]>
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 | ||
12 | from yt_dlp.aes import aes_cbc_decrypt | |
13 | from yt_dlp.compat import ( | |
14 | compat_b64decode, | |
15 | compat_cookiejar_Cookie, | |
edf65256 | 16 | compat_pycrypto_AES |
982ee69a MB |
17 | ) |
18 | from yt_dlp.utils import ( | |
063c409d | 19 | bug_reports_message, |
982ee69a MB |
20 | bytes_to_intlist, |
21 | expand_path, | |
22 | intlist_to_bytes, | |
23 | process_communicate_or_kill, | |
24 | YoutubeDLCookieJar, | |
25 | ) | |
26 | ||
767b02a9 MB |
27 | try: |
28 | import sqlite3 | |
29 | SQLITE_AVAILABLE = True | |
30 | except ImportError: | |
31 | # although sqlite3 is part of the standard library, it is possible to compile python without | |
32 | # sqlite support. See: https://github.com/yt-dlp/yt-dlp/issues/544 | |
33 | SQLITE_AVAILABLE = False | |
34 | ||
35 | ||
982ee69a MB |
36 | try: |
37 | import keyring | |
38 | KEYRING_AVAILABLE = True | |
063c409d | 39 | KEYRING_UNAVAILABLE_REASON = f'due to unknown reasons{bug_reports_message()}' |
982ee69a MB |
40 | except ImportError: |
41 | KEYRING_AVAILABLE = False | |
063c409d | 42 | KEYRING_UNAVAILABLE_REASON = ( |
43 | 'as the `keyring` module is not installed. ' | |
44 | 'Please install by running `python3 -m pip install keyring`. ' | |
45 | 'Depending on your platform, additional packages may be required ' | |
46 | 'to access the keyring; see https://pypi.org/project/keyring') | |
47 | except Exception as _err: | |
48 | KEYRING_AVAILABLE = False | |
49 | KEYRING_UNAVAILABLE_REASON = 'as the `keyring` module could not be initialized: %s' % _err | |
982ee69a MB |
50 | |
51 | ||
52 | CHROMIUM_BASED_BROWSERS = {'brave', 'chrome', 'chromium', 'edge', 'opera', 'vivaldi'} | |
53 | SUPPORTED_BROWSERS = CHROMIUM_BASED_BROWSERS | {'firefox', 'safari'} | |
54 | ||
55 | ||
56 | class YDLLogger: | |
57 | def __init__(self, ydl=None): | |
58 | self._ydl = ydl | |
59 | ||
60 | def debug(self, message): | |
61 | if self._ydl: | |
62 | self._ydl.write_debug(message) | |
63 | ||
64 | def info(self, message): | |
65 | if self._ydl: | |
66 | self._ydl.to_screen(f'[Cookies] {message}') | |
67 | ||
68 | def warning(self, message, only_once=False): | |
69 | if self._ydl: | |
70 | self._ydl.report_warning(message, only_once) | |
71 | ||
72 | def error(self, message): | |
73 | if self._ydl: | |
74 | self._ydl.report_error(message) | |
75 | ||
76 | ||
77 | def load_cookies(cookie_file, browser_specification, ydl): | |
78 | cookie_jars = [] | |
79 | if browser_specification is not None: | |
80 | browser_name, profile = _parse_browser_specification(*browser_specification) | |
81 | cookie_jars.append(extract_cookies_from_browser(browser_name, profile, YDLLogger(ydl))) | |
82 | ||
83 | if cookie_file is not None: | |
84 | cookie_file = expand_path(cookie_file) | |
85 | jar = YoutubeDLCookieJar(cookie_file) | |
86 | if os.access(cookie_file, os.R_OK): | |
87 | jar.load(ignore_discard=True, ignore_expires=True) | |
88 | cookie_jars.append(jar) | |
89 | ||
90 | return _merge_cookie_jars(cookie_jars) | |
91 | ||
92 | ||
93 | def extract_cookies_from_browser(browser_name, profile=None, logger=YDLLogger()): | |
94 | if browser_name == 'firefox': | |
95 | return _extract_firefox_cookies(profile, logger) | |
96 | elif browser_name == 'safari': | |
97 | return _extract_safari_cookies(profile, logger) | |
98 | elif browser_name in CHROMIUM_BASED_BROWSERS: | |
99 | return _extract_chrome_cookies(browser_name, profile, logger) | |
100 | else: | |
101 | raise ValueError('unknown browser: {}'.format(browser_name)) | |
102 | ||
103 | ||
104 | def _extract_firefox_cookies(profile, logger): | |
105 | logger.info('Extracting cookies from firefox') | |
767b02a9 MB |
106 | if not SQLITE_AVAILABLE: |
107 | logger.warning('Cannot extract cookies from firefox without sqlite3 support. ' | |
108 | 'Please use a python interpreter compiled with sqlite3 support') | |
109 | return YoutubeDLCookieJar() | |
982ee69a MB |
110 | |
111 | if profile is None: | |
