diff options
Diffstat (limited to 'test')
| -rw-r--r-- | test/test_YoutubeDL.py | 4 | ||||
| -rw-r--r-- | test/test_cookies.py | 4 | ||||
| -rw-r--r-- | test/test_networking.py | 632 | ||||
| -rw-r--r-- | test/test_socks.py | 33 | ||||
| -rw-r--r-- | test/test_traversal.py | 444 | ||||
| -rw-r--r-- | test/test_utils.py | 383 | ||||
| -rw-r--r-- | test/test_websockets.py | 53 |
7 files changed, 1004 insertions, 549 deletions
diff --git a/test/test_YoutubeDL.py b/test/test_YoutubeDL.py index 6be47af..5242cf8 100644 --- a/test/test_YoutubeDL.py +++ b/test/test_YoutubeDL.py @@ -183,7 +183,7 @@ class TestFormatSelection(unittest.TestCase): ] info_dict = _make_result(formats) - ydl = YDL({'format': 'best'}) + ydl = YDL({'format': 'best', 'format_sort': ['abr', 'ext']}) ydl.sort_formats(info_dict) ydl.process_ie_result(copy.deepcopy(info_dict)) downloaded = ydl.downloaded_info_dicts[0] @@ -195,7 +195,7 @@ class TestFormatSelection(unittest.TestCase): downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['format_id'], 'mp3-64') - ydl = YDL({'prefer_free_formats': True}) + ydl = YDL({'prefer_free_formats': True, 'format_sort': ['abr', 'ext']}) ydl.sort_formats(info_dict) ydl.process_ie_result(copy.deepcopy(info_dict)) downloaded = ydl.downloaded_info_dicts[0] diff --git a/test/test_cookies.py b/test/test_cookies.py index 5282ef6..bd61f30 100644 --- a/test/test_cookies.py +++ b/test/test_cookies.py @@ -1,5 +1,5 @@ +import datetime as dt import unittest -from datetime import datetime, timezone from yt_dlp import cookies from yt_dlp.cookies import ( @@ -138,7 +138,7 @@ class TestCookies(unittest.TestCase): self.assertEqual(cookie.name, 'foo') self.assertEqual(cookie.value, 'test%20%3Bcookie') self.assertFalse(cookie.secure) - expected_expiration = datetime(2021, 6, 18, 21, 39, 19, tzinfo=timezone.utc) + expected_expiration = dt.datetime(2021, 6, 18, 21, 39, 19, tzinfo=dt.timezone.utc) self.assertEqual(cookie.expires, int(expected_expiration.timestamp())) def test_pbkdf2_sha1(self): diff --git a/test/test_networking.py b/test/test_networking.py index 628f1f1..b50f70d 100644 --- a/test/test_networking.py +++ b/test/test_networking.py @@ -27,9 +27,10 @@ import zlib from email.message import Message from http.cookiejar import CookieJar +from test.conftest import validate_and_send from test.helper import FakeYDL, http_server_port, verify_address_availability from yt_dlp.cookies import YoutubeDLCookieJar -from yt_dlp.dependencies import brotli, requests, urllib3 +from yt_dlp.dependencies import brotli, curl_cffi, requests, urllib3 from yt_dlp.networking import ( HEADRequest, PUTRequest, @@ -50,10 +51,13 @@ from yt_dlp.networking.exceptions import ( TransportError, UnsupportedRequest, ) +from yt_dlp.networking.impersonate import ( + ImpersonateRequestHandler, + ImpersonateTarget, +) +from yt_dlp.utils import YoutubeDLError from yt_dlp.utils._utils import _YDLLogger as FakeLogger -from yt_dlp.utils.networking import HTTPHeaderDict - -from test.conftest import validate_and_send +from yt_dlp.utils.networking import HTTPHeaderDict, std_headers TEST_DIR = os.path.dirname(os.path.abspath(__file__)) @@ -75,6 +79,7 @@ def _build_proxy_handler(name): class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler): protocol_version = 'HTTP/1.1' + default_request_version = 'HTTP/1.1' def log_message(self, format, *args): pass @@ -112,6 +117,8 @@ class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler): def _read_data(self): if 'Content-Length' in self.headers: return self.rfile.read(int(self.headers['Content-Length'])) + else: + return b'' def do_POST(self): data = self._read_data() + str(self.headers).encode() @@ -195,7 +202,8 @@ class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler): self._headers() elif self.path.startswith('/308-to-headers'): self.send_response(308) - self.send_header('Location', '/headers') + # redirect to "localhost" for testing cookie redirection handling + self.send_header('Location', f'http://localhost:{self.connection.getsockname()[1]}/headers') self.send_header('Content-Length', '0') self.end_headers() elif self.path == '/trailing_garbage': @@ -310,7 +318,7 @@ class TestRequestHandlerBase: class TestHTTPRequestHandler(TestRequestHandlerBase): - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_verify_cert(self, handler): with handler() as rh: with pytest.raises(CertificateVerifyError): @@ -321,7 +329,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): assert r.status == 200 r.close() - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_ssl_error(self, handler): # HTTPS server with too old TLS version # XXX: is there a better way to test this than to create a new server? @@ -335,11 +343,11 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): https_server_thread.start() with handler(verify=False) as rh: - with pytest.raises(SSLError, match=r'ssl(?:v3|/tls) alert handshake failure') as exc_info: + with pytest.raises(SSLError, match=r'(?i)ssl(?:v3|/tls).alert.handshake.failure') as exc_info: validate_and_send(rh, Request(f'https://127.0.0.1:{https_port}/headers')) assert not issubclass(exc_info.type, CertificateVerifyError) - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_percent_encode(self, handler): with handler() as rh: # Unicode characters should be encoded with uppercase percent-encoding @@ -351,7 +359,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): assert res.status == 200 res.close() - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) @pytest.mark.parametrize('path', [ '/a/b/./../../headers', '/redirect_dotsegments', @@ -367,6 +375,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): assert res.url == f'http://127.0.0.1:{self.http_port}/headers' res.close() + # Not supported by CurlCFFI (non-standard) @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) def test_unicode_path_redirection(self, handler): with handler() as rh: @@ -374,7 +383,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): assert r.url == f'http://127.0.0.1:{self.http_port}/%E4%B8%AD%E6%96%87.html' r.close() - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_raise_http_error(self, handler): with handler() as rh: for bad_status in (400, 500, 599, 302): @@ -384,7 +393,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): # Should not raise an error validate_and_send(rh, Request('http://127.0.0.1:%d/gen_200' % self.http_port)).close() - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_response_url(self, handler): with handler() as rh: # Response url should be that of the last url in redirect chain @@ -395,62 +404,50 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): assert res2.url == f'http://127.0.0.1:{self.http_port}/gen_200' res2.close() - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) - def test_redirect(self, handler): + # Covers some basic cases we expect some level of consistency between request handlers for + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) + @pytest.mark.parametrize('redirect_status,method,expected', [ + # A 303 must either use GET or HEAD for subsequent request + (303, 'POST', ('', 'GET', False)), + (303, 'HEAD', ('', 'HEAD', False)), + + # 301 and 302 turn POST only into a GET + (301, 'POST', ('', 'GET', False)), + (301, 'HEAD', ('', 'HEAD', False)), + (302, 'POST', ('', 'GET', False)), + (302, 'HEAD', ('', 'HEAD', False)), + + # 307 and 308 should not change method + (307, 'POST', ('testdata', 'POST', True)), + (308, 'POST', ('testdata', 'POST', True)), + (307, 'HEAD', ('', 'HEAD', False)), + (308, 'HEAD', ('', 'HEAD', False)), + ]) + def test_redirect(self, handler, redirect_status, method, expected): with handler() as rh: - def do_req(redirect_status, method, assert_no_content=False): - data = b'testdata' if method in ('POST', 'PUT') else None - res = validate_and_send( - rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_{redirect_status}', method=method, data=data)) - - headers = b'' - data_sent = b'' - if data is not None: - data_sent += res.read(len(data)) - if data_sent != data: - headers += data_sent - data_sent = b'' - - headers += res.read() - - if assert_no_content or data is None: - assert b'Content-Type' not in headers - assert b'Content-Length' not in headers - else: - assert b'Content-Type' in headers - assert b'Content-Length' in headers - - return data_sent.decode(), res.headers.get('method', '') - - # A 303 must either use GET or HEAD for subsequent request - assert do_req(303, 'POST', True) == ('', 'GET') - assert do_req(303, 'HEAD') == ('', 'HEAD') - - assert do_req(303, 'PUT', True) == ('', 'GET') - - # 301 and 302 turn POST only into a GET - assert do_req(301, 'POST', True) == ('', 'GET') - assert do_req(301, 'HEAD') == ('', 'HEAD') - assert do_req(302, 'POST', True) == ('', 'GET') - assert do_req(302, 'HEAD') == ('', 'HEAD') - - assert do_req(301, 'PUT') == ('testdata', 'PUT') - assert do_req(302, 'PUT') == ('testdata', 'PUT') + data = b'testdata' if method == 'POST' else None + headers = {} + if data is not None: + headers['Content-Type'] = 'application/test' + res = validate_and_send( + rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_{redirect_status}', method=method, data=data, + headers=headers)) - # 307 and 308 should not change method - for m in ('POST', 'PUT'): - assert do_req(307, m) == ('testdata', m) - assert do_req(308, m) == ('testdata', m) + headers = b'' + data_recv = b'' + if data is not None: + data_recv += res.read(len(data)) + if data_recv != data: + headers += data_recv + data_recv = b'' - assert do_req(307, 'HEAD') == ('', 'HEAD') - assert do_req(308, 'HEAD') == ('', 'HEAD') + headers += res.read() - # These should not redirect and instead raise an HTTPError - for code in (300, 304, 305, 306): - with pytest.raises(HTTPError): - do_req(code, 'GET') + assert expected[0] == data_recv.decode() + assert expected[1] == res.headers.get('method') + assert expected[2] == ('content-length' in headers.decode().lower()) - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_request_cookie_header(self, handler): # We should accept a Cookie header being passed as in normal headers and handle it appropriately. with handler() as rh: @@ -459,16 +456,17 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): rh, Request( f'http://127.0.0.1:{self.http_port}/headers', headers={'Cookie': 'test=test'})).read().decode() - assert 'Cookie: test=test' in res + assert 'cookie: test=test' in res.lower() # Specified Cookie header should be removed on any redirect res = validate_and_send( rh, Request( f'http://127.0.0.1:{self.http_port}/308-to-headers', - headers={'Cookie': 'test=test'})).read().decode() - assert 'Cookie: test=test' not in res + headers={'Cookie': 'test=test2'})).read().decode() + assert 'cookie: test=test2' not in res.lower() # Specified Cookie header should override global cookiejar for that request + # Whether cookies from the cookiejar is applied on the redirect is considered undefined for now cookiejar = YoutubeDLCookieJar() cookiejar.set_cookie(http.cookiejar.Cookie( version=0, name='test', value='ytdlp', port=None, port_specified=False, @@ -478,23 +476,23 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): with handler(cookiejar=cookiejar) as rh: data = validate_and_send( - rh, Request(f'http://127.0.0.1:{self.http_port}/headers', headers={'cookie': 'test=test'})).read() - assert b'Cookie: test=ytdlp' not in data - assert b'Cookie: test=test' in data + rh, Request(f'http://127.0.0.1:{self.http_port}/headers', headers={'cookie': 'test=test3'})).read() + assert b'cookie: test=ytdlp' not in data.lower() + assert b'cookie: test=test3' in data.lower() - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_redirect_loop(self, handler): with handler() as rh: with pytest.raises(HTTPError, match='redirect