administration/yt-dlp
1
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-09 10:50:38 +00:00
Code Issues Projects Releases Wiki Activity

[networking] Remove dot segments during URL normalization (#7662)

Browse Source 
This implements RFC3986 5.2.4 remove_dot_segments during the URL normalization process.

Closes #3355, #6526

Authored by: coletdjnz
This commit is contained in:
coletdjnz 2023-07-29 10:40:20 +12:00 committed by GitHub
parent a15fcd299e
commit 4bf912282a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
8 changed files with 104 additions and 36 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
test/test_networking.py
View File

@ -173,6 +173,12 @@ class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler):
self.send_header('Location', self.path)
self.send_header('Content-Length', '0')
self.end_headers()
elif self.path == '/redirect_dotsegments':
self.send_response(301)
# redirect to /headers but with dot segments before
self.send_header('Location', '/a/b/./../../headers')
self.send_header('Content-Length', '0')
self.end_headers()
elif self.path.startswith('/redirect_'):
self._redirect()
elif self.path.startswith('/method'):
@ -355,6 +361,21 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert res.status == 200
res.close()
@pytest.mark.parametrize('handler', ['Urllib'], indirect=True)
def test_remove_dot_segments(self, handler):
with handler() as rh:
# This isn't a comprehensive test,
# but it should be enough to check whether the handler is removing dot segments
res = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/a/b/./../../headers'))
assert res.status == 200
assert res.url == f'http://127.0.0.1:{self.http_port}/headers'
res.close()
res = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_dotsegments'))
assert res.status == 200
assert res.url == f'http://127.0.0.1:{self.http_port}/headers'
res.close()
@pytest.mark.parametrize('handler', ['Urllib'], indirect=True)
def test_unicode_path_redirection(self, handler):
with handler() as rh:

42
test/test_utils.py
View File

@ -47,8 +47,6 @@ from yt_dlp.utils import (
encode_base_n,
encode_compat_str,
encodeFilename,
escape_rfc3986,
escape_url,
expand_path,
extract_attributes,
extract_basic_auth,
@ -132,7 +130,12 @@ from yt_dlp.utils import (
xpath_text,
xpath_with_ns,
)
from yt_dlp.utils.networking import HTTPHeaderDict
from yt_dlp.utils.networking import (
HTTPHeaderDict,
escape_rfc3986,
normalize_url,
remove_dot_segments,
)
class TestUtil(unittest.TestCase):
@ -933,24 +936,45 @@ class TestUtil(unittest.TestCase):
self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
def test_escape_url(self):
def test_normalize_url(self):
self.assertEqual(
escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
normalize_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
)
self.assertEqual(
escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
normalize_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
)
self.assertEqual(
escape_url('http://тест.рф/фрагмент'),
normalize_url('http://тест.рф/фрагмент'),
'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
)
self.assertEqual(
escape_url('http://тест.рф/абв?абв=абв#абв'),
normalize_url('http://тест.рф/абв?абв=абв#абв'),
'http://xn--e1aybc.xn--p1ai/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
)
self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
self.assertEqual(normalize_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
self.assertEqual(normalize_url('http://www.example.com/../a/b/../c/./d.html'), 'http://www.example.com/a/c/d.html')
def test_remove_dot_segments(self):
self.assertEqual(remove_dot_segments('/a/b/c/./../../g'), '/a/g')
self.assertEqual(remove_dot_segments('mid/content=5/../6'), 'mid/6')
self.assertEqual(remove_dot_segments('/ad/../cd'), '/cd')
self.assertEqual(remove_dot_segments('/ad/../cd/'), '/cd/')
self.assertEqual(remove_dot_segments('/..'), '/')
self.assertEqual(remove_dot_segments('/./'), '/')
self.assertEqual(remove_dot_segments('/./a'), '/a')
self.assertEqual(remove_dot_segments('/abc/./.././d/././e/.././f/./../../ghi'), '/ghi')
self.assertEqual(remove_dot_segments('/'), '/')
self.assertEqual(remove_dot_segments('/t'), '/t')
self.assertEqual(remove_dot_segments('t'), 't')
self.assertEqual(remove_dot_segments(''), '')
self.assertEqual(remove_dot_segments('/../a/b/c'), '/a/b/c')
self.assertEqual(remove_dot_segments('../a'), 'a')
self.assertEqual(remove_dot_segments('./a'), 'a')
self.assertEqual(remove_dot_segments('.'), '')
self.assertEqual(remove_dot_segments('////'), '////')
def test_js_to_json_vars_strings(self):
self.assertDictEqual(

6
yt_dlp/cookies.py
View File

@ -33,7 +33,6 @@ from .minicurses import MultilinePrinter, QuietMultilinePrinter
from .utils import (
Popen,
error_to_str,
escape_url,
expand_path,
is_path_like,
sanitize_url,
@ -42,6 +41,7 @@ from .utils import (
write_string,
)
from .utils._utils import _YDLLogger
from .utils.networking import normalize_url
CHROMIUM_BASED_BROWSERS = {'brave', 'chrome', 'chromium', 'edge', 'opera', 'vivaldi'}
SUPPORTED_BROWSERS = CHROMIUM_BASED_BROWSERS | {'firefox', 'safari'}
@ -1308,7 +1308,7 @@ class YoutubeDLCookieJar(http.cookiejar.MozillaCookieJar):
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)))
cookie_req = urllib.request.Request(normalize_url(sanitize_url(url)))
self.add_cookie_header(cookie_req)
return cookie_req.get_header('Cookie')
@ -1317,7 +1317,7 @@ class YoutubeDLCookieJar(http.cookiejar.MozillaCookieJar):
# Policy `_now` attribute must be set before calling `_cookies_for_request`
# Ref: https://github.com/python/cpython/blob/3.7/Lib/http/cookiejar.py#L1360
self._policy._now = self._now = int(time.time())
return self._cookies_for_request(urllib.request.Request(escape_url(sanitize_url(url))))
return self._cookies_for_request(urllib.request.Request(normalize_url(sanitize_url(url))))
def clear(self, *args, **kwargs):
with contextlib.suppress(KeyError):

