diff options
Diffstat (limited to 'searx')
32 files changed, 839 insertions, 348 deletions
diff --git a/searx/autocomplete.py b/searx/autocomplete.py index 75992a1d8..cf2deb62f 100644 --- a/searx/autocomplete.py +++ b/searx/autocomplete.py @@ -20,10 +20,11 @@ from lxml import etree from json import loads from urllib.parse import urlencode -from requests import RequestException +from httpx import HTTPError + from searx import settings -from searx.poolrequests import get as http_get +from searx.network import get as http_get from searx.exceptions import SearxEngineResponseException @@ -136,5 +137,5 @@ def search_autocomplete(backend_name, query, lang): try: return backend(query, lang) - except (RequestException, SearxEngineResponseException): + except (HTTPError, SearxEngineResponseException): return [] diff --git a/searx/engines/__init__.py b/searx/engines/__init__.py index 2238ea1b9..95eda6dde 100644 --- a/searx/engines/__init__.py +++ b/searx/engines/__init__.py @@ -27,7 +27,7 @@ from searx import settings from searx import logger from searx.data import ENGINES_LANGUAGES from searx.exceptions import SearxEngineResponseException -from searx.poolrequests import get, get_proxy_cycles +from searx.network import get, initialize as initialize_network, set_context_network_name from searx.utils import load_module, match_language, get_engine_from_settings, gen_useragent @@ -89,8 +89,6 @@ def load_engine(engine_data): engine.categories = [] else: engine.categories = list(map(str.strip, param_value.split(','))) - elif param_name == 'proxies': - engine.proxies = get_proxy_cycles(param_value) else: setattr(engine, param_name, param_value) @@ -289,9 +287,11 @@ def load_engines(engine_list): def initialize_engines(engine_list): load_engines(engine_list) + initialize_network(engine_list, settings['outgoing']) def engine_init(engine_name, init_fn): try: + set_context_network_name(engine_name) init_fn(get_engine_from_settings(engine_name)) except SearxEngineResponseException as exc: logger.warn('%s engine: Fail to initialize // %s', engine_name, exc) diff --git a/searx/engines/dictzone.py b/searx/engines/dictzone.py index 2483c0805..eaa8b6ab4 100644 --- a/searx/engines/dictzone.py +++ b/searx/engines/dictzone.py @@ -52,7 +52,7 @@ def response(resp): to_results.append(to_result.text_content()) results.append({ - 'url': urljoin(resp.url, '?%d' % k), + 'url': urljoin(str(resp.url), '?%d' % k), 'title': from_result.text_content(), 'content': '; '.join(to_results) }) diff --git a/searx/engines/duckduckgo.py b/searx/engines/duckduckgo.py index ae1e36686..3c086f81b 100644 --- a/searx/engines/duckduckgo.py +++ b/searx/engines/duckduckgo.py @@ -6,7 +6,7 @@ from lxml.html import fromstring from json import loads from searx.utils import extract_text, match_language, eval_xpath, dict_subset -from searx.poolrequests import get +from searx.network import get # about about = { diff --git a/searx/engines/duckduckgo_images.py b/searx/engines/duckduckgo_images.py index 305eb1ca1..0daaf41e9 100644 --- a/searx/engines/duckduckgo_images.py +++ b/searx/engines/duckduckgo_images.py @@ -8,7 +8,7 @@ from urllib.parse import urlencode from searx.exceptions import SearxEngineAPIException from searx.engines.duckduckgo import get_region_code from searx.engines.duckduckgo import _fetch_supported_languages, supported_languages_url # NOQA # pylint: disable=unused-import -from searx.poolrequests import get +from searx.network import get # about about = { diff --git a/searx/engines/elasticsearch.py b/searx/engines/elasticsearch.py index da7f98074..db84a5c13 100644 --- a/searx/engines/elasticsearch.py +++ b/searx/engines/elasticsearch.py @@ -4,7 +4,6 @@ """ from json import loads, dumps -from requests.auth import HTTPBasicAuth from searx.exceptions import SearxEngineAPIException @@ -32,7 +31,7 @@ def request(query, params): return params if username and password: - params['auth'] = HTTPBasicAuth(username, password) + params['auth'] = (username, password) params['url'] = search_url params['method'] = 'GET' diff --git a/searx/engines/gigablast.py b/searx/engines/gigablast.py index 248991df9..bbd9e20d2 100644 --- a/searx/engines/gigablast.py +++ b/searx/engines/gigablast.py @@ -8,7 +8,7 @@ import re from json import loads from urllib.parse import urlencode # from searx import logger -from searx.poolrequests import get +from searx.network import get # about about = { diff --git a/searx/engines/google.py b/searx/engines/google.py index 8f5b700eb..a4aee5c20 100644 --- a/searx/engines/google.py +++ b/searx/engines/google.py @@ -10,7 +10,7 @@ Definitions`_. # pylint: disable=invalid-name, missing-function-docstring -from urllib.parse import urlencode, urlparse +from urllib.parse import urlencode from lxml import html from searx import logger from searx.utils import match_language, extract_text, eval_xpath, eval_xpath_list, eval_xpath_getindex @@ -186,8 +186,7 @@ def get_lang_info(params, lang_list, custom_aliases): return ret_val def detect_google_sorry(resp): - resp_url = urlparse(resp.url) - if resp_url.netloc == 'sorry.google.com' or resp_url.path.startswith('/sorry'): + if resp.url.host == 'sorry.google.com' or resp.url.path.startswith('/sorry'): raise SearxEngineCaptchaException() diff --git a/searx/engines/pubmed.py b/searx/engines/pubmed.py index da02f91ca..5d88d398e 100644 --- a/searx/engines/pubmed.py +++ b/searx/engines/pubmed.py @@ -7,7 +7,7 @@ from flask_babel import gettext from lxml import etree from datetime import datetime from urllib.parse import urlencode -from searx.poolrequests import get +from searx.network import get # about about = { diff --git a/searx/engines/qwant.py b/searx/engines/qwant.py index 13dcf1250..d01dc0acc 100644 --- a/searx/engines/qwant.py +++ b/searx/engines/qwant.py @@ -8,7 +8,7 @@ from json import loads from urllib.parse import urlencode from searx.utils import html_to_text, match_language from searx.exceptions import SearxEngineAPIException, SearxEngineCaptchaException -from searx.raise_for_httperror import raise_for_httperror +from searx.network import raise_for_httperror # about about = { diff --git a/searx/engines/seznam.py b/searx/engines/seznam.py index faceb0550..042088dbe 100644 --- a/searx/engines/seznam.py +++ b/searx/engines/seznam.py @@ -3,9 +3,9 @@ Seznam """ -from urllib.parse import urlencode, urlparse +from urllib.parse import urlencode from lxml import html -from searx.poolrequests import get +from searx.network import get from searx.exceptions import SearxEngineAccessDeniedException from searx.utils import ( extract_text, @@ -46,8 +46,7 @@ def request(query, params): def response(resp): - resp_url = urlparse(resp.url) - if resp_url.path.startswith('/verify'): + if resp.url.path.startswith('/verify'): raise SearxEngineAccessDeniedException() results = [] diff --git a/searx/engines/soundcloud.py b/searx/engines/soundcloud.py index b3e3383bd..a6f923855 100644 --- a/searx/engines/soundcloud.py +++ b/searx/engines/soundcloud.py @@ -9,7 +9,7 @@ from lxml import html from dateutil import parser from urllib.parse import quote_plus, urlencode from searx import logger -from searx.poolrequests import get as http_get +from searx.network import get as http_get # about about = { diff --git a/searx/engines/spotify.py b/searx/engines/spotify.py index 0ad8bfe32..6816fe672 100644 --- a/searx/engines/spotify.py +++ b/searx/engines/spotify.py @@ -5,9 +5,10 @@ from json import loads from urllib.parse import urlencode -import requests import base64 +from searx.network import post as http_post + # about about = { "website": 'https://www.spotify.com', @@ -38,7 +39,7 @@ def request(query, params): params['url'] = search_url.format(query=urlencode({'q': query}), offset=offset) - r = requests.post( + r = http_post( 'https://accounts.spotify.com/api/token', data={'grant_type': 'client_credentials'}, headers={'Authorization': 'Basic ' + base64.b64encode( diff --git a/searx/engines/stackoverflow.py b/searx/engines/stackoverflow.py index 91eaa68e9..8fc2cdb3a 100644 --- a/searx/engines/stackoverflow.py +++ b/searx/engines/stackoverflow.py @@ -3,7 +3,7 @@ Stackoverflow (IT) """ -from urllib.parse import urlencode, urljoin, urlparse +from urllib.parse import urlencode, urljoin from lxml import html from searx.utils import extract_text from searx.exceptions import SearxEngineCaptchaException @@ -41,8 +41,7 @@ def request(query, params): # get response from search-request def response(resp): - resp_url = urlparse(resp.url) - if resp_url.path.startswith('/nocaptcha'): + if resp.url.path.startswith('/nocaptcha'): raise SearxEngineCaptchaException() results = [] diff --git a/searx/engines/wikidata.py b/searx/engines/wikidata.py index c8e4cfae6..ddcce9085 100644 --- a/searx/engines/wikidata.py +++ b/searx/engines/wikidata.py @@ -12,7 +12,7 @@ from babel.dates import format_datetime, format_date, format_time, get_datetime_ from searx import logger from searx.data import WIKIDATA_UNITS -from searx.poolrequests import post, get +from searx.network import post, get from searx.utils import match_language, searx_useragent, get_string_replaces_function from searx.external_urls import get_external_url, get_earth_coordinates_url, area_to_osm_zoom from searx.engines.wikipedia import _fetch_supported_languages, supported_languages_url # NOQA # pylint: disable=unused-import diff --git a/searx/engines/wikipedia.py b/searx/engines/wikipedia.py index 3ad8748fb..5e34db9a7 100644 --- a/searx/engines/wikipedia.py +++ b/searx/engines/wikipedia.py @@ -7,7 +7,7 @@ from urllib.parse import quote from json import loads from lxml.html import fromstring from searx.utils import match_language, searx_useragent -from searx.raise_for_httperror import raise_for_httperror +from searx.network import raise_for_httperror # about about = { diff --git a/searx/engines/wolframalpha_noapi.py b/searx/engines/wolframalpha_noapi.py index 8e427d575..1f2cfa4e6 100644 --- a/searx/engines/wolframalpha_noapi.py +++ b/searx/engines/wolframalpha_noapi.py @@ -7,7 +7,7 @@ from json import loads from time import time from urllib.parse import urlencode -from searx.poolrequests import get as http_get +from searx.network import get as http_get # about about = { diff --git a/searx/engines/wordnik.py b/searx/engines/wordnik.py index 3abe9efa2..4bfeb4070 100644 --- a/searx/engines/wordnik.py +++ b/searx/engines/wordnik.py @@ -6,7 +6,7 @@ from lxml.html import fromstring from searx import logger from searx.utils import extract_text -from searx.raise_for_httperror import raise_for_httperror +from searx.network import raise_for_httperror logger = logger.getChild('Wordnik engine') diff --git a/searx/engines/yacy.py b/searx/engines/yacy.py index c194ca451..fbd99c47b 100644 --- a/searx/engines/yacy.py +++ b/searx/engines/yacy.py @@ -7,7 +7,7 @@ from json import loads from dateutil import parser from urllib.parse import urlencode -from requests.auth import HTTPDigestAuth +from httpx import DigestAuth from searx.utils import html_to_text @@ -56,7 +56,7 @@ def request(query, params): search_type=search_type) if http_digest_auth_user and http_digest_auth_pass: - params['auth'] = HTTPDigestAuth(http_digest_auth_user, http_digest_auth_pass) + params['auth'] = DigestAuth(http_digest_auth_user, http_digest_auth_pass) # add language tag if specified if params['language'] != 'all': diff --git a/searx/engines/yggtorrent.py b/searx/engines/yggtorrent.py index 8dfc0a0f2..f5af91f46 100644 --- a/searx/engines/yggtorrent.py +++ b/searx/engines/yggtorrent.py @@ -8,7 +8,7 @@ from operator import itemgetter from datetime import datetime from urllib.parse import quote from searx.utils import extract_text, get_torrent_size -from searx.poolrequests import get as http_get +from searx.network import get as http_get # about about = { diff --git a/searx/metrology/error_recorder.py b/searx/metrology/error_recorder.py index f533e4e8b..167d1c8aa 100644 --- a/searx/metrology/error_recorder.py +++ b/searx/metrology/error_recorder.py @@ -3,7 +3,7 @@ import inspect import logging from json import JSONDecodeError from urllib.parse import urlparse -from requests.exceptions import RequestException +from httpx import HTTPError, HTTPStatusError from searx.exceptions import (SearxXPathSyntaxException, SearxEngineXPathException, SearxEngineAPIException, SearxEngineAccessDeniedException) from searx import logger @@ -60,28 +60,28 @@ def get_trace(traces): return traces[-1] -def get_hostname(exc: RequestException) -> typing.Optional[None]: +def get_hostname(exc: HTTPError) -> typing.Optional[None]: url = exc.request.url if url is None and exc.response is not None: url = exc.response.url return urlparse(url).netloc -def get_request_exception_messages(exc: RequestException)\ +def get_request_exception_messages(exc: HTTPError)\ -> typing.Tuple[typing.Optional[str], typing.Optional[str], typing.Optional[str]]: url = None status_code = None reason = None hostname = None - if exc.request is not None: + if hasattr(exc, 'request') and exc.request is not None: url = exc.request.url - if url is None and exc.response is not None: + if url is None and hasattr(exc, 'response') and exc.respones is not None: url = exc.response.url if url is not None: - hostname = str(urlparse(url).netloc) - if exc.response is not None: + hostname = url.host + if isinstance(exc, HTTPStatusError): status_code = str(exc.response.status_code) - reason = exc.response.reason + reason = exc.response.reason_phrase return (status_code, reason, hostname) @@ -92,7 +92,7 @@ def get_messages(exc, filename) -> typing.Tuple: return (str(exc), ) if isinstance(exc, ValueError) and 'lxml' in filename: return (str(exc), ) - if isinstance(exc, RequestException): + if isinstance(exc, HTTPError): return get_request_exception_messages(exc) if isinstance(exc, SearxXPathSyntaxException): return (exc.xpath_str, exc.message) diff --git a/searx/network/__init__.py