[rh:requests] Add handler for requests HTTP library (#3668)

Adds support for HTTPS proxies and persistent connections (keep-alive) Closes #1890 Resolves #4070 Resolves ytdl-org/youtube-dl#32549 Resolves ytdl-org/youtube-dl#14523 Resolves ytdl-org/youtube-dl#13734 Authored by: coletdjnz, Grub4K, bashonly
yt-dlp · Oct 13, 2023 · 8a8b545 · 8a8b545
1 parent 700444c
commit 8a8b545
Show file tree

Hide file tree

Showing 14 changed files with 619 additions and 79 deletions.
diff --git a/.github/workflows/core.yml b/.github/workflows/core.yml
@@ -32,7 +32,7 @@ jobs:
       uses: actions/setup-python@v4
       with:
         python-version: ${{ matrix.python-version }}
-    - name: Install pytest
+    - name: Install dependencies
       run: pip install pytest -r requirements.txt
     - name: Run tests
       continue-on-error: False

diff --git a/README.md b/README.md
@@ -157,14 +157,15 @@ Some of yt-dlp's default options are different from that of youtube-dl and youtu
 * yt-dlp's sanitization of invalid characters in filenames is different/smarter than in youtube-dl. You can use `--compat-options filename-sanitization` to revert to youtube-dl's behavior
 * yt-dlp tries to parse the external downloader outputs into the standard progress output if possible (Currently implemented: [~~aria2c~~](https://github.com/yt-dlp/yt-dlp/issues/5931)). You can use `--compat-options no-external-downloader-progress` to get the downloader output as-is
 * yt-dlp versions between 2021.09.01 and 2023.01.02 applies `--match-filter` to nested playlists. This was an unintentional side-effect of [8f18ac](https://github.com/yt-dlp/yt-dlp/commit/8f18aca8717bb0dd49054555af8d386e5eda3a88) and is fixed in [d7b460](https://github.com/yt-dlp/yt-dlp/commit/d7b460d0e5fc710950582baed2e3fc616ed98a80). Use `--compat-options playlist-match-filter` to revert this
+* yt-dlp uses modern http client backends such as `requests`. Use `--compat-options prefer-legacy-http-handler` to prefer the legacy http handler (`urllib`) to be used for standard http requests.
 
 For ease of use, a few more compat options are available:
 
 * `--compat-options all`: Use all compat options (Do NOT use)
 * `--compat-options youtube-dl`: Same as `--compat-options all,-multistreams,-playlist-match-filter`
 * `--compat-options youtube-dlc`: Same as `--compat-options all,-no-live-chat,-no-youtube-channel-redirect,-playlist-match-filter`
 * `--compat-options 2021`: Same as `--compat-options 2022,no-certifi,filename-sanitization,no-youtube-prefer-utc-upload-date`
-* `--compat-options 2022`: Same as `--compat-options playlist-match-filter,no-external-downloader-progress`. Use this to enable all future compat options
+* `--compat-options 2022`: Same as `--compat-options playlist-match-filter,no-external-downloader-progress,prefer-legacy-http-handler`. Use this to enable all future compat options
 
 
 # INSTALLATION
@@ -274,6 +275,7 @@ While all the other dependencies are optional, `ffmpeg` and `ffprobe` are highly
 * [**certifi**](https://github.com/certifi/python-certifi)\* - Provides Mozilla's root certificate bundle. Licensed under [MPLv2](https://github.com/certifi/python-certifi/blob/master/LICENSE)
 * [**brotli**](https://github.com/google/brotli)\* or [**brotlicffi**](https://github.com/python-hyper/brotlicffi) - [Brotli](https://en.wikipedia.org/wiki/Brotli) content encoding support. Both licensed under MIT <sup>[1](https://github.com/google/brotli/blob/master/LICENSE) [2](https://github.com/python-hyper/brotlicffi/blob/master/LICENSE) </sup>
 * [**websockets**](https://github.com/aaugustin/websockets)\* - For downloading over websocket. Licensed under [BSD-3-Clause](https://github.com/aaugustin/websockets/blob/main/LICENSE)
+* [**requests**](https://github.com/psf/requests)\* - HTTP library. For HTTPS proxy and persistent connections support. Licensed under [Apache-2.0](https://github.com/psf/requests/blob/main/LICENSE)
 