112 | search_root = _firefox_browser_dir() | |
113 | elif _is_path(profile): | |
114 | search_root = profile | |
115 | else: | |
116 | search_root = os.path.join(_firefox_browser_dir(), profile) | |
117 | ||
118 | cookie_database_path = _find_most_recently_used_file(search_root, 'cookies.sqlite') | |
119 | if cookie_database_path is None: | |
120 | raise FileNotFoundError('could not find firefox cookies database in {}'.format(search_root)) | |
526d74ec | 121 | logger.debug('Extracting cookies from: "{}"'.format(cookie_database_path)) |
982ee69a MB |
122 | |
123 | with tempfile.TemporaryDirectory(prefix='youtube_dl') as tmpdir: | |
124 | cursor = None | |
125 | try: | |
126 | cursor = _open_database_copy(cookie_database_path, tmpdir) | |
127 | cursor.execute('SELECT host, name, value, path, expiry, isSecure FROM moz_cookies') | |
128 | jar = YoutubeDLCookieJar() | |
129 | for host, name, value, path, expiry, is_secure in cursor.fetchall(): | |
130 | cookie = compat_cookiejar_Cookie( | |
131 | version=0, name=name, value=value, port=None, port_specified=False, | |
132 | domain=host, domain_specified=bool(host), domain_initial_dot=host.startswith('.'), | |
133 | path=path, path_specified=bool(path), secure=is_secure, expires=expiry, discard=False, | |
134 | comment=None, comment_url=None, rest={}) | |
135 | jar.set_cookie(cookie) | |
136 | logger.info('Extracted {} cookies from firefox'.format(len(jar))) | |
137 | return jar | |
138 | finally: | |
139 | if cursor is not None: | |
140 | cursor.connection.close() | |
141 | ||
142 | ||
143 | def _firefox_browser_dir(): | |
144 | if sys.platform in ('linux', 'linux2'): | |
145 | return os.path.expanduser('~/.mozilla/firefox') | |
146 | elif sys.platform == 'win32': | |
147 | return os.path.expandvars(r'%APPDATA%\Mozilla\Firefox\Profiles') | |
148 | elif sys.platform == 'darwin': | |
149 | return os.path.expanduser('~/Library/Application Support/Firefox') | |
150 | else: | |
151 | raise ValueError('unsupported platform: {}'.format(sys.platform)) | |
152 | ||
153 | ||
154 | def _get_chromium_based_browser_settings(browser_name): | |
155 | # https://chromium.googlesource.com/chromium/src/+/HEAD/docs/user_data_dir.md | |
156 | if sys.platform in ('linux', 'linux2'): | |
157 | config = _config_home() | |
158 | browser_dir = { | |
159 | 'brave': os.path.join(config, 'BraveSoftware/Brave-Browser'), | |
160 | 'chrome': os.path.join(config, 'google-chrome'), | |
161 | 'chromium': os.path.join(config, 'chromium'), | |
162 | 'edge': os.path.join(config, 'microsoft-edge'), | |
163 | 'opera': os.path.join(config, 'opera'), | |
164 | 'vivaldi': os.path.join(config, 'vivaldi'), | |
165 | }[browser_name] | |
166 | ||
167 | elif sys.platform == 'win32': | |
168 | appdata_local = os.path.expandvars('%LOCALAPPDATA%') | |
169 | appdata_roaming = os.path.expandvars('%APPDATA%') | |
170 | browser_dir = { | |
171 | 'brave': os.path.join(appdata_local, r'BraveSoftware\Brave-Browser\User Data'), | |
172 | 'chrome': os.path.join(appdata_local, r'Google\Chrome\User Data'), | |
173 | 'chromium': os.path.join(appdata_local, r'Chromium\User Data'), | |
174 | 'edge': os.path.join(appdata_local, r'Microsoft\Edge\User Data'), | |
175 | 'opera': os.path.join(appdata_roaming, r'Opera Software\Opera Stable'), | |
176 | 'vivaldi': os.path.join(appdata_local, r'Vivaldi\User Data'), | |
177 | }[browser_name] | |
178 | ||
179 | elif sys.platform == 'darwin': | |
180 | appdata = os.path.expanduser('~/Library/Application Support') | |
181 | browser_dir = { | |
182 | 'brave': os.path.join(appdata, 'BraveSoftware/Brave-Browser'), | |
183 | 'chrome': os.path.join(appdata, 'Google/Chrome'), | |
184 | 'chromium': os.path.join(appdata, 'Chromium'), | |
185 | 'edge': os.path.join(appdata, 'Microsoft Edge'), | |
186 | 'opera': os.path.join(appdata, 'com.operasoftware.Opera'), | |
187 | 'vivaldi': os.path.join(appdata, 'Vivaldi'), | |
188 | }[browser_name] | |
189 | ||
190 | else: | |
191 | raise ValueError('unsupported platform: {}'.format(sys.platform)) | |