loop'): validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_loop')) - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_incompleteread(self, handler): with handler(timeout=2) as rh: - with pytest.raises(IncompleteRead): + with pytest.raises(IncompleteRead, match='13 bytes read, 234221 more expected'): validate_and_send(rh, Request('http://127.0.0.1:%d/incompleteread' % self.http_port)).read() - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_cookies(self, handler): cookiejar = YoutubeDLCookieJar() cookiejar.set_cookie(http.cookiejar.Cookie( @@ -503,47 +501,66 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): with handler(cookiejar=cookiejar) as rh: data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read() - assert b'Cookie: test=ytdlp' in data + assert b'cookie: test=ytdlp' in data.lower() # Per request with handler() as rh: data = validate_and_send( rh, Request(f'http://127.0.0.1:{self.http_port}/headers', extensions={'cookiejar': cookiejar})).read() - assert b'Cookie: test=ytdlp' in data + assert b'cookie: test=ytdlp' in data.lower() - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_headers(self, handler): with handler(headers=HTTPHeaderDict({'test1': 'test', 'test2': 'test2'})) as rh: # Global Headers - data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read() - assert b'Test1: test' in data + data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read().lower() + assert b'test1: test' in data # Per request headers, merged with global data = validate_and_send(rh, Request( - f'http://127.0.0.1:{self.http_port}/headers', headers={'test2': 'changed', 'test3': 'test3'})).read() - assert b'Test1: test' in data - assert b'Test2: changed' in data - assert b'Test2: test2' not in data - assert b'Test3: test3' in data - - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) - def test_timeout(self, handler): + f'http://127.0.0.1:{self.http_port}/headers', headers={'test2': 'changed', 'test3': 'test3'})).read().lower() + assert b'test1: test' in data + assert b'test2: changed' in data + assert b'test2: test2' not in data + assert b'test3: test3' in data + + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) + def test_read_timeout(self, handler): with handler() as rh: # Default timeout is 20 seconds, so this should go through validate_and_send( - rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_3')) + rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1')) - with handler(timeout=0.5) as rh: + with handler(timeout=0.1) as rh: with pytest.raises(TransportError): validate_and_send( - rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1')) + rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_5')) # Per request timeout, should override handler timeout validate_and_send( rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1', extensions={'timeout': 4})) - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) + def test_connect_timeout(self, handler): + # nothing should be listening on this port + connect_timeout_url = 'http://10.255.255.255' + with handler(timeout=0.01) as rh: + now = time.time() + with pytest.raises(TransportError): + validate_and_send( + rh, Request(connect_timeout_url)) + assert 0.01 <= time.time() - now < 20 + + with handler() as rh: + with pytest.raises(TransportError): + # Per request timeout, should override handler timeout + now = time.time() + validate_and_send( + rh, Request(connect_timeout_url, extensions={'timeout': 0.01})) + assert 0.01 <= time.time() - now < 20 + + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_source_address(self, handler): source_address = f'127.0.0.{random.randint(5, 255)}' # on some systems these loopback addresses we need for testing may not be available @@ -554,6 +571,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): rh, Request(f'http://127.0.0.1:{self.http_port}/source_address')).read().decode() assert source_address == data + # Not supported by CurlCFFI @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) def test_gzip_trailing_garbage(self, handler): with handler() as rh: @@ -571,7 +589,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): assert res.headers.get('Content-Encoding') == 'br' assert res.read() == b'<html><video src="/vid.mp4" /></html>' - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_deflate(self, handler): with handler() as rh: res = validate_and_send( @@ -581,7 +599,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): assert res.headers.get('Content-Encoding') == 'deflate' assert res.read() == b'<html><video src="/vid.mp4" /></html>' - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_gzip(self, handler): with handler() as rh: res = validate_and_send( @@ -591,7 +609,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): assert res.headers.get('Content-Encoding') == 'gzip' assert res.read() == b'<html><video src="/vid.mp4" /></html>' - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_multiple_encodings(self, handler): with handler() as rh: for pair in ('gzip,deflate', 'deflate, gzip', 'gzip, gzip', 'deflate, deflate'): @@ -602,17 +620,18 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): assert res.headers.get('Content-Encoding') == pair assert res.read() == b'<html><video src="/vid.mp4" /></html>' + # Not supported by curl_cffi @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) def test_unsupported_encoding(self, handler): with handler() as rh: res = validate_and_send( rh, Request( f'http://127.0.0.1:{self.http_port}/content-encoding', - headers={'ytdl-encoding': 'unsupported'})) + headers={'ytdl-encoding': 'unsupported', 'Accept-Encoding': '*'})) assert res.headers.get('Content-Encoding') == 'unsupported' assert res.read() == b'raw' - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_read(self, handler): with handler() as rh: res = validate_and_send( @@ -620,9 +639,12 @@ class TestHTTPRequestHandler(TestRequestHandlerBase): assert res.readable() assert res.read(1) == b'H' assert res.read(3) == b'ost' + assert res.read().decode().endswith('\n\n') + assert res.read() == b'' class TestHTTPProxy(TestRequestHandlerBase): + # Note: this only tests http urls over non-CONNECT proxy @classmethod def setup_class(cls): super().setup_class() @@ -642,7 +664,7 @@ class TestHTTPProxy(TestRequestHandlerBase): cls.geo_proxy_thread.daemon = True cls.geo_proxy_thread.start() - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_http_proxy(self, handler): http_proxy = f'http://127.0.0.1:{self.proxy_port}' geo_proxy = f'http://127.0.0.1:{self.geo_port}' @@ -668,7 +690,7 @@ class TestHTTPProxy(TestRequestHandlerBase): assert res != f'normal: {real_url}' assert 'Accept' in res - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_noproxy(self, handler): with handler(proxies={'proxy': f'http://127.0.0.1:{self.proxy_port}'}) as rh: # NO_PROXY @@ -678,7 +700,7 @@ class TestHTTPProxy(TestRequestHandlerBase): 'utf-8') assert 'Accept' in nop_response - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_allproxy(self, handler): url = 'http://foo.com/bar' with handler() as rh: @@ -686,7 +708,7 @@ class TestHTTPProxy(TestRequestHandlerBase): 'utf-8') assert response == f'normal: {url}' - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_http_proxy_with_idn(self, handler): with handler(proxies={ 'http': f'http://127.0.0.1:{self.proxy_port}', @@ -698,7 +720,6 @@ class TestHTTPProxy(TestRequestHandlerBase): class TestClientCertificate: - @classmethod def setup_class(cls): certfn = os.path.join(TEST_DIR, 'testcert.pem') @@ -724,27 +745,27 @@ class TestClientCertificate: ) as rh: validate_and_send(rh, Request(f'https://127.0.0.1:{self.port}/video.html')).read().decode() - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_certificate_combined_nopass(self, handler): self._run_test(handler, client_cert={ 'client_certificate': os.path.join(self.certdir, 'clientwithkey.crt'), }) - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_certificate_nocombined_nopass(self, handler): self._run_test(handler, client_cert={ 'client_certificate': os.path.join(self.certdir, 'client.crt'), 'client_certificate_key': os.path.join(self.certdir, 'client.key'), }) - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_certificate_combined_pass(self, handler): self._run_test(handler, client_cert={ 'client_certificate': os.path.join(self.certdir, 'clientwithencryptedkey.crt'), 'client_certificate_password': 'foobar', }) - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_certificate_nocombined_pass(self, handler): self._run_test(handler, client_cert={ 'client_certificate': os.path.join(self.certdir, 'client.crt'), @@ -753,6 +774,18 @@ class TestClientCertificate: }) +@pytest.mark.parametrize('handler', ['CurlCFFI'], indirect=True) +class TestHTTPImpersonateRequestHandler(TestRequestHandlerBase): + def test_supported_impersonate_targets(self, handler): + with handler(headers=std_headers) as rh: + # note: this assumes the impersonate request handler supports the impersonate extension + for target in rh.supported_targets: + res = validate_and_send(rh, Request( + f'http://127.0.0.1:{self.http_port}/headers', extensions={'impersonate': target})) + assert res.status == 200 + assert std_headers['user-agent'].lower() not in res.read().decode().lower() + + class TestRequestHandlerMisc: """Misc generic tests for request handlers, not related to request or validation testing""" @pytest.mark.parametrize('handler,logger_name', [ @@ -931,6 +964,172 @@ class TestRequestsRequestHandler(TestRequestHandlerBase): assert called +@pytest.mark.parametrize('handler', ['CurlCFFI'], indirect=True) +class TestCurlCFFIRequestHandler(TestRequestHandlerBase): + + @pytest.mark.parametrize('params,extensions', [ + ({}, {'impersonate': ImpersonateTarget('chrome')}), + ({'impersonate': ImpersonateTarget('chrome', '110')}, {}), + ({'impersonate': ImpersonateTarget('chrome', '99')}, {'impersonate': ImpersonateTarget('chrome', '110')}), + ]) + def test_impersonate(self, handler, params, extensions): + with handler(headers=std_headers, **params) as rh: + res = validate_and_send( + rh, Request(f'http://127.0.0.1:{self.http_port}/headers', extensions=extensions)).read().decode() + assert 'sec-ch-ua: "Chromium";v="110"' in res + # Check that user agent is added over ours + assert 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/110.0.0.0 Safari/537.36' in res + + def test_headers(self, handler): + with handler(headers=std_headers) as rh: + # Ensure curl-impersonate overrides our standard headers (usually added + res = validate_and_send( + rh, Request(f'http://127.0.0.1:{self.http_port}/headers', extensions={ + 'impersonate': ImpersonateTarget('safari')}, headers={'x-custom': 'test', 'sec-fetch-mode': 'custom'})).read().decode().lower() + + assert std_headers['user-agent'].lower() not in res + assert std_headers['accept-language'].lower() not in res + assert std_headers['sec-fetch-mode'].lower() not in res + # other than UA, custom headers that differ from std_headers should be kept + assert 'sec-fetch-mode: custom' in res + assert 'x-custom: test' in res + # but when not impersonating don't remove std_headers + res = validate_and_send( + rh, Request(f'http://127.0.0.1:{self.http_port}/headers', headers={'x-custom': 'test'})).read().decode().lower() + # std_headers should be present + for k, v in std_headers.items(): + assert f'{k}: {v}'.lower() in res + + @pytest.mark.parametrize('raised,expected,match', [ + (lambda: curl_cffi.requests.errors.RequestsError( + '', code=curl_cffi.const.CurlECode.PARTIAL_FILE), IncompleteRead, None), + (lambda: curl_cffi.requests.errors.RequestsError( + '', code=curl_cffi.const.CurlECode.OPERATION_TIMEDOUT), TransportError, None), + (lambda: curl_cffi.requests.errors.RequestsError( + '', code=curl_cffi.const.CurlECode.RECV_ERROR), TransportError, None), + ]) + def