 ### Metadata
 

diff --git a/requirements.txt b/requirements.txt
@@ -4,3 +4,5 @@ websockets
 brotli; platform_python_implementation=='CPython'
 brotlicffi; platform_python_implementation!='CPython'
 certifi
+requests>=2.31.0,<3
+urllib3>=1.26.17,<3
diff --git a/setup.py b/setup.py
@@ -62,7 +62,14 @@ def py2exe_params():
             'compressed': 1,
             'optimize': 2,
             'dist_dir': './dist',
-            'excludes': ['Crypto', 'Cryptodome'],  # py2exe cannot import Crypto
+            'excludes': [
+                # py2exe cannot import Crypto
+                'Crypto',
+                'Cryptodome',
+                # py2exe appears to confuse this with our socks library.
+                # We don't use pysocks and urllib3.contrib.socks would fail to import if tried.
+                'urllib3.contrib.socks'
+            ],
             'dll_excludes': ['w9xpopen.exe', 'crypt32.dll'],
             # Modules that are only imported dynamically must be added here
             'includes': ['yt_dlp.compat._legacy', 'yt_dlp.compat._deprecated',

diff --git a/test/test_networking.py b/test/test_networking.py
diff --git a/test/test_socks.py b/test/test_socks.py
@@ -263,15 +263,15 @@ def ctx(request):
 
 
 class TestSocks4Proxy:
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks4_no_auth(self, handler, ctx):
         with handler() as rh:
             with ctx.socks_server(Socks4ProxyHandler) as server_address:
                 response = ctx.socks_info_request(
                     rh, proxies={'all': f'socks4://{server_address}'})
                 assert response['version'] == 4
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks4_auth(self, handler, ctx):
         with handler() as rh:
             with ctx.socks_server(Socks4ProxyHandler, user_id='user') as server_address:
@@ -281,15 +281,15 @@ def test_socks4_auth(self, handler, ctx):
                     rh, proxies={'all': f'socks4://user:@{server_address}'})
                 assert response['version'] == 4
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks4a_ipv4_target(self, handler, ctx):
         with ctx.socks_server(Socks4ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks4a://{server_address}'}) as rh:
                 response = ctx.socks_info_request(rh, target_domain='127.0.0.1')
                 assert response['version'] == 4
                 assert (response['ipv4_address'] == '127.0.0.1') != (response['domain_address'] == '127.0.0.1')
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks4a_domain_target(self, handler, ctx):
         with ctx.socks_server(Socks4ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks4a://{server_address}'}) as rh:
@@ -298,7 +298,7 @@ def test_socks4a_domain_target(self, handler, ctx):
                 assert response['ipv4_address'] is None
                 assert response['domain_address'] == 'localhost'
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_ipv4_client_source_address(self, handler, ctx):
         with ctx.socks_server(Socks4ProxyHandler) as server_address:
             source_address = f'127.0.0.{random.randint(5, 255)}'
@@ -308,7 +308,7 @@ def test_ipv4_client_source_address(self, handler, ctx):
                 assert response['client_address'][0] == source_address
                 assert response['version'] == 4
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     @pytest.mark.parametrize('reply_code', [
         Socks4CD.REQUEST_REJECTED_OR_FAILED,
         Socks4CD.REQUEST_REJECTED_CANNOT_CONNECT_TO_IDENTD,
@@ -320,7 +320,7 @@ def test_socks4_errors(self, handler, ctx, reply_code):
                 with pytest.raises(ProxyError):
                     ctx.socks_info_request(rh)
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_ipv6_socks4_proxy(self, handler, ctx):
         with ctx.socks_server(Socks4ProxyHandler, bind_ip='::1') as server_address:
             with handler(proxies={'all': f'socks4://{server_address}'}) as rh:
@@ -329,7 +329,7 @@ def test_ipv6_socks4_proxy(self, handler, ctx):
                 assert response['ipv4_address'] == '127.0.0.1'
                 assert response['version'] == 4
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_timeout(self, handler, ctx):
         with ctx.socks_server(Socks4ProxyHandler, sleep=2) as server_address:
             with handler(proxies={'all': f'socks4://{server_address}'}, timeout=0.5) as rh:
@@ -339,15 +339,15 @@ def test_timeout(self, handler, ctx):
 