192 | ||
193 | # Linux keyring names can be determined by snooping on dbus while opening the browser in KDE: | |
194 | # dbus-monitor "interface='org.kde.KWallet'" "type=method_return" | |
195 | keyring_name = { | |
196 | 'brave': 'Brave', | |
197 | 'chrome': 'Chrome', | |
198 | 'chromium': 'Chromium', | |
29b208f6 | 199 | 'edge': 'Microsoft Edge' if sys.platform == 'darwin' else 'Chromium', |
982ee69a MB |
200 | 'opera': 'Opera' if sys.platform == 'darwin' else 'Chromium', |
201 | 'vivaldi': 'Vivaldi' if sys.platform == 'darwin' else 'Chrome', | |
202 | }[browser_name] | |
203 | ||
204 | browsers_without_profiles = {'opera'} | |
205 | ||
206 | return { | |
207 | 'browser_dir': browser_dir, | |
208 | 'keyring_name': keyring_name, | |
209 | 'supports_profiles': browser_name not in browsers_without_profiles | |
210 | } | |
211 | ||
212 | ||
213 | def _extract_chrome_cookies(browser_name, profile, logger): | |
214 | logger.info('Extracting cookies from {}'.format(browser_name)) | |
767b02a9 MB |
215 | |
216 | if not SQLITE_AVAILABLE: | |
217 | logger.warning(('Cannot extract cookies from {} without sqlite3 support. ' | |
218 | 'Please use a python interpreter compiled with sqlite3 support').format(browser_name)) | |
219 | return YoutubeDLCookieJar() | |
220 | ||
982ee69a MB |
221 | config = _get_chromium_based_browser_settings(browser_name) |
222 | ||
223 | if profile is None: | |
224 | search_root = config['browser_dir'] | |
225 | elif _is_path(profile): | |
226 | search_root = profile | |
227 | config['browser_dir'] = os.path.dirname(profile) if config['supports_profiles'] else profile | |
228 | else: | |
229 | if config['supports_profiles']: | |
230 | search_root = os.path.join(config['browser_dir'], profile) | |
231 | else: | |
232 | logger.error('{} does not support profiles'.format(browser_name)) | |
233 | search_root = config['browser_dir'] | |
234 | ||
235 | cookie_database_path = _find_most_recently_used_file(search_root, 'Cookies') | |
236 | if cookie_database_path is None: | |
237 | raise FileNotFoundError('could not find {} cookies database in "{}"'.format(browser_name, search_root)) | |
526d74ec | 238 | logger.debug('Extracting cookies from: "{}"'.format(cookie_database_path)) |
982ee69a MB |
239 | |
240 | decryptor = get_cookie_decryptor(config['browser_dir'], config['keyring_name'], logger) | |
241 | ||
242 | with tempfile.TemporaryDirectory(prefix='youtube_dl') as tmpdir: | |
243 | cursor = None | |
244 | try: | |
245 | cursor = _open_database_copy(cookie_database_path, tmpdir) | |
246 | cursor.connection.text_factory = bytes | |
247 | column_names = _get_column_names(cursor, 'cookies') | |
248 | secure_column = 'is_secure' if 'is_secure' in column_names else 'secure' | |
249 | cursor.execute('SELECT host_key, name, value, encrypted_value, path, ' | |
250 | 'expires_utc, {} FROM cookies'.format(secure_column)) | |
251 | jar = YoutubeDLCookieJar() | |
252 | failed_cookies = 0 | |
253 | for host_key, name, value, encrypted_value, path, expires_utc, is_secure in cursor.fetchall(): | |
254 | host_key = host_key.decode('utf-8') | |
255 | name = name.decode('utf-8') | |
256 | value = value.decode('utf-8') | |
257 | path = path.decode('utf-8') | |
258 | ||
259 | if not value and encrypted_value: | |
260 | value = decryptor.decrypt(encrypted_value) | |
261 | if value is None: | |
262 | failed_cookies += 1 | |
263 | continue | |
264 | ||
265 | cookie = compat_cookiejar_Cookie( | |
266 | version=0, name=name, value=value, port=None, port_specified=False, | |
267 | domain=host_key, domain_specified=bool(host_key), domain_initial_dot=host_key.startswith('.'), | |
268 | path=path, path_specified=bool(path), secure=is_secure, expires=expires_utc, discard=False, | |
269 | comment=None, comment_url=None, rest={}) | |
270 | jar.set_cookie(cookie) | |
271 | if failed_cookies > 0: | |
272 | failed_message = ' ({} could not be decrypted)'.format(failed_cookies) | |
273 | else: | |
274 | failed_message = '' | |