test_response_error_mapping(self, handler, monkeypatch, raised, expected, match): + import curl_cffi.requests + + from yt_dlp.networking._curlcffi import CurlCFFIResponseAdapter + curl_res = curl_cffi.requests.Response() + res = CurlCFFIResponseAdapter(curl_res) + + def mock_read(*args, **kwargs): + try: + raise raised() + except Exception as e: + e.response = curl_res + raise + monkeypatch.setattr(res.fp, 'read', mock_read) + + with pytest.raises(expected, match=match) as exc_info: + res.read() + + assert exc_info.type is expected + + @pytest.mark.parametrize('raised,expected,match', [ + (lambda: curl_cffi.requests.errors.RequestsError( + '', code=curl_cffi.const.CurlECode.OPERATION_TIMEDOUT), TransportError, None), + (lambda: curl_cffi.requests.errors.RequestsError( + '', code=curl_cffi.const.CurlECode.PEER_FAILED_VERIFICATION), CertificateVerifyError, None), + (lambda: curl_cffi.requests.errors.RequestsError( + '', code=curl_cffi.const.CurlECode.SSL_CONNECT_ERROR), SSLError, None), + (lambda: curl_cffi.requests.errors.RequestsError( + '', code=curl_cffi.const.CurlECode.TOO_MANY_REDIRECTS), HTTPError, None), + (lambda: curl_cffi.requests.errors.RequestsError( + '', code=curl_cffi.const.CurlECode.PROXY), ProxyError, None), + ]) + def test_request_error_mapping(self, handler, monkeypatch, raised, expected, match): + import curl_cffi.requests + curl_res = curl_cffi.requests.Response() + curl_res.status_code = 301 + + with handler() as rh: + original_get_instance = rh._get_instance + + def mock_get_instance(*args, **kwargs): + instance = original_get_instance(*args, **kwargs) + + def request(*_, **__): + try: + raise raised() + except Exception as e: + e.response = curl_res + raise + monkeypatch.setattr(instance, 'request', request) + return instance + + monkeypatch.setattr(rh, '_get_instance', mock_get_instance) + + with pytest.raises(expected) as exc_info: + rh.send(Request('http://fake')) + + assert exc_info.type is expected + + def test_response_reader(self, handler): + class FakeResponse: + def __init__(self, raise_error=False): + self.raise_error = raise_error + self.closed = False + + def iter_content(self): + yield b'foo' + yield b'bar' + yield b'z' + if self.raise_error: + raise Exception('test') + + def close(self): + self.closed = True + + from yt_dlp.networking._curlcffi import CurlCFFIResponseReader + + res = CurlCFFIResponseReader(FakeResponse()) + assert res.readable + assert res.bytes_read == 0 + assert res.read(1) == b'f' + assert res.bytes_read == 3 + assert res._buffer == b'oo' + + assert res.read(2) == b'oo' + assert res.bytes_read == 3 + assert res._buffer == b'' + + assert res.read(2) == b'ba' + assert res.bytes_read == 6 + assert res._buffer == b'r' + + assert res.read(3) == b'rz' + assert res.bytes_read == 7 + assert res._buffer == b'' + assert res.closed + assert res._response.closed + + # should handle no size param + res2 = CurlCFFIResponseReader(FakeResponse()) + assert res2.read() == b'foobarz' + assert res2.bytes_read == 7 + assert res2._buffer == b'' + assert res2.closed + + # should close on an exception + res3 = CurlCFFIResponseReader(FakeResponse(raise_error=True)) + with pytest.raises(Exception, match='test'): + res3.read() + assert res3._buffer == b'' + assert res3.bytes_read == 7 + assert res3.closed + + # buffer should be cleared on close + res4 = CurlCFFIResponseReader(FakeResponse()) + res4.read(2) + assert res4._buffer == b'o' + res4.close() + assert res4.closed + assert res4._buffer == b'' + + def run_validation(handler, error, req, **handler_kwargs): with handler(**handler_kwargs) as rh: if error: @@ -975,6 +1174,10 @@ class TestRequestHandlerValidation: ('ws', False, {}), ('wss', False, {}), ]), + ('CurlCFFI', [ + ('http', False, {}), + ('https', False, {}), + ]), (NoCheckRH, [('http', False, {})]), (ValidationRH, [('http', UnsupportedRequest, {})]) ] @@ -998,6 +1201,14 @@ class TestRequestHandlerValidation: ('socks5', False), ('socks5h', False), ]), + ('CurlCFFI', 'http', [ + ('http', False), + ('https', False), + ('socks4', False), + ('socks4a', False), + ('socks5', False), + ('socks5h', False), + ]), (NoCheckRH, 'http', [('http', False)]), (HTTPSupportedRH, 'http', [('http', UnsupportedRequest)]), ('Websockets', 'ws', [('http', UnsupportedRequest)]), @@ -1015,6 +1226,10 @@ class TestRequestHandlerValidation: ('all', False), ('unrelated', False), ]), + ('CurlCFFI', [ + ('all', False), + ('unrelated', False), + ]), (NoCheckRH, [('all', False)]), (HTTPSupportedRH, [('all', UnsupportedRequest)]), (HTTPSupportedRH, [('no', UnsupportedRequest)]), @@ -1036,6 +1251,19 @@ class TestRequestHandlerValidation: ({'timeout': 'notatimeout'}, AssertionError), ({'unsupported': 'value'}, UnsupportedRequest), ]), + ('CurlCFFI', 'http', [ + ({'cookiejar': 'notacookiejar'}, AssertionError), + ({'cookiejar': YoutubeDLCookieJar()}, False), + ({'timeout': 1}, False), + ({'timeout': 'notatimeout'}, AssertionError), + ({'unsupported': 'value'}, UnsupportedRequest), + ({'impersonate': ImpersonateTarget('badtarget', None, None, None)}, UnsupportedRequest), + ({'impersonate': 123}, AssertionError), + ({'impersonate': ImpersonateTarget('chrome', None, None, None)}, False), + ({'impersonate': ImpersonateTarget(None, None, None, None)}, False), + ({'impersonate': ImpersonateTarget()}, False), + ({'impersonate': 'chrome'}, AssertionError) + ]), (NoCheckRH, 'http', [ ({'cookiejar': 'notacookiejar'}, False), ({'somerandom': 'test'}, False), # but any extension is allowed through @@ -1055,7 +1283,7 @@ class TestRequestHandlerValidation: def test_url_scheme(self, handler, scheme, fail, handler_kwargs): run_validation(handler, fail, Request(f'{scheme}://'), **(handler_kwargs or {})) - @pytest.mark.parametrize('handler,fail', [('Urllib', False), ('Requests', False)], indirect=['handler']) + @pytest.mark.parametrize('handler,fail', [('Urllib', False), ('Requests', False), ('CurlCFFI', False)], indirect=['handler']) def test_no_proxy(self, handler, fail): run_validation(handler, fail, Request('http://', proxies={'no': '127.0.0.1,github.com'})) run_validation(handler, fail, Request('http://'), proxies={'no': '127.0.0.1,github.com'}) @@ -1078,13 +1306,13 @@ class TestRequestHandlerValidation: run_validation(handler, fail, Request(f'{req_scheme}://', proxies={req_scheme: f'{scheme}://example.com'})) run_validation(handler, fail, Request(f'{req_scheme}://'), proxies={req_scheme: f'{scheme}://example.com'}) - @pytest.mark.parametrize('handler', ['Urllib', HTTPSupportedRH, 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', HTTPSupportedRH, 'Requests', 'CurlCFFI'], indirect=True) def test_empty_proxy(self, handler): run_validation(handler, False, Request('http://', proxies={'http': None})) run_validation(handler, False, Request('http://'), proxies={'http': None}) @pytest.mark.parametrize('proxy_url', ['//example.com', 'example.com', '127.0.0.1', '/a/b/c']) - @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True) + @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True) def test_invalid_proxy_url(self, handler, proxy_url): run_validation(handler, UnsupportedRequest, Request('http://', proxies={'http': proxy_url})) @@ -1113,6 +1341,10 @@ class FakeResponse(Response): class FakeRH(RequestHandler): + def __init__(self, *args, **params): + self.params = params + super().__init__(*args, **params) + def _validate(self, request): return @@ -1271,15 +1503,10 @@ class TestYoutubeDLNetworking: ('', {'all': '__noproxy__'}), (None, {'http': 'http://127.0.0.1:8081', 'https': 'http://127.0.0.1:8081'}) # env, set https ]) - def test_proxy(self, proxy, expected): - old_http_proxy = os.environ.get('HTTP_PROXY') - try: - os.environ['HTTP_PROXY'] = 'http://127.0.0.1:8081' # ensure that provided proxies override env - with FakeYDL({'proxy': proxy}) as ydl: - assert ydl.proxies == expected - finally: - if old_http_proxy: - os.environ['HTTP_PROXY'] = old_http_proxy + def test_proxy(self, proxy, expected, monkeypatch): + monkeypatch.setenv('HTTP_PROXY', 'http://127.0.0.1:8081') + with FakeYDL({'proxy': proxy}) as ydl: + assert ydl.proxies == expected def test_compat_request(self): with FakeRHYDL() as ydl: @@ -1331,6 +1558,95 @@ class TestYoutubeDLNetworking: with pytest.raises(SSLError, match='testerror'): ydl.urlopen('ssl://testerror') + def test_unsupported_impersonate_target(self): + class FakeImpersonationRHYDL(FakeYDL): + def __init__(self, *args, **kwargs): + class HTTPRH(RequestHandler): + def _send(self, request: Request): + pass + _SUPPORTED_URL_SCHEMES = ('http',) + _SUPPORTED_PROXY_SCHEMES = None + + super().__init__(*args, **kwargs) + self._request_director = self.build_request_director([HTTPRH]) + + with FakeImpersonationRHYDL() as ydl: + with pytest.raises( + RequestError, + match=r'Impersonate target "test" is not available' + ): + ydl.urlopen(Request('http://', extensions={'impersonate': ImpersonateTarget('test', None, None, None)})) + + def test_unsupported_impersonate_extension(self): + class FakeHTTPRHYDL(FakeYDL): + def __init__(self, *args, **kwargs): + class IRH(ImpersonateRequestHandler): + def _send(self, request: Request): + pass + + _SUPPORTED_URL_SCHEMES = ('http',) + _SUPPORTED_IMPERSONATE_TARGET_MAP = {ImpersonateTarget('abc',): 'test'} + _SUPPORTED_PROXY_SCHEMES = None + + super().__init__(*args, **kwargs) + self._request_director = self.build_request_director([IRH]) + + with FakeHTTPRHYDL() as ydl: + with pytest.raises( + RequestError, + match=r'Impersonate target "test" is not available' + ): + ydl.urlopen(Request('http://', extensions={'impersonate': ImpersonateTarget('test', None, None, None)})) + + def test_raise_impersonate_error(self): + with pytest.raises( + YoutubeDLError, + match=r'Impersonate target "test" is not available' + ): + FakeYDL({'impersonate': ImpersonateTarget('test', None, None, None)}) + + def test_pass_impersonate_param(self, monkeypatch): + + class IRH(ImpersonateRequestHandler): + def _send(self, request: Request): + pass + + _SUPPORTED_URL_SCHEMES = ('http',) + _SUPPORTED_IMPERSONATE_TARGET_MAP = {ImpersonateTarget('abc'): 'test'} + + # Bypass the check on initialize + brh = FakeYDL.build_request_director + monkeypatch.setattr(FakeYDL, 'build_request_director', lambda cls, handlers, preferences=None: brh(cls, handlers=[IRH])) + + with FakeYDL({ + 'impersonate': ImpersonateTarget('abc', None, None, None) + }) as ydl: + rh = self.build_handler(ydl, IRH) + assert rh.impersonate == ImpersonateTarget('abc', None, None, None) + + def test_get_impersonate_targets(self): + handlers = [] + for target_client in ('abc', 'xyz', 'asd'): + class TestRH(ImpersonateRequestHandler): + def _send(self, request: Request): + pass + _SUPPORTED_URL_SCHEMES = ('http',) + _SUPPORTED_IMPERSONATE_TARGET_MAP = {ImpersonateTarget(target_client,): 'test'} + RH_KEY = target_client + RH_NAME = target_client + handlers.append(TestRH) + + with FakeYDL() as ydl: + ydl._request_director = ydl.build_request_director(handlers) + assert set(ydl._get_available_impersonate_targets()) == { + (ImpersonateTarget('xyz'), 'xyz'), + (ImpersonateTarget('abc'), 'abc'), + (ImpersonateTarget('asd'), 'asd') + } + assert ydl._impersonate_target_available(ImpersonateTarget('abc')) + assert ydl._impersonate_target_available(ImpersonateTarget()) + assert not ydl._impersonate_target_available(ImpersonateTarget('zxy')) + @pytest.mark.parametrize('proxy_key,proxy_url,expected', [ ('http', '__noproxy__', None), ('no', '127.0.0.1,foo.bar', '127.0.0.1,foo.bar'), @@ -1341,23 +1657,17 @@ class TestYoutubeDLNetworking: ('http', 'socks4://example.com', 'socks4://example.com'), ('unrelated', '/bad/proxy', '/bad/proxy'), # clean_proxies should ignore bad proxies ]) - def test_clean_proxy(self, proxy_key, proxy_url, expected): + def