]>
Commit | Line | Data |
---|---|---|
982ee69a MB |
1 | import ctypes |
2 | import json | |
3 | import os | |
4 | import shutil | |
982ee69a MB |
5 | import struct |
6 | import subprocess | |
7 | import sys | |
8 | import tempfile | |
9 | from datetime import datetime, timedelta, timezone | |
10 | from hashlib import pbkdf2_hmac | |
11 | ||
09906f55 ÁS |
12 | from .aes import aes_cbc_decrypt_bytes, aes_gcm_decrypt_and_verify_bytes |
13 | from .compat import ( | |
982ee69a MB |
14 | compat_b64decode, |
15 | compat_cookiejar_Cookie, | |
16 | ) | |
09906f55 | 17 | from .utils import ( |
063c409d | 18 | bug_reports_message, |
982ee69a | 19 | expand_path, |
982ee69a MB |
20 | process_communicate_or_kill, |
21 | YoutubeDLCookieJar, | |
22 | ) | |
23 | ||
767b02a9 MB |
24 | try: |
25 | import sqlite3 | |
26 | SQLITE_AVAILABLE = True | |
27 | except ImportError: | |
28 | # although sqlite3 is part of the standard library, it is possible to compile python without | |
29 | # sqlite support. See: https://github.com/yt-dlp/yt-dlp/issues/544 | |
30 | SQLITE_AVAILABLE = False | |
31 | ||
32 | ||
982ee69a MB |
33 | try: |
34 | import keyring | |
35 | KEYRING_AVAILABLE = True | |
063c409d | 36 | KEYRING_UNAVAILABLE_REASON = f'due to unknown reasons{bug_reports_message()}' |
982ee69a MB |
37 | except ImportError: |
38 | KEYRING_AVAILABLE = False | |
063c409d | 39 | KEYRING_UNAVAILABLE_REASON = ( |
40 | 'as the `keyring` module is not installed. ' | |
41 | 'Please install by running `python3 -m pip install keyring`. ' | |
42 | 'Depending on your platform, additional packages may be required ' | |
43 | 'to access the keyring; see https://pypi.org/project/keyring') | |
44 | except Exception as _err: | |
45 | KEYRING_AVAILABLE = False | |
46 | KEYRING_UNAVAILABLE_REASON = 'as the `keyring` module could not be initialized: %s' % _err | |
982ee69a MB |
47 | |
48 | ||
49 | CHROMIUM_BASED_BROWSERS = {'brave', 'chrome', 'chromium', 'edge', 'opera', 'vivaldi'} | |
50 | SUPPORTED_BROWSERS = CHROMIUM_BASED_BROWSERS | {'firefox', 'safari'} | |
51 | ||
52 | ||
53 | class YDLLogger: | |
54 | def __init__(self, ydl=None): | |
55 | self._ydl = ydl | |
56 | ||
57 | def debug(self, message): | |
58 | if self._ydl: | |
59 | self._ydl.write_debug(message) | |
60 | ||
61 | def info(self, message): | |
62 | if self._ydl: | |
63 | self._ydl.to_screen(f'[Cookies] {message}') | |
64 | ||
65 | def warning(self, message, only_once=False): | |
66 | if self._ydl: | |
67 | self._ydl.report_warning(message, only_once) | |
68 | ||
69 | def error(self, message): | |
70 | if self._ydl: | |
71 | self._ydl.report_error(message) | |
72 | ||
73 | ||
74 | def load_cookies(cookie_file, browser_specification, ydl): | |
75 | cookie_jars = [] | |
76 | if browser_specification is not None: | |
77 | browser_name, profile = _parse_browser_specification(*browser_specification) | |
78 | cookie_jars.append(extract_cookies_from_browser(browser_name, profile, YDLLogger(ydl))) | |
79 | ||
80 | if cookie_file is not None: | |
81 | cookie_file = expand_path(cookie_file) | |
82 | jar = YoutubeDLCookieJar(cookie_file) | |
83 | if os.access(cookie_file, os.R_OK): | |
84 | jar.load(ignore_discard=True, ignore_expires=True) | |
85 | cookie_jars.append(jar) | |
86 | ||
87 | return _merge_cookie_jars(cookie_jars) | |
88 | ||
89 | ||
90 | def extract_cookies_from_browser(browser_name, profile=None, logger=YDLLogger()): | |
91 | if browser_name == 'firefox': | |
92 | return _extract_firefox_cookies(profile, logger) | |
93 | elif browser_name == 'safari': | |
94 | return _extract_safari_cookies(profile, logger) | |
95 | elif browser_name in CHROMIUM_BASED_BROWSERS: | |
96 | return _extract_chrome_cookies(browser_name, profile, logger) | |
97 | else: | |
98 | raise ValueError('unknown browser: {}'.format(browser_name)) | |
99 | ||
100 | ||
101 | def _extract_firefox_cookies(profile, logger): | |
102 | logger.info('Extracting cookies from firefox') | |
767b02a9 MB |
103 | if not SQLITE_AVAILABLE: |