275 | logger.info('Extracted {} cookies from {}{}'.format(len(jar), browser_name, failed_message)) | |
276 | return jar | |
277 | finally: | |
278 | if cursor is not None: | |
279 | cursor.connection.close() | |
280 | ||
281 | ||
282 | class ChromeCookieDecryptor: | |
283 | """ | |
284 | Overview: | |
285 | ||
286 | Linux: | |
287 | - cookies are either v10 or v11 | |
288 | - v10: AES-CBC encrypted with a fixed key | |
289 | - v11: AES-CBC encrypted with an OS protected key (keyring) | |
290 | - v11 keys can be stored in various places depending on the activate desktop environment [2] | |
291 | ||
292 | Mac: | |
293 | - cookies are either v10 or not v10 | |
294 | - v10: AES-CBC encrypted with an OS protected key (keyring) and more key derivation iterations than linux | |
295 | - not v10: 'old data' stored as plaintext | |
296 | ||
297 | Windows: | |
298 | - cookies are either v10 or not v10 | |
299 | - v10: AES-GCM encrypted with a key which is encrypted with DPAPI | |
300 | - not v10: encrypted with DPAPI | |
301 | ||
302 | Sources: | |
303 | - [1] https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/ | |
304 | - [2] https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/key_storage_linux.cc | |
305 | - KeyStorageLinux::CreateService | |
306 | """ | |
307 | ||
308 | def decrypt(self, encrypted_value): | |
309 | raise NotImplementedError | |
310 | ||
311 | ||
312 | def get_cookie_decryptor(browser_root, browser_keyring_name, logger): | |
313 | if sys.platform in ('linux', 'linux2'): | |
314 | return LinuxChromeCookieDecryptor(browser_keyring_name, logger) | |
315 | elif sys.platform == 'darwin': | |
316 | return MacChromeCookieDecryptor(browser_keyring_name, logger) | |
317 | elif sys.platform == 'win32': | |
318 | return WindowsChromeCookieDecryptor(browser_root, logger) | |
319 | else: | |
320 | raise NotImplementedError('Chrome cookie decryption is not supported ' | |
321 | 'on this platform: {}'.format(sys.platform)) | |
322 | ||
323 | ||
324 | class LinuxChromeCookieDecryptor(ChromeCookieDecryptor): | |
325 | def __init__(self, browser_keyring_name, logger): | |
326 | self._logger = logger | |
327 | self._v10_key = self.derive_key(b'peanuts') | |
328 | if KEYRING_AVAILABLE: | |
329 | self._v11_key = self.derive_key(_get_linux_keyring_password(browser_keyring_name)) | |
330 | else: | |
331 | self._v11_key = None | |
332 | ||
333 | @staticmethod | |
334 | def derive_key(password): | |
335 | # values from | |
336 | # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_linux.cc | |
337 | return pbkdf2_sha1(password, salt=b'saltysalt', iterations=1, key_length=16) | |
338 | ||
339 | def decrypt(self, encrypted_value): | |
340 | version = encrypted_value[:3] | |
341 | ciphertext = encrypted_value[3:] | |
342 | ||
343 | if version == b'v10': | |
344 | return _decrypt_aes_cbc(ciphertext, self._v10_key, self._logger) | |
345 | ||
346 | elif version == b'v11': | |
347 | if self._v11_key is None: | |
063c409d | 348 | self._logger.warning(f'cannot decrypt cookie {KEYRING_UNAVAILABLE_REASON}', only_once=True) |
982ee69a MB |
349 | return None |
350 | return _decrypt_aes_cbc(ciphertext, self._v11_key, self._logger) | |
351 | ||
352 | else: | |
353 | return None | |
354 | ||
355 | ||
356 | class MacChromeCookieDecryptor(ChromeCookieDecryptor): | |
357 | def __init__(self, browser_keyring_name, logger): | |
358 | self._logger = logger | |
359 | password = _get_mac_keyring_password(browser_keyring_name) | |
360 | self._v10_key = None if password is None else self.derive_key(password) | |
361 | ||
362 | @staticmethod | |
363 | def derive_key(password): | |
364 | # values from | |
365 | # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_mac.mm | |
366 | return pbkdf2_sha1(password, salt=b'saltysalt', iterations=1003, key_length=16) | |
367 | ||
368 | def decrypt(self, encrypted_value): | |
369 | version = encrypted_value[:3] | |
370 | ciphertext = encrypted_value[3:] | |
371 | ||
372 | if version == b'v10': | |
