else:
 386             self._cookie_counts['other'] += 1
 387             return None
 388
 389
 390 class MacChromeCookieDecryptor(ChromeCookieDecryptor):
 391     def __init__(self, browser_keyring_name, logger):
 392         self._logger = logger
 393         password = _get_mac_keyring_password(browser_keyring_name, logger)
 394         self._v10_key = None if password is None else self.derive_key(password)
 395         self._cookie_counts = {'v10': 0, 'other': 0}
 396
 397     @staticmethod
 398     def derive_key(password):
 399         # values from
 400         # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_mac.mm
 401         return pbkdf2_sha1(password, salt=b'saltysalt', iterations=1003, key_length=16)
 402
 403     def decrypt(self, encrypted_value):
 404         version = encrypted_value[:3]
 405         ciphertext = encrypted_value[3:]
 406
 407         if version == b'v10':
 408             self._cookie_counts['v10'] += 1
 409             if self._v10_key is None:
 410                 self._logger.warning('cannot decrypt v10 cookies: no key found', only_once=True)
 411                 return None
 412
 413             return _decrypt_aes_cbc(ciphertext, self._v10_key, self._logger)
 414
 415         else:
 416             self._cookie_counts['other'] += 1
 417             # other prefixes are considered 'old data' which were stored as plaintext
 418             # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_mac.mm
 419             return encrypted_value
 420
 421
 422 class WindowsChromeCookieDecryptor(ChromeCookieDecryptor):
 423     def __init__(self, browser_root, logger):
 424         self._logger = logger
 425         self._v10_key = _get_windows_v10_key(browser_root, logger)
 426         self._cookie_counts = {'v10': 0, 'other': 0}
 427
 428     def decrypt(self, encrypted_value):
 429         version = encrypted_value[:3]
 430         ciphertext = encrypted_value[3:]
 431
 432         if version == b'v10':
 433             self._cookie_counts['v10'] += 1
 434             if self._v10_key is None:
 435                 self._logger.warning('cannot decrypt v10 cookies: no key found', only_once=True)
 436                 return None
 437
 438             # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_win.cc
 439             #   kNonceLength
 440             nonce_length = 96 // 8
 441             # boringssl
 442             #   EVP_AEAD_AES_GCM_TAG_LEN
 443             authentication_tag_length = 16
 444
 445             raw_ciphertext = ciphertext
 446             nonce = raw_ciphertext[:nonce_length]
 447             ciphertext = raw_ciphertext[nonce_length:-authentication_tag_length]
 448             authentication_tag = raw_ciphertext[-authentication_tag_length:]
 449
 450             return _decrypt_aes_gcm(ciphertext, self._v10_key, nonce, authentication_tag, self._logger)
 451
 452         else:
 453             self._cookie_counts['other'] += 1
 454             # any other prefix means the data is DPAPI encrypted
 455             # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_win.cc
 456             return _decrypt_windows_dpapi(encrypted_value, self._logger).decode()
 457
 458
 459 def _extract_safari_cookies(profile, logger):
 460     if profile is not None:
 461         logger.error('safari does not support profiles')
 462     if sys.platform != 'darwin':
 463         raise ValueError(f'unsupported platform: {sys.platform}')
 464
 465     cookies_path = os.path.expanduser('~/Library/Cookies/Cookies.binarycookies')
 466
 467     if not os.path.isfile(cookies_path):
 468         logger.debug('Trying secondary cookie location')
 469         cookies_path = os.path.expanduser('~/Library/Containers/com.apple.Safari/Data/Library/Cookies/Cookies.binarycookies')
 470         if not os.path.isfile(cookies_path):
 471             raise FileNotFoundError('could not find safari cookies database')
 472
 473     with open(cookies_path, 'rb') as f:
 474         cookies_data = f.read()
 475
 476     jar = parse_safari_cookies(cookies_data, logger=logger)
 477     logger.info(f'Extracted {len(jar)} cookies from safari')
 478     return jar
 479
 480
 481 class ParserError(Exception):
 482     pass
 483
 484
 485 class DataParser:
 486     def __init__(self, data, logger):
 487         self._data = data
 488         self.cursor = 0
 489         self._logger = logger
 490
 491     def read_bytes(self, num_bytes):
 492         if num_bytes < 0:
 493             raise ParserError(f'invalid read of {num_bytes} bytes')