104 | logger.warning('Cannot extract cookies from firefox without sqlite3 support. ' | |
105 | 'Please use a python interpreter compiled with sqlite3 support') | |
106 | return YoutubeDLCookieJar() | |
982ee69a MB |
107 | |
108 | if profile is None: | |
109 | search_root = _firefox_browser_dir() | |
110 | elif _is_path(profile): | |
111 | search_root = profile | |
112 | else: | |
113 | search_root = os.path.join(_firefox_browser_dir(), profile) | |
114 | ||
115 | cookie_database_path = _find_most_recently_used_file(search_root, 'cookies.sqlite') | |
116 | if cookie_database_path is None: | |
117 | raise FileNotFoundError('could not find firefox cookies database in {}'.format(search_root)) | |
526d74ec | 118 | logger.debug('Extracting cookies from: "{}"'.format(cookie_database_path)) |
982ee69a MB |
119 | |
120 | with tempfile.TemporaryDirectory(prefix='youtube_dl') as tmpdir: | |
121 | cursor = None | |
122 | try: | |
123 | cursor = _open_database_copy(cookie_database_path, tmpdir) | |
124 | cursor.execute('SELECT host, name, value, path, expiry, isSecure FROM moz_cookies') | |
125 | jar = YoutubeDLCookieJar() | |
126 | for host, name, value, path, expiry, is_secure in cursor.fetchall(): | |
127 | cookie = compat_cookiejar_Cookie( | |
128 | version=0, name=name, value=value, port=None, port_specified=False, | |
129 | domain=host, domain_specified=bool(host), domain_initial_dot=host.startswith('.'), | |
130 | path=path, path_specified=bool(path), secure=is_secure, expires=expiry, discard=False, | |
131 | comment=None, comment_url=None, rest={}) | |
132 | jar.set_cookie(cookie) | |
133 | logger.info('Extracted {} cookies from firefox'.format(len(jar))) | |
134 | return jar | |
135 | finally: | |
136 | if cursor is not None: | |
137 | cursor.connection.close() | |
138 | ||
139 | ||
140 | def _firefox_browser_dir(): | |
141 | if sys.platform in ('linux', 'linux2'): | |
142 | return os.path.expanduser('~/.mozilla/firefox') | |
143 | elif sys.platform == 'win32': | |
144 | return os.path.expandvars(r'%APPDATA%\Mozilla\Firefox\Profiles') | |
145 | elif sys.platform == 'darwin': | |
146 | return os.path.expanduser('~/Library/Application Support/Firefox') | |
147 | else: | |
148 | raise ValueError('unsupported platform: {}'.format(sys.platform)) | |
149 | ||
150 | ||
151 | def _get_chromium_based_browser_settings(browser_name): | |
152 | # https://chromium.googlesource.com/chromium/src/+/HEAD/docs/user_data_dir.md | |
153 | if sys.platform in ('linux', 'linux2'): | |
154 | config = _config_home() | |
155 | browser_dir = { | |
156 | 'brave': os.path.join(config, 'BraveSoftware/Brave-Browser'), | |
157 | 'chrome': os.path.join(config, 'google-chrome'), | |
158 | 'chromium': os.path.join(config, 'chromium'), | |
159 | 'edge': os.path.join(config, 'microsoft-edge'), | |
160 | 'opera': os.path.join(config, 'opera'), | |
161 | 'vivaldi': os.path.join(config, 'vivaldi'), | |
162 | }[browser_name] | |
163 | ||
164 | elif sys.platform == 'win32': | |
165 | appdata_local = os.path.expandvars('%LOCALAPPDATA%') | |
166 | appdata_roaming = os.path.expandvars('%APPDATA%') | |
167 | browser_dir = { | |
168 | 'brave': os.path.join(appdata_local, r'BraveSoftware\Brave-Browser\User Data'), | |
169 | 'chrome': os.path.join(appdata_local, r'Google\Chrome\User Data'), | |
170 | 'chromium': os.path.join(appdata_local, r'Chromium\User Data'), | |
171 | 'edge': os.path.join(appdata_local, r'Microsoft\Edge\User Data'), | |
172 | 'opera': os.path.join(appdata_roaming, r'Opera Software\Opera Stable'), | |
173 | 'vivaldi': os.path.join(appdata_local, r'Vivaldi\User Data'), | |
174 | }[browser_name] | |
175 | ||
176 | elif sys.platform == 'darwin': | |
177 | appdata = os.path.expanduser('~/Library/Application Support') | |
178 | browser_dir = { | |
179 | 'brave': os.path.join(appdata, 'BraveSoftware/Brave-Browser'), | |
180 | 'chrome': os.path.join(appdata, 'Google/Chrome'), | |
181 | 'chromium': os.path.join(appdata, 'Chromium'), | |