7
yt_dlp/networking/_urllib.py
View File

@ -41,7 +41,8 @@ from .exceptions import (
from ..dependencies import brotli
from ..socks import ProxyError as SocksProxyError
from ..socks import sockssocket
from ..utils import escape_url, update_url_query
from ..utils import update_url_query
from ..utils.networking import normalize_url
SUPPORTED_ENCODINGS = ['gzip', 'deflate']
CONTENT_DECODE_ERRORS = [zlib.error, OSError]
@ -179,7 +180,7 @@ class HTTPHandler(urllib.request.AbstractHTTPHandler):
# Since redirects are also affected (e.g. http://www.southpark.de/alle-episoden/s18e09)
# the code of this workaround has been moved here from YoutubeDL.urlopen()
url = req.get_full_url()
url_escaped = escape_url(url)
url_escaped = normalize_url(url)
# Substitute URL if any change after escaping
if url != url_escaped:
@ -212,7 +213,7 @@ class HTTPHandler(urllib.request.AbstractHTTPHandler):
if location:
# As of RFC 2616 default charset is iso-8859-1 that is respected by python 3
location = location.encode('iso-8859-1').decode()
location_escaped = escape_url(location)
location_escaped = normalize_url(location)
if location != location_escaped:
del resp.headers['Location']
resp.headers['Location'] = location_escaped

5
yt_dlp/networking/common.py
View File

@ -27,10 +27,9 @@ from ..utils import (
classproperty,
deprecation_warning,
error_to_str,
escape_url,
update_url_query,
)
from ..utils.networking import HTTPHeaderDict
from ..utils.networking import HTTPHeaderDict, normalize_url
if typing.TYPE_CHECKING:
RequestData = bytes | Iterable[bytes] | typing.IO | None
@ -372,7 +371,7 @@ class Request:
raise TypeError('url must be a string')
elif url.startswith('//'):
url = 'http:' + url
self._url = escape_url(url)
self._url = normalize_url(url)
@property
def method(self):

4
yt_dlp/utils/_legacy.py
View File

@ -8,6 +8,8 @@ import urllib.request
import zlib
from ._utils import Popen, decode_base_n, preferredencoding
from .networking import escape_rfc3986 # noqa: F401
from .networking import normalize_url as escape_url # noqa: F401
from .traversal import traverse_obj
from ..dependencies import certifi, websockets
from ..networking._helper import make_ssl_context
@ -197,7 +199,7 @@ def request_to_url(req):
def sanitized_Request(url, *args, **kwargs):
from ..utils import escape_url, extract_basic_auth, sanitize_url
from ..utils import extract_basic_auth, sanitize_url
url, auth_header = extract_basic_auth(escape_url(sanitize_url(url)))
if auth_header is not None:
headers = args[1] if len(args) >= 2 else kwargs.setdefault('headers', {})

17
yt_dlp/utils/_utils.py
View File

@ -2464,23 +2464,6 @@ def lowercase_escape(s):
s)
def escape_rfc3986(s):
"""Escape non-ASCII characters as suggested by RFC 3986"""
return urllib.parse.quote(s, b"%/;:@&=+$,!~*'()?#[]")
def escape_url(url):
"""Escape URL as suggested by RFC 3986"""
url_parsed = urllib.parse.urlparse(url)
return url_parsed._replace(
netloc=url_parsed.netloc.encode('idna').decode('ascii'),
path=escape_rfc3986(url_parsed.path),
params=escape_rfc3986(url_parsed.params),
query=escape_rfc3986(url_parsed.query),
fragment=escape_rfc3986(url_parsed.fragment)
).geturl()
def parse_qs(url, **kwargs):
return urllib.parse.parse_qs(urllib.parse.urlparse(url).query, **kwargs)

38
yt_dlp/utils/networking.py
View File

@ -121,3 +121,41 @@ def clean_headers(headers: HTTPHeaderDict):
if 'Youtubedl-No-Compression' in headers: # compat
del headers['Youtubedl-No-Compression']
headers['Accept-Encoding'] = 'identity'
def remove_dot_segments(path):
# Implements RFC3986 5.2.4 remote_dot_segments
# Pseudo-code: https://tools.ietf.org/html/rfc3986#section-5.2.4
# https://github.com/urllib3/urllib3/blob/ba49f5c4e19e6bca6827282feb77a3c9f937e64b/src/urllib3/util/url.py#L263
output = []
segments = path.split('/')
for s in segments:
if s == '.':
continue
elif s == '..':
if output:
output.pop()
else:
output.append(s)
if not segments[0] and (not output or output[0]):
output.insert(0, '')
if segments[-1] in ('.', '..'):
output.append('')
return '/'.join(output)
def escape_rfc3986(s):
"""Escape non-ASCII characters as suggested by RFC 3986"""
return urllib.parse.quote(s, b"%/;:@&=+$,!~*'()?#[]")
def normalize_url(url):
"""Normalize URL as suggested by RFC 3986"""
url_parsed = urllib.parse.urlparse(url)
return url_parsed._replace(
netloc=url_parsed.netloc.encode('idna').decode('ascii'),
path=escape_rfc3986(remove_dot_segments(url_parsed.path)),
params=escape_rfc3986(url_parsed.params),
query=escape_rfc3986(url_parsed.query),
fragment=escape_rfc3986(url_parsed.fragment)
).geturl()