test_clean_proxy(self, proxy_key, proxy_url, expected, monkeypatch): # proxies should be cleaned in urlopen() with FakeRHYDL() as ydl: req = ydl.urlopen(Request('test://', proxies={proxy_key: proxy_url})).request assert req.proxies[proxy_key] == expected # and should also be cleaned when building the handler - env_key = f'{proxy_key.upper()}_PROXY' - old_env_proxy = os.environ.get(env_key) - try: - os.environ[env_key] = proxy_url # ensure that provided proxies override env - with FakeYDL() as ydl: - rh = self.build_handler(ydl) - assert rh.proxies[proxy_key] == expected - finally: - if old_env_proxy: - os.environ[env_key] = old_env_proxy + monkeypatch.setenv(f'{proxy_key.upper()}_PROXY', proxy_url) + with FakeYDL() as ydl: + rh = self.build_handler(ydl) + assert rh.proxies[proxy_key] == expected def test_clean_proxy_header(self): with FakeRHYDL() as ydl: @@ -1629,3 +1939,71 @@ class TestResponse: assert res.geturl() == res.url assert res.info() is res.headers assert res.getheader('test') == res.get_header('test') + + +class TestImpersonateTarget: + @pytest.mark.parametrize('target_str,expected', [ + ('abc', ImpersonateTarget('abc', None, None, None)), + ('abc-120_esr', ImpersonateTarget('abc', '120_esr', None, None)), + ('abc-120:xyz', ImpersonateTarget('abc', '120', 'xyz', None)), + ('abc-120:xyz-5.6', ImpersonateTarget('abc', '120', 'xyz', '5.6')), + ('abc:xyz', ImpersonateTarget('abc', None, 'xyz', None)), + ('abc:', ImpersonateTarget('abc', None, None, None)), + ('abc-120:', ImpersonateTarget('abc', '120', None, None)), + (':xyz', ImpersonateTarget(None, None, 'xyz', None)), + (':xyz-6.5', ImpersonateTarget(None, None, 'xyz', '6.5')), + (':', ImpersonateTarget(None, None, None, None)), + ('', ImpersonateTarget(None, None, None, None)), + ]) + def test_target_from_str(self, target_str, expected): + assert ImpersonateTarget.from_str(target_str) == expected + + @pytest.mark.parametrize('target_str', [ + '-120', ':-12.0', '-12:-12', '-:-', + '::', 'a-c-d:', 'a-c-d:e-f-g', 'a:b:' + ]) + def test_target_from_invalid_str(self, target_str): + with pytest.raises(ValueError): + ImpersonateTarget.from_str(target_str) + + @pytest.mark.parametrize('target,expected', [ + (ImpersonateTarget('abc', None, None, None), 'abc'), + (ImpersonateTarget('abc', '120', None, None), 'abc-120'), + (ImpersonateTarget('abc', '120', 'xyz', None), 'abc-120:xyz'), + (ImpersonateTarget('abc', '120', 'xyz', '5'), 'abc-120:xyz-5'), + (ImpersonateTarget('abc', None, 'xyz', None), 'abc:xyz'), + (ImpersonateTarget('abc', '120', None, None), 'abc-120'), + (ImpersonateTarget('abc', '120', 'xyz', None), 'abc-120:xyz'), + (ImpersonateTarget('abc', None, 'xyz'), 'abc:xyz'), + (ImpersonateTarget(None, None, 'xyz', '6.5'), ':xyz-6.5'), + (ImpersonateTarget('abc', ), 'abc'), + (ImpersonateTarget(None, None, None, None), ''), + ]) + def test_str(self, target, expected): + assert str(target) == expected + + @pytest.mark.parametrize('args', [ + ('abc', None, None, '5'), + ('abc', '120', None, '5'), + (None, '120', None, None), + (None, '120', None, '5'), + (None, None, None, '5'), + (None, '120', 'xyz', '5'), + ]) + def test_invalid_impersonate_target(self, args): + with pytest.raises(ValueError): + ImpersonateTarget(*args) + + @pytest.mark.parametrize('target1,target2,is_in,is_eq', [ + (ImpersonateTarget('abc', None, None, None), ImpersonateTarget('abc', None, None, None), True, True), + (ImpersonateTarget('abc', None, None, None), ImpersonateTarget('abc', '120', None, None), True, False), + (ImpersonateTarget('abc', None, 'xyz', 'test'), ImpersonateTarget('abc', '120', 'xyz', None), True, False), + (ImpersonateTarget('abc', '121', 'xyz', 'test'), ImpersonateTarget('abc', '120', 'xyz', 'test'), False, False), + (ImpersonateTarget('abc'), ImpersonateTarget('abc', '120', 'xyz', 'test'), True, False), + (ImpersonateTarget('abc', '120', 'xyz', 'test'), ImpersonateTarget('abc'), True, False), + (ImpersonateTarget(), ImpersonateTarget('abc', '120', 'xyz'), True, False), + (ImpersonateTarget(), ImpersonateTarget(), True, True), + ]) + def test_impersonate_target_in(self, target1, target2, is_in, is_eq): + assert (target1 in target2) is is_in + assert (target1 == target2) is is_eq diff --git a/test/test_socks.py b/test/test_socks.py index cb22b61..43d612d 100644 --- a/test/test_socks.py +++ b/test/test_socks.py @@ -286,8 +286,14 @@ def ctx(request): return CTX_MAP[request.param]() +@pytest.mark.parametrize( + 'handler,ctx', [ + ('Urllib', 'http'), + ('Requests', 'http'), + ('Websockets', 'ws'), + ('CurlCFFI', 'http') + ], indirect=True) class TestSocks4Proxy: - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True) def test_socks4_no_auth(self, handler, ctx): with handler() as rh: with ctx.socks_server(Socks4ProxyHandler) as server_address: @@ -295,7 +301,6 @@ class TestSocks4Proxy: rh, proxies={'all': f'socks4://{server_address}'}) assert response['version'] == 4 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True) def test_socks4_auth(self, handler, ctx): with handler() as rh: with ctx.socks_server(Socks4ProxyHandler, user_id='user') as server_address: @@ -305,7 +310,6 @@ class TestSocks4Proxy: rh, proxies={'all': f'socks4://user:@{server_address}'}) assert response['version'] == 4 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -313,7 +317,6 @@ class TestSocks4Proxy: 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'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -322,7 +325,6 @@ class TestSocks4Proxy: assert response['ipv4_address'] is None assert response['domain_address'] == 'localhost' - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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)}' @@ -333,7 +335,6 @@ class TestSocks4Proxy: assert response['client_address'][0] == source_address assert response['version'] == 4 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True) @pytest.mark.parametrize('reply_code', [ Socks4CD.REQUEST_REJECTED_OR_FAILED, Socks4CD.REQUEST_REJECTED_CANNOT_CONNECT_TO_IDENTD, @@ -345,7 +346,6 @@ class TestSocks4Proxy: with pytest.raises(ProxyError): ctx.socks_info_request(rh) - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -354,7 +354,6 @@ class TestSocks4Proxy: assert response['ipv4_address'] == '127.0.0.1' assert response['version'] == 4 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -362,9 +361,15 @@ class TestSocks4Proxy: ctx.socks_info_request(rh) +@pytest.mark.parametrize( + 'handler,ctx', [ + ('Urllib', 'http'), + ('Requests', 'http'), + ('Websockets', 'ws'), + ('CurlCFFI', 'http') + ], indirect=True) class TestSocks5Proxy: - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -372,7 +377,6 @@ class TestSocks5Proxy: assert response['auth_methods'] == [0x0] assert response['version'] == 5 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -385,7 +389,6 @@ class TestSocks5Proxy: assert response['auth_methods'] == [Socks5Auth.AUTH_NONE, Socks5Auth.AUTH_USER_PASS] assert response['version'] == 5 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -393,7 +396,6 @@ class TestSocks5Proxy: assert response['ipv4_address'] == '127.0.0.1' assert response['version'] == 5 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -401,7 +403,6 @@ class TestSocks5Proxy: assert (response['ipv4_address'] == '127.0.0.1') != (response['ipv6_address'] == '::1') assert response['version'] == 5 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -410,7 +411,6 @@ class TestSocks5Proxy: assert response['domain_address'] == 'localhost' assert response['version'] == 5 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -419,7 +419,6 @@ class TestSocks5Proxy: assert response['domain_address'] is None assert response['version'] == 5 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -427,7 +426,6 @@ class TestSocks5Proxy: assert response['ipv6_address'] == '::1' assert response['version'] == 5 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], 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: @@ -438,7 +436,6 @@ class TestSocks5Proxy: # 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'), ('Requests', 'http'), ('Websockets', 'ws')], 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)}' @@ -448,7 +445,6 @@ class TestSocks5Proxy: assert response['client_address'][0] == source_address assert response['version'] == 5 - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True) @pytest.mark.parametrize('reply_code', [ Socks5Reply.GENERAL_FAILURE, Socks5Reply.CONNECTION_NOT_ALLOWED, @@ -465,7 +461,6 @@ class TestSocks5Proxy: with pytest.raises(ProxyError): ctx.socks_info_request(rh) - @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Websockets', 'ws')], indirect=True) def test_timeout(self, handler, ctx): with ctx.socks_server(Socks5ProxyHandler, sleep=2) as server_address: with handler(proxies={'all': f'socks5://{server_address}'}, timeout=1) as rh: diff --git a/test/test_traversal.py b/test/test_traversal.py new file mode 100644 index 0000000..9b2a27b --- /dev/null +++ b/test/test_traversal.py @@ -0,0 +1,444 @@ +import http.cookies +import re +import xml.etree.ElementTree + +import pytest + +from yt_dlp.utils import dict_get, int_or_none, str_or_none +from yt_dlp.utils.traversal import traverse_obj + +_TEST_DATA = { + 100: 100, + 1.2: 1.2, + 'str': 'str', + 'None': None, + '...': ..., + 'urls': [ + {'index': 0, 'url': 'https://www.example.com/0'}, + {'index': 1, 'url': 'https://www.example.com/1'}, + ], + 'data': ( + {'index': 2}, + {'index': 3}, + ), + 'dict': {}, +} + + +class TestTraversal: + def test_traversal_base(self): + assert traverse_obj(_TEST_DATA, ('str',)) == 'str', \ + 'allow tuple path' + assert traverse_obj(_TEST_DATA, ['str']) == 'str', \ + 'allow list path' + assert traverse_obj(_TEST_DATA, (value for value in ("str",))) == 'str', \ + 'allow iterable path' + assert traverse_obj(_TEST_DATA, 'str') == 'str', \ + 'single items should be treated as a path' + assert traverse_obj(_TEST_DATA, 100) == 100, \ + 'allow int path' + assert traverse_obj(_TEST_DATA, 1.2) == 1.2, \ + 'allow float path' + assert traverse_obj(_TEST_DATA, None) == _TEST_DATA, \ + '`None` should not perform any modification' + + def test_traversal_ellipsis(self): + assert traverse_obj(_TEST_DATA, ...) == [x for x in _TEST_DATA.values() if x not in (None, {})], \ + '`...` should give all non discarded values' + assert traverse_obj(_TEST_DATA, ('urls', 0, ...)) == list(_TEST_DATA['urls'][0].values()), \ + '`...` selection for dicts should select all values' + assert traverse_obj(_TEST_DATA, (..., ..., 'url')) == ['https://www.example.com/0', 'https://www.example.com/1'], \ + 'nested `...` queries should work' + assert traverse_obj(_TEST_DATA, (..., ..., 'index')) == list(range(4)), \ + '`...` query result should be flattened' + assert traverse_obj(iter(range(4)), ...) == list(range(4)), \ + '`...