b/searx/network/__init__.py new file mode 100644 index 000000000..dbd31c781 --- /dev/null +++ b/searx/network/__init__.py @@ -0,0 +1,189 @@ +# SPDX-License-Identifier: AGPL-3.0-or-later + +import asyncio +import threading +import concurrent.futures +from time import time + +import httpx +import h2.exceptions + +from .network import get_network, initialize +from .client import LOOP +from .raise_for_httperror import raise_for_httperror + +# queue.SimpleQueue: Support Python 3.6 +try: + from queue import SimpleQueue +except ImportError: + from queue import Empty + from collections import deque + + class SimpleQueue: + """Minimal backport of queue.SimpleQueue""" + + def __init__(self): + self._queue = deque() + self._count = threading.Semaphore(0) + + def put(self, item): + self._queue.append(item) + self._count.release() + + def get(self): + if not self._count.acquire(True): + raise Empty + return self._queue.popleft() + + +THREADLOCAL = threading.local() + + +def reset_time_for_thread(): + THREADLOCAL.total_time = 0 + + +def get_time_for_thread(): + return THREADLOCAL.total_time + + +def set_timeout_for_thread(timeout, start_time=None): + THREADLOCAL.timeout = timeout + THREADLOCAL.start_time = start_time + + +def set_context_network_name(network_name): + THREADLOCAL.network = get_network(network_name) + + +def get_context_network(): + try: + return THREADLOCAL.network + except AttributeError: + return get_network() + + +def request(method, url, **kwargs): + """same as requests/requests/api.py request(...)""" + time_before_request = time() + + # timeout (httpx) + if 'timeout' in kwargs: + timeout = kwargs['timeout'] + else: + timeout = getattr(THREADLOCAL, 'timeout', None) + if timeout is not None: + kwargs['timeout'] = timeout + + # 2 minutes timeout for the requests without timeout + timeout = timeout or 120 + + # ajdust actual timeout + timeout += 0.2 # overhead + start_time = getattr(THREADLOCAL, 'start_time', time_before_request) + if start_time: + timeout -= time() - start_time + + # raise_for_error + check_for_httperror = True + if 'raise_for_httperror' in kwargs: + check_for_httperror = kwargs['raise_for_httperror'] + del kwargs['raise_for_httperror'] + + # requests compatibility + if isinstance(url, bytes): + url = url.decode() + + # network + network = get_context_network() + + # do request + future = asyncio.run_coroutine_threadsafe(network.request(method, url, **kwargs), LOOP) + try: + response = future.result(timeout) + except concurrent.futures.TimeoutError as e: + raise httpx.TimeoutException('Timeout', request=None) from e + + # requests compatibility + # see also https://www.python-httpx.org/compatibility/#checking-for-4xx5xx-responses + response.ok = not response.is_error + + # update total_time. + # See get_time_for_thread() and reset_time_for_thread() + if hasattr(THREADLOCAL, 'total_time'): + time_after_request = time() + THREADLOCAL.total_time += time_after_request - time_before_request + + # raise an exception + if check_for_httperror: + raise_for_httperror(response) + + return response + + +def get(url, **kwargs): + kwargs.setdefault('allow_redirects', True) + return request('get', url, **kwargs) + + +def options(url, **kwargs): + kwargs.setdefault('allow_redirects', True) + return request('options', url, **kwargs) + + +def head(url, **kwargs): + kwargs.setdefault('allow_redirects', False) + return request('head', url, **kwargs) + + +def post(url, data=None, **kwargs): + return request('post', url, data=data, **kwargs) + + +def put(url, data=None, **kwargs): + return request('put', url, data=data, **kwargs) + + +def patch(url, data=None, **kwargs): + return request('patch', url, data=data, **kwargs) + + +def delete(url, **kwargs): + return request('delete', url, **kwargs) + + +async def stream_chunk_to_queue(network, q, method, url, **kwargs): + try: + async with network.stream(method, url, **kwargs) as response: + q.put(response) + async for chunk in response.aiter_bytes(65536): + if len(chunk) > 0: + q.put(chunk) + except (httpx.HTTPError, OSError, h2.exceptions.ProtocolError) as e: + q.put(e) + finally: + q.put(None) + + +def stream(method, url, **kwargs): + """Replace httpx.stream. + + Usage: + stream = poolrequests.stream(...) + response = next(stream) + for chunk in stream: + ... + + httpx.Client.stream requires to write the httpx.HTTPTransport version of the + the httpx.AsyncHTTPTransport declared above. + """ + q = SimpleQueue() + future = asyncio.run_coroutine_threadsafe(stream_chunk_to_queue(get_network(), q, method, url, **kwargs), + LOOP) + chunk_or_exception = q.get() + while chunk_or_exception is not None: + if isinstance(chunk_or_exception, Exception): + raise chunk_or_exception + yield chunk_or_exception + chunk_or_exception = q.get() + return future.result() diff --git a/searx/network/client.py b/searx/network/client.py new file mode 100644 index 000000000..631e36f8f --- /dev/null +++ b/searx/network/client.py @@ -0,0 +1,214 @@ +# SPDX-License-Identifier: AGPL-3.0-or-later + +import asyncio +import logging +import threading + +import httpcore +import httpx +from httpx_socks import AsyncProxyTransport +from python_socks import parse_proxy_url +import python_socks._errors + +from searx import logger + +# Optional uvloop (support Python 3.6) +try: + import uvloop +except ImportError: + pass +else: + uvloop.install() + + +logger = logger.getChild('searx.http.client') +LOOP = None +TRANSPORT_KWARGS = { + 'backend': 'asyncio', + 'trust_env': False, +} + + +async def close_connections_for_url(connection_pool: httpcore.AsyncConnectionPool, url: httpcore._utils.URL): + origin = httpcore._utils.url_to_origin(url) + logger.debug('Drop connections for %r', origin) + connections_to_close = connection_pool._connections_for_origin(origin) + for connection in connections_to_close: + await connection_pool._remove_from_pool(connection) + try: + await connection.aclose() + except httpcore.NetworkError as e: + logger.warning('Error closing an existing connection', exc_info=e) + + +class AsyncHTTPTransportNoHttp(httpcore.AsyncHTTPTransport): + """Block HTTP request""" + + async def arequest(self, method, url, headers=None, stream=None, ext=None): + raise httpcore.UnsupportedProtocol("HTTP protocol is disabled") + + +class AsyncProxyTransportFixed(AsyncProxyTransport): + """Fix httpx_socks.AsyncProxyTransport + + Map python_socks exceptions to httpcore.ProxyError + + Map socket.gaierror to httpcore.ConnectError + + Note: keepalive_expiry is ignored, AsyncProxyTransport should call: + * self._keepalive_sweep() + * self._response_closed(self, connection) + + Note: AsyncProxyTransport inherit from AsyncConnectionPool + + Note: the API is going to change on httpx 0.18.0 + see https://github.com/encode/httpx/pull/1522 + """ + + async def arequest(self, method, url, headers=None, stream=None, ext=None): + retry = 2 + while retry > 0: + retry -= 1 + try: + return await