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     def decrypt(self, encrypted_value):
 340         raise NotImplementedError('Must be implemented by sub classes')
 341
 342     @property
 343     def _cookie_counts(self):
 344         raise NotImplementedError('Must be implemented by sub classes')
 345
 346
 347 def get_cookie_decryptor(browser_root, browser_keyring_name, logger, *, keyring=None):
 348     if sys.platform in ('linux', 'linux2'):
 349         return LinuxChromeCookieDecryptor(browser_keyring_name, logger, keyring=keyring)
 350     elif sys.platform == 'darwin':
 351         return MacChromeCookieDecryptor(browser_keyring_name, logger)
 352     elif sys.platform == 'win32':
 353         return WindowsChromeCookieDecryptor(browser_root, logger)
 354     else:
 355         raise NotImplementedError(f'Chrome cookie decryption is not supported on this platform: {sys.platform}')
 356
 357
 358 class LinuxChromeCookieDecryptor(ChromeCookieDecryptor):
 359     def __init__(self, browser_keyring_name, logger, *, keyring=None):
 360         self._logger = logger
 361         self._v10_key = self.derive_key(b'peanuts')
 362         password = _get_linux_keyring_password(browser_keyring_name, keyring, logger)
 363         self._v11_key = None if password is None else self.derive_key(password)
 364         self._cookie_counts = {'v10': 0, 'v11': 0, 'other': 0}
 365
 366     @staticmethod
 367     def derive_key(password):
 368         # values from
 369         # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_linux.cc
 370         return pbkdf2_sha1(password, salt=b'saltysalt', iterations=1, key_length=16)
 371
 372     def decrypt(self, encrypted_value):
 373         version = encrypted_value[:3]
 374         ciphertext = encrypted_value[3:]
 375
 376         if version == b'v10':
 377             self._cookie_counts['v10'] += 1
 378             return _decrypt_aes_cbc(ciphertext, self._v10_key, self._logger)
 379
 380         elif version == b'v11':
 381             self._cookie_counts['v11'] += 1
 382             if self._v11_key is None:
 383                 self._logger.warning('cannot decrypt v11 cookies: no key found', only_once=True)
 384                 return None
 385             return _decrypt_aes_cbc(ciphertext, self._v11_key, self._logger)
 386
 387         else:
 388             self._cookie_counts['other'] += 1
 389             return None
 390
 391
 392 class MacChromeCookieDecryptor(ChromeCookieDecryptor):
 393     def __init__(self, browser_keyring_name, logger):
 394         self._logger = logger
 395         password = _get_mac_keyring_password(browser_keyring_name, logger)
 396         self._v10_key = None if password is None else self.derive_key(password)
 397         self._cookie_counts = {'v10': 0, 'other': 0}
 398
 399     @staticmethod
 400     def derive_key(password):
 401         # values from
 402         # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_mac.mm
 403         return pbkdf2_sha1(password, salt=b'saltysalt', iterations=1003, key_length=16)
 404
 405     def decrypt(self, encrypted_value):
 406         version = encrypted_value[:3]
 407         ciphertext = encrypted_value[3:]
 408
 409         if version == b'v10':
 410             self._cookie_counts['v10'] += 1
 411             if self._v10_key is None:
 412                 self._logger.warning('cannot decrypt v10 cookies: no key found', only_once=True)
 413                 return None
 414
 415             return _decrypt_aes_cbc(ciphertext, self._v10_key, self._logger)
 416
 417         else:
 418             self._cookie_counts['other'] += 1
 419             # other prefixes are considered 'old data' which were stored as plaintext
 420             # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_mac.mm
 421             return encrypted_value
 422
 423
 424 class WindowsChromeCookieDecryptor(ChromeCookieDecryptor):
 425     def __init__(self, browser_root, logger):
 426         self._logger = logger
