X-Git-Url: https://jfr.im/git/yt-dlp.git/blobdiff_plain/08916a49c777cb6e000eec092881eb93ec22076c..b87e01c123fd560b6a674ce00f45a9459d82d98a:/yt_dlp/cookies.py diff --git a/yt_dlp/cookies.py b/yt_dlp/cookies.py index 4cafb522e..eb6a2656b 100644 --- a/yt_dlp/cookies.py +++ b/yt_dlp/cookies.py @@ -1,7 +1,9 @@ import base64 +import collections import contextlib import http.cookiejar import http.cookies +import io import json import os import re @@ -11,6 +13,7 @@ import sys import tempfile import time +import urllib.request from datetime import datetime, timedelta, timezone from enum import Enum, auto from hashlib import pbkdf2_hmac @@ -29,11 +32,14 @@ from .minicurses import MultilinePrinter, QuietMultilinePrinter from .utils import ( Popen, - YoutubeDLCookieJar, error_to_str, + escape_url, expand_path, is_path_like, + sanitize_url, + str_or_none, try_call, + write_string, ) CHROMIUM_BASED_BROWSERS = {'brave', 'chrome', 'chromium', 'edge', 'opera', 'vivaldi'} @@ -1091,3 +1097,139 @@ def load(self, data): else: morsel = None + + +class YoutubeDLCookieJar(http.cookiejar.MozillaCookieJar): + """ + See [1] for cookie file format. + + 1. https://curl.haxx.se/docs/http-cookies.html + """ + _HTTPONLY_PREFIX = '#HttpOnly_' + _ENTRY_LEN = 7 + _HEADER = '''# Netscape HTTP Cookie File +# This file is generated by yt-dlp. Do not edit. + +''' + _CookieFileEntry = collections.namedtuple( + 'CookieFileEntry', + ('domain_name', 'include_subdomains', 'path', 'https_only', 'expires_at', 'name', 'value')) + + def __init__(self, filename=None, *args, **kwargs): + super().__init__(None, *args, **kwargs) + if is_path_like(filename): + filename = os.fspath(filename) + self.filename = filename + + @staticmethod + def _true_or_false(cndn): + return 'TRUE' if cndn else 'FALSE' + + @contextlib.contextmanager + def open(self, file, *, write=False): + if is_path_like(file): + with open(file, 'w' if write else 'r', encoding='utf-8') as f: + yield f + else: + if write: + file.truncate(0) + yield file + + def _really_save(self, f, ignore_discard=False, ignore_expires=False): + now = time.time() + for cookie in self: + if (not ignore_discard and cookie.discard + or not ignore_expires and cookie.is_expired(now)): + continue + name, value = cookie.name, cookie.value + if value is None: + # cookies.txt regards 'Set-Cookie: foo' as a cookie + # with no name, whereas http.cookiejar regards it as a + # cookie with no value. + name, value = '', name + f.write('%s\n' % '\t'.join(( + cookie.domain, + self._true_or_false(cookie.domain.startswith('.')), + cookie.path, + self._true_or_false(cookie.secure), + str_or_none(cookie.expires, default=''), + name, value + ))) + + def save(self, filename=None, *args, **kwargs): + """ + Save cookies to a file. + Code is taken from CPython 3.6 + https://github.com/python/cpython/blob/8d999cbf4adea053be6dbb612b9844635c4dfb8e/Lib/http/cookiejar.py#L2091-L2117 """ + + if filename is None: + if self.filename is not None: + filename = self.filename + else: + raise ValueError(http.cookiejar.MISSING_FILENAME_TEXT) + + # Store session cookies with `expires` set to 0 instead of an empty string + for cookie in self: + if cookie.expires is None: + cookie.expires = 0 + + with self.open(filename, write=True) as f: + f.write(self._HEADER) + self._really_save(f, *args, **kwargs) + + def load(self, filename=None, ignore_discard=False, ignore_expires=False): + """Load cookies from a file.""" + if filename is None: + if self.filename is not None: + filename = self.filename + else: + raise ValueError(http.cookiejar.MISSING_FILENAME_TEXT) + + def prepare_line(line): + if line.startswith(self._HTTPONLY_PREFIX): + line = line[len(self._HTTPONLY_PREFIX):] + # comments and empty lines are fine + if line.startswith('#') or not line.strip(): + return line + cookie_list = line.split('\t') + if len(cookie_list) != self._ENTRY_LEN: + raise http.cookiejar.LoadError('invalid length %d' % len(cookie_list)) + cookie = self._CookieFileEntry(*cookie_list) + if cookie.expires_at and not cookie.expires_at.isdigit(): + raise http.cookiejar.LoadError('invalid expires at %s' % cookie.expires_at) + return line + + cf = io.StringIO() + with self.open(filename) as f: + for line in f: + try: + cf.write(prepare_line(line)) + except http.cookiejar.LoadError as e: + if f'{line.strip()} '[0] in '[{"': + raise http.cookiejar.LoadError( + 'Cookies file must be Netscape formatted, not JSON. See ' + 'https://github.com/yt-dlp/yt-dlp/wiki/FAQ#how-do-i-pass-cookies-to-yt-dlp') + write_string(f'WARNING: skipping cookie file entry due to {e}: {line!r}\n') + continue + cf.seek(0) + self._really_load(cf, filename, ignore_discard, ignore_expires) + # Session cookies are denoted by either `expires` field set to + # an empty string or 0. MozillaCookieJar only recognizes the former + # (see [1]). So we need force the latter to be recognized as session + # cookies on our own. + # Session cookies may be important for cookies-based authentication, + # e.g. usually, when user does not check 'Remember me' check box while + # logging in on a site, some important cookies are stored as session + # cookies so that not recognizing them will result in failed login. + # 1. https://bugs.python.org/issue17164 + for cookie in self: + # Treat `expires=0` cookies as session cookies + if cookie.expires == 0: + cookie.expires = None + cookie.discard = True + + def get_cookie_header(self, url): + """Generate a Cookie HTTP header for a given url""" + cookie_req = urllib.request.Request(escape_url(sanitize_url(url))) + self.add_cookie_header(cookie_req) + return cookie_req.get_header('Cookie')