super().arequest(method, url, headers, stream, ext) + except (python_socks._errors.ProxyConnectionError, + python_socks._errors.ProxyTimeoutError, + python_socks._errors.ProxyError) as e: + raise httpcore.ProxyError(e) + except OSError as e: + # socket.gaierror when DNS resolution fails + raise httpcore.NetworkError(e) + except httpcore.RemoteProtocolError as e: + # in case of httpcore.RemoteProtocolError: Server disconnected + await close_connections_for_url(self, url) + logger.warning('httpcore.RemoteProtocolError: retry', exc_info=e) + # retry + except (httpcore.NetworkError, httpcore.ProtocolError) as e: + # httpcore.WriteError on HTTP/2 connection leaves a new opened stream + # then each new request creates a new stream and raise the same WriteError + await close_connections_for_url(self, url) + raise e + + +class AsyncHTTPTransportFixed(httpx.AsyncHTTPTransport): + """Fix httpx.AsyncHTTPTransport""" + + async def arequest(self, method, url, headers=None, stream=None, ext=None): + retry = 2 + while retry > 0: + retry -= 1 + try: + return await super().arequest(method, url, headers, stream, ext) + except OSError as e: + # socket.gaierror when DNS resolution fails + raise httpcore.ConnectError(e) + except httpcore.CloseError as e: + # httpcore.CloseError: [Errno 104] Connection reset by peer + # raised by _keepalive_sweep() + # from https://github.com/encode/httpcore/blob/4b662b5c42378a61e54d673b4c949420102379f5/httpcore/_backends/asyncio.py#L198 # noqa + await close_connections_for_url(self._pool, url) + logger.warning('httpcore.CloseError: retry', exc_info=e) + # retry + except httpcore.RemoteProtocolError as e: + # in case of httpcore.RemoteProtocolError: Server disconnected + await close_connections_for_url(self._pool, url) + logger.warning('httpcore.RemoteProtocolError: retry', exc_info=e) + # retry + except (httpcore.ProtocolError, httpcore.NetworkError) as e: + await close_connections_for_url(self._pool, url) + raise e + + +def get_transport_for_socks_proxy(verify, http2, local_address, proxy_url, limit, retries): + global LOOP, TRANSPORT_KWARGS + # support socks5h (requests compatibility): + # https://requests.readthedocs.io/en/master/user/advanced/#socks + # socks5:// hostname is resolved on client side + # socks5h:// hostname is resolved on proxy side + rdns = False + socks5h = 'socks5h://' + if proxy_url.startswith(socks5h): + proxy_url = 'socks5://' + proxy_url[len(socks5h):] + rdns = True + + proxy_type, proxy_host, proxy_port, proxy_username, proxy_password = parse_proxy_url(proxy_url) + + return AsyncProxyTransportFixed(proxy_type=proxy_type, proxy_host=proxy_host, proxy_port=proxy_port, + username=proxy_username, password=proxy_password, + rdns=rdns, + loop=LOOP, + verify=verify, + http2=http2, + local_address=local_address, + max_connections=limit.max_connections, + max_keepalive_connections=limit.max_keepalive_connections, + keepalive_expiry=limit.keepalive_expiry, + retries=retries, + **TRANSPORT_KWARGS) + + +def get_transport(verify, http2, local_address, proxy_url, limit, retries): + return AsyncHTTPTransportFixed(verify=verify, + http2=http2, + local_address=local_address, + proxy=httpx._config.Proxy(proxy_url) if proxy_url else None, + limits=limit, + retries=retries, + **TRANSPORT_KWARGS) + + +def iter_proxies(proxies): + # https://www.python-httpx.org/compatibility/#proxy-keys + if isinstance(proxies, str): + yield 'all://', proxies + elif isinstance(proxies, dict): + for pattern, proxy_url in proxies.items(): + yield pattern, proxy_url + + +def new_client(enable_http, verify, enable_http2, + max_connections, max_keepalive_connections, keepalive_expiry, + proxies, local_address, retries, max_redirects): + limit = httpx.Limits(max_connections=max_connections, + max_keepalive_connections=max_keepalive_connections, + keepalive_expiry=keepalive_expiry) + # See https://www.python-httpx.org/advanced/#routing + mounts = {} + for pattern, proxy_url in iter_proxies(proxies): + if not enable_http and (pattern == 'http' or pattern.startswith('http://')): + continue + if proxy_url.startswith('socks4://') \ + or proxy_url.startswith('socks5://') \ + or proxy_url.startswith('socks5h://'): + mounts[pattern] = get_transport_for_socks_proxy(verify, enable_http2, local_address, proxy_url, limit, + retries) + else: + mounts[pattern] = get_transport(verify, enable_http2, local_address, proxy_url, limit, retries) + + if not enable_http: + mounts['http://'] = AsyncHTTPTransportNoHttp() + + transport = get_transport(verify, enable_http2, local_address, None, limit, retries) + return httpx.AsyncClient(transport=transport, mounts=mounts, max_redirects=max_redirects) + + +def init(): + # log + for logger_name in ('hpack.hpack', 'hpack.table'): + logging.getLogger(logger_name).setLevel(logging.WARNING) + + # loop + def loop_thread(): + global LOOP + LOOP = asyncio.new_event_loop() + LOOP.run_forever() + + th = threading.Thread( + target=loop_thread, + name='asyncio_loop', + daemon=True, + ) + th.start() + + +init() diff --git a/searx/network/network.py b/searx/network/network.py new file mode 100644 index 000000000..f50acf595 --- /dev/null +++ b/searx/network/network.py @@ -0,0 +1,302 @@ +# SPDX-License-Identifier: AGPL-3.0-or-later + +import atexit +import asyncio +import ipaddress +from itertools import cycle + +import httpx + +from .client import new_client, LOOP + + +DEFAULT_NAME = '__DEFAULT__' +NETWORKS = {} +# requests compatibility when reading proxy settings from settings.yml +PROXY_PATTERN_MAPPING = { + 'http': 'http://', + 'https': 'https://', + 'socks4': 'socks4://', + 'socks5': 'socks5://', + 'socks5h': 'socks5h://', + 'http:': 'http://', + 'https:': 'https://', + 'socks4:': 'socks4://', + 'socks5:': 'socks5://', + 'socks5h:': 'socks5h://', +} + +ADDRESS_MAPPING = { + 'ipv4': '0.0.0.0', + 'ipv6': '::' +} + + +class Network: + + __slots__ = ('enable_http', 'verify', 'enable_http2', + 'max_connections', 'max_keepalive_connections', 'keepalive_expiry', + 'local_addresses', 'proxies', 'max_redirects', 'retries', 'retry_on_http_error', + '_local_addresses_cycle', '_proxies_cycle', '_clients') + + def __init__(self, + enable_http=True, + verify=True, + enable_http2=False, + max_connections=None, + max_keepalive_connections=None, + keepalive_expiry=None, + proxies=None, + local_addresses=None, + retries=0, + retry_on_http_error=None, + max_redirects=30): + self.enable_http = enable_http + self.verify = verify + self.enable_http2 = enable_http2 + self.max_connections = max_connections + self.max_keepalive_connections = max_keepalive_connections + self.keepalive_expiry = keepalive_expiry + self.proxies = proxies + self.local_addresses = local_addresses + self.retries = retries + self.retry_on_http_error = retry_on_http_error + self.max_redirects = max_redirects + self._local_addresses_cycle = self.get_ipaddress_cycle() + self._proxies_cycle = self.get_proxy_cycles() + self._clients = {} + self.check_parameters() + + def check_parameters(self): + for address in self.iter_ipaddresses(): + if '/' in address: + ipaddress.ip_network(address, False) + else: + ipaddress.ip_address(address) + + if self.proxies is not None and not isinstance(self.proxies, (str, dict)): + raise ValueError('proxies type has to be str, dict or None') + + def iter_ipaddresses(self): + local_addresses = self.local_addresses + if not local_addresses: + return + elif isinstance(local_addresses, str): + local_addresses = [local_addresses] + for address in local_addresses: + yield address + + def get_ipaddress_cycle(self): + while True: + count = 0 + for address in self.iter_ipaddresses(): + if '/' in address: + for a in ipaddress.ip_network(address, False).hosts(): + yield str(a) + count += 1 + else: + a = ipaddress.ip_address(address) + yield str(a) + count += 1 + if count == 0: + yield None + + def iter_proxies(self): + if not self.proxies: + return + # https://www.python-httpx.org/compatibility/#proxy-keys + if isinstance(self.proxies, str): + yield 'all://', [self.proxies] + else: + for pattern, proxy_url in self.proxies.items(): + pattern = PROXY_PATTERN_MAPPING.get(pattern, pattern) + if isinstance(proxy_url, str): + proxy_url = [proxy_url] + yield pattern, proxy_url + + def get_proxy_cycles(self): + proxy_settings = {} + for pattern, proxy_urls in self.iter_proxies(): + proxy_settings[pattern] = cycle(proxy_urls) + while True: + yield tuple((pattern, next(proxy_url_cycle)) for pattern, proxy_url_cycle in proxy_settings.items()) + + def get_client(self, verify=None, max_redirects=None): + verify = self.verify if verify is None else verify + max_redirects = self.max_redirects if max_redirects is None else max_redirects + local_address = next(self._local_addresses_cycle) + proxies = next(self._proxies_cycle) # is a tuple so it can be part of the key + key = (verify, max_redirects, local_address, proxies) + if key not in self._clients or self._clients[key].is_closed: + self._clients[key] = new_client(self.enable_http, + verify, + self.enable_http2, + self.max_connections, + self.max_keepalive_connections, + self.keepalive_expiry, + dict(proxies), + local_address, + 0, + max_redirects) + return self._clients[key] + + async def aclose(self): + async def close_client(client): + try: + await client.aclose() + except httpx.HTTPError: + pass + await asyncio.gather(*[close_client(client) for client in self._clients.values()], return_exceptions=False) + + @staticmethod + def get_kwargs_clients(kwargs): + kwargs_clients = {} + if 'verify' in kwargs: + kwargs_clients['verify'] = kwargs.pop('verify') + if 'max_redirects' in kwargs: + kwargs_clients['max_redirects'] = kwargs.pop('max_redirects') + return kwargs_clients + + def is_valid_respones(self, response): + if (self.retry_on_http_error is True and 400 <= response.status_code <= 599) \ + or (isinstance(self.retry_on_http_error, list) and response.status_code in self.retry_on_http_error) \ + or (isinstance(self.retry_on_http_error, int) and response.status_code == self.retry_on_http_error): + return False + return True + + async def request(self, method, url, **kwargs): + retries = self.retries + while retries >= 0: # pragma: no cover + kwargs_clients = Network.get_kwargs_clients(kwargs) + client = self.get_client(**kwargs_clients) + try: + response = await client.request(method, url, **kwargs) + if self.is_valid_respones(response) or retries <= 0: + return response + except (httpx.RequestError, httpx.HTTPStatusError) as e: + if retries <= 0: + raise e + retries -= 1 + + def stream(self, method, url, **kwargs): + retries = self.retries + while retries >= 0: # pragma: no cover + kwargs_clients = Network.get_kwargs_clients(kwargs) + client = self.get_client(**kwargs_clients) + try: + response = client.stream(method, url, **kwargs) + if self.is_valid_respones(response) or retries <= 0: + return response + except (httpx.RequestError, httpx.HTTPStatusError) as e: + if retries <= 0: + raise e + retries -= 1 + + @classmethod + async def aclose_all(cls): + await asyncio.gather(*[network.aclose() for network in NETWORKS.values()], return_exceptions=False) + + +def get_network(name=None): + global NETWORKS + return NETWORKS[name or DEFAULT_NAME] + + +def initialize(settings_engines=None, settings_outgoing=None): + from searx.engines import engines + from searx import settings + + global NETWORKS + + settings_engines = settings_engines or settings.get('engines') + settings_outgoing = settings_outgoing or settings.get('outgoing') + + # default parameters for AsyncHTTPTransport + # see https://github.com/encode/httpx/blob/e05a5372eb6172287458b37447c30f650047e1b8/httpx/_transports/default.py#L108-L121 # noqa + default_params = { + 'enable_http': False, + 'verify': True, + 'enable_http2': settings_outgoing.get('enable_http2', True), + # Magic number kept from previous code + 'max_connections': settings_outgoing.get('pool_connections', 100), + # Picked from constructor + 'max_keepalive_connections': settings_outgoing.get('pool_maxsize', 10), + # + 'keepalive_expiry': settings_outgoing.get('keepalive_expiry', 5.0), + 'local_addresses': settings_outgoing.get('source_ips'), + 'proxies': settings_outgoing.get('proxies'), + # default maximum redirect + # from https://github.com/psf/requests/blob/8c211a96cdbe9fe320d63d9e1ae15c5c07e179f8/requests/models.py#L55 + 'max_redirects': settings_outgoing.get('max_redirects', 30), + # + 'retries': settings_outgoing.get('retries', 0), + 'retry_on_http_error': None, + } + + def new_network(params): + nonlocal default_params + result = {} + result.update(default_params) + result.update(params) + return Network(**result) + + def iter_networks(): + nonlocal settings_engines + for engine_spec in settings_engines: + engine_name = engine_spec['name'] + engine = engines.get(engine_name) + if engine is None: + continue + network = getattr(engine, 'network', None) + yield engine_name, engine, network + + if NETWORKS: + done() + NETWORKS.clear() + NETWORKS[DEFAULT_NAME] = new_network({}) + NETWORKS['ipv4'] = new_network({'local_addresses': '0.0.0.0'}) + NETWORKS['ipv6'] = new_network({'local_addresses': '::'}) + + # define networks from outgoing.networks + for network_name, network in settings_outgoing.get('networks', {}).items(): + NETWORKS[network_name] = new_network(network) + + # define networks from engines.[i].network (except references) + for engine_name, engine, network in iter_networks(): + if network is None: + network = {} + for attribute_name, attribute_value in default_params.items(): + if hasattr(engine, attribute_name): + network[attribute_name] = getattr(engine, attribute_name) + else: + network[attribute_name] = attribute_value + NETWORKS[engine_name] = new_network(network) + elif isinstance(network, dict): + NETWORKS[engine_name] = new_network(network) + + # define networks from engines.