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