 class TestSocks5Proxy:
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks5_no_auth(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
                 response = ctx.socks_info_request(rh)
                 assert response['auth_methods'] == [0x0]
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks5_user_pass(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler, auth=('test', 'testpass')) as server_address:
             with handler() as rh:
@@ -360,23 +360,23 @@ def test_socks5_user_pass(self, handler, ctx):
                 assert response['auth_methods'] == [Socks5Auth.AUTH_NONE, Socks5Auth.AUTH_USER_PASS]
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks5_ipv4_target(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
                 response = ctx.socks_info_request(rh, target_domain='127.0.0.1')
                 assert response['ipv4_address'] == '127.0.0.1'
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks5_domain_target(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
                 response = ctx.socks_info_request(rh, target_domain='localhost')
                 assert (response['ipv4_address'] == '127.0.0.1') != (response['ipv6_address'] == '::1')
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks5h_domain_target(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5h://{server_address}'}) as rh:
@@ -385,7 +385,7 @@ def test_socks5h_domain_target(self, handler, ctx):
                 assert response['domain_address'] == 'localhost'
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks5h_ip_target(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5h://{server_address}'}) as rh:
@@ -394,15 +394,15 @@ def test_socks5h_ip_target(self, handler, ctx):
                 assert response['domain_address'] is None
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_socks5_ipv6_destination(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
                 response = ctx.socks_info_request(rh, target_domain='[::1]')
                 assert response['ipv6_address'] == '::1'
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_ipv6_socks5_proxy(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler, bind_ip='::1') as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@@ -413,7 +413,7 @@ def test_ipv6_socks5_proxy(self, handler, ctx):
 
     # XXX: is there any feasible way of testing IPv6 source addresses?
     # Same would go for non-proxy source_address test...
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     def test_ipv4_client_source_address(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             source_address = f'127.0.0.{random.randint(5, 255)}'
@@ -422,7 +422,7 @@ def test_ipv4_client_source_address(self, handler, ctx):
                 assert response['client_address'][0] == source_address
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http')], indirect=True)
+    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http')], indirect=True)
     @pytest.mark.parametrize('reply_code', [
         Socks5Reply.GENERAL_FAILURE,
         Socks5Reply.CONNECTION_NOT_ALLOWED,

diff --git a/yt_dlp/YoutubeDL.py b/yt_dlp/YoutubeDL.py
@@ -3968,7 +3968,7 @@ def get_encoding(stream):
         })) or 'none'))
 
         write_debug(f'Proxy map: {self.proxies}')
-        # write_debug(f'Request Handlers: {", ".join(rh.RH_NAME for rh in self._request_director.handlers.values())}')
+        write_debug(f'Request Handlers: {", ".join(rh.RH_NAME for rh in self._request_director.handlers.values())}')
         for plugin_type, plugins in {'Extractor': plugin_ies, 'Post-Processor': plugin_pps}.items():
             display_list = ['%s%s' % (
                 klass.__name__, '' if klass.__name__ == name else f' as {name}')
@@ -4057,6 +4057,9 @@ def urlopen(self, req):
                     raise RequestError(
                         'file:// URLs are disabled by default in yt-dlp for security reasons. '
                         'Use --enable-file-urls to enable at your own risk.', cause=ue) from ue
+                if 'unsupported proxy type: "https"' in ue.msg.lower():
+                    raise RequestError(
+                        'To use an HTTPS proxy for this request, one of the following dependencies needs to be installed: requests')
             raise
         except SSLError as e:
             if 'UNSAFE_LEGACY_RENEGOTIATION_DISABLED' in str(e):
@@ -4099,6 +4102,8 @@ def build_request_director(self, handlers, preferences=None):
                 }),
             ))
         director.preferences.update(preferences or [])
+        if 'prefer-legacy-http-handler' in self.params['compat_opts']:
+            director.preferences.add(lambda rh, _: 500 if rh.RH_KEY == 'Urllib' else 0)
         return director
 