` should accept iterables' + + def test_traversal_function(self): + filter_func = lambda x, y: x == 'urls' and isinstance(y, list) + assert traverse_obj(_TEST_DATA, filter_func) == [_TEST_DATA['urls']], \ + 'function as query key should perform a filter based on (key, value)' + assert traverse_obj(_TEST_DATA, lambda _, x: isinstance(x[0], str)) == ['str'], \ + 'exceptions in the query function should be catched' + assert traverse_obj(iter(range(4)), lambda _, x: x % 2 == 0) == [0, 2], \ + 'function key should accept iterables' + # Wrong function signature should raise (debug mode) + with pytest.raises(Exception): + traverse_obj(_TEST_DATA, lambda a: ...) + with pytest.raises(Exception): + traverse_obj(_TEST_DATA, lambda a, b, c: ...) + + def test_traversal_set(self): + # transformation/type, like `expected_type` + assert traverse_obj(_TEST_DATA, (..., {str.upper}, )) == ['STR'], \ + 'Function in set should be a transformation' + assert traverse_obj(_TEST_DATA, (..., {str})) == ['str'], \ + 'Type in set should be a type filter' + assert traverse_obj(_TEST_DATA, (..., {str, int})) == [100, 'str'], \ + 'Multiple types in set should be a type filter' + assert traverse_obj(_TEST_DATA, {dict}) == _TEST_DATA, \ + 'A single set should be wrapped into a path' + assert traverse_obj(_TEST_DATA, (..., {str.upper})) == ['STR'], \ + 'Transformation function should not raise' + expected = [x for x in map(str_or_none, _TEST_DATA.values()) if x is not None] + assert traverse_obj(_TEST_DATA, (..., {str_or_none})) == expected, \ + 'Function in set should be a transformation' + assert traverse_obj(_TEST_DATA, ('fail', {lambda _: 'const'})) == 'const', \ + 'Function in set should always be called' + # Sets with length < 1 or > 1 not including only types should raise + with pytest.raises(Exception): + traverse_obj(_TEST_DATA, set()) + with pytest.raises(Exception): + traverse_obj(_TEST_DATA, {str.upper, str}) + + def test_traversal_slice(self): + _SLICE_DATA = [0, 1, 2, 3, 4] + + assert traverse_obj(_TEST_DATA, ('dict', slice(1))) is None, \ + 'slice on a dictionary should not throw' + assert traverse_obj(_SLICE_DATA, slice(1)) == _SLICE_DATA[:1], \ + 'slice key should apply slice to sequence' + assert traverse_obj(_SLICE_DATA, slice(1, 2)) == _SLICE_DATA[1:2], \ + 'slice key should apply slice to sequence' + assert traverse_obj(_SLICE_DATA, slice(1, 4, 2)) == _SLICE_DATA[1:4:2], \ + 'slice key should apply slice to sequence' + + def test_traversal_alternatives(self): + assert traverse_obj(_TEST_DATA, 'fail', 'str') == 'str', \ + 'multiple `paths` should be treated as alternative paths' + assert traverse_obj(_TEST_DATA, 'str', 100) == 'str', \ + 'alternatives should exit early' + assert traverse_obj(_TEST_DATA, 'fail', 'fail') is None, \ + 'alternatives should return `default` if exhausted' + assert traverse_obj(_TEST_DATA, (..., 'fail'), 100) == 100, \ + 'alternatives should track their own branching return' + assert traverse_obj(_TEST_DATA, ('dict', ...), ('data', ...)) == list(_TEST_DATA['data']), \ + 'alternatives on empty objects should search further' + + def test_traversal_branching_nesting(self): + assert traverse_obj(_TEST_DATA, ('urls', (3, 0), 'url')) == ['https://www.example.com/0'], \ + 'tuple as key should be treated as branches' + assert traverse_obj(_TEST_DATA, ('urls', [3, 0], 'url')) == ['https://www.example.com/0'], \ + 'list as key should be treated as branches' + assert traverse_obj(_TEST_DATA, ('urls', ((1, 'fail'), (0, 'url')))) == ['https://www.example.com/0'], \ + 'double nesting in path should be treated as paths' + assert traverse_obj(['0', [1, 2]], [(0, 1), 0]) == [1], \ + 'do not fail early on branching' + expected = ['https://www.example.com/0', 'https://www.example.com/1'] + assert traverse_obj(_TEST_DATA, ('urls', ((0, ('fail', 'url')), (1, 'url')))) == expected, \ + 'tripple nesting in path should be treated as branches' + assert traverse_obj(_TEST_DATA, ('urls', ('fail', (..., 'url')))) == expected, \ + 'ellipsis as branch path start gets flattened' + + def test_traversal_dict(self): + assert traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}) == {0: 100, 1: 1.2}, \ + 'dict key should result in a dict with the same keys' + expected = {0: 'https://www.example.com/0'} + assert traverse_obj(_TEST_DATA, {0: ('urls', 0, 'url')}) == expected, \ + 'dict key should allow paths' + expected = {0: ['https://www.example.com/0']} + assert traverse_obj(_TEST_DATA, {0: ('urls', (3, 0), 'url')}) == expected, \ + 'tuple in dict path should be treated as branches' + assert traverse_obj(_TEST_DATA, {0: ('urls', ((1, 'fail'), (0, 'url')))}) == expected, \ + 'double nesting in dict path should be treated as paths' + expected = {0: ['https://www.example.com/1', 'https://www.example.com/0']} + assert traverse_obj(_TEST_DATA, {0: ('urls', ((1, ('fail', 'url')), (0, 'url')))}) == expected, \ + 'tripple nesting in dict path should be treated as branches' + assert traverse_obj(_TEST_DATA, {0: 'fail'}) == {}, \ + 'remove `None` values when top level dict key fails' + assert traverse_obj(_TEST_DATA, {0: 'fail'}, default=...) == {0: ...}, \ + 'use `default` if key fails and `default`' + assert traverse_obj(_TEST_DATA, {0: 'dict'}) == {}, \ + 'remove empty values when dict key' + assert traverse_obj(_TEST_DATA, {0: 'dict'}, default=...) == {0: ...}, \ + 'use `default` when dict key and `default`' + assert traverse_obj(_TEST_DATA, {0: {0: 'fail'}}) == {}, \ + 'remove empty values when nested dict key fails' + assert traverse_obj(None, {0: 'fail'}) == {}, \ + 'default to dict if pruned' + assert traverse_obj(None, {0: 'fail'}, default=...) == {0: ...}, \ + 'default to dict if pruned and default is given' + assert traverse_obj(_TEST_DATA, {0: {0: 'fail'}}, default=...) == {0: {0: ...}}, \ + 'use nested `default` when nested dict key fails and `default`' + assert traverse_obj(_TEST_DATA, {0: ('dict', ...)}) == {}, \ + 'remove key if branch in dict key not successful' + + def test_traversal_default(self): + _DEFAULT_DATA = {'None': None, 'int': 0, 'list': []} + + assert traverse_obj(_DEFAULT_DATA, 'fail') is None, \ + 'default value should be `None`' + assert traverse_obj(_DEFAULT_DATA, 'fail', 'fail', default=...) == ..., \ + 'chained fails should result in default' + assert traverse_obj(_DEFAULT_DATA, 'None', 'int') == 0, \ + 'should not short cirquit on `None`' + assert traverse_obj(_DEFAULT_DATA, 'fail', default=1) == 1, \ + 'invalid dict key should result in `default`' + assert traverse_obj(_DEFAULT_DATA, 'None', default=1) == 1, \ + '`None` is a deliberate sentinel and should become `default`' + assert traverse_obj(_DEFAULT_DATA, ('list', 10)) is None, \ + '`IndexError` should result in `default`' + assert traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=1) == 1, \ + 'if branched but not successful return `default` if defined, not `[]`' + assert traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=None) is None, \ + 'if branched but not successful return `default` even if `default` is `None`' + assert traverse_obj(_DEFAULT_DATA, (..., 'fail')) == [], \ + 'if branched but not successful return `[]`, not `default`' + assert traverse_obj(_DEFAULT_DATA, ('list', ...)) == [], \ + 'if branched but object is empty return `[]`, not `default`' + assert traverse_obj(None, ...) == [], \ + 'if branched but object is `None` return `[]`, not `default`' + assert traverse_obj({0: None}, (0, ...)) == [], \ + 'if branched but state is `None` return `[]`, not `default`' + + @pytest.mark.parametrize('path', [ + ('fail', ...), + (..., 'fail'), + 100 * ('fail',) + (...,), + (...,) + 100 * ('fail',), + ]) + def test_traversal_branching(self, path): + assert traverse_obj({}, path) == [], \ + 'if branched but state is `None`, return `[]` (not `default`)' + assert traverse_obj({}, 'fail', path) == [], \ + 'if branching in last alternative and previous did not match, return `[]` (not `default`)' + assert traverse_obj({0: 'x'}, 0, path) == 'x', \ + 'if branching in last alternative and previous did match, return single value' + assert traverse_obj({0: 'x'}, path, 0) == 'x', \ + 'if branching in first alternative and non-branching path does match, return single value' + assert traverse_obj({}, path, 'fail') is None, \ + 'if branching in first alternative and non-branching path does not match, return `default`' + + def test_traversal_expected_type(self): + _EXPECTED_TYPE_DATA = {'str': 'str', 'int': 0} + + assert traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=str) == 'str', \ + 'accept matching `expected_type` type' + assert traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=int) is None, \ + 'reject non matching `expected_type` type' + assert traverse_obj(_EXPECTED_TYPE_DATA, 'int', expected_type=lambda x: str(x)) == '0', \ + 'transform type using type function' + assert traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=lambda _: 1 / 0) is None, \ + 'wrap expected_type fuction in try_call' + assert traverse_obj(_EXPECTED_TYPE_DATA, ..., expected_type=str) == ['str'], \ + 'eliminate items that expected_type fails on' + assert traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}, expected_type=int) == {0: 100}, \ + 'type as expected_type should filter dict values' + assert traverse_obj(_TEST_DATA, {0: 100, 1: 1.2, 2: 'None'}, expected_type=str_or_none) == {0: '100', 1: '1.2'}, \ + 'function as expected_type should transform dict values' + assert traverse_obj(_TEST_DATA, ({0: 1.2}, 0, {int_or_none}), expected_type=int) == 1, \ + 'expected_type should not filter non final dict values' + assert traverse_obj(_TEST_DATA, {0: {0: 100, 1: 'str'}}, expected_type=int) == {0: {0: 100}}, \ + 'expected_type should transform deep dict values' + assert traverse_obj(_TEST_DATA, [({0: '...'}, {0: '...'})], expected_type=type(...)) == [{0: ...}, {0: ...}], \ + 'expected_type should transform branched dict values' + assert traverse_obj({1: {3: 4}}, [(1, 2), 3], expected_type=int) == [4], \ + 'expected_type regression for type matching in tuple branching' + assert traverse_obj(_TEST_DATA, ['data', ...], expected_type=int) == [], \ + 'expected_type regression for type matching in dict result' + + def test_traversal_get_all(self): + _GET_ALL_DATA = {'key': [0, 1, 2]} + + assert traverse_obj(_GET_ALL_DATA, ('key', ...), get_all=False) == 0, \ + 'if not `get_all`, return only first matching value' + assert traverse_obj(_GET_ALL_DATA, ..., get_all=False) == [0, 1, 2], \ + 'do not overflatten if not `get_all`' + + def test_traversal_casesense(self): + _CASESENSE_DATA = { + 'KeY': 'value0', + 0: { + 'KeY': 'value1', + 0: {'KeY': 'value2'}, + }, + } + + assert traverse_obj(_CASESENSE_DATA, 'key') is None, \ + 'dict keys should be case sensitive unless `casesense`' + assert traverse_obj(_CASESENSE_DATA, 'keY', casesense=False) == 'value0', \ + 'allow non matching key case if `casesense`' + assert traverse_obj(_CASESENSE_DATA, [0, ('keY',)], casesense=False) == ['value1'], \ + 'allow non matching key case in branch if `casesense`' + assert traverse_obj(_CASESENSE_DATA, [0, ([0, 'keY'],)], casesense=False) == ['value2'], \ + 'allow non matching key case in branch path if `casesense`' + + def test_traversal_traverse_string(self): + _TRAVERSE_STRING_DATA = {'str': 'str', 1.2: 1.2} + + assert traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0)) is None, \ + 'do not traverse into string if not `traverse_string`' + assert traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0), traverse_string=True) == 's', \ + 'traverse into string if `traverse_string`' + assert traverse_obj(_TRAVERSE_STRING_DATA, (1.2, 1), traverse_string=True) == '.', \ + 'traverse into converted data if `traverse_string`' + assert traverse_obj(_TRAVERSE_STRING_DATA, ('str', ...), traverse_string=True) == 'str', \ + '`...` should result in string (same value) if `traverse_string`' + assert traverse_obj(_TRAVERSE_STRING_DATA, ('str', slice(0, None, 2)), traverse_string=True) == 'sr', \ + '`slice` should result in string if `traverse_string`' + assert traverse_obj(_TRAVERSE_STRING_DATA, ('str', lambda i, v: i or v == "s"), traverse_string=True) == 'str', \ + 'function should result in string if `traverse_string`' + assert traverse_obj(_TRAVERSE_STRING_DATA, ('str', (0, 2)), traverse_string=True) == ['s', 'r'], \ + 'branching should result in list if `traverse_string`' + assert traverse_obj({}, (0, ...), traverse_string=True) == [], \ + 'branching should result in list if `traverse_string`' + assert traverse_obj({}, (0, lambda x, y: True), traverse_string=True) == [], \ + 'branching should result in list if `traverse_string`' + assert traverse_obj({}, (0, slice(1)), traverse_string=True) == [], \ + 'branching should result in list if `traverse_string`' + + def test_traversal_re(self): + mobj = re.fullmatch(r'0(12)(?P<group>3)(4)?', '0123') + assert traverse_obj(mobj, ...) == [x for x in mobj.groups() if x is not None], \ + '`...