182 | 'edge': os.path.join(appdata, 'Microsoft Edge'), | |
183 | 'opera': os.path.join(appdata, 'com.operasoftware.Opera'), | |
184 | 'vivaldi': os.path.join(appdata, 'Vivaldi'), | |
185 | }[browser_name] | |
186 | ||
187 | else: | |
188 | raise ValueError('unsupported platform: {}'.format(sys.platform)) | |
189 | ||
190 | # Linux keyring names can be determined by snooping on dbus while opening the browser in KDE: | |
191 | # dbus-monitor "interface='org.kde.KWallet'" "type=method_return" | |
192 | keyring_name = { | |
193 | 'brave': 'Brave', | |
194 | 'chrome': 'Chrome', | |
195 | 'chromium': 'Chromium', | |
29b208f6 | 196 | 'edge': 'Microsoft Edge' if sys.platform == 'darwin' else 'Chromium', |
982ee69a MB |
197 | 'opera': 'Opera' if sys.platform == 'darwin' else 'Chromium', |
198 | 'vivaldi': 'Vivaldi' if sys.platform == 'darwin' else 'Chrome', | |
199 | }[browser_name] | |
200 | ||
201 | browsers_without_profiles = {'opera'} | |
202 | ||
203 | return { | |
204 | 'browser_dir': browser_dir, | |
205 | 'keyring_name': keyring_name, | |
206 | 'supports_profiles': browser_name not in browsers_without_profiles | |
207 | } | |
208 | ||
209 | ||
210 | def _extract_chrome_cookies(browser_name, profile, logger): | |
211 | logger.info('Extracting cookies from {}'.format(browser_name)) | |
767b02a9 MB |
212 | |
213 | if not SQLITE_AVAILABLE: | |
214 | logger.warning(('Cannot extract cookies from {} without sqlite3 support. ' | |
215 | 'Please use a python interpreter compiled with sqlite3 support').format(browser_name)) | |
216 | return YoutubeDLCookieJar() | |
217 | ||
982ee69a MB |
218 | config = _get_chromium_based_browser_settings(browser_name) |
219 | ||
220 | if profile is None: | |
221 | search_root = config['browser_dir'] | |
222 | elif _is_path(profile): | |
223 | search_root = profile | |
224 | config['browser_dir'] = os.path.dirname(profile) if config['supports_profiles'] else profile | |
225 | else: | |
226 | if config['supports_profiles']: | |
227 | search_root = os.path.join(config['browser_dir'], profile) | |
228 | else: | |
229 | logger.error('{} does not support profiles'.format(browser_name)) | |
230 | search_root = config['browser_dir'] | |
231 | ||
232 | cookie_database_path = _find_most_recently_used_file(search_root, 'Cookies') | |
233 | if cookie_database_path is None: | |
234 | raise FileNotFoundError('could not find {} cookies database in "{}"'.format(browser_name, search_root)) | |
526d74ec | 235 | logger.debug('Extracting cookies from: "{}"'.format(cookie_database_path)) |
982ee69a MB |
236 | |
237 | decryptor = get_cookie_decryptor(config['browser_dir'], config['keyring_name'], logger) | |
238 | ||
239 | with tempfile.TemporaryDirectory(prefix='youtube_dl') as tmpdir: | |
240 | cursor = None | |
241 | try: | |
242 | cursor = _open_database_copy(cookie_database_path, tmpdir) | |
243 | cursor.connection.text_factory = bytes | |
244 | column_names = _get_column_names(cursor, 'cookies') | |
245 | secure_column = 'is_secure' if 'is_secure' in column_names else 'secure' | |
246 | cursor.execute('SELECT host_key, name, value, encrypted_value, path, ' | |
247 | 'expires_utc, {} FROM cookies'.format(secure_column)) | |
248 | jar = YoutubeDLCookieJar() | |
249 | failed_cookies = 0 | |
250 | for host_key, name, value, encrypted_value, path, expires_utc, is_secure in cursor.fetchall(): | |
251 | host_key = host_key.decode('utf-8') | |
252 | name = name.decode('utf-8') | |
253 | value = value.decode('utf-8') | |
254 | path = path.decode('utf-8') | |
255 | ||
256 | if not value and encrypted_value: | |
257 | value = decryptor.decrypt(encrypted_value) | |
258 | if value is None: | |
259 | failed_cookies += 1 | |
260 | continue | |
261 | ||
262 | cookie = compat_cookiejar_Cookie( | |
263 | version=0, name=name, value=value, port=None, port_specified=False, | |
264 | domain=host_key, domain_specified=bool(host_key), domain_initial_dot=host_key.startswith('.'), | |