373 | if self._v10_key is None: | |
374 | self._logger.warning('cannot decrypt v10 cookies: no key found', only_once=True) | |
375 | return None | |
376 | ||
377 | return _decrypt_aes_cbc(ciphertext, self._v10_key, self._logger) | |
378 | ||
379 | else: | |
380 | # other prefixes are considered 'old data' which were stored as plaintext | |
381 | # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_mac.mm | |
382 | return encrypted_value | |
383 | ||
384 | ||
385 | class WindowsChromeCookieDecryptor(ChromeCookieDecryptor): | |
386 | def __init__(self, browser_root, logger): | |
387 | self._logger = logger | |
388 | self._v10_key = _get_windows_v10_key(browser_root, logger) | |
389 | ||
390 | def decrypt(self, encrypted_value): | |
391 | version = encrypted_value[:3] | |
392 | ciphertext = encrypted_value[3:] | |
393 | ||
394 | if version == b'v10': | |
395 | if self._v10_key is None: | |
396 | self._logger.warning('cannot decrypt v10 cookies: no key found', only_once=True) | |
397 | return None | |
edf65256 | 398 | elif not compat_pycrypto_AES: |
982ee69a MB |
399 | self._logger.warning('cannot decrypt cookie as the `pycryptodome` module is not installed. ' |
400 | 'Please install by running `python3 -m pip install pycryptodome`', | |
401 | only_once=True) | |
402 | return None | |
403 | ||
404 | # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_win.cc | |
405 | # kNonceLength | |
406 | nonce_length = 96 // 8 | |
407 | # boringssl | |
408 | # EVP_AEAD_AES_GCM_TAG_LEN | |
409 | authentication_tag_length = 16 | |
410 | ||
411 | raw_ciphertext = ciphertext | |
412 | nonce = raw_ciphertext[:nonce_length] | |
413 | ciphertext = raw_ciphertext[nonce_length:-authentication_tag_length] | |
414 | authentication_tag = raw_ciphertext[-authentication_tag_length:] | |
415 | ||
416 | return _decrypt_aes_gcm(ciphertext, self._v10_key, nonce, authentication_tag, self._logger) | |
417 | ||
418 | else: | |
419 | # any other prefix means the data is DPAPI encrypted | |
420 | # https://chromium.googlesource.com/chromium/src/+/refs/heads/main/components/os_crypt/os_crypt_win.cc | |
421 | return _decrypt_windows_dpapi(encrypted_value, self._logger).decode('utf-8') | |
422 | ||
423 | ||
424 | def _extract_safari_cookies(profile, logger): | |
425 | if profile is not None: | |
426 | logger.error('safari does not support profiles') | |
427 | if sys.platform != 'darwin': | |
428 | raise ValueError('unsupported platform: {}'.format(sys.platform)) | |
429 | ||
430 | cookies_path = os.path.expanduser('~/Library/Cookies/Cookies.binarycookies') | |
431 | ||
432 | if not os.path.isfile(cookies_path): | |
433 | raise FileNotFoundError('could not find safari cookies database') | |
434 | ||
435 | with open(cookies_path, 'rb') as f: | |
436 | cookies_data = f.read() | |
437 | ||
438 | jar = parse_safari_cookies(cookies_data, logger=logger) | |
439 | logger.info('Extracted {} cookies from safari'.format(len(jar))) | |
440 | return jar | |
441 | ||
442 | ||
443 | class ParserError(Exception): | |
444 | pass | |
445 | ||
446 | ||
447 | class DataParser: | |
448 | def __init__(self, data, logger): | |
449 | self._data = data | |
450 | self.cursor = 0 | |
451 | self._logger = logger | |
452 | ||
453 | def read_bytes(self, num_bytes): | |
454 | if num_bytes < 0: | |
455 | raise ParserError('invalid read of {} bytes'.format(num_bytes)) | |
456 | end = self.cursor + num_bytes | |
457 | if end > len(self._data): | |
458 | raise ParserError('reached end of input') | |
459 | data = self._data[self.cursor:end] | |
460 | self.cursor = end | |
461 | return data | |
462 | ||
463 | def expect_bytes(self, expected_value, message): | |
464 | value = self.read_bytes(len(expected_value)) | |
465 | if value != expected_value: | |
466 | raise ParserError('unexpected value: {} != {} ({})'.format(value, expected_value, message)) | |
467 | ||
468 | def read_uint(self, big_endian=False): | |
469 | data_format = '>I' if big_endian else '<I' | |
470 | return struct.unpack(data_format, self.read_bytes(4))[0] | |