` on a `re.Match` should give its `groups()`' + assert traverse_obj(mobj, lambda k, _: k in (0, 2)) == ['0123', '3'], \ + 'function on a `re.Match` should give groupno, value starting at 0' + assert traverse_obj(mobj, 'group') == '3', \ + 'str key on a `re.Match` should give group with that name' + assert traverse_obj(mobj, 2) == '3', \ + 'int key on a `re.Match` should give group with that name' + assert traverse_obj(mobj, 'gRoUp', casesense=False) == '3', \ + 'str key on a `re.Match` should respect casesense' + assert traverse_obj(mobj, 'fail') is None, \ + 'failing str key on a `re.Match` should return `default`' + assert traverse_obj(mobj, 'gRoUpS', casesense=False) is None, \ + 'failing str key on a `re.Match` should return `default`' + assert traverse_obj(mobj, 8) is None, \ + 'failing int key on a `re.Match` should return `default`' + assert traverse_obj(mobj, lambda k, _: k in (0, 'group')) == ['0123', '3'], \ + 'function on a `re.Match` should give group name as well' + + def test_traversal_xml_etree(self): + etree = xml.etree.ElementTree.fromstring('''<?xml version="1.0"?> + <data> + <country name="Liechtenstein"> + <rank>1</rank> + <year>2008</year> + <gdppc>141100</gdppc> + <neighbor name="Austria" direction="E"/> + <neighbor name="Switzerland" direction="W"/> + </country> + <country name="Singapore"> + <rank>4</rank> + <year>2011</year> + <gdppc>59900</gdppc> + <neighbor name="Malaysia" direction="N"/> + </country> + <country name="Panama"> + <rank>68</rank> + <year>2011</year> + <gdppc>13600</gdppc> + <neighbor name="Costa Rica" direction="W"/> + <neighbor name="Colombia" direction="E"/> + </country> + </data>''') + assert traverse_obj(etree, '') == etree, \ + 'empty str key should return the element itself' + assert traverse_obj(etree, 'country') == list(etree), \ + 'str key should lead all children with that tag name' + assert traverse_obj(etree, ...) == list(etree), \ + '`...` as key should return all children' + assert traverse_obj(etree, lambda _, x: x[0].text == '4') == [etree[1]], \ + 'function as key should get element as value' + assert traverse_obj(etree, lambda i, _: i == 1) == [etree[1]], \ + 'function as key should get index as key' + assert traverse_obj(etree, 0) == etree[0], \ + 'int key should return the nth child' + expected = ['Austria', 'Switzerland', 'Malaysia', 'Costa Rica', 'Colombia'] + assert traverse_obj(etree, './/neighbor/@name') == expected, \ + '`@<attribute>` at end of path should give that attribute' + assert traverse_obj(etree, '//neighbor/@fail') == [None, None, None, None, None], \ + '`@<nonexistant>` at end of path should give `None`' + assert traverse_obj(etree, ('//neighbor/@', 2)) == {'name': 'Malaysia', 'direction': 'N'}, \ + '`@` should give the full attribute dict' + assert traverse_obj(etree, '//year/text()') == ['2008', '2011', '2011'], \ + '`text()` at end of path should give the inner text' + assert traverse_obj(etree, '//*[@direction]/@direction') == ['E', 'W', 'N', 'W', 'E'], \ + 'full Python xpath features should be supported' + assert traverse_obj(etree, (0, '@name')) == 'Liechtenstein', \ + 'special transformations should act on current element' + assert traverse_obj(etree, ('country', 0, ..., 'text()', {int_or_none})) == [1, 2008, 141100], \ + 'special transformations should act on current element' + + def test_traversal_unbranching(self): + assert traverse_obj(_TEST_DATA, [(100, 1.2), all]) == [100, 1.2], \ + '`all` should give all results as list' + assert traverse_obj(_TEST_DATA, [(100, 1.2), any]) == 100, \ + '`any` should give the first result' + assert traverse_obj(_TEST_DATA, [100, all]) == [100], \ + '`all` should give list if non branching' + assert traverse_obj(_TEST_DATA, [100, any]) == 100, \ + '`any` should give single item if non branching' + assert traverse_obj(_TEST_DATA, [('dict', 'None', 100), all]) == [100], \ + '`all` should filter `None` and empty dict' + assert traverse_obj(_TEST_DATA, [('dict', 'None', 100), any]) == 100, \ + '`any` should filter `None` and empty dict' + assert traverse_obj(_TEST_DATA, [{ + 'all': [('dict', 'None', 100, 1.2), all], + 'any': [('dict', 'None', 100, 1.2), any], + }]) == {'all': [100, 1.2], 'any': 100}, \ + '`all`/`any` should apply to each dict path separately' + assert traverse_obj(_TEST_DATA, [{ + 'all': [('dict', 'None', 100, 1.2), all], + 'any': [('dict', 'None', 100, 1.2), any], + }], get_all=False) == {'all': [100, 1.2], 'any': 100}, \ + '`all`/`any` should apply to dict regardless of `get_all`' + assert traverse_obj(_TEST_DATA, [('dict', 'None', 100, 1.2), all, {float}]) is None, \ + '`all` should reset branching status' + assert traverse_obj(_TEST_DATA, [('dict', 'None', 100, 1.2), any, {float}]) is None, \ + '`any` should reset branching status' + assert traverse_obj(_TEST_DATA, [('dict', 'None', 100, 1.2), all, ..., {float}]) == [1.2], \ + '`all` should allow further branching' + assert traverse_obj(_TEST_DATA, [('dict', 'None', 'urls', 'data'), any, ..., 'index']) == [0, 1], \ + '`any` should allow further branching' + + def test_traversal_morsel(self): + values = { + 'expires': 'a', + 'path': 'b', + 'comment': 'c', + 'domain': 'd', + 'max-age': 'e', + 'secure': 'f', + 'httponly': 'g', + 'version': 'h', + 'samesite': 'i', + } + morsel = http.cookies.Morsel() + morsel.set('item_key', 'item_value', 'coded_value') + morsel.update(values) + values['key'] = 'item_key' + values['value'] = 'item_value' + + for key, value in values.items(): + assert traverse_obj(morsel, key) == value, \ + 'Morsel should provide access to all values' + assert traverse_obj(morsel, ...) == list(values.values()), \ + '`...` should yield all values' + assert traverse_obj(morsel, lambda k, v: True) == list(values.values()), \ + 'function key should yield all values' + assert traverse_obj(morsel, [(None,), any]) == morsel, \ + 'Morsel should not be implicitly changed to dict on usage' + + +class TestDictGet: + def test_dict_get(self): + FALSE_VALUES = { + 'none': None, + 'false': False, + 'zero': 0, + 'empty_string': '', + 'empty_list': [], + } + d = {**FALSE_VALUES, 'a': 42} + assert dict_get(d, 'a') == 42 + assert dict_get(d, 'b') is None + assert dict_get(d, 'b', 42) == 42 + assert dict_get(d, ('a',)) == 42 + assert dict_get(d, ('b', 'a')) == 42 + assert dict_get(d, ('b', 'c', 'a', 'd')) == 42 + assert dict_get(d, ('b', 'c')) is None + assert dict_get(d, ('b', 'c'), 42) == 42 + for key, false_value in FALSE_VALUES.items(): + assert dict_get(d, ('b', 'c', key)) is None + assert dict_get(d, ('b', 'c', key), skip_false_values=False) == false_value diff --git a/test/test_utils.py b/test/test_utils.py index a3073f0..ddf0a7c 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -2,7 +2,6 @@ # Allow direct execution import os -import re import sys import unittest import warnings @@ -45,7 +44,6 @@ from yt_dlp.utils import ( determine_ext, determine_file_encoding, dfxp2srt, - dict_get, encode_base_n, encode_compat_str, encodeFilename, @@ -106,13 +104,11 @@ from yt_dlp.utils import ( sanitize_url, shell_quote, smuggle_url, - str_or_none, str_to_int, strip_jsonp, strip_or_none, subtitles_filename, timeconvert, - traverse_obj, try_call, unescapeHTML, unified_strdate, @@ -755,28 +751,6 @@ class TestUtil(unittest.TestCase): self.assertRaises( ValueError, multipart_encode, {b'field': b'value'}, boundary='value') - def test_dict_get(self): - FALSE_VALUES = { - 'none': None, - 'false': False, - 'zero': 0, - 'empty_string': '', - 'empty_list': [], - } - d = FALSE_VALUES.copy() - d['a'] = 42 - self.assertEqual(dict_get(d, 'a'), 42) - self.assertEqual(dict_get(d, 'b'), None) - self.assertEqual(dict_get(d, 'b', 42), 42) - self.assertEqual(dict_get(d, ('a', )), 42) - self.assertEqual(dict_get(d, ('b', 'a', )), 42) - self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42) - self.assertEqual(dict_get(d, ('b', 'c', )), None) - self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42) - for key, false_value in FALSE_VALUES.items(): - self.assertEqual(dict_get(d, ('b', 'c', key, )), None) - self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value) - def test_merge_dicts(self): self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2}) self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1}) @@ -2039,359 +2013,6 @@ Line 1 warnings.simplefilter('ignore') self.assertEqual(variadic('spam', allowed_types=[dict]), 'spam') - def test_traverse_obj(self): - _TEST_DATA = { - 100: 100, - 1.2: 1.2, - 'str': 'str', - 'None': None, - '...': ..., - 'urls': [ - {'index': 0, 'url': 'https://www.example.com/0'}, - {'index': 1, 'url': 'https://www.example.com/1'}, - ], - 'data': ( - {'index': 2}, - {'index': 3}, - ), - 'dict': {}, - } - - # Test base functionality - self.assertEqual(traverse_obj(_TEST_DATA, ('str',)), 'str', - msg='allow tuple path') - self.assertEqual(traverse_obj(_TEST_DATA, ['str']), 'str', - msg='allow list path') - self.assertEqual(traverse_obj(_TEST_DATA, (value for value in ("str",))), 'str', - msg='allow iterable path') - self.assertEqual(traverse_obj(_TEST_DATA, 'str'), 'str', - msg='single items should be treated as a path') - self.assertEqual(traverse_obj(_TEST_DATA, None), _TEST_DATA) - self.assertEqual(traverse_obj(_TEST_DATA, 100), 100) - self.assertEqual(traverse_obj(_TEST_DATA, 1.2), 1.2) - - # Test Ellipsis behavior - self.assertCountEqual(traverse_obj(_TEST_DATA, ...), - (item for item in _TEST_DATA.values() if item not in (None, {})), - msg='`...` should give all non discarded values') - self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', 0, ...)), _TEST_DATA['urls'][0].values(), - msg='`...` selection for dicts should select all values') - self.assertEqual(traverse_obj(_TEST_DATA, (..., ..., 'url')), - ['https://www.example.com/0', 'https://www.example.com/1'], - msg='nested `...` queries should work') - self.assertCountEqual(traverse_obj(_TEST_DATA, (..., ..., 'index')), range(4), - msg='`...` query result should be flattened') - self.assertEqual(traverse_obj(iter(range(4)), ...), list(range(4)), - msg='`...