From ef103ba80a8494d01c6dc8ce8e408957d4424b0b Mon Sep 17 00:00:00 2001 From: Grant Lanham Date: Sat, 22 Jun 2024 14:02:16 -0400 Subject: [PATCH] Implement google/brave switch in Mullvad Leta cleanup Import annontations --- searx/engines/mullvad_leta.py | 34 +++++++++++++++++++++++++++++++--- searx/settings.yml | 1 + 2 files changed, 32 insertions(+), 3 deletions(-) diff --git a/searx/engines/mullvad_leta.py b/searx/engines/mullvad_leta.py index 6e46163e3..a1e59d93b 100644 --- a/searx/engines/mullvad_leta.py +++ b/searx/engines/mullvad_leta.py @@ -20,6 +20,8 @@ Otherwise, follow instructions provided by Mullvad for enabling the VPN on Linux update of SearXNG! """ +from __future__ import annotations + from typing import TYPE_CHECKING from httpx import Response from lxml import html @@ -37,6 +39,8 @@ traits: EngineTraits use_cache: bool = True # non-cache use only has 100 searches per day! +leta_engine: str = 'google' + search_url = "https://leta.mullvad.net" # about @@ -61,6 +65,11 @@ time_range_dict = { "year": "y1", } +available_leta_engines = [ + 'google', # first will be default if provided engine is invalid + 'brave', +] + def is_vpn_connected(dom: html.HtmlElement) -> bool: """Returns true if the VPN is connected, False otherwise""" @@ -80,11 +89,22 @@ def assign_headers(headers: dict) -> dict: def request(query: str, params: dict): country = traits.get_region(params.get('searxng_locale', 'all'), traits.all_locale) # type: ignore + result_engine = leta_engine + if leta_engine not in available_leta_engines: + result_engine = available_leta_engines[0] + logger.warning( + 'Configured engine "%s" not one of the available engines %s, defaulting to "%s"', + leta_engine, + available_leta_engines, + result_engine, + ) + params['url'] = search_url params['method'] = 'POST' params['data'] = { "q": query, "gl": country if country is str else '', + 'engine': result_engine, } # pylint: disable=undefined-variable if use_cache: @@ -107,8 +127,8 @@ def request(query: str, params: dict): return params -def extract_result(dom_result: html.HtmlElement): - [a_elem, h3_elem, p_elem] = eval_xpath_list(dom_result, 'div/div/*') +def extract_result(dom_result: list[html.HtmlElement]): + [a_elem, h3_elem, p_elem] = dom_result return { 'url': extract_text(a_elem.text), 'title': extract_text(h3_elem), @@ -116,6 +136,14 @@ def extract_result(dom_result: html.HtmlElement): } +def extract_results(search_results: html.HtmlElement): + for search_result in search_results: + dom_result = eval_xpath_list(search_result, 'div/div/*') + # sometimes an info box pops up, will need to filter that out + if len(dom_result) == 3: + yield extract_result(dom_result) + + def response(resp: Response): """Checks if connected to Mullvad VPN, then extracts the search results from the DOM resp: requests response object""" @@ -124,7 +152,7 @@ def response(resp: Response): if not is_vpn_connected(dom): raise SearxEngineResponseException('Not connected to Mullvad VPN') search_results = eval_xpath(dom.body, '//main/div[2]/div') - return [extract_result(sr) for sr in search_results] + return list(extract_results(search_results)) def fetch_traits(engine_traits: EngineTraits): diff --git a/searx/settings.yml b/searx/settings.yml index 748e5ac0e..0b804c542 100644 --- a/searx/settings.yml +++ b/searx/settings.yml @@ -1234,6 +1234,7 @@ engines: # read https://docs.searxng.org/dev/engines/online/mullvad_leta.html # - name: mullvadleta # engine: mullvad_leta + # leta_engine: google # choose one of the following: google, brave # use_cache: true # Only 100 non-cache searches per day, suggested only for private instances # search_url: https://leta.mullvad.net # categories: [general, web]