[i].network (references) + for engine_name, engine, network in iter_networks(): + if isinstance(network, str): + NETWORKS[engine_name] = NETWORKS[network] + + +@atexit.register +def done(): + """Close all HTTP client + + Avoid a warning at exit + see https://github.com/encode/httpx/blob/1a6e254f72d9fd5694a1c10a28927e193ab4f76b/httpx/_client.py#L1785 + + Note: since Network.aclose has to be async, it is not possible to call this method on Network.__del__ + So Network.aclose is called here using atexit.register + """ + try: + if LOOP: + future = asyncio.run_coroutine_threadsafe(Network.aclose_all(), LOOP) + # wait 3 seconds to close the HTTP clients + future.result(3) + finally: + NETWORKS.clear() + + +NETWORKS[DEFAULT_NAME] = Network() diff --git a/searx/raise_for_httperror.py b/searx/network/raise_for_httperror.py index bd12df9a9..bd12df9a9 100644 --- a/searx/raise_for_httperror.py +++ b/searx/network/raise_for_httperror.py diff --git a/searx/poolrequests.py b/searx/poolrequests.py deleted file mode 100644 index ab327251b..000000000 --- a/searx/poolrequests.py +++ /dev/null @@ -1,235 +0,0 @@ -import sys -from time import time -from itertools import cycle -from threading import local - -import requests - -from searx import settings -from searx import logger -from searx.raise_for_httperror import raise_for_httperror - - -logger = logger.getChild('poolrequests') - - -try: - import ssl - if ssl.OPENSSL_VERSION_INFO[0:3] < (1, 0, 2): - # https://github.com/certifi/python-certifi#1024-bit-root-certificates - logger.critical('You are using an old openssl version({0}), please upgrade above 1.0.2!' - .format(ssl.OPENSSL_VERSION)) - sys.exit(1) -except ImportError: - ssl = None -if not getattr(ssl, "HAS_SNI", False): - try: - import OpenSSL # pylint: disable=unused-import - except ImportError: - logger.critical("ssl doesn't support SNI and the pyopenssl module is not installed.\n" - "Some HTTPS connections will fail") - sys.exit(1) - - -class HTTPAdapterWithConnParams(requests.adapters.HTTPAdapter): - - def __init__(self, pool_connections=requests.adapters.DEFAULT_POOLSIZE, - pool_maxsize=requests.adapters.DEFAULT_POOLSIZE, - max_retries=requests.adapters.DEFAULT_RETRIES, - pool_block=requests.adapters.DEFAULT_POOLBLOCK, - **conn_params): - if max_retries == requests.adapters.DEFAULT_RETRIES: - self.max_retries = requests.adapters.Retry(0, read=False) - else: - self.max_retries = requests.adapters.Retry.from_int(max_retries) - self.config = {} - self.proxy_manager = {} - - super().__init__() - - self._pool_connections = pool_connections - self._pool_maxsize = pool_maxsize - self._pool_block = pool_block - self._conn_params = conn_params - - self.init_poolmanager(pool_connections, pool_maxsize, block=pool_block, **conn_params) - - def __setstate__(self, state): - # Can't handle by adding 'proxy_manager' to self.__attrs__ because - # because self.poolmanager uses a lambda function, which isn't pickleable. - self.proxy_manager = {} - self.config = {} - - for attr, value in state.items(): - setattr(self, attr, value) - - self.init_poolmanager(self._pool_connections, self._pool_maxsize, - block=self._pool_block, **self._conn_params) - - -threadLocal = local() -connect = settings['outgoing'].get('pool_connections', 100) # Magic number kept from previous code -maxsize = settings['outgoing'].get('pool_maxsize', requests.adapters.DEFAULT_POOLSIZE) # Picked from constructor -if settings['outgoing'].get('source_ips'): - http_adapters = cycle(HTTPAdapterWithConnParams(pool_connections=connect, pool_maxsize=maxsize, - source_address=(source_ip, 0)) - for source_ip in settings['outgoing']['source_ips']) - https_adapters = cycle(HTTPAdapterWithConnParams(pool_connections=connect, pool_maxsize=maxsize, - source_address=(source_ip, 0)) - for source_ip in settings['outgoing']['source_ips']) -else: - http_adapters = cycle((HTTPAdapterWithConnParams(pool_connections=connect, pool_maxsize=maxsize), )) - https_adapters = cycle((HTTPAdapterWithConnParams(pool_connections=connect, pool_maxsize=maxsize), )) - - -class SessionSinglePool(requests.Session): - - def __init__(self): - super().__init__() - - # reuse the same adapters - self.adapters.clear() - - https_adapter = threadLocal.__dict__.setdefault('https_adapter', next(https_adapters)) - self.mount('https://', https_adapter) - if get_enable_http_protocol(): - http_adapter = threadLocal.__dict__.setdefault('http_adapter', next(http_adapters)) - self.mount('http://', http_adapter) - - def close(self): - """Call super, but clear adapters since there are managed globaly""" - self.adapters.clear() - super().close() - - -def set_timeout_for_thread(timeout, start_time=None): - threadLocal.timeout = timeout - threadLocal.start_time = start_time - - -def set_enable_http_protocol(enable_http): - threadLocal.enable_http = enable_http - - -def get_enable_http_protocol(): - try: - return threadLocal.enable_http - except AttributeError: - return False - - -def reset_time_for_thread(): - threadLocal.total_time = 0 - - -def get_time_for_thread(): - return threadLocal.total_time - - -def get_proxy_cycles(proxy_settings): - if not proxy_settings: - return None - # Backwards compatibility for single proxy in settings.yml - for protocol, proxy in proxy_settings.items(): - if isinstance(proxy, str): - proxy_settings[protocol] = [proxy] - - for protocol in proxy_settings: - proxy_settings[protocol] = cycle(proxy_settings[protocol]) - return proxy_settings - - -GLOBAL_PROXY_CYCLES = get_proxy_cycles(settings['outgoing'].get('proxies')) - - -def get_proxies(proxy_cycles): - if proxy_cycles: - return {protocol: next(proxy_cycle) for protocol, proxy_cycle in proxy_cycles.items()} - return None - - -def get_global_proxies(): - return get_proxies(GLOBAL_PROXY_CYCLES) - - -def request(method, url, **kwargs): - """same as requests/requests/api.py request(...)""" - time_before_request = time() - - # session start - session = SessionSinglePool() - - # proxies - if not kwargs.get('proxies'): - kwargs['proxies'] = get_global_proxies() - - # timeout - if 'timeout' in kwargs: - timeout = kwargs['timeout'] - else: - timeout = getattr(threadLocal, 'timeout', None) - if timeout is not None: - kwargs['timeout'] = timeout - - # raise_for_error - check_for_httperror = True - if 'raise_for_httperror' in kwargs: - check_for_httperror = kwargs['raise_for_httperror'] - del kwargs['raise_for_httperror'] - - # do request - response = session.request(method=method, url=url, **kwargs) - - time_after_request = time() - - # is there a timeout for this engine ? - if timeout is not None: - timeout_overhead = 0.2 # seconds - # start_time = when the user request started - start_time = getattr(threadLocal, 'start_time', time_before_request) - search_duration = time_after_request - start_time - if search_duration > timeout + timeout_overhead: - raise requests.exceptions.Timeout(response=response) - - # session end - session.close() - - if hasattr(threadLocal, 