` should accept iterables') - - # Test function as key - self.assertEqual(traverse_obj(_TEST_DATA, lambda x, y: x == 'urls' and isinstance(y, list)), - [_TEST_DATA['urls']], - msg='function as query key should perform a filter based on (key, value)') - self.assertCountEqual(traverse_obj(_TEST_DATA, lambda _, x: isinstance(x[0], str)), {'str'}, - msg='exceptions in the query function should be catched') - self.assertEqual(traverse_obj(iter(range(4)), lambda _, x: x % 2 == 0), [0, 2], - msg='function key should accept iterables') - if __debug__: - with self.assertRaises(Exception, msg='Wrong function signature should raise in debug'): - traverse_obj(_TEST_DATA, lambda a: ...) - with self.assertRaises(Exception, msg='Wrong function signature should raise in debug'): - traverse_obj(_TEST_DATA, lambda a, b, c: ...) - - # Test set as key (transformation/type, like `expected_type`) - self.assertEqual(traverse_obj(_TEST_DATA, (..., {str.upper}, )), ['STR'], - msg='Function in set should be a transformation') - self.assertEqual(traverse_obj(_TEST_DATA, (..., {str})), ['str'], - msg='Type in set should be a type filter') - self.assertEqual(traverse_obj(_TEST_DATA, {dict}), _TEST_DATA, - msg='A single set should be wrapped into a path') - self.assertEqual(traverse_obj(_TEST_DATA, (..., {str.upper})), ['STR'], - msg='Transformation function should not raise') - self.assertEqual(traverse_obj(_TEST_DATA, (..., {str_or_none})), - [item for item in map(str_or_none, _TEST_DATA.values()) if item is not None], - msg='Function in set should be a transformation') - self.assertEqual(traverse_obj(_TEST_DATA, ('fail', {lambda _: 'const'})), 'const', - msg='Function in set should always be called') - if __debug__: - with self.assertRaises(Exception, msg='Sets with length != 1 should raise in debug'): - traverse_obj(_TEST_DATA, set()) - with self.assertRaises(Exception, msg='Sets with length != 1 should raise in debug'): - traverse_obj(_TEST_DATA, {str.upper, str}) - - # Test `slice` as a key - _SLICE_DATA = [0, 1, 2, 3, 4] - self.assertEqual(traverse_obj(_TEST_DATA, ('dict', slice(1))), None, - msg='slice on a dictionary should not throw') - self.assertEqual(traverse_obj(_SLICE_DATA, slice(1)), _SLICE_DATA[:1], - msg='slice key should apply slice to sequence') - self.assertEqual(traverse_obj(_SLICE_DATA, slice(1, 2)), _SLICE_DATA[1:2], - msg='slice key should apply slice to sequence') - self.assertEqual(traverse_obj(_SLICE_DATA, slice(1, 4, 2)), _SLICE_DATA[1:4:2], - msg='slice key should apply slice to sequence') - - # Test alternative paths - self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'str'), 'str', - msg='multiple `paths` should be treated as alternative paths') - self.assertEqual(traverse_obj(_TEST_DATA, 'str', 100), 'str', - msg='alternatives should exit early') - self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'fail'), None, - msg='alternatives should return `default` if exhausted') - self.assertEqual(traverse_obj(_TEST_DATA, (..., 'fail'), 100), 100, - msg='alternatives should track their own branching return') - self.assertEqual(traverse_obj(_TEST_DATA, ('dict', ...), ('data', ...)), list(_TEST_DATA['data']), - msg='alternatives on empty objects should search further') - - # Test branch and path nesting - self.assertEqual(traverse_obj(_TEST_DATA, ('urls', (3, 0), 'url')), ['https://www.example.com/0'], - msg='tuple as key should be treated as branches') - self.assertEqual(traverse_obj(_TEST_DATA, ('urls', [3, 0], 'url')), ['https://www.example.com/0'], - msg='list as key should be treated as branches') - self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ((1, 'fail'), (0, 'url')))), ['https://www.example.com/0'], - msg='double nesting in path should be treated as paths') - self.assertEqual(traverse_obj(['0', [1, 2]], [(0, 1), 0]), [1], - msg='do not fail early on branching') - self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', ((1, ('fail', 'url')), (0, 'url')))), - ['https://www.example.com/0', 'https://www.example.com/1'], - msg='tripple nesting in path should be treated as branches') - self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ('fail', (..., 'url')))), - ['https://www.example.com/0', 'https://www.example.com/1'], - msg='ellipsis as branch path start gets flattened') - - # Test dictionary as key - self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}), {0: 100, 1: 1.2}, - msg='dict key should result in a dict with the same keys') - self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', 0, 'url')}), - {0: 'https://www.example.com/0'}, - msg='dict key should allow paths') - self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', (3, 0), 'url')}), - {0: ['https://www.example.com/0']}, - msg='tuple in dict path should be treated as branches') - self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, 'fail'), (0, 'url')))}), - {0: ['https://www.example.com/0']}, - msg='double nesting in dict path should be treated as paths') - self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, ('fail', 'url')), (0, 'url')))}), - {0: ['https://www.example.com/1', 'https://www.example.com/0']}, - msg='tripple nesting in dict path should be treated as branches') - self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}), {}, - msg='remove `None` values when top level dict key fails') - self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}, default=...), {0: ...}, - msg='use `default` if key fails and `default`') - self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}), {}, - msg='remove empty values when dict key') - self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}, default=...), {0: ...}, - msg='use `default` when dict key and `default`') - self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 'fail'}}), {}, - msg='remove empty values when nested dict key fails') - self.assertEqual(traverse_obj(None, {0: 'fail'}), {}, - msg='default to dict if pruned') - self.assertEqual(traverse_obj(None, {0: 'fail'}, default=...), {0: ...}, - msg='default to dict if pruned and default is given') - self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 'fail'}}, default=...), {0: {0: ...}}, - msg='use nested `default` when nested dict key fails and `default`') - self.assertEqual(traverse_obj(_TEST_DATA, {0: ('dict', ...)}), {}, - msg='remove key if branch in dict key not successful') - - # Testing default parameter behavior - _DEFAULT_DATA = {'None': None, 'int': 0, 'list': []} - self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail'), None, - msg='default value should be `None`') - self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', 'fail', default=...), ..., - msg='chained fails should result in default') - self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', 'int'), 0, - msg='should not short cirquit on `None`') - self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', default=1), 1, - msg='invalid dict key should result in `default`') - self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', default=1), 1, - msg='`None` is a deliberate sentinel and should become `default`') - self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', 10)), None, - msg='`IndexError` should result in `default`') - self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=1), 1, - msg='if branched but not successful return `default` if defined, not `[]`') - self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=None), None, - msg='if branched but not successful return `default` even if `default` is `None`') - self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail')), [], - msg='if branched but not successful return `[]`, not `default`') - self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', ...)), [], - msg='if branched but object is empty return `[]`, not `default`') - self.assertEqual(traverse_obj(None, ...), [], - msg='if branched but object is `None` return `[]`, not `default`') - self.assertEqual(traverse_obj({0: None}, (0, ...)), [], - msg='if branched but state is `None` return `[]`, not `default`') - - branching_paths = [ - ('fail', ...), - (..., 'fail'), - 100 * ('fail',) + (...,), - (...,) + 100 * ('fail',), - ] - for branching_path in branching_paths: - self.assertEqual(traverse_obj({}, branching_path), [], - msg='if branched but state is `None`, return `[]` (not `default`)') - self.assertEqual(traverse_obj({}, 'fail', branching_path), [], - msg='if branching in last alternative and previous did not match, return `[]` (not `default`)') - self.assertEqual(traverse_obj({0: 'x'}, 0, branching_path), 'x', - msg='if branching in last alternative and previous did match, return single value') - self.assertEqual(traverse_obj({0: 'x'}, branching_path, 0), 'x', - msg='if branching in first alternative and non-branching path does match, return single value') - self.assertEqual(traverse_obj({}, branching_path, 'fail'), None, - msg='if branching in first alternative and non-branching path does not match, return `default`') - - # Testing expected_type behavior - _EXPECTED_TYPE_DATA = {'str': 'str', 'int': 0} - self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=str), - 'str', msg='accept matching `expected_type` type') - self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=int), - None, msg='reject non matching `expected_type` type') - self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'int', expected_type=lambda x: str(x)), - '0', msg='transform type using type function') - self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=lambda _: 1 / 0), - None, msg='wrap expected_type fuction in try_call') - self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, ..., expected_type=str), - ['str'], msg='eliminate items that expected_type fails on') - self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}, expected_type=int), - {0: 100}, msg='type as expected_type should filter dict values') - self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2, 2: 'None'}, expected_type=str_or_none), - {0: '100', 1: '1.2'}, msg='function as expected_type should transform dict values') - self.assertEqual(traverse_obj(_TEST_DATA, ({0: 1.2}, 0, {int_or_none}), expected_type=int), - 1, msg='expected_type should not filter non final dict values') - self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 100, 1: 'str'}}, expected_type=int), - {0: {0: 100}}, msg='expected_type should transform deep dict values') - self.assertEqual(traverse_obj(_TEST_DATA, [({0: '...'}, {0: '...'})], expected_type=type(...)), - [{0: ...}, {0: ...}], msg='expected_type should transform branched dict values') - self.assertEqual(traverse_obj({1: {3: 4}}, [(1, 2), 3], expected_type=int), - [4], msg='expected_type regression for type matching in tuple branching') - self.assertEqual(traverse_obj(_TEST_DATA, ['data', ...], expected_type=int), - [], msg='expected_type regression for type matching in dict result') - - # Test get_all behavior - _GET_ALL_DATA = {'key': [0, 1, 2]} - self.assertEqual(traverse_obj(_GET_ALL_DATA, ('key', ...), get_all=False), 0, - msg='if not `get_all`, return only first matching value') - self.assertEqual(traverse_obj(_GET_ALL_DATA, ..., get_all=False), [0, 1, 2], - msg='do not overflatten if not `get_all`') - - # Test casesense behavior - _CASESENSE_DATA = { - 'KeY': 'value0', - 0: { - 'KeY': 'value1', - 0: {'KeY': 'value2'}, - }, - } - self.assertEqual(traverse_obj(_CASESENSE_DATA, 'key'), None, - msg='dict keys should be case sensitive unless `casesense`') - self.assertEqual(traverse_obj(_CASESENSE_DATA, 'keY', - casesense=False), 'value0', - msg='allow non matching key case if `casesense`') - self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ('keY',)), - casesense=False), ['value1'], - msg='allow non matching key case in branch if `casesense`') - self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ((0, 'keY'),)), - casesense=False), ['value2'], - msg='allow non matching key case in branch path if `casesense`') - - # Test traverse_string behavior - _TRAVERSE_STRING_DATA = {'str': 'str', 1.2: 1.2} - self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0)), None, - msg='do not traverse into string if not `traverse_string`') - self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0), - traverse_string=True), 's', - msg='traverse into string if `traverse_string`') - self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, (1.2, 1), - traverse_string=True), '.', - msg='traverse into converted data if `traverse_string`') - self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', ...), - traverse_string=True), 'str', - msg='`...` should result in string (same value) if `traverse_string`') - self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', slice(0, None, 2)), - traverse_string=True), 'sr', - msg='`slice` should result in string if `traverse_string`') - self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', lambda i, v: i or v == "s"), - traverse_string=True), 'str', - msg='function should result in string if `traverse_string`') - self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', (0, 2)), - traverse_string=True), ['s', 'r'], - msg='branching should result in list if `traverse_string`') - self.assertEqual(traverse_obj({}, (0, ...), traverse_string=True), [], - msg='branching should result in list if `traverse_string`') - self.assertEqual(traverse_obj({}, (0, lambda x, y: True), traverse_string=True), [], - msg='branching should result in list if `traverse_string`') - self.assertEqual(traverse_obj({}, (0, slice(1)), traverse_string=True), [], - msg='branching should result in list if `traverse_string`') - - # Test re.Match as input obj - mobj = re.fullmatch(r'0(12)(?P<group>3)(4)?', '0123') - self.assertEqual(traverse_obj(mobj, ...), [x for x in mobj.groups() if x is not None], - msg='`...