265 | path=path, path_specified=bool(path), secure=is_secure, expires=expires_utc, discard=False, | |
266 | comment=None, comment_url=None, rest={}) | |
267 | jar.set_cookie(cookie) | |
268 | if failed_cookies > 0: | |
269 | failed_message = ' ({} could not be decrypted)'.format(failed_cookies) | |
270 | else: | |
271 | failed_message = '' | |
272 | logger.info('Extracted {} cookies from {}{}'.format(len(jar), browser_name, failed_message)) | |
273 | return jar | |
274 | finally: | |
275 | if cursor is not None: | |
276 | cursor.connection.close() | |
277 | ||
278 | ||
279 | class ChromeCookieDecryptor: | |
280 | """ | |
281 | Overview: | |
282 | ||
283 | Linux: | |
284 | - cookies are either v10 or v11 | |
285 | - v10: AES-CBC encrypted with a fixed key | |
286 | - v11: AES-CBC encrypted with an OS protected key (keyring) | |
287 | - v11 keys can be stored in various places depending on the activate desktop environment [2] | |
288 | ||
289 | Mac: | |
290 | - cookies are either v10 or not v10 | |
291 | - v10: AES-CBC encrypted with an OS protected key (keyring) and more key derivation iterations than linux | |
292 | - not v10: 'old data' stored as plaintext | |
293 | ||
294 | Windows: | |
295 | - cookies are either v10 or not v10 | |
296 | - v10: AES-GCM encrypted with a key which is encrypted with DPAPI | |
297 | - not v10: encrypted with DPAPI | |
298 | ||
299 | Sources: | |
300 | - [1] https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/ | |
301 | - [2] https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/key_storage_linux.cc | |
302 | - KeyStorageLinux::CreateService | |
303 | """ | |
304 | ||
305 | def decrypt(self, encrypted_value): | |
306 | raise NotImplementedError | |
307 | ||
308 | ||
309 | def get_cookie_decryptor(browser_root, browser_keyring_name, logger): | |
310 | if sys.platform in ('linux', 'linux2'): | |
311 | return LinuxChromeCookieDecryptor(browser_keyring_name, logger) | |
312 | elif sys.platform == 'darwin': | |
313 | return MacChromeCookieDecryptor(browser_keyring_name, logger) | |
314 | elif sys.platform == 'win32': | |
315 | return WindowsChromeCookieDecryptor(browser_root, logger) | |
316 | else: | |
317 | raise NotImplementedError('Chrome cookie decryption is not supported ' | |
318 | 'on this platform: {}'.format(sys.platform)) | |
319 | ||
320 | ||
321 | class LinuxChromeCookieDecryptor(ChromeCookieDecryptor): | |
322 | def __init__(self, browser_keyring_name, logger): | |
323 | self._logger = logger | |
324 | self._v10_key = self.derive_key(b'peanuts') | |
325 | if KEYRING_AVAILABLE: | |
326 | self._v11_key = self.derive_key(_get_linux_keyring_password(browser_keyring_name)) | |
327 | else: | |
328 | self._v11_key = None | |
329 | ||
330 | @staticmethod | |
331 | def derive_key(password): | |
332 | # values from | |
333 | # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_linux.cc | |
334 | return pbkdf2_sha1(password, salt=b'saltysalt', iterations=1, key_length=16) | |
335 | ||
336 | def decrypt(self, encrypted_value): | |
337 | version = encrypted_value[:3] | |
338 | ciphertext = encrypted_value[3:] | |
339 | ||
340 | if version == b'v10': | |
341 | return _decrypt_aes_cbc(ciphertext, self._v10_key, self._logger) | |
342 | ||
343 | elif version == b'v11': | |
344 | if self._v11_key is None: | |
063c409d | 345 | self._logger.warning(f'cannot decrypt cookie {KEYRING_UNAVAILABLE_REASON}', only_once=True) |
982ee69a MB |
346 | return None |
347 | return _decrypt_aes_cbc(ciphertext, self._v11_key, self._logger) | |
348 | ||
349 | else: | |
350 | return None | |
351 | ||
352 | ||
353 | class MacChromeCookieDecryptor(ChromeCookieDecryptor): | |
354 | def __init__(self, browser_keyring_name, logger): | |
355 | self._logger = logger | |
356 | password = _get_mac_keyring_password(browser_keyring_name) | |
357 | self._v10_key = None if password is None else self.derive_key(password) | |
358 | ||
359 | @staticmethod | |
360 | def derive_key(password): | |
361 | # values from | |