471 | ||
472 | def read_double(self, big_endian=False): | |
473 | data_format = '>d' if big_endian else '<d' | |
474 | return struct.unpack(data_format, self.read_bytes(8))[0] | |
475 | ||
476 | def read_cstring(self): | |
477 | buffer = [] | |
478 | while True: | |
479 | c = self.read_bytes(1) | |
480 | if c == b'\x00': | |
481 | return b''.join(buffer).decode('utf-8') | |
482 | else: | |
483 | buffer.append(c) | |
484 | ||
485 | def skip(self, num_bytes, description='unknown'): | |
486 | if num_bytes > 0: | |
487 | self._logger.debug('skipping {} bytes ({}): {}'.format( | |
488 | num_bytes, description, self.read_bytes(num_bytes))) | |
489 | elif num_bytes < 0: | |
490 | raise ParserError('invalid skip of {} bytes'.format(num_bytes)) | |
491 | ||
492 | def skip_to(self, offset, description='unknown'): | |
493 | self.skip(offset - self.cursor, description) | |
494 | ||
495 | def skip_to_end(self, description='unknown'): | |
496 | self.skip_to(len(self._data), description) | |
497 | ||
498 | ||
499 | def _mac_absolute_time_to_posix(timestamp): | |
500 | return int((datetime(2001, 1, 1, 0, 0, tzinfo=timezone.utc) + timedelta(seconds=timestamp)).timestamp()) | |
501 | ||
502 | ||
503 | def _parse_safari_cookies_header(data, logger): | |
504 | p = DataParser(data, logger) | |
505 | p.expect_bytes(b'cook', 'database signature') | |
506 | number_of_pages = p.read_uint(big_endian=True) | |
507 | page_sizes = [p.read_uint(big_endian=True) for _ in range(number_of_pages)] | |
508 | return page_sizes, p.cursor | |
509 | ||
510 | ||
511 | def _parse_safari_cookies_page(data, jar, logger): | |
512 | p = DataParser(data, logger) | |
513 | p.expect_bytes(b'\x00\x00\x01\x00', 'page signature') | |
514 | number_of_cookies = p.read_uint() | |
515 | record_offsets = [p.read_uint() for _ in range(number_of_cookies)] | |
516 | if number_of_cookies == 0: | |
517 | logger.debug('a cookies page of size {} has no cookies'.format(len(data))) | |
518 | return | |
519 | ||
520 | p.skip_to(record_offsets[0], 'unknown page header field') | |
521 | ||
522 | for record_offset in record_offsets: | |
523 | p.skip_to(record_offset, 'space between records') | |
524 | record_length = _parse_safari_cookies_record(data[record_offset:], jar, logger) | |
525 | p.read_bytes(record_length) | |
526 | p.skip_to_end('space in between pages') | |
527 | ||
528 | ||
529 | def _parse_safari_cookies_record(data, jar, logger): | |
530 | p = DataParser(data, logger) | |
531 | record_size = p.read_uint() | |
532 | p.skip(4, 'unknown record field 1') | |
533 | flags = p.read_uint() | |
534 | is_secure = bool(flags & 0x0001) | |
535 | p.skip(4, 'unknown record field 2') | |
536 | domain_offset = p.read_uint() | |
537 | name_offset = p.read_uint() | |
538 | path_offset = p.read_uint() | |
539 | value_offset = p.read_uint() | |
540 | p.skip(8, 'unknown record field 3') | |
541 | expiration_date = _mac_absolute_time_to_posix(p.read_double()) | |
542 | _creation_date = _mac_absolute_time_to_posix(p.read_double()) # noqa: F841 | |
543 | ||
544 | try: | |
545 | p.skip_to(domain_offset) | |
546 | domain = p.read_cstring() | |
547 | ||
548 | p.skip_to(name_offset) | |
549 | name = p.read_cstring() | |
550 | ||
551 | p.skip_to(path_offset) | |
552 | path = p.read_cstring() | |
553 | ||
554 | p.skip_to(value_offset) | |
555 | value = p.read_cstring() | |
556 | except UnicodeDecodeError: | |
f9be9cb9 | 557 | logger.warning('failed to parse cookie because UTF-8 decoding failed', only_once=True) |
982ee69a MB |
558 | return record_size |
559 | ||
560 | p.skip_to(record_size, 'space at the end of the record') | |
561 | ||
562 | cookie = compat_cookiejar_Cookie( | |
563 | version=0, name=name, value=value, port=None, port_specified=False, | |
564 | domain=domain, domain_specified=bool(domain), domain_initial_dot=domain.startswith('.'), | |
565 | path=path, path_specified=bool(path), secure=is_secure, expires=expiration_date, discard=False, | |
566 | comment=None, comment_url=None, rest={}) | |
