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