 494         end = self.cursor + num_bytes
 495         if end > len(self._data):
 496             raise ParserError('reached end of input')
 497         data = self._data[self.cursor:end]
 498         self.cursor = end
 499         return data
 500
 501     def expect_bytes(self, expected_value, message):
 502         value = self.read_bytes(len(expected_value))
 503         if value != expected_value:
 504             raise ParserError(f'unexpected value: {value} != {expected_value} ({message})')
 505
 506     def read_uint(self, big_endian=False):
 507         data_format = '>I' if big_endian else '<I'
 508         return struct.unpack(data_format, self.read_bytes(4))[0]
 509
 510     def read_double(self, big_endian=False):
 511         data_format = '>d' if big_endian else '<d'
 512         return struct.unpack(data_format, self.read_bytes(8))[0]
 513
 514     def read_cstring(self):
 515         buffer = []
 516         while True:
 517             c = self.read_bytes(1)
 518             if c == b'\x00':
 519                 return b''.join(buffer).decode()
 520             else:
 521                 buffer.append(c)
 522
 523     def skip(self, num_bytes, description='unknown'):
 524         if num_bytes > 0:
 525             self._logger.debug(f'skipping {num_bytes} bytes ({description}): {self.read_bytes(num_bytes)!r}')
 526         elif num_bytes < 0:
 527             raise ParserError(f'invalid skip of {num_bytes} bytes')
 528
 529     def skip_to(self, offset, description='unknown'):
 530         self.skip(offset - self.cursor, description)
 531
 532     def skip_to_end(self, description='unknown'):
 533         self.skip_to(len(self._data), description)
 534
 535
 536 def _mac_absolute_time_to_posix(timestamp):
 537     return int((datetime(2001, 1, 1, 0, 0, tzinfo=timezone.utc) + timedelta(seconds=timestamp)).timestamp())
 538
 539
 540 def _parse_safari_cookies_header(data, logger):
 541     p = DataParser(data, logger)
 542     p.expect_bytes(b'cook', 'database signature')
 543     number_of_pages = p.read_uint(big_endian=True)
 544     page_sizes = [p.read_uint(big_endian=True) for _ in range(number_of_pages)]
 545     return page_sizes, p.cursor
 546
 547
 548 def _parse_safari_cookies_page(data, jar, logger):
 549     p = DataParser(data, logger)
 550     p.expect_bytes(b'\x00\x00\x01\x00', 'page signature')
 551     number_of_cookies = p.read_uint()
 552     record_offsets = [p.read_uint() for _ in range(number_of_cookies)]
 553     if number_of_cookies == 0:
 554         logger.debug(f'a cookies page of size {len(data)} has no cookies')
 555         return
 556
 557     p.skip_to(record_offsets[0], 'unknown page header field')
 558
 559     with _create_progress_bar(logger) as progress_bar:
 560         for i, record_offset in enumerate(record_offsets):
 561             progress_bar.print(f'Loading cookie {i: 6d}/{number_of_cookies: 6d}')
 562             p.skip_to(record_offset, 'space between records')
 563             record_length = _parse_safari_cookies_record(data[record_offset:], jar, logger)
 564             p.read_bytes(record_length)
 565     p.skip_to_end('space in between pages')
 566
 567
 568 def _parse_safari_cookies_record(data, jar, logger):
 569     p = DataParser(data, logger)
 570     record_size = p.read_uint()
 571     p.skip(4, 'unknown record field 1')
 572     flags = p.read_uint()
 573     is_secure = bool(flags & 0x0001)
 574     p.skip(4, 'unknown record field 2')
 575     domain_offset = p.read_uint()
 576     name_offset = p.read_uint()
 577     path_offset = p.read_uint()
 578     value_offset = p.read_uint()
 579     p.skip(8, 'unknown record field 3')
 580     expiration_date = _mac_absolute_time_to_posix(p.read_double())
 581     _creation_date = _mac_absolute_time_to_posix(p.read_double())  # noqa: F841
 582
 583     try:
 584         p.skip_to(domain_offset)
 585         domain = p.read_cstring()
 586
 587         p.skip_to(name_offset)
 588         name = p.read_cstring()
 589
 590         p.skip_to(path_offset)
 591         path = p.read_cstring()
 592
 593         p.skip_to(value_offset)
 594         value = p.read_cstring()
 595     except UnicodeDecodeError:
 596         logger.warning('failed to parse Safari cookie because UTF-8 decoding failed', only_once=True)
 597         return record_size
 598
 599     p.skip_to(record_size, 'space at the end of the record')
 600
 601     cookie = compat_cookiejar_Cookie(
 602         version=0, name=name, value=value, port=None, port_specified=False,