'total_time'): - threadLocal.total_time += time_after_request - time_before_request - - # raise an exception - if check_for_httperror: - raise_for_httperror(response) - - return response - - -def get(url, **kwargs): - kwargs.setdefault('allow_redirects', True) - return request('get', url, **kwargs) - - -def options(url, **kwargs): - kwargs.setdefault('allow_redirects', True) - return request('options', url, **kwargs) - - -def head(url, **kwargs): - kwargs.setdefault('allow_redirects', False) - return request('head', url, **kwargs) - - -def post(url, data=None, **kwargs): - return request('post', url, data=data, **kwargs) - - -def put(url, data=None, **kwargs): - return request('put', url, data=data, **kwargs) - - -def patch(url, data=None, **kwargs): - return request('patch', url, data=data, **kwargs) - - -def delete(url, **kwargs): - return request('delete', url, **kwargs) diff --git a/searx/search/checker/impl.py b/searx/search/checker/impl.py index ad45440ea..e54b3f68d 100644 --- a/searx/search/checker/impl.py +++ b/searx/search/checker/impl.py @@ -11,9 +11,9 @@ from urllib.parse import urlparse import re from langdetect import detect_langs from langdetect.lang_detect_exception import LangDetectException -import requests.exceptions +import httpx -from searx import poolrequests, logger +from searx import network, logger from searx.results import ResultContainer from searx.search.models import SearchQuery, EngineRef from searx.search.processors import EngineProcessor @@ -75,8 +75,8 @@ def _is_url_image(image_url): while retry > 0: a = time() try: - poolrequests.set_timeout_for_thread(10.0, time()) - r = poolrequests.get(image_url, timeout=10.0, allow_redirects=True, headers={ + network.set_timeout_for_thread(10.0, time()) + r = network.get(image_url, timeout=10.0, allow_redirects=True, headers={ 'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:84.0) Gecko/20100101 Firefox/84.0', 'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8', 'Accept-Language': 'en-US;q=0.5,en;q=0.3', @@ -90,10 +90,10 @@ def _is_url_image(image_url): if r.headers["content-type"].startswith('image/'): return True return False - except requests.exceptions.Timeout: + except httpx.TimeoutException: logger.error('Timeout for %s: %i', image_url, int(time() - a)) retry -= 1 - except requests.exceptions.RequestException: + except httpx.HTTPError: logger.exception('Exception for %s', image_url) return False diff --git a/searx/search/processors/online.py b/searx/search/processors/online.py index 1fc6444ad..66719ea9b 100644 --- a/searx/search/processors/online.py +++ b/searx/search/processors/online.py @@ -1,12 +1,12 @@ # SPDX-License-Identifier: AGPL-3.0-or-later -from urllib.parse import urlparse from time import time import threading +import asyncio -import requests.exceptions +import httpx -import searx.poolrequests as poolrequests +import searx.network from searx.engines import settings from searx import logger from searx.utils import gen_useragent @@ -64,10 +64,6 @@ class OnlineProcessor(EngineProcessor): auth=params['auth'] ) - # setting engine based proxies - if hasattr(self.engine, 'proxies'): - request_args['proxies'] = poolrequests.get_proxies(self.engine.proxies) - # max_redirects max_redirects = params.get('max_redirects') if max_redirects: @@ -85,9 +81,9 @@ class OnlineProcessor(EngineProcessor): # specific type of request (GET or POST) if params['method'] == 'GET': - req = poolrequests.get + req = searx.network.get else: - req = poolrequests.post + req = searx.network.post request_args['data'] = params['data'] @@ -99,8 +95,8 @@ class OnlineProcessor(EngineProcessor): # unexpected redirect : record an error # but the engine might still return valid results. status_code = str(response.status_code or '') - reason = response.reason or '' - hostname = str(urlparse(response.url or '').netloc) + reason = response.reason_phrase or '' + hostname = response.url.host record_error(self.engine_name, '{} redirects, maximum: {}'.format(len(response.history), soft_max_redirects), (status_code, reason, hostname)) @@ -128,14 +124,14 @@ class OnlineProcessor(EngineProcessor): def search(self, query, params, result_container, start_time, timeout_limit): # set timeout for all HTTP requests - poolrequests.set_timeout_for_thread(timeout_limit, start_time=start_time) + searx.network.set_timeout_for_thread(timeout_limit, start_time=start_time) # reset the HTTP total time - poolrequests.reset_time_for_thread() - # enable HTTP only if explicitly enabled - poolrequests.set_enable_http_protocol(self.engine.enable_http) + searx.network.reset_time_for_thread() + # set the network + searx.network.set_context_network_name(self.engine_name) # suppose everything will be alright - requests_exception = False + http_exception = False suspended_time = None try: @@ -149,7 +145,7 @@ class OnlineProcessor(EngineProcessor): # update engine time when there is no exception engine_time = time() - start_time - page_load_time = poolrequests.get_time_for_thread() + page_load_time = searx.network.get_time_for_thread() result_container.add_timing(self.engine_name, engine_time, page_load_time) with threading.RLock(): self.engine.stats['engine_time'] += engine_time @@ -162,27 +158,27 @@ class OnlineProcessor(EngineProcessor): # Timing engine_time = time() - start_time - page_load_time = poolrequests.get_time_for_thread() + page_load_time = searx.network.get_time_for_thread() result_container.add_timing(self.engine_name, engine_time, page_load_time) # Record the errors with threading.RLock(): self.engine.stats['errors'] += 1 - if (issubclass(e.__class__, requests.exceptions.Timeout)): + if (issubclass(e.__class__, (httpx.TimeoutException, asyncio.TimeoutError))): result_container.add_unresponsive_engine(self.engine_name, 'HTTP timeout') # requests timeout (connect or read) logger.error("engine {0} : HTTP requests timeout" "(search duration : {1} s, timeout: {2} s) : {3}" .format(self.engine_name, engine_time, timeout_limit, e.__class__.__name__)) - requests_exception = True - elif (issubclass(e.__class__, requests.exceptions.RequestException)): + http_exception = True + elif (issubclass(e.__class__, (httpx.HTTPError, httpx.StreamError))): result_container.add_unresponsive_engine(self.engine_name, 'HTTP error') # other requests exception logger.exception("engine {0} : requests exception" "(search duration : {1} s, timeout: {2} s) : {3}" .format(self.engine_name, engine_time, timeout_limit, e)) - requests_exception = True + http_exception = True elif (issubclass(e.