362 | # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_mac.mm | |
363 | return pbkdf2_sha1(password, salt=b'saltysalt', iterations=1003, key_length=16) | |
364 | ||
365 | def decrypt(self, encrypted_value): | |
366 | version = encrypted_value[:3] | |
367 | ciphertext = encrypted_value[3:] | |
368 | ||
369 | if version == b'v10': | |
370 | if self._v10_key is None: | |
371 | self._logger.warning('cannot decrypt v10 cookies: no key found', only_once=True) | |
372 | return None | |
373 | ||
374 | return _decrypt_aes_cbc(ciphertext, self._v10_key, self._logger) | |
375 | ||
376 | else: | |
377 | # other prefixes are considered 'old data' which were stored as plaintext | |
378 | # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_mac.mm | |
379 | return encrypted_value | |
380 | ||
381 | ||
382 | class WindowsChromeCookieDecryptor(ChromeCookieDecryptor): | |
383 | def __init__(self, browser_root, logger): | |
384 | self._logger = logger | |
385 | self._v10_key = _get_windows_v10_key(browser_root, logger) | |
386 | ||
387 | def decrypt(self, encrypted_value): | |
388 | version = encrypted_value[:3] | |
389 | ciphertext = encrypted_value[3:] | |
390 | ||
391 | if version == b'v10': | |
392 | if self._v10_key is None: | |
393 | self._logger.warning('cannot decrypt v10 cookies: no key found', only_once=True) | |
394 | return None | |
982ee69a MB |
395 | |
396 | # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_win.cc | |
397 | # kNonceLength | |
398 | nonce_length = 96 // 8 | |
399 | # boringssl | |
400 | # EVP_AEAD_AES_GCM_TAG_LEN | |
401 | authentication_tag_length = 16 | |
402 | ||
403 | raw_ciphertext = ciphertext | |
404 | nonce = raw_ciphertext[:nonce_length] | |
405 | ciphertext = raw_ciphertext[nonce_length:-authentication_tag_length] | |
406 | authentication_tag = raw_ciphertext[-authentication_tag_length:] | |
407 | ||
408 | return _decrypt_aes_gcm(ciphertext, self._v10_key, nonce, authentication_tag, self._logger) | |
409 | ||
410 | else: | |
411 | # any other prefix means the data is DPAPI encrypted | |
412 | # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_win.cc | |
413 | return _decrypt_windows_dpapi(encrypted_value, self._logger).decode('utf-8') | |
414 | ||
415 | ||
416 | def _extract_safari_cookies(profile, logger): | |
417 | if profile is not None: | |
418 | logger.error('safari does not support profiles') | |
419 | if sys.platform != 'darwin': | |
420 | raise ValueError('unsupported platform: {}'.format(sys.platform)) | |
421 | ||
422 | cookies_path = os.path.expanduser('~/Library/Cookies/Cookies.binarycookies') | |
423 | ||
424 | if not os.path.isfile(cookies_path): | |
425 | raise FileNotFoundError('could not find safari cookies database') | |
426 | ||
427 | with open(cookies_path, 'rb') as f: | |
428 | cookies_data = f.read() | |
429 | ||
430 | jar = parse_safari_cookies(cookies_data, logger=logger) | |
431 | logger.info('Extracted {} cookies from safari'.format(len(jar))) | |
432 | return jar | |
433 | ||
434 | ||
435 | class ParserError(Exception): | |
436 | pass | |
437 | ||
438 | ||
439 | class DataParser: | |
440 | def __init__(self, data, logger): | |
441 | self._data = data | |
442 | self.cursor = 0 | |
443 | self._logger = logger | |
444 | ||
445 | def read_bytes(self, num_bytes): | |
446 | if num_bytes < 0: | |
447 | raise ParserError('invalid read of {} bytes'.format(num_bytes)) | |
448 | end = self.cursor + num_bytes | |
449 | if end > len(self._data): | |
450 | raise ParserError('reached end of input') | |
451 | data = self._data[self.cursor:end] | |
452 | self.cursor = end | |
453 | return data | |
454 | ||
455 | def expect_bytes(self, expected_value, message): | |
456 | value = self.read_bytes(len(expected_value)) | |
457 | if value != expected_value: | |
458 | raise ParserError('unexpected value: {} != {} ({})'.format(value, expected_value, message)) | |
459 | ||
460 | def read_uint(self, big_endian=False): | |
461 | data_format = '>I' if big_endian else '<I' | |