 427         self._v10_key = _get_windows_v10_key(browser_root, logger)
 428         self._cookie_counts = {'v10': 0, 'other': 0}
 429
 430     def decrypt(self, encrypted_value):
 431         version = encrypted_value[:3]
 432         ciphertext = encrypted_value[3:]
 433
 434         if version == b'v10':
 435             self._cookie_counts['v10'] += 1
 436             if self._v10_key is None:
 437                 self._logger.warning('cannot decrypt v10 cookies: no key found', only_once=True)
 438                 return None
 439
 440             # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_win.cc
 441             #   kNonceLength
 442             nonce_length = 96 // 8
 443             # boringssl
 444             #   EVP_AEAD_AES_GCM_TAG_LEN
 445             authentication_tag_length = 16
 446
 447             raw_ciphertext = ciphertext
 448             nonce = raw_ciphertext[:nonce_length]
 449             ciphertext = raw_ciphertext[nonce_length:-authentication_tag_length]
 450             authentication_tag = raw_ciphertext[-authentication_tag_length:]
 451
 452             return _decrypt_aes_gcm(ciphertext, self._v10_key, nonce, authentication_tag, self._logger)
 453
 454         else:
 455             self._cookie_counts['other'] += 1
 456             # any other prefix means the data is DPAPI encrypted
 457             # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_win.cc
 458             return _decrypt_windows_dpapi(encrypted_value, self._logger).decode()
 459
 460
 461 def _extract_safari_cookies(profile, logger):
 462     if profile is not None:
 463         logger.error('safari does not support profiles')
 464     if sys.platform != 'darwin':
 465         raise ValueError(f'unsupported platform: {sys.platform}')
 466
 467     cookies_path = os.path.expanduser('~/Library/Cookies/Cookies.binarycookies')
 468
 469     if not os.path.isfile(cookies_path):
 470         logger.debug('Trying secondary cookie location')
 471         cookies_path = os.path.expanduser('~/Library/Containers/com.apple.Safari/Data/Library/Cookies/Cookies.binarycookies')
 472         if not os.path.isfile(cookies_path):
 473             raise FileNotFoundError('could not find safari cookies database')
 474
 475     with open(cookies_path, 'rb') as f:
 476         cookies_data = f.read()
 477
 478     jar = parse_safari_cookies(cookies_data, logger=logger)
 479     logger.info(f'Extracted {len(jar)} cookies from safari')
 480     return jar
 481
 482
 483 class ParserError(Exception):
 484     pass
 485
 486
 487 class DataParser:
 488     def __init__(self, data, logger):
 489         self._data = data
 490         self.cursor = 0
 491         self._logger = logger
 492
 493     def read_bytes(self, num_bytes):
 494         if num_bytes < 0:
 495             raise ParserError(f'invalid read of {num_bytes} bytes')
 496         end = self.cursor + num_bytes
 497         if end > len(self._data):
 498             raise ParserError('reached end of input')
 499         data = self._data[self.cursor:end]
 500         self.cursor = end
 501         return data
 502
 503     def expect_bytes(self, expected_value, message):
 504         value = self.read_bytes(len(expected_value))
 505         if value != expected_value:
 506             raise ParserError(f'unexpected value: {value} != {expected_value} ({message})')
 507
 508     def read_uint(self, big_endian=False):
 509         data_format = '>I' if big_endian else '<I'
 510         return struct.unpack(data_format, self.read_bytes(4))[0]
 511
 512     def read_double(self, big_endian=False):
 513         data_format = '>d' if big_endian else '<d'
 514         return struct.unpack(data_format, self.read_bytes(8))[0]
 515
 516     def read_cstring(self):
 517         buffer = []
 518         while True:
 519             c = self.read_bytes(1)
 520             if c == b'\x00':
 521                 return b''.join(buffer).decode()
 522             else:
 523                 buffer.append(c)
 524
 525     def skip(self, num_bytes, description='unknown'):
 526         if num_bytes > 0:
 527             self._logger.debug(f'skipping {num_bytes} bytes ({description}): {self.read_bytes(num_bytes)!r}')
 528         elif num_bytes < 0:
 529             raise ParserError(f'invalid skip of {num_bytes} bytes')
 530
 531     def skip_to(self, offset, description='unknown'):
 532         self.skip(offset - self.cursor, description)
 533
 534     def skip_to_end(self, description='unknown'):
 535         self.skip_to(len(self._data), description)
 536
 537
 538 def _mac_absolute_time_to_posix(timestamp):
 539     return int((datetime(2001, 1, 1, 0, 0, tzinfo=timezone.utc) + timedelta(seconds=timestamp)).timestamp())
 540
 541
 542 def _parse_safari_cookies_header(data, logger):
 543     p = DataParser(data, logger)
 544     p.expect_bytes(b'cook', 'database signature')
 545     number_of_pages = p.read_uint(big_endian=True)
 546     page_sizes = [p.read_uint(big_endian=True) for _ in range(number_of_pages)]
 547     return page_sizes, p.cursor
 548
 549
 550 def _parse_safari_cookies_page(data, jar, logger):
 551     p = DataParser(data, logger)
 552     p.expect_bytes(b'\x00\x00\x01\x00', 'page signature')
 553     number_of_cookies = p.read_uint()
 554     record_offsets = [p.read_uint() for _ in range(number_of_cookies)]
 555     if number_of_cookies == 0:
 556         logger.debug(f'a cookies page of size {len(data)} has no cookies')
 557         return
 558
 559     p.skip_to(record_offsets[0], 'unknown page header field')
 560
 561     with _create_progress_bar(logger) as progress_bar:
 562         for i, record_offset in enumerate(record_offsets):
 563             progress_bar.print(f'Loading cookie {i: 6d}/{number_of_cookies: 6d}')
 564             p.skip_to(record_offset, 'space between records')
 565             record_length = _parse_safari_cookies_record(data[record_offset:], jar, logger)
 566             p.read_bytes(record_length)
 567     p.skip_to_end('space in between pages')
 568
 569
 570 def _parse_safari_cookies_record(data, jar, logger):
 571     p = DataParser(data, logger)
 572     record_size = p.read_uint()
 573     p.skip(4, 'unknown record field 1')
 574     flags = p.read_uint()
 575     is_secure = bool(flags & 0x0001)
 576     p.skip(4, 'unknown record field 2')
 577     domain_offset = p.read_uint()
 578     name_offset = p.read_uint()
 579     path_offset = p.read_uint()
 580     value_offset = p.read_uint()
 581     p.skip(8, 'unknown record field 3')
 582     expiration_date = _mac_absolute_time_to_posix(p.read_double())
 583     _creation_date = _mac_absolute_time_to_posix(p.read_double())  # noqa: F841
 584
 585     try:
 586         p.skip_to(domain_offset)
 587         domain = p.read_cstring()
 588
 589         p.skip_to(name_offset)
 590         name = p.read_cstring()
 591
 592         p.skip_to(path_offset)
 593         path = p.read_cstring()
 594
 595         p.skip_to(value_offset)
 596         value = p.read_cstring()
 597     except UnicodeDecodeError:
 598         logger.warning('failed to parse Safari cookie because UTF-8 decoding failed', only_once=True)
 599         return record_size
 600
 601     p.skip_to(record_size, 'space at the end of the record')
 602
 603     cookie = compat_cookiejar_Cookie(
 604         version=0, name=name, value=value, port=None, port_specified=False,
 605         domain=domain, domain_specified=bool(domain), domain_initial_dot=domain.startswith('.'),
 606         path=path, path_specified=bool(path), secure=is_secure, expires=expiration_date, discard=False,
 607         comment=None, comment_url=None, rest={})
 608     jar.set_cookie(cookie)
 609     return record_size
 610
 611
 612 def parse_safari_cookies(data, jar=None, logger=YDLLogger()):
 613     """
 614     References:
 615         - https://github.com/libyal/dtformats/blob/main/documentation/Safari%20Cookies.asciidoc