__class__, SearxEngineCaptchaException)): result_container.add_unresponsive_engine(self.engine_name, 'CAPTCHA required') logger.exception('engine {0} : CAPTCHA'.format(self.engine_name)) @@ -206,7 +202,7 @@ class OnlineProcessor(EngineProcessor): # suspend the engine if there is an HTTP error # or suspended_time is defined with threading.RLock(): - if requests_exception or suspended_time: + if http_exception or suspended_time: # update continuous_errors / suspend_end_time self.engine.continuous_errors += 1 if suspended_time is None: diff --git a/searx/settings.yml b/searx/settings.yml index dde2e249c..1940739bc 100644 --- a/searx/settings.yml +++ b/searx/settings.yml @@ -67,19 +67,17 @@ ui: # key : !!binary "your_morty_proxy_key" outgoing: # communication with search engines - request_timeout : 2.0 # default timeout in seconds, can be override by engine + request_timeout : 3.0 # default timeout in seconds, can be override by engine # max_request_timeout: 10.0 # the maximum timeout in seconds useragent_suffix : "" # suffix of searx_useragent, could contain informations like an email address to the administrator - pool_connections : 100 # Number of different hosts - pool_maxsize : 10 # Number of simultaneous requests by host + pool_connections : 100 # The maximum number of concurrent connections that may be established. + pool_maxsize : 20 # Allow the connection pool to maintain keep-alive connections below this point. + enable_http2: True # See https://www.python-httpx.org/http2/ # uncomment below section if you want to use a proxy # see https://2.python-requests.org/en/latest/user/advanced/#proxies # SOCKS proxies are also supported: see https://2.python-requests.org/en/latest/user/advanced/#socks # proxies: -# http: -# - http://proxy1:8080 -# - http://proxy2:8080 -# https: +# all://: # - http://proxy1:8080 # - http://proxy2:8080 # using_tor_proxy : True @@ -89,6 +87,7 @@ outgoing: # communication with search engines # source_ips: # - 1.1.1.1 # - 1.1.1.2 +# - fe80::/126 # External plugin configuration # See http://searx.github.io/searx/dev/plugins.html for more details @@ -858,11 +857,13 @@ engines: engine : qwant shortcut : qwi categories : images + network: qwant - name : qwant news engine : qwant shortcut : qwn categories : news + network: qwant # - name: library # engine: recoll diff --git a/searx/testing.py b/searx/testing.py index ec253cb3d..51ca92bd0 100644 --- a/searx/testing.py +++ b/searx/testing.py @@ -10,8 +10,8 @@ import traceback from os.path import dirname, join, abspath, realpath -from unittest import TestCase from splinter import Browser +import aiounittest class SearxTestLayer: @@ -82,7 +82,7 @@ def run_robot_tests(tests): test(browser) -class SearxTestCase(TestCase): +class SearxTestCase(aiounittest.AsyncTestCase): """Base test case for non-robot tests.""" layer = SearxTestLayer diff --git a/searx/utils.py b/searx/utils.py index 3172ad8f3..55a386bd5 100644 --- a/searx/utils.py +++ b/searx/utils.py @@ -45,7 +45,7 @@ def searx_useragent(): """Return the searx User Agent""" return 'searx/{searx_version} {suffix}'.format( searx_version=VERSION_STRING, - suffix=settings['outgoing'].get('useragent_suffix', '')) + suffix=settings['outgoing'].get('useragent_suffix', '')).strip() def gen_useragent(os=None): diff --git a/searx/webapp.py b/searx/webapp.py index 072f140ca..8c59b8f3a 100755 --- a/searx/webapp.py +++ b/searx/webapp.py @@ -26,12 +26,26 @@ if __name__ == '__main__': from os.path import realpath, dirname sys.path.append(realpath(dirname(realpath(__file__)) + '/../')) +# set Unix thread name +try: + import setproctitle +except ImportError: + pass +else: + import threading + old_thread_init = threading.Thread.__init__ + + def new_thread_init(self, *args, **kwargs): + old_thread_init(self, *args, **kwargs) + setproctitle.setthreadtitle(self._name) + threading.Thread.__init__ = new_thread_init + import hashlib import hmac import json import os -import requests +import httpx from searx import logger logger = logger.getChild('webapp') @@ -79,7 +93,7 @@ from searx.plugins import plugins from searx.plugins.oa_doi_rewrite import get_doi_resolver from searx.preferences import Preferences, ValidationException, LANGUAGE_CODES from searx.answerers import answerers -from searx.poolrequests import get_global_proxies +from searx.network import stream as http_stream from searx.answerers import ask from searx.metrology.error_recorder import errors_per_engines @@ -890,50 +904,62 @@ def _is_selected_language_supported(engine, preferences): @app.route('/image_proxy', methods=['GET']) def image_proxy(): - url = request.args.get('url').encode() + url = request.args.get('url') if not url: return '', 400 - h = new_hmac(settings['server']['secret_key'], url) + h = new_hmac(settings['server']['secret_key'], url.encode()) if h != request.args.get('h'): return '', 400 - headers = dict_subset(request.headers, {'If-Modified-Since', 'If-None-Match'}) - headers['User-Agent'] = gen_useragent() - - resp = requests.get(url, - stream=True, - timeout=settings['outgoing']['request_timeout'], - headers=headers, - proxies=get_global_proxies()) + maximum_size = 5 * 1024 * 1024 - if resp.status_code == 304: - return '', resp.status_code - - if resp.status_code != 200: - logger.debug('image-proxy: wrong response code: {0}'.format(resp.status_code)) - if resp.status_code >= 400: + try: + headers = dict_subset(request.headers, {'If-Modified-Since', 'If-None-Match'}) + headers['User-Agent'] = gen_useragent() + stream = http_stream( + method='GET', + url=url, + headers=headers, + timeout=settings['outgoing']['request_timeout'], + allow_redirects=True, + max_redirects=20) + + resp = next(stream) + content_length = resp.headers.get('Content-Length') + if content_length and content_length.isdigit() and int(content_length) > maximum_size: + return 'Max size', 400 + + if resp.status_code == 304: return '', resp.status_code - return '', 400 - if not resp.headers.get('content-type', '').startswith('image/'): - logger.debug('image-proxy: wrong content-type: {0}'.format(resp.headers.get('content-type'))) - return '', 400 + if resp.status_code != 200: + logger.debug('image-proxy: wrong response code: {0}'.format(resp.status_code)) + if resp.status_code >= 400: + return '', resp.status_code + return '', 400 - img = b'' - chunk_counter = 0 + if not resp.headers.get('content-type', '').startswith('image/'): + logger.debug('image-proxy: wrong content-type: {0}'.format(resp.headers.get('content-type'))) + return '', 400 - for chunk in resp.iter_content(1024 * 1024): - chunk_counter += 1 - if chunk_counter > 5: - return '', 502 # Bad gateway - file is too big (>5M) - img += chunk + headers = dict_subset(resp.headers, {'Content-Length', 'Length', 'Date', 'Last-Modified', 'Expires', 'Etag'}) - headers = dict_subset(resp.headers, {'Content-Length', 'Length', 'Date', 'Last-Modified', 'Expires', 'Etag'}) + total_length = 0 - return Response(img, mimetype=resp.headers['content-type'], headers=headers) + def forward_chunk(): + nonlocal total_length + for chunk in stream: + total_length += len(chunk) + if total_length > maximum_size: + break + yield chunk + + return Response(forward_chunk(), mimetype=resp.headers['Content-Type'], headers=headers) + except httpx.HTTPError: + return '', 400 @app.route('/stats', methods=['GET']) |