462 | return struct.unpack(data_format, self.read_bytes(4))[0] | |
463 | ||
464 | def read_double(self, big_endian=False): | |
465 | data_format = '>d' if big_endian else '<d' | |
466 | return struct.unpack(data_format, self.read_bytes(8))[0] | |
467 | ||
468 | def read_cstring(self): | |
469 | buffer = [] | |
470 | while True: | |
471 | c = self.read_bytes(1) | |
472 | if c == b'\x00': | |
473 | return b''.join(buffer).decode('utf-8') | |
474 | else: | |
475 | buffer.append(c) | |
476 | ||
477 | def skip(self, num_bytes, description='unknown'): | |
478 | if num_bytes > 0: | |
479 | self._logger.debug('skipping {} bytes ({}): {}'.format( | |
480 | num_bytes, description, self.read_bytes(num_bytes))) | |
481 | elif num_bytes < 0: | |
482 | raise ParserError('invalid skip of {} bytes'.format(num_bytes)) | |
483 | ||
484 | def skip_to(self, offset, description='unknown'): | |
485 | self.skip(offset - self.cursor, description) | |
486 | ||
487 | def skip_to_end(self, description='unknown'): | |
488 | self.skip_to(len(self._data), description) | |
489 | ||
490 | ||
491 | def _mac_absolute_time_to_posix(timestamp): | |
492 | return int((datetime(2001, 1, 1, 0, 0, tzinfo=timezone.utc) + timedelta(seconds=timestamp)).timestamp()) | |
493 | ||
494 | ||
495 | def _parse_safari_cookies_header(data, logger): | |
496 | p = DataParser(data, logger) | |
497 | p.expect_bytes(b'cook', 'database signature') | |
498 | number_of_pages = p.read_uint(big_endian=True) | |
499 | page_sizes = [p.read_uint(big_endian=True) for _ in range(number_of_pages)] | |
500 | return page_sizes, p.cursor | |
501 | ||
502 | ||
503 | def _parse_safari_cookies_page(data, jar, logger): | |
504 | p = DataParser(data, logger) | |
505 | p.expect_bytes(b'\x00\x00\x01\x00', 'page signature') | |
506 | number_of_cookies = p.read_uint() | |
507 | record_offsets = [p.read_uint() for _ in range(number_of_cookies)] | |
508 | if number_of_cookies == 0: | |
509 | logger.debug('a cookies page of size {} has no cookies'.format(len(data))) | |
510 | return | |
511 | ||
512 | p.skip_to(record_offsets[0], 'unknown page header field') | |
513 | ||
514 | for record_offset in record_offsets: | |
515 | p.skip_to(record_offset, 'space between records') | |
516 | record_length = _parse_safari_cookies_record(data[record_offset:], jar, logger) | |
517 | p.read_bytes(record_length) | |
518 | p.skip_to_end('space in between pages') | |
519 | ||
520 | ||
521 | def _parse_safari_cookies_record(data, jar, logger): | |
522 | p = DataParser(data, logger) | |
523 | record_size = p.read_uint() | |
524 | p.skip(4, 'unknown record field 1') | |
525 | flags = p.read_uint() | |
526 | is_secure = bool(flags & 0x0001) | |
527 | p.skip(4, 'unknown record field 2') | |
528 | domain_offset = p.read_uint() | |
529 | name_offset = p.read_uint() | |
530 | path_offset = p.read_uint() | |
531 | value_offset = p.read_uint() | |
532 | p.skip(8, 'unknown record field 3') | |
533 | expiration_date = _mac_absolute_time_to_posix(p.read_double()) | |
534 | _creation_date = _mac_absolute_time_to_posix(p.read_double()) # noqa: F841 | |
535 | ||
536 | try: | |
537 | p.skip_to(domain_offset) | |
538 | domain = p.read_cstring() | |
539 | ||
540 | p.skip_to(name_offset) | |
541 | name = p.read_cstring() | |
542 | ||
543 | p.skip_to(path_offset) | |
544 | path = p.read_cstring() | |
545 | ||
546 | p.skip_to(value_offset) | |
547 | value = p.read_cstring() | |
548 | except UnicodeDecodeError: | |
f9be9cb9 | 549 | logger.warning('failed to parse cookie because UTF-8 decoding failed', only_once=True) |
982ee69a MB |
550 | return record_size |
551 | ||
552 | p.skip_to(record_size, 'space at the end of the record') | |
553 | ||
554 | cookie = compat_cookiejar_Cookie( | |
555 | version=0, name=name, value=value, port=None, port_specified=False, | |
556 | domain=domain, domain_specified=bool(domain), domain_initial_dot=domain.startswith('.'), | |
557 | path=path, path_specified=bool(path), secure=is_secure, expires=expiration_date, discard=False, | |