` on a `re.Match` should give its `groups()`') - self.assertEqual(traverse_obj(mobj, lambda k, _: k in (0, 2)), ['0123', '3'], - msg='function on a `re.Match` should give groupno, value starting at 0') - self.assertEqual(traverse_obj(mobj, 'group'), '3', - msg='str key on a `re.Match` should give group with that name') - self.assertEqual(traverse_obj(mobj, 2), '3', - msg='int key on a `re.Match` should give group with that name') - self.assertEqual(traverse_obj(mobj, 'gRoUp', casesense=False), '3', - msg='str key on a `re.Match` should respect casesense') - self.assertEqual(traverse_obj(mobj, 'fail'), None, - msg='failing str key on a `re.Match` should return `default`') - self.assertEqual(traverse_obj(mobj, 'gRoUpS', casesense=False), None, - msg='failing str key on a `re.Match` should return `default`') - self.assertEqual(traverse_obj(mobj, 8), None, - msg='failing int key on a `re.Match` should return `default`') - self.assertEqual(traverse_obj(mobj, lambda k, _: k in (0, 'group')), ['0123', '3'], - msg='function on a `re.Match` should give group name as well') - - # Test xml.etree.ElementTree.Element as input obj - etree = xml.etree.ElementTree.fromstring('''<?xml version="1.0"?> - <data> - <country name="Liechtenstein"> - <rank>1</rank> - <year>2008</year> - <gdppc>141100</gdppc> - <neighbor name="Austria" direction="E"/> - <neighbor name="Switzerland" direction="W"/> - </country> - <country name="Singapore"> - <rank>4</rank> - <year>2011</year> - <gdppc>59900</gdppc> - <neighbor name="Malaysia" direction="N"/> - </country> - <country name="Panama"> - <rank>68</rank> - <year>2011</year> - <gdppc>13600</gdppc> - <neighbor name="Costa Rica" direction="W"/> - <neighbor name="Colombia" direction="E"/> - </country> - </data>''') - self.assertEqual(traverse_obj(etree, ''), etree, - msg='empty str key should return the element itself') - self.assertEqual(traverse_obj(etree, 'country'), list(etree), - msg='str key should lead all children with that tag name') - self.assertEqual(traverse_obj(etree, ...), list(etree), - msg='`...` as key should return all children') - self.assertEqual(traverse_obj(etree, lambda _, x: x[0].text == '4'), [etree[1]], - msg='function as key should get element as value') - self.assertEqual(traverse_obj(etree, lambda i, _: i == 1), [etree[1]], - msg='function as key should get index as key') - self.assertEqual(traverse_obj(etree, 0), etree[0], - msg='int key should return the nth child') - self.assertEqual(traverse_obj(etree, './/neighbor/@name'), - ['Austria', 'Switzerland', 'Malaysia', 'Costa Rica', 'Colombia'], - msg='`@<attribute>` at end of path should give that attribute') - self.assertEqual(traverse_obj(etree, '//neighbor/@fail'), [None, None, None, None, None], - msg='`@<nonexistant>` at end of path should give `None`') - self.assertEqual(traverse_obj(etree, ('//neighbor/@', 2)), {'name': 'Malaysia', 'direction': 'N'}, - msg='`@` should give the full attribute dict') - self.assertEqual(traverse_obj(etree, '//year/text()'), ['2008', '2011', '2011'], - msg='`text()` at end of path should give the inner text') - self.assertEqual(traverse_obj(etree, '//*[@direction]/@direction'), ['E', 'W', 'N', 'W', 'E'], - msg='full Python xpath features should be supported') - self.assertEqual(traverse_obj(etree, (0, '@name')), 'Liechtenstein', - msg='special transformations should act on current element') - self.assertEqual(traverse_obj(etree, ('country', 0, ..., 'text()', {int_or_none})), [1, 2008, 141100], - msg='special transformations should act on current element') - def test_http_header_dict(self): headers = HTTPHeaderDict() headers['ytdl-test'] = b'0' @@ -2448,6 +2069,10 @@ Line 1 # Test escaping assert run_shell(['echo', 'test"&']) == '"test""&"\n' + assert run_shell(['echo', '%CMDCMDLINE:~-1%&']) == '"%CMDCMDLINE:~-1%&"\n' + assert run_shell(['echo', 'a\nb']) == '"a"\n"b"\n' + assert run_shell(['echo', '"']) == '""""\n' + assert run_shell(['echo', '\\']) == '\\\n' # Test if delayed expansion is disabled assert run_shell(['echo', '^!']) == '"^!"\n' assert run_shell('echo "^!"') == '"^!"\n' diff --git a/test/test_websockets.py b/test/test_websockets.py index 13b3a1e..b294b09 100644 --- a/test/test_websockets.py +++ b/test/test_websockets.py @@ -32,8 +32,6 @@ from yt_dlp.networking.exceptions import ( ) from yt_dlp.utils.networking import HTTPHeaderDict -from test.conftest import validate_and_send - TEST_DIR = os.path.dirname(os.path.abspath(__file__)) @@ -66,7 +64,9 @@ def process_request(self, request): def create_websocket_server(**ws_kwargs): import websockets.sync.server - wsd = websockets.sync.server.serve(websocket_handler, '127.0.0.1', 0, process_request=process_request, **ws_kwargs) + wsd = websockets.sync.server.serve( + websocket_handler, '127.0.0.1', 0, + process_request=process_request, open_timeout=2, **ws_kwargs) ws_port = wsd.socket.getsockname()[1] ws_server_thread = threading.Thread(target=wsd.serve_forever) ws_server_thread.daemon = True @@ -100,6 +100,19 @@ def create_mtls_wss_websocket_server(): return create_websocket_server(ssl_context=sslctx) +def ws_validate_and_send(rh, req): + rh.validate(req) + max_tries = 3 + for i in range(max_tries): + try: + return rh.send(req) + except TransportError as e: + if i < (max_tries - 1) and 'connection closed during handshake' in str(e): + # websockets server sometimes hangs on new connections + continue + raise + + @pytest.mark.skipif(not websockets, reason='websockets must be installed to test websocket request handlers') class TestWebsSocketRequestHandlerConformance: @classmethod @@ -119,7 +132,7 @@ class TestWebsSocketRequestHandlerConformance: @pytest.mark.parametrize('handler', ['Websockets'], indirect=True) def test_basic_websockets(self, handler): with handler() as rh: - ws = validate_and_send(rh, Request(self.ws_base_url)) + ws = ws_validate_and_send(rh, Request(self.ws_base_url)) assert 'upgrade' in ws.headers assert ws.status == 101 ws.send('foo') @@ -131,7 +144,7 @@ class TestWebsSocketRequestHandlerConformance: @pytest.mark.parametrize('handler', ['Websockets'], indirect=True) def test_send_types(self, handler, msg, opcode): with handler() as rh: - ws = validate_and_send(rh, Request(self.ws_base_url)) + ws = ws_validate_and_send(rh, Request(self.ws_base_url)) ws.send(msg) assert int(ws.recv()) == opcode ws.close() @@ -140,10 +153,10 @@ class TestWebsSocketRequestHandlerConformance: def test_verify_cert(self, handler): with handler() as rh: with pytest.raises(CertificateVerifyError): - validate_and_send(rh, Request(self.wss_base_url)) + ws_validate_and_send(rh, Request(self.wss_base_url)) with handler(verify=False) as rh: - ws = validate_and_send(rh, Request(self.wss_base_url)) + ws = ws_validate_and_send(rh, Request(self.wss_base_url)) assert ws.status == 101 ws.close() @@ -151,7 +164,7 @@ class TestWebsSocketRequestHandlerConformance: def test_ssl_error(self, handler): with handler(verify=False) as rh: with pytest.raises(SSLError, match=r'ssl(?:v3|/tls) alert handshake failure') as exc_info: - validate_and_send(rh, Request(self.bad_wss_host)) + ws_validate_and_send(rh, Request(self.bad_wss_host)) assert not issubclass(exc_info.type, CertificateVerifyError) @pytest.mark.parametrize('handler', ['Websockets'], indirect=True) @@ -163,7 +176,7 @@ class TestWebsSocketRequestHandlerConformance: ]) def test_percent_encode(self, handler, path, expected): with handler() as rh: - ws = validate_and_send(rh, Request(f'{self.ws_base_url}{path}')) + ws = ws_validate_and_send(rh, Request(f'{self.ws_base_url}{path}')) ws.send('path') assert ws.recv() == expected assert ws.status == 101 @@ -174,7 +187,7 @@ class TestWebsSocketRequestHandlerConformance: with handler() as rh: # This isn't a comprehensive test, # but it should be enough to check whether the handler is removing dot segments - ws = validate_and_send(rh, Request(f'{self.ws_base_url}/a/b/./../../test')) + ws = ws_validate_and_send(rh, Request(f'{self.ws_base_url}/a/b/./../../test')) assert ws.status == 101 ws.send('path') assert ws.recv() == '/test' @@ -187,7 +200,7 @@ class TestWebsSocketRequestHandlerConformance: def test_raise_http_error(self, handler, status): with handler() as rh: with pytest.raises(HTTPError) as exc_info: - validate_and_send(rh, Request(f'{self.ws_base_url}/gen_{status}')) + ws_validate_and_send(rh, Request(f'{self.ws_base_url}/gen_{status}')) assert exc_info.value.status == status @pytest.mark.parametrize('handler', ['Websockets'], indirect=True) @@ -198,7 +211,7 @@ class TestWebsSocketRequestHandlerConformance: def test_timeout(self, handler, params, extensions): with handler(**params) as rh: with pytest.raises(TransportError): - validate_and_send(rh, Request(self.ws_base_url, extensions=extensions)) + ws_validate_and_send(rh, Request(self.ws_base_url, extensions=extensions)) @pytest.mark.parametrize('handler', ['Websockets'], indirect=True) def test_cookies(self, handler): @@ -210,18 +223,18 @@ class TestWebsSocketRequestHandlerConformance: comment_url=None, rest={})) with handler(cookiejar=cookiejar) as rh: - ws = validate_and_send(rh, Request(self.ws_base_url)) + ws = ws_validate_and_send(rh, Request(self.ws_base_url)) ws.send('headers') assert json.loads(ws.recv())['cookie'] == 'test=ytdlp' ws.close() with handler() as rh: - ws = validate_and_send(rh, Request(self.ws_base_url)) + ws = ws_validate_and_send(rh, Request(self.ws_base_url)) ws.send('headers') assert 'cookie' not in json.loads(ws.recv()) ws.close() - ws = validate_and_send(rh, Request(self.ws_base_url, extensions={'cookiejar': cookiejar})) + ws = ws_validate_and_send(rh, Request(self.ws_base_url, extensions={'cookiejar': cookiejar})) ws.send('headers') assert json.loads(ws.recv())['cookie'] == 'test=ytdlp' ws.close() @@ -231,7 +244,7 @@ class TestWebsSocketRequestHandlerConformance: source_address = f'127.0.0.{random.randint(5, 255)}' verify_address_availability(source_address) with handler(source_address=source_address) as rh: - ws = validate_and_send(rh, Request(self.ws_base_url)) + ws = ws_validate_and_send(rh, Request(self.ws_base_url)) ws.send('source_address') assert source_address == ws.recv() ws.close() @@ -240,7 +253,7 @@ class TestWebsSocketRequestHandlerConformance: def test_response_url(self, handler): with handler() as rh: url = f'{self.ws_base_url}/something' - ws = validate_and_send(rh, Request(url)) + ws = ws_validate_and_send(rh, Request(url)) assert ws.url == url ws.close() @@ -248,14 +261,14 @@ class TestWebsSocketRequestHandlerConformance: def test_request_headers(self, handler): with handler(headers=HTTPHeaderDict({'test1': 'test', 'test2': 'test2'})) as rh: # Global Headers - ws = validate_and_send(rh, Request(self.ws_base_url)) + ws = ws_validate_and_send(rh, Request(self.ws_base_url)) ws.send('headers') headers = HTTPHeaderDict(json.loads(ws.recv())) assert headers['test1'] == 'test' ws.close() # Per request headers, merged with global - ws = validate_and_send(rh, Request( + ws = ws_validate_and_send(rh, Request( self.ws_base_url, headers={'test2': 'changed', 'test3': 'test3'})) ws.send('headers') headers = HTTPHeaderDict(json.loads(ws.recv())) @@ -288,7 +301,7 @@ class TestWebsSocketRequestHandlerConformance: verify=False, client_cert=client_cert ) as rh: - validate_and_send(rh, Request(self.mtls_wss_base_url)).close() + ws_validate_and_send(rh, Request(self.mtls_wss_base_url)).close() def create_fake_ws_connection(raised): |