 616             - this data appears to be out of date but the important parts of the database structure is the same
 617             - there are a few bytes here and there which are skipped during parsing
 618     """
 619     if jar is None:
 620         jar = YoutubeDLCookieJar()
 621     page_sizes, body_start = _parse_safari_cookies_header(data, logger)
 622     p = DataParser(data[body_start:], logger)
 623     for page_size in page_sizes:
 624         _parse_safari_cookies_page(p.read_bytes(page_size), jar, logger)
 625     p.skip_to_end('footer')
 626     return jar
 627
 628
 629 class _LinuxDesktopEnvironment(Enum):
 630     """
 631     https://chromium.googlesource.com/chromium/src/+/refs/heads/main/base/nix/xdg_util.h
 632     DesktopEnvironment
 633     """
 634     OTHER = auto()
 635     CINNAMON = auto()
 636     GNOME = auto()
 637     KDE = auto()
 638     PANTHEON = auto()
 639     UNITY = auto()
 640     XFCE = auto()
 641
 642
 643 class _LinuxKeyring(Enum):
 644     """
 645     https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/key_storage_util_linux.h
 646     SelectedLinuxBackend
 647     """
 648     KWALLET = auto()
 649     GNOMEKEYRING = auto()
 650     BASICTEXT = auto()
 651
 652
 653 SUPPORTED_KEYRINGS = _LinuxKeyring.__members__.keys()
 654
 655
 656 def _get_linux_desktop_environment(env):
 657     """
 658     https://chromium.googlesource.com/chromium/src/+/refs/heads/main/base/nix/xdg_util.cc
 659     GetDesktopEnvironment
 660     """
 661     xdg_current_desktop = env.get('XDG_CURRENT_DESKTOP', None)
 662     desktop_session = env.get('DESKTOP_SESSION', None)
 663     if xdg_current_desktop is not None:
 664         xdg_current_desktop = xdg_current_desktop.split(':')[0].strip()
 665
 666         if xdg_current_desktop == 'Unity':
 667             if desktop_session is not None and 'gnome-fallback' in desktop_session:
 668                 return _LinuxDesktopEnvironment.GNOME
 669             else:
 670                 return _LinuxDesktopEnvironment.UNITY
 671         elif xdg_current_desktop == 'GNOME':
 672             return _LinuxDesktopEnvironment.GNOME
 673         elif xdg_current_desktop == 'X-Cinnamon':
 674             return _LinuxDesktopEnvironment.CINNAMON
 675         elif xdg_current_desktop == 'KDE':
 676             return _LinuxDesktopEnvironment.KDE
 677         elif xdg_current_desktop == 'Pantheon':
 678             return _LinuxDesktopEnvironment.PANTHEON
 679         elif xdg_current_desktop == 'XFCE':
 680             return _LinuxDesktopEnvironment.XFCE
 681     elif desktop_session is not None:
 682         if desktop_session in ('mate', 'gnome'):
 683             return _LinuxDesktopEnvironment.GNOME
 684         elif 'kde' in desktop_session:
 685             return _LinuxDesktopEnvironment.KDE
 686         elif 'xfce' in desktop_session:
 687             return _LinuxDesktopEnvironment.XFCE
 688     else:
 689         if 'GNOME_DESKTOP_SESSION_ID' in env:
 690             return _LinuxDesktopEnvironment.GNOME
 691         elif 'KDE_FULL_SESSION' in env:
 692             return _LinuxDesktopEnvironment.KDE
 693     return _LinuxDesktopEnvironment.OTHER
 694
 695
 696 def _choose_linux_keyring(logger):
 697     """
 698     https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/key_storage_util_linux.cc
 699     SelectBackend
 700     """
 701     desktop_environment = _get_linux_desktop_environment(os.environ)
 702     logger.debug(f'detected desktop environment: {desktop_environment.name}')
 703     if desktop_environment == _LinuxDesktopEnvironment.KDE:
 704         linux_keyring = _LinuxKeyring.KWALLET
 705     elif desktop_environment == _LinuxDesktopEnvironment.OTHER:
 706         linux_keyring = _LinuxKeyring.BASICTEXT
 707     else:
 708         linux_keyring = _LinuxKeyring.GNOMEKEYRING
 709     return linux_keyring
 710
 711
 712 def _get_kwallet_network_wallet(logger):
 713     """ The name of the wallet used to store network passwords.