558 | comment=None, comment_url=None, rest={}) | |
559 | jar.set_cookie(cookie) | |
560 | return record_size | |
561 | ||
562 | ||
563 | def parse_safari_cookies(data, jar=None, logger=YDLLogger()): | |
564 | """ | |
565 | References: | |
566 | - https://github.com/libyal/dtformats/blob/main/documentation/Safari%20Cookies.asciidoc | |
567 | - this data appears to be out of date but the important parts of the database structure is the same | |
568 | - there are a few bytes here and there which are skipped during parsing | |
569 | """ | |
570 | if jar is None: | |
571 | jar = YoutubeDLCookieJar() | |
572 | page_sizes, body_start = _parse_safari_cookies_header(data, logger) | |
573 | p = DataParser(data[body_start:], logger) | |
574 | for page_size in page_sizes: | |
575 | _parse_safari_cookies_page(p.read_bytes(page_size), jar, logger) | |
576 | p.skip_to_end('footer') | |
577 | return jar | |
578 | ||
579 | ||
580 | def _get_linux_keyring_password(browser_keyring_name): | |
581 | password = keyring.get_password('{} Keys'.format(browser_keyring_name), | |
582 | '{} Safe Storage'.format(browser_keyring_name)) | |
583 | if password is None: | |
584 | # this sometimes occurs in KDE because chrome does not check hasEntry and instead | |
585 | # just tries to read the value (which kwallet returns "") whereas keyring checks hasEntry | |
586 | # to verify this: | |
587 | # dbus-monitor "interface='org.kde.KWallet'" "type=method_return" | |
588 | # while starting chrome. | |
589 | # this may be a bug as the intended behaviour is to generate a random password and store | |
590 | # it, but that doesn't matter here. | |
591 | password = '' | |
592 | return password.encode('utf-8') | |
593 | ||
594 | ||
595 | def _get_mac_keyring_password(browser_keyring_name): | |
596 | if KEYRING_AVAILABLE: | |
597 | password = keyring.get_password('{} Safe Storage'.format(browser_keyring_name), browser_keyring_name) | |
598 | return password.encode('utf-8') | |
599 | else: | |
600 | proc = subprocess.Popen(['security', 'find-generic-password', | |
601 | '-w', # write password to stdout | |
602 | '-a', browser_keyring_name, # match 'account' | |
603 | '-s', '{} Safe Storage'.format(browser_keyring_name)], # match 'service' | |
604 | stdout=subprocess.PIPE, | |
605 | stderr=subprocess.DEVNULL) | |
606 | try: | |
607 | stdout, stderr = process_communicate_or_kill(proc) | |
608 | return stdout | |
609 | except BaseException: | |
610 | return None | |
611 | ||
612 | ||
613 | def _get_windows_v10_key(browser_root, logger): | |
614 | path = _find_most_recently_used_file(browser_root, 'Local State') | |
615 | if path is None: | |
616 | logger.error('could not find local state file') | |
617 | return None | |
618 | with open(path, 'r') as f: | |
619 | data = json.load(f) | |
620 | try: | |
621 | base64_key = data['os_crypt']['encrypted_key'] | |
622 | except KeyError: | |
623 | logger.error('no encrypted key in Local State') | |
624 | return None | |
625 | encrypted_key = compat_b64decode(base64_key) | |
626 | prefix = b'DPAPI' | |
627 | if not encrypted_key.startswith(prefix): | |
628 | logger.error('invalid key') | |
629 | return None | |
630 | return _decrypt_windows_dpapi(encrypted_key[len(prefix):], logger) | |
631 | ||
632 | ||
633 | def pbkdf2_sha1(password, salt, iterations, key_length): | |
634 | return pbkdf2_hmac('sha1', password, salt, iterations, key_length) | |
635 | ||
636 | ||
637 | def _decrypt_aes_cbc(ciphertext, key, logger, initialization_vector=b' ' * 16): | |
09906f55 | 638 | plaintext = aes_cbc_decrypt_bytes(ciphertext, key, initialization_vector) |
982ee69a MB |
639 | padding_length = plaintext[-1] |
640 | try: | |
09906f55 | 641 | return plaintext[:-padding_length].decode('utf-8') |
982ee69a | 642 | except UnicodeDecodeError: |
f9be9cb9 | 643 | logger.warning('failed to decrypt cookie because UTF-8 decoding failed. Possibly the key is wrong?', only_once=True) |
982ee69a MB |
644 | return None |
645 | ||