567 | jar.set_cookie(cookie) | |
568 | return record_size | |
569 | ||
570 | ||
571 | def parse_safari_cookies(data, jar=None, logger=YDLLogger()): | |
572 | """ | |
573 | References: | |
574 | - https://github.com/libyal/dtformats/blob/main/documentation/Safari%20Cookies.asciidoc | |
575 | - this data appears to be out of date but the important parts of the database structure is the same | |
576 | - there are a few bytes here and there which are skipped during parsing | |
577 | """ | |
578 | if jar is None: | |
579 | jar = YoutubeDLCookieJar() | |
580 | page_sizes, body_start = _parse_safari_cookies_header(data, logger) | |
581 | p = DataParser(data[body_start:], logger) | |
582 | for page_size in page_sizes: | |
583 | _parse_safari_cookies_page(p.read_bytes(page_size), jar, logger) | |
584 | p.skip_to_end('footer') | |
585 | return jar | |
586 | ||
587 | ||
588 | def _get_linux_keyring_password(browser_keyring_name): | |
589 | password = keyring.get_password('{} Keys'.format(browser_keyring_name), | |
590 | '{} Safe Storage'.format(browser_keyring_name)) | |
591 | if password is None: | |
592 | # this sometimes occurs in KDE because chrome does not check hasEntry and instead | |
593 | # just tries to read the value (which kwallet returns "") whereas keyring checks hasEntry | |
594 | # to verify this: | |
595 | # dbus-monitor "interface='org.kde.KWallet'" "type=method_return" | |
596 | # while starting chrome. | |
597 | # this may be a bug as the intended behaviour is to generate a random password and store | |
598 | # it, but that doesn't matter here. | |
599 | password = '' | |
600 | return password.encode('utf-8') | |
601 | ||
602 | ||
603 | def _get_mac_keyring_password(browser_keyring_name): | |
604 | if KEYRING_AVAILABLE: | |
605 | password = keyring.get_password('{} Safe Storage'.format(browser_keyring_name), browser_keyring_name) | |
606 | return password.encode('utf-8') | |
607 | else: | |
608 | proc = subprocess.Popen(['security', 'find-generic-password', | |
609 | '-w', # write password to stdout | |
610 | '-a', browser_keyring_name, # match 'account' | |
611 | '-s', '{} Safe Storage'.format(browser_keyring_name)], # match 'service' | |
612 | stdout=subprocess.PIPE, | |
613 | stderr=subprocess.DEVNULL) | |
614 | try: | |
615 | stdout, stderr = process_communicate_or_kill(proc) | |
616 | return stdout | |
617 | except BaseException: | |
618 | return None | |
619 | ||
620 | ||
621 | def _get_windows_v10_key(browser_root, logger): | |
622 | path = _find_most_recently_used_file(browser_root, 'Local State') | |
623 | if path is None: | |
624 | logger.error('could not find local state file') | |
625 | return None | |
626 | with open(path, 'r') as f: | |
627 | data = json.load(f) | |
628 | try: | |
629 | base64_key = data['os_crypt']['encrypted_key'] | |
630 | except KeyError: | |
631 | logger.error('no encrypted key in Local State') | |
632 | return None | |
633 | encrypted_key = compat_b64decode(base64_key) | |
634 | prefix = b'DPAPI' | |
635 | if not encrypted_key.startswith(prefix): | |
636 | logger.error('invalid key') | |
637 | return None | |
638 | return _decrypt_windows_dpapi(encrypted_key[len(prefix):], logger) | |
639 | ||
640 | ||
641 | def pbkdf2_sha1(password, salt, iterations, key_length): | |
642 | return pbkdf2_hmac('sha1', password, salt, iterations, key_length) | |
643 | ||
644 | ||
645 | def _decrypt_aes_cbc(ciphertext, key, logger, initialization_vector=b' ' * 16): | |
646 | plaintext = aes_cbc_decrypt(bytes_to_intlist(ciphertext), | |
647 | bytes_to_intlist(key), | |
648 | bytes_to_intlist(initialization_vector)) | |
649 | padding_length = plaintext[-1] | |
650 | try: | |
651 | return intlist_to_bytes(plaintext[:-padding_length]).decode('utf-8') | |
652 | except UnicodeDecodeError: | |
f9be9cb9 | 653 | logger.warning('failed to decrypt cookie because UTF-8 decoding failed. Possibly the key is wrong?', only_once=True) |
982ee69a MB |
654 | return None |
655 | ||
656 | ||