 714
 715     https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/kwallet_dbus.cc
 716     KWalletDBus::NetworkWallet
 717     which does a dbus call to the following function:
 718     https://api.kde.org/frameworks/kwallet/html/classKWallet_1_1Wallet.html
 719     Wallet::NetworkWallet
 720     """
 721     default_wallet = 'kdewallet'
 722     try:
 723         proc = Popen([
 724             'dbus-send', '--session', '--print-reply=literal',
 725             '--dest=org.kde.kwalletd5',
 726             '/modules/kwalletd5',
 727             'org.kde.KWallet.networkWallet'
 728         ], stdout=subprocess.PIPE, stderr=subprocess.DEVNULL)
 729
 730         stdout, stderr = proc.communicate_or_kill()
 731         if proc.returncode != 0:
 732             logger.warning('failed to read NetworkWallet')
 733             return default_wallet
 734         else:
 735             network_wallet = stdout.decode().strip()
 736             logger.debug(f'NetworkWallet = "{network_wallet}"')
 737             return network_wallet
 738     except Exception as e:
 739         logger.warning(f'exception while obtaining NetworkWallet: {e}')
 740         return default_wallet
 741
 742
 743 def _get_kwallet_password(browser_keyring_name, logger):
 744     logger.debug('using kwallet-query to obtain password from kwallet')
 745
 746     if shutil.which('kwallet-query') is None:
 747         logger.error('kwallet-query command not found. KWallet and kwallet-query '
 748                      'must be installed to read from KWallet. kwallet-query should be'
 749                      'included in the kwallet package for your distribution')
 750         return b''
 751
 752     network_wallet = _get_kwallet_network_wallet(logger)
 753
 754     try:
 755         proc = Popen([
 756             'kwallet-query',
 757             '--read-password', f'{browser_keyring_name} Safe Storage',
 758             '--folder', f'{browser_keyring_name} Keys',
 759             network_wallet
 760         ], stdout=subprocess.PIPE, stderr=subprocess.DEVNULL)
 761
 762         stdout, stderr = proc.communicate_or_kill()
 763         if proc.returncode != 0:
 764             logger.error(f'kwallet-query failed with return code {proc.returncode}. Please consult '
 765                          'the kwallet-query man page for details')
 766             return b''
 767         else:
 768             if stdout.lower().startswith(b'failed to read'):
 769                 logger.debug('failed to read password from kwallet. Using empty string instead')
 770                 # this sometimes occurs in KDE because chrome does not check hasEntry and instead
 771                 # just tries to read the value (which kwallet returns "") whereas kwallet-query
 772                 # checks hasEntry. To verify this:
 773                 # dbus-monitor "interface='org.kde.KWallet'" "type=method_return"
 774                 # while starting chrome.
 775                 # this may be a bug as the intended behaviour is to generate a random password and store
 776                 # it, but that doesn't matter here.
 777                 return b''
 778             else:
 779                 logger.debug('password found')
 780                 if stdout[-1:] == b'\n':
 781                     stdout = stdout[:-1]
 782                 return stdout
 783     except Exception as e:
 784         logger.warning(f'exception running kwallet-query: {error_to_str(e)}')
 785         return b''
 786
 787
 788 def _get_gnome_keyring_password(browser_keyring_name, logger):
 789     if not secretstorage:
 790         logger.error(f'secretstorage not available {_SECRETSTORAGE_UNAVAILABLE_REASON}')
 791         return b''
 792     # the Gnome keyring does not seem to organise keys in the same way as KWallet,
 793     # using `dbus-monitor` during startup, it can be observed that chromium lists all keys
 794     # and presumably searches for its key in the list. It appears that we must do the same.