     def encode(self, s):

diff --git a/yt_dlp/__pyinstaller/hook-yt_dlp.py b/yt_dlp/__pyinstaller/hook-yt_dlp.py
@@ -21,7 +21,9 @@ def get_hidden_imports():
     yield from ('yt_dlp.compat._legacy', 'yt_dlp.compat._deprecated')
     yield from ('yt_dlp.utils._legacy', 'yt_dlp.utils._deprecated')
     yield pycryptodome_module()
-    yield from collect_submodules('websockets')
+    # Only `websockets` is required, others are collected just in case
+    for module in ('websockets', 'requests', 'urllib3'):
+        yield from collect_submodules(module)
     # These are auto-detected, but explicitly add them just in case
     yield from ('mutagen', 'brotli', 'certifi')
 

diff --git a/yt_dlp/dependencies/__init__.py b/yt_dlp/dependencies/__init__.py
@@ -58,6 +58,15 @@
     # See https://github.com/yt-dlp/yt-dlp/issues/2633
     websockets = None
 
+try:
+    import urllib3
+except ImportError:
+    urllib3 = None
+
+try:
+    import requests
+except ImportError:
+    requests = None
 
 try:
     import xattr  # xattr or pyxattr

diff --git a/yt_dlp/networking/__init__.py b/yt_dlp/networking/__init__.py
@@ -1,4 +1,6 @@
 # flake8: noqa: F401
+import warnings
+
 from .common import (
     HEADRequest,
     PUTRequest,
@@ -11,3 +13,11 @@
 # isort: split
 # TODO: all request handlers should be safely imported
 from . import _urllib
+from ..utils import bug_reports_message
+
+try:
+    from . import _requests
+except ImportError:
+    pass
+except Exception as e:
+    warnings.warn(f'Failed to import "requests" request handler: {e}' + bug_reports_message())
diff --git a/yt_dlp/networking/_helper.py b/yt_dlp/networking/_helper.py
@@ -11,7 +11,7 @@
 
 from .exceptions import RequestError, UnsupportedRequest
 from ..dependencies import certifi
-from ..socks import ProxyType
+from ..socks import ProxyType, sockssocket
 from ..utils import format_field, traverse_obj
 
 if typing.TYPE_CHECKING:
@@ -224,6 +224,24 @@ def _socket_connect(ip_addr, timeout, source_address):
         raise
 
 
+def create_socks_proxy_socket(dest_addr, proxy_args, proxy_ip_addr, timeout, source_address):
+    af, socktype, proto, canonname, sa = proxy_ip_addr
+    sock = sockssocket(af, socktype, proto)
+    try:
+        connect_proxy_args = proxy_args.copy()
+        connect_proxy_args.update({'addr': sa[0], 'port': sa[1]})
+        sock.setproxy(**connect_proxy_args)
+        if timeout is not socket._GLOBAL_DEFAULT_TIMEOUT:  # noqa: E721
+            sock.settimeout(timeout)
+        if source_address:
+            sock.bind(source_address)
+        sock.connect(dest_addr)
+        return sock
+    except socket.error:
+        sock.close()
+        raise
+
+
 def create_connection(
     address,
     timeout=socket._GLOBAL_DEFAULT_TIMEOUT,