mirror of
https://github.com/searxng/searxng.git
synced 2025-04-23 20:04:07 +00:00
[fix] engine: re-implement mullvad leta integration
This commit is contained in:
Grant Lanham
parent
391bb1268d
commit
5a061d443a
2 changed files with 69 additions and 134 deletions
|
|
@ -1,34 +1,14 @@
|
|||
# SPDX-License-Identifier: AGPL-3.0-or-later
|
||||
|
||||
"""This is the implementation of the Mullvad-Leta meta-search engine.
|
||||
|
||||
This engine **REQUIRES** that searxng operate within a Mullvad VPN
|
||||
|
||||
If using docker, consider using gluetun for easily connecting to the Mullvad
|
||||
|
||||
- https://github.com/qdm12/gluetun
|
||||
|
||||
Otherwise, follow instructions provided by Mullvad for enabling the VPN on Linux
|
||||
|
||||
- https://mullvad.net/en/help/install-mullvad-app-linux
|
||||
|
||||
.. hint::
|
||||
|
||||
The :py:obj:`EngineTraits` is empty by default. Maintainers have to run
|
||||
``make data.traits`` (in the Mullvad VPN / :py:obj:`fetch_traits`) and rebase
|
||||
the modified JSON file ``searx/data/engine_traits.json`` on every single
|
||||
update of SearXNG!
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import TYPE_CHECKING
|
||||
from typing import TYPE_CHECKING, TypedDict
|
||||
from urllib.parse import urlencode
|
||||
from httpx import Response
|
||||
from lxml import html
|
||||
from searx.enginelib.traits import EngineTraits
|
||||
from searx.locales import region_tag, get_official_locales
|
||||
from searx.utils import eval_xpath, extract_text, eval_xpath_list
|
||||
from searx.exceptions import SearxEngineResponseException
|
||||
|
||||
if TYPE_CHECKING:
|
||||
import logging
|
||||
|
|
@ -37,15 +17,13 @@ if TYPE_CHECKING:
|
|||
|
||||
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"
|
||||
base_url = "https://leta.mullvad.net"
|
||||
|
||||
# about
|
||||
about = {
|
||||
"website": search_url,
|
||||
"website": base_url,
|
||||
"wikidata_id": 'Q47008412', # the Mullvad id - not leta, but related
|
||||
"official_api_documentation": 'https://leta.mullvad.net/faq',
|
||||
"use_official_api": False,
|
||||
|
|
@ -56,13 +34,13 @@ about = {
|
|||
# engine dependent config
|
||||
categories = ['general', 'web']
|
||||
paging = True
|
||||
max_page = 50
|
||||
max_page = 10
|
||||
time_range_support = True
|
||||
time_range_dict = {
|
||||
"day": "d1",
|
||||
"week": "w1",
|
||||
"month": "m1",
|
||||
"year": "y1",
|
||||
"day": "d",
|
||||
"week": "w",
|
||||
"month": "m",
|
||||
"year": "y",
|
||||
}
|
||||
|
||||
available_leta_engines = [
|
||||
|
|
@ -71,19 +49,32 @@ available_leta_engines = [
|
|||
]
|
||||
|
||||
|
||||
def is_vpn_connected(dom: html.HtmlElement) -> bool:
|
||||
"""Returns true if the VPN is connected, False otherwise"""
|
||||
connected_text = extract_text(eval_xpath(dom, '//main/div/p[1]'))
|
||||
return connected_text != 'You are not connected to Mullvad VPN.'
|
||||
class DataNodeQueryMetaDataIndices(TypedDict):
|
||||
"""Indices into query metadata"""
|
||||
|
||||
success: int
|
||||
q: int # pylint: disable=invalid-name
|
||||
country: int
|
||||
language: int
|
||||
lastUpdated: int
|
||||
engine: int
|
||||
items: int
|
||||
infobox: int
|
||||
news: int
|
||||
timestamp: int
|
||||
altered: int
|
||||
page: int
|
||||
next: int # if -1, there no more results are available
|
||||
previous: int
|
||||
|
||||
|
||||
def assign_headers(headers: dict) -> dict:
|
||||
"""Assigns the headers to make a request to Mullvad Leta"""
|
||||
headers['Accept'] = "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,*/*;q=0.8"
|
||||
headers['Content-Type'] = "application/x-www-form-urlencoded"
|
||||
headers['Host'] = "leta.mullvad.net"
|
||||
headers['Origin'] = "https://leta.mullvad.net"
|
||||
return headers
|
||||
class DataNodeResultIndices(TypedDict):
|
||||
"""Indices into query resultsdata"""
|
||||
|
||||
link: int
|
||||
snippet: int
|
||||
title: int
|
||||
favicon: int
|
||||
|
||||
|
||||
def request(query: str, params: dict):
|
||||
|
|
@ -99,109 +90,55 @@ def request(query: str, params: dict):
|
|||
result_engine,
|
||||
)
|
||||
|
||||
params['url'] = search_url
|
||||
params['method'] = 'POST'
|
||||
params['data'] = {
|
||||
"q": query,
|
||||
"gl": country if country is str else '',
|
||||
params['method'] = 'GET'
|
||||
|
||||
args = {
|
||||
'q': query,
|
||||
'engine': result_engine,
|
||||
'x-sveltekit-invalidated': "001", # hardcoded from all requests seen
|
||||
}
|
||||
# pylint: disable=undefined-variable
|
||