 603         domain=domain, domain_specified=bool(domain), domain_initial_dot=domain.startswith('.'),
 604         path=path, path_specified=bool(path), secure=is_secure, expires=expiration_date, discard=False,
 605         comment=None, comment_url=None, rest={})
 606     jar.set_cookie(cookie)
 607     return record_size
 608
 609
 610 def parse_safari_cookies(data, jar=None, logger=YDLLogger()):
 611     """
 612     References:
 613         - https://github.com/libyal/dtformats/blob/main/documentation/Safari%20Cookies.asciidoc
 614             - this data appears to be out of date but the important parts of the database structure is the same
 615             - there are a few bytes here and there which are skipped during parsing
 616     """
 617     if jar is None:
 618         jar = YoutubeDLCookieJar()
 619     page_sizes, body_start = _parse_safari_cookies_header(data, logger)
 620     p = DataParser(data[body_start:], logger)
 621     for page_size in page_sizes:
 622         _parse_safari_cookies_page(p.read_bytes(page_size), jar, logger)
 623     p.skip_to_end('footer')
 624     return jar
 625
 626
 627 class _LinuxDesktopEnvironment(Enum):
 628     """
 629     https://chromium.googlesource.com/chromium/src/+/refs/heads/main/base/nix/xdg_util.h
 630     DesktopEnvironment
 631     """
 632     OTHER = auto()
 633     CINNAMON = auto()
 634     GNOME = auto()
 635     KDE = auto()
 636     PANTHEON = auto()
 637     UNITY = auto()
 638     XFCE = auto()
 639
 640
 641 class _LinuxKeyring(Enum):
 642     """
 643     https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/key_storage_util_linux.h
 644     SelectedLinuxBackend
 645     """
 646     KWALLET = auto()
 647     GNOMEKEYRING = auto()
 648     BASICTEXT = auto()
 649
 650
 651 SUPPORTED_KEYRINGS = _LinuxKeyring.__members__.keys()
 652
 653
 654 def _get_linux_desktop_environment(env):
 655     """
 656     https://chromium.googlesource.com/chromium/src/+/refs/heads/main/base/nix/xdg_util.cc
 657     GetDesktopEnvironment
 658     """
 659     xdg_current_desktop = env.get('XDG_CURRENT_DESKTOP', None)
 660     desktop_session = env.get('DESKTOP_SESSION', None)
 661     if xdg_current_desktop is not None:
 662         xdg_current_desktop = xdg_current_desktop.split(':')[0].strip()
 663
 664         if xdg_current_desktop == 'Unity':
 665             if desktop_session is not None and 'gnome-fallback' in desktop_session:
 666                 return _LinuxDesktopEnvironment.GNOME
 667             else:
 668                 return _LinuxDesktopEnvironment.UNITY
 669         elif xdg_current_desktop == 'GNOME':
 670             return _LinuxDesktopEnvironment.GNOME
 671         elif xdg_current_desktop == 'X-Cinnamon':
 672             return _LinuxDesktopEnvironment.CINNAMON
 673         elif xdg_current_desktop == 'KDE':
 674             return _LinuxDesktopEnvironment.KDE
 675         elif xdg_current_desktop == 'Pantheon':
 676             return _LinuxDesktopEnvironment.PANTHEON
 677         elif xdg_current_desktop == 'XFCE':
 678             return _LinuxDesktopEnvironment.XFCE
 679     elif desktop_session is not None:
 680         if desktop_session in ('mate', 'gnome'):
 681             return _LinuxDesktopEnvironment.GNOME
 682         elif 'kde' in desktop_session:
 683             return _LinuxDesktopEnvironment.KDE
 684         elif 'xfce' in desktop_session:
 685             return _LinuxDesktopEnvironment.XFCE
 686     else:
 687         if 'GNOME_DESKTOP_SESSION_ID' in env:
 688             return _LinuxDesktopEnvironment.GNOME
 689         elif 'KDE_FULL_SESSION' in env:
 690             return _LinuxDesktopEnvironment.KDE
 691     return _LinuxDesktopEnvironment.OTHER
 692
 693
 694 def _choose_linux_keyring(logger):
 695     """
 696     https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/key_storage_util_linux.cc
 697     SelectBackend
 698     """
 699     desktop_environment = _get_linux_desktop_environment(os.environ)
 700     logger.debug(f'detected desktop environment: {desktop_environment.name}')
 701     if desktop_environment == _LinuxDesktopEnvironment.KDE:
 702         linux_keyring = _LinuxKeyring.KWALLET
 703     elif desktop_environment == _LinuxDesktopEnvironment.OTHER:
 704         linux_keyring = _LinuxKeyring.BASICTEXT
 705     else:
 706         linux_keyring = _LinuxKeyring.GNOMEKEYRING
 707     return linux_keyring
 708
 709
 710 def _get_kwallet_network_wallet(logger):
 711     """ The name of the wallet used to store network passwords.