657 | def _decrypt_aes_gcm(ciphertext, key, nonce, authentication_tag, logger): | |
edf65256 | 658 | cipher = compat_pycrypto_AES.new(key, compat_pycrypto_AES.MODE_GCM, nonce) |
982ee69a MB |
659 | try: |
660 | plaintext = cipher.decrypt_and_verify(ciphertext, authentication_tag) | |
661 | except ValueError: | |
f9be9cb9 | 662 | logger.warning('failed to decrypt cookie because the MAC check failed. Possibly the key is wrong?', only_once=True) |
982ee69a MB |
663 | return None |
664 | ||
665 | try: | |
666 | return plaintext.decode('utf-8') | |
667 | except UnicodeDecodeError: | |
f9be9cb9 | 668 | logger.warning('failed to decrypt cookie because UTF-8 decoding failed. Possibly the key is wrong?', only_once=True) |
982ee69a MB |
669 | return None |
670 | ||
671 | ||
672 | def _decrypt_windows_dpapi(ciphertext, logger): | |
673 | """ | |
674 | References: | |
675 | - https://docs.microsoft.com/en-us/windows/win32/api/dpapi/nf-dpapi-cryptunprotectdata | |
676 | """ | |
677 | from ctypes.wintypes import DWORD | |
678 | ||
679 | class DATA_BLOB(ctypes.Structure): | |
680 | _fields_ = [('cbData', DWORD), | |
681 | ('pbData', ctypes.POINTER(ctypes.c_char))] | |
682 | ||
683 | buffer = ctypes.create_string_buffer(ciphertext) | |
684 | blob_in = DATA_BLOB(ctypes.sizeof(buffer), buffer) | |
685 | blob_out = DATA_BLOB() | |
686 | ret = ctypes.windll.crypt32.CryptUnprotectData( | |
687 | ctypes.byref(blob_in), # pDataIn | |
688 | None, # ppszDataDescr: human readable description of pDataIn | |
689 | None, # pOptionalEntropy: salt? | |
690 | None, # pvReserved: must be NULL | |
691 | None, # pPromptStruct: information about prompts to display | |
692 | 0, # dwFlags | |
693 | ctypes.byref(blob_out) # pDataOut | |
694 | ) | |
695 | if not ret: | |
f9be9cb9 | 696 | logger.warning('failed to decrypt with DPAPI', only_once=True) |
982ee69a MB |
697 | return None |
698 | ||
699 | result = ctypes.string_at(blob_out.pbData, blob_out.cbData) | |
700 | ctypes.windll.kernel32.LocalFree(blob_out.pbData) | |
701 | return result | |
702 | ||
703 | ||
704 | def _config_home(): | |
705 | return os.environ.get('XDG_CONFIG_HOME', os.path.expanduser('~/.config')) | |
706 | ||
707 | ||
708 | def _open_database_copy(database_path, tmpdir): | |
709 | # cannot open sqlite databases if they are already in use (e.g. by the browser) | |
710 | database_copy_path = os.path.join(tmpdir, 'temporary.sqlite') | |
711 | shutil.copy(database_path, database_copy_path) | |
712 | conn = sqlite3.connect(database_copy_path) | |
713 | return conn.cursor() | |
714 | ||
715 | ||
716 | def _get_column_names(cursor, table_name): | |
717 | table_info = cursor.execute('PRAGMA table_info({})'.format(table_name)).fetchall() | |
718 | return [row[1].decode('utf-8') for row in table_info] | |
719 | ||
720 | ||
721 | def _find_most_recently_used_file(root, filename): | |
722 | # if there are multiple browser profiles, take the most recently used one | |
723 | paths = [] | |
724 | for root, dirs, files in os.walk(root): | |
725 | for file in files: | |
726 | if file == filename: | |
727 | paths.append(os.path.join(root, file)) | |
728 | return None if not paths else max(paths, key=lambda path: os.lstat(path).st_mtime) | |
729 | ||
730 | ||
731 | def _merge_cookie_jars(jars): | |
732 | output_jar = YoutubeDLCookieJar() | |
733 | for jar in jars: | |
734 | for cookie in jar: | |
735 | output_jar.set_cookie(cookie) | |
736 | if jar.filename is not None: | |
737 | output_jar.filename = jar.filename | |
738 | return output_jar | |
739 | ||
740 | ||
741 | def _is_path(value): | |
742 | return os.path.sep in value | |
743 | ||
744 | ||
745 | def _parse_browser_specification(browser_name, profile=None): | |
ca46b941 | 746 | browser_name = browser_name.lower() |
982ee69a MB |
747 | if browser_name not in SUPPORTED_BROWSERS: |
748 | raise ValueError(f'unsupported browser: "{browser_name}"') | |
749 | if profile is not None and _is_path(profile): | |
750 | profile = os.path.expanduser(profile) | |
751 | return browser_name, profile |