646 | ||
647 | def _decrypt_aes_gcm(ciphertext, key, nonce, authentication_tag, logger): | |
982ee69a | 648 | try: |
09906f55 | 649 | plaintext = aes_gcm_decrypt_and_verify_bytes(ciphertext, key, authentication_tag, nonce) |
982ee69a | 650 | except ValueError: |
f9be9cb9 | 651 | logger.warning('failed to decrypt cookie because the MAC check failed. Possibly the key is wrong?', only_once=True) |
982ee69a MB |
652 | return None |
653 | ||
654 | try: | |
655 | return plaintext.decode('utf-8') | |
656 | except UnicodeDecodeError: | |
f9be9cb9 | 657 | logger.warning('failed to decrypt cookie because UTF-8 decoding failed. Possibly the key is wrong?', only_once=True) |
982ee69a MB |
658 | return None |
659 | ||
660 | ||
661 | def _decrypt_windows_dpapi(ciphertext, logger): | |
662 | """ | |
663 | References: | |
664 | - https://docs.microsoft.com/en-us/windows/win32/api/dpapi/nf-dpapi-cryptunprotectdata | |
665 | """ | |
666 | from ctypes.wintypes import DWORD | |
667 | ||
668 | class DATA_BLOB(ctypes.Structure): | |
669 | _fields_ = [('cbData', DWORD), | |
670 | ('pbData', ctypes.POINTER(ctypes.c_char))] | |
671 | ||
672 | buffer = ctypes.create_string_buffer(ciphertext) | |
673 | blob_in = DATA_BLOB(ctypes.sizeof(buffer), buffer) | |
674 | blob_out = DATA_BLOB() | |
675 | ret = ctypes.windll.crypt32.CryptUnprotectData( | |
676 | ctypes.byref(blob_in), # pDataIn | |
677 | None, # ppszDataDescr: human readable description of pDataIn | |
678 | None, # pOptionalEntropy: salt? | |
679 | None, # pvReserved: must be NULL | |
680 | None, # pPromptStruct: information about prompts to display | |
681 | 0, # dwFlags | |
682 | ctypes.byref(blob_out) # pDataOut | |
683 | ) | |
684 | if not ret: | |
f9be9cb9 | 685 | logger.warning('failed to decrypt with DPAPI', only_once=True) |
982ee69a MB |
686 | return None |
687 | ||
688 | result = ctypes.string_at(blob_out.pbData, blob_out.cbData) | |
689 | ctypes.windll.kernel32.LocalFree(blob_out.pbData) | |
690 | return result | |
691 | ||
692 | ||
693 | def _config_home(): | |
694 | return os.environ.get('XDG_CONFIG_HOME', os.path.expanduser('~/.config')) | |
695 | ||
696 | ||
697 | def _open_database_copy(database_path, tmpdir): | |
698 | # cannot open sqlite databases if they are already in use (e.g. by the browser) | |
699 | database_copy_path = os.path.join(tmpdir, 'temporary.sqlite') | |
700 | shutil.copy(database_path, database_copy_path) | |
701 | conn = sqlite3.connect(database_copy_path) | |
702 | return conn.cursor() | |
703 | ||
704 | ||
705 | def _get_column_names(cursor, table_name): | |
706 | table_info = cursor.execute('PRAGMA table_info({})'.format(table_name)).fetchall() | |
707 | return [row[1].decode('utf-8') for row in table_info] | |
708 | ||
709 | ||
710 | def _find_most_recently_used_file(root, filename): | |
711 | # if there are multiple browser profiles, take the most recently used one | |
712 | paths = [] | |
713 | for root, dirs, files in os.walk(root): | |
714 | for file in files: | |
715 | if file == filename: | |
716 | paths.append(os.path.join(root, file)) | |
717 | return None if not paths else max(paths, key=lambda path: os.lstat(path).st_mtime) | |
718 | ||
719 | ||
720 | def _merge_cookie_jars(jars): | |
721 | output_jar = YoutubeDLCookieJar() | |
722 | for jar in jars: | |
723 | for cookie in jar: | |
724 | output_jar.set_cookie(cookie) | |
725 | if jar.filename is not None: | |
726 | output_jar.filename = jar.filename | |
727 | return output_jar | |
728 | ||
729 | ||
730 | def _is_path(value): | |
731 | return os.path.sep in value | |
732 | ||
733 | ||
734 | def _parse_browser_specification(browser_name, profile=None): | |
ca46b941 | 735 | browser_name = browser_name.lower() |
982ee69a MB |
736 | if browser_name not in SUPPORTED_BROWSERS: |
737 | raise ValueError(f'unsupported browser: "{browser_name}"') | |
738 | if profile is not None and _is_path(profile): | |
739 | profile = os.path.expanduser(profile) | |
740 | return browser_name, profile |