 712
 713     https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/kwallet_dbus.cc
 714     KWalletDBus::NetworkWallet
 715     which does a dbus call to the following function:
 716     https://api.kde.org/frameworks/kwallet/html/classKWallet_1_1Wallet.html
 717     Wallet::NetworkWallet
 718     """
 719     default_wallet = 'kdewallet'
 720     try:
 721         proc = Popen([
 722             'dbus-send', '--session', '--print-reply=literal',
 723             '--dest=org.kde.kwalletd5',
 724             '/modules/kwalletd5',
 725             'org.kde.KWallet.networkWallet'
 726         ], stdout=subprocess.PIPE, stderr=subprocess.DEVNULL)
 727
 728         stdout, stderr = proc.communicate_or_kill()
 729         if proc.returncode != 0:
 730             logger.warning('failed to read NetworkWallet')
 731             return default_wallet
 732         else:
 733             network_wallet = stdout.decode().strip()
 734             logger.debug(f'NetworkWallet = "{network_wallet}"')
 735             return network_wallet
 736     except Exception as e:
 737         logger.warning(f'exception while obtaining NetworkWallet: {e}')
 738         return default_wallet
 739
 740
 741 def _get_kwallet_password(browser_keyring_name, logger):
 742     logger.debug('using kwallet-query to obtain password from kwallet')
 743
 744     if shutil.which('kwallet-query') is None:
 745         logger.error('kwallet-query command not found. KWallet and kwallet-query '
 746                      'must be installed to read from KWallet. kwallet-query should be'
 747                      'included in the kwallet package for your distribution')
 748         return b''
 749
 750     network_wallet = _get_kwallet_network_wallet(logger)
 751
 752     try:
 753         proc = Popen([
 754             'kwallet-query',
 755             '--read-password', f'{browser_keyring_name} Safe Storage',
 756             '--folder', f'{browser_keyring_name} Keys',
 757             network_wallet
 758         ], stdout=subprocess.PIPE, stderr=subprocess.DEVNULL)
 759
 760         stdout, stderr = proc.communicate_or_kill()
 761         if proc.returncode != 0:
 762             logger.error(f'kwallet-query failed with return code {proc.returncode}. Please consult '
 763                          'the kwallet-query man page for details')
 764             return b''
 765         else:
 766             if stdout.lower().startswith(b'failed to read'):
 767                 logger.debug('failed to read password from kwallet. Using empty string instead')
 768                 # this sometimes occurs in KDE because chrome does not check hasEntry and instead
 769                 # just tries to read the value (which kwallet returns "") whereas kwallet-query
 770                 # checks hasEntry. To verify this:
 771                 # dbus-monitor "interface='org.kde.KWallet'" "type=method_return"
 772                 # while starting chrome.
 773                 # this may be a bug as the intended behaviour is to generate a random password and store
 774                 # it, but that doesn't matter here.
 775                 return b''
 776             else:
 777                 logger.debug('password found')
 778                 if stdout[-1:] == b'\n':
 779                     stdout = stdout[:-1]
 780                 return stdout
 781     except Exception as e:
 782         logger.warning(f'exception running kwallet-query: {error_to_str(e)}')
 783         return b''
 784
 785
 786 def _get_gnome_keyring_password(browser_keyring_name, logger):
 787     if not secretstorage:
 788         logger.error(f'secretstorage not available {_SECRETSTORAGE_UNAVAILABLE_REASON}')
 789         return b''
 790     # the Gnome keyring does not seem to organise keys in the same way as KWallet,
 791     # using `dbus-monitor` during startup, it can be observed that chromium lists all keys
 792     # and presumably searches for its key in the list. It appears that we must do the same.
 793     # https://github.com/jaraco/keyring/issues/556
 794     with contextlib.closing(secretstorage.dbus_init()) as con:
 795         col = secretstorage.get_default_collection(con)
 796         for item in col.get_all_items():
 797             if item.get_label() == f'{browser_keyring_name} Safe Storage':
 798                 return item.get_secret()
 799         else:
 800             logger.error('failed to read from keyring')
 801             return b''
 802
 803
 804 def _get_linux_keyring_password(browser_keyring_name, keyring, logger):
 805     # note: chrome/chromium can be run with the following flags to determine which keyring backend
 806     # it has chosen to use
 807     # chromium --enable-logging=stderr --v=1 2>&1 | grep key_storage_
 808     # Chromium supports a flag: --password-store=<basic|gnome|kwallet> so the automatic detection
 809     # will not be sufficient in all cases.
 810
 811     keyring = _LinuxKeyring[keyring] if keyring else _choose_linux_keyring(logger)
 812     logger.debug(f'Chosen keyring: {keyring.name}')
 813
 814     if keyring == _LinuxKeyring.KWALLET:
 815         return _get_kwallet_password(browser_keyring_name, logger)
 816     elif keyring == _LinuxKeyring.GNOMEKEYRING:
 817         return _get_gnome_keyring_password(browser_keyring_name, logger)
 818     elif keyring == _LinuxKeyring.BASICTEXT:
 819         # when basic text is chosen, all cookies are stored as v10 (so no keyring password is required)
 820         return None
 821     assert False, f'Unknown keyring {keyring}'
 822
 823
 824 def _get_mac_keyring_password(browser_keyring_name, logger):
 825     logger.debug('using find-generic-password to obtain password from OSX keychain')
 826     try:
 827         proc = Popen(
 828             ['security', 'find-generic-password',
 829              '-w',  # write password to stdout
 830              '-a', browser_keyring_name,  # match 'account'
 831              '-s', f'{browser_keyring_name} Safe Storage'],  # match 'service'
 832             stdout=subprocess.PIPE, stderr=subprocess.DEVNULL)
 833
 834         stdout, stderr = proc.communicate_or_kill()
 835         if stdout[-1:] == b'\n':
 836             stdout = stdout[:-1]
 837         return stdout
 838     except Exception as e:
 839         logger.warning(f'exception running find-generic-password: {error_to_str(e)}')
 840         return None
 841
 842
 843 def _get_windows_v10_key(browser_root, logger):
 844     path = _find_most_recently_used_file(browser_root, 'Local State', logger)
 845     if path is None:
 846         logger.error('could not find local state file')
 847         return None
 848     logger.debug(f'Found local state file at "{path}"')
 849     with open(path, encoding='utf8') as f:
 850         data = json.load(f)
 851     try:
 852         base64_key = data['os_crypt']['encrypted_key']
 853     except KeyError:
 854         logger.error('no encrypted key in Local State')
 855         return None
 856     encrypted_key = compat_b64decode(base64_key)
 857     prefix = b'DPAPI'
 858     if not encrypted_key.startswith(prefix):
 859         logger.error('invalid key')
 860         return None
 861     return _decrypt_windows_dpapi(encrypted_key[len(prefix):], logger)
 862
 863
 864 def pbkdf2_sha1(password, salt, iterations, key_length):
 865     return pbkdf2_hmac('sha1', password, salt, iterations, key_length)
 866
 867
 868 def _decrypt_aes_cbc(ciphertext, key, logger, initialization_vector=b' ' * 16):
 869     plaintext = unpad_pkcs7(aes_cbc_decrypt_bytes(ciphertext, key, initialization_vector))
 870     try:
 871         return plaintext.decode()
 872     except UnicodeDecodeError:
 873         logger.warning('failed to decrypt cookie (AES-CBC) because UTF-8 decoding failed. Possibly the key is wrong?', only_once=True)
 874         return None
 875
 876
 877 def _decrypt_aes_gcm(ciphertext, key, nonce, authentication_tag, logger):
 878     try:
 879         plaintext = aes_gcm_decrypt_and_verify_bytes(ciphertext, key, authentication_tag, nonce)
 880     except ValueError:
 881         logger.warning('failed to decrypt cookie (AES-GCM) because the MAC check failed. Possibly the key is wrong?', only_once=True)
 882         return None
 883
 884     try:
 885         return plaintext.decode()
 886     except UnicodeDecodeError:
 887         logger.warning('failed to decrypt cookie (AES-GCM) because UTF-8 decoding failed. Possibly the key is wrong?', only_once=True)
 888         return None
 889
 890
 891 def _decrypt_windows_dpapi(ciphertext, logger):
 892     """
 893     References:
 894         - https://docs.microsoft.com/en-us/windows/win32/api/dpapi/nf-dpapi-cryptunprotectdata
 895     """
 896     from ctypes.wintypes import DWORD
 897
 898     class DATA_BLOB(ctypes.Structure):
 899         _fields_ = [('cbData', DWORD),
 900                     ('pbData', ctypes.POINTER(ctypes.c_char))]
 901
 902     buffer = ctypes.create_string_buffer(ciphertext)
 903     blob_in = DATA_BLOB(ctypes.sizeof(buffer), buffer)
 904     blob_out = DATA_BLOB()
 905     ret = ctypes.windll.crypt32.CryptUnprotectData(
 906         ctypes.byref(blob_in),  # pDataIn
 907         None,  # ppszDataDescr: human readable description of pDataIn
 908         None,  # pOptionalEntropy: salt?
 909         None,  # pvReserved: must be NULL
 910         None,  # pPromptStruct: information about prompts to display
 911         0,  # dwFlags
 912         ctypes.byref(blob_out)  # pDataOut
 913     )
 914     if not ret:
 915         logger.warning('failed to decrypt with DPAPI', only_once=True)
 916         return None
 917
 918     result = ctypes.string_at(blob_out.pbData, blob_out.cbData)
 919     ctypes.windll.kernel32.LocalFree(blob_out.pbData)
 920     return result
 921
 922
 923 def _config_home():
 924     return os.environ.get('XDG_CONFIG_HOME', os.path.expanduser('~/.config'))
 925
 926
 927 def _open_database_copy(database_path, tmpdir):
 928     # cannot open sqlite databases if they are already in use (e.g. by the browser)
 929     database_copy_path = os.path.join(tmpdir, 'temporary.sqlite')
 930     shutil.copy(database_path, database_copy_path)
 931     conn = sqlite3.connect(database_copy_path)
 932     return conn.cursor()
 933
 934
 935 def _get_column_names(cursor, table_name):
 936     table_info = cursor.execute(f'PRAGMA table_info({table_name})').fetchall()
 937     return [row[1].decode() for row in table_info]
 938
 939
 940 def _find_most_recently_used_file(root, filename, logger):
 941     # if there are multiple browser profiles, take the most recently used one
 942     i, paths = 0, []
 943     with _create_progress_bar(logger) as progress_bar:
 944         for curr_root, dirs, files in os.walk(root):
 945             for file in files:
 946                 i += 1
 947                 progress_bar.print(f'Searching for "{filename}": {i: 6d} files searched')
 948                 if file == filename:
 949                     paths.append(os.path.join(curr_root, file))
 950     return None if not paths else max(paths, key=lambda path: os.lstat(path).st_mtime)
 951
 952
 953 def _merge_cookie_jars(jars):
 954     output_jar = YoutubeDLCookieJar()
 955     for jar in jars:
 956         for cookie in jar:
 957             output_jar.set_cookie(cookie)
 958         if jar.filename is not None:
 959             output_jar.filename = jar.filename
 960     return output_jar
 961
 962
 963 def _is_path(value):
 964     return os.path.sep in value
 965
 966
 967 def _parse_browser_specification(browser_name, profile=None, keyring=None):
 968     if browser_name not in SUPPORTED_BROWSERS:
 969         raise ValueError(f'unsupported browser: "{browser_name}"')
 970     if keyring not in (None, *SUPPORTED_KEYRINGS):
 971         raise ValueError(f'unsupported keyring: "{keyring}"')
 972     if profile is not None and _is_path(profile):
 973         profile = os.path.expanduser(profile)
 974     return browser_name, profile, keyring