 795     # https://github.com/jaraco/keyring/issues/556
 796     with contextlib.closing(secretstorage.dbus_init()) as con:
 797         col = secretstorage.get_default_collection(con)
 798         for item in col.get_all_items():
 799             if item.get_label() == f'{browser_keyring_name} Safe Storage':
 800                 return item.get_secret()
 801         else:
 802             logger.error('failed to read from keyring')
 803             return b''
 804
 805
 806 def _get_linux_keyring_password(browser_keyring_name, keyring, logger):
 807     # note: chrome/chromium can be run with the following flags to determine which keyring backend
 808     # it has chosen to use
 809     # chromium --enable-logging=stderr --v=1 2>&1 | grep key_storage_
 810     # Chromium supports a flag: --password-store=<basic|gnome|kwallet> so the automatic detection
 811     # will not be sufficient in all cases.
 812
 813     keyring = _LinuxKeyring[keyring] if keyring else _choose_linux_keyring(logger)
 814     logger.debug(f'Chosen keyring: {keyring.name}')
 815
 816     if keyring == _LinuxKeyring.KWALLET:
 817         return _get_kwallet_password(browser_keyring_name, logger)
 818     elif keyring == _LinuxKeyring.GNOMEKEYRING:
 819         return _get_gnome_keyring_password(browser_keyring_name, logger)
 820     elif keyring == _LinuxKeyring.BASICTEXT:
 821         # when basic text is chosen, all cookies are stored as v10 (so no keyring password is required)
 822         return None
 823     assert False, f'Unknown keyring {keyring}'
 824
 825
 826 def _get_mac_keyring_password(browser_keyring_name, logger):
 827     logger.debug('using find-generic-password to obtain password from OSX keychain')
 828     try:
 829         proc = Popen(
 830             ['security', 'find-generic-password',
 831              '-w',  # write password to stdout
 832              '-a', browser_keyring_name,  # match 'account'
 833              '-s', f'{browser_keyring_name} Safe Storage'],  # match 'service'
 834             stdout=subprocess.PIPE, stderr=subprocess.DEVNULL)
 835
 836         stdout, stderr = proc.communicate_or_kill()
 837         if stdout[-1:] == b'\n':
 838             stdout = stdout[:-1]
 839         return stdout
 840     except Exception as e:
 841         logger.warning(f'exception running find-generic-password: {error_to_str(e)}')
 842         return None
 843
 844
 845 def _get_windows_v10_key(browser_root, logger):
 846     path = _find_most_recently_used_file(browser_root, 'Local State', logger)
 847     if path is None:
 848         logger.error('could not find local state file')
 849         return None
 850     logger.debug(f'Found local state file at "{path}"')
 851     with open(path, encoding='utf8') as f:
 852         data = json.load(f)
 853     try:
 854         base64_key = data['os_crypt']['encrypted_key']
 855     except KeyError:
 856         logger.error('no encrypted key in Local State')
 857         return None
 858     encrypted_key = compat_b64decode(base64_key)
 859     prefix = b'DPAPI'
 860     if not encrypted_key.startswith(prefix):
 861         logger.error('invalid key')
 862         return None
 863     return _decrypt_windows_dpapi(encrypted_key[len(prefix):], logger)
 864
 865
 866 def pbkdf2_sha1(password, salt, iterations, key_length):
 867     return pbkdf2_hmac('sha1', password, salt, iterations, key_length)
 868
 869
 870 def _decrypt_aes_cbc(ciphertext, key, logger, initialization_vector=b' ' * 16):
 871     plaintext = unpad_pkcs7(aes_cbc_decrypt_bytes(ciphertext, key, initialization_vector))
 872     try:
 873         return plaintext.decode()
 874     except UnicodeDecodeError:
 875         logger.warning('failed to decrypt cookie (AES-CBC) because UTF-8 decoding failed. Possibly the key is wrong?', only_once=True)
 876         return None
 877
 878
 879 def _decrypt_aes_gcm(ciphertext, key, nonce, authentication_tag, logger):
 880     try:
 881         plaintext = aes_gcm_decrypt_and_verify_bytes(ciphertext, key, authentication_tag, nonce)
 882     except ValueError:
 883         logger.warning('failed to decrypt cookie (AES-GCM) because the MAC check failed. Possibly the key is wrong?', only_once=True)
 884         return None
 885
 886     try:
 887         return plaintext.decode()
 888     except UnicodeDecodeError:
 889         logger.warning('failed to decrypt cookie (AES-GCM) because UTF-8 decoding failed. Possibly the key is wrong?', only_once=True)
 890         return None
 891
 892
 893 def _decrypt_windows_dpapi(ciphertext, logger):
 894     """
 895     References:
 896         - https://docs.microsoft.com/en-us/windows/win32/api/dpapi/nf-dpapi-cryptunprotectdata
 897     """
 898     from ctypes.wintypes import DWORD
 899
 900     class DATA_BLOB(ctypes.Structure):
 901         _fields_ = [('cbData', DWORD),
 902                     ('pbData', ctypes.POINTER(ctypes.c_char))]
 903
 904     buffer = ctypes.create_string_buffer(ciphertext)
 905     blob_in = DATA_BLOB(ctypes.sizeof(buffer), buffer)
 906     blob_out = DATA_BLOB()
 907     ret = ctypes.windll.crypt32.CryptUnprotectData(
 908         ctypes.byref(blob_in),  # pDataIn
 909         None,  # ppszDataDescr: human readable description of pDataIn
 910         None,  # pOptionalEntropy: salt?
 911         None,  # pvReserved: must be NULL
 912         None,  # pPromptStruct: information about prompts to display
 913         0,  # dwFlags
 914         ctypes.byref(blob_out)  # pDataOut
 915     )
 916     if not ret:
 917         logger.warning('failed to decrypt with DPAPI', only_once=True)
 918         return None
 919
 920     result = ctypes.string_at(blob_out.pbData, blob_out.cbData)
 921     ctypes.windll.kernel32.LocalFree(blob_out.pbData)
 922     return result
 923
 924
 925 def _config_home():
 926     return os.environ.get('XDG_CONFIG_HOME', os.path.expanduser('~/.config'))
 927
 928
 929 def _open_database_copy(database_path, tmpdir):
 930     # cannot open sqlite databases if they are already in use (e.g. by the browser)
 931     database_copy_path = os.path.join(tmpdir, 'temporary.sqlite')
 932     shutil.copy(database_path, database_copy_path)
 933     conn = sqlite3.connect(database_copy_path)
 934     return conn.cursor()
 935
 936
 937 def _get_column_names(cursor, table_name):
 938     table_info = cursor.execute(f'PRAGMA table_info({table_name})').fetchall()
 939     return [row[1].decode() for row in table_info]
 940
 941
 942 def _find_most_recently_used_file(root, filename, logger):
 943     # if there are multiple browser profiles, take the most recently used one
 944     i, paths = 0, []
 945     with _create_progress_bar(logger) as progress_bar:
 946         for curr_root, dirs, files in os.walk(root):
 947             for file in files:
 948                 i += 1
 949                 progress_bar.print(f'Searching for "{filename}": {i: 6d} files searched')
 950                 if file == filename:
 951                     paths.append(os.path.join(curr_root, file))
 952     return None if not paths else max(paths, key=lambda path: os.lstat(path).st_mtime)
 953
 954
 955 def _merge_cookie_jars(jars):
 956     output_jar = YoutubeDLCookieJar()
 957     for jar in jars:
 958         for cookie in jar:
 959             output_jar.set_cookie(cookie)
 960         if jar.filename is not None:
 961             output_jar.filename = jar.filename
 962     return output_jar
 963
 964
 965 def _is_path(value):
 966     return os.path.sep in value
 967
 968
 969 def _parse_browser_specification(browser_name, profile=None, keyring=None):
 970     if browser_name not in SUPPORTED_BROWSERS:
 971         raise ValueError(f'unsupported browser: "{browser_name}"')
 972     if keyring not in (None, *SUPPORTED_KEYRINGS):
 973         raise ValueError(f'unsupported keyring: "{keyring}"')
 974     if profile is not None and _is_path(profile):
 975         profile = os.path.expanduser(profile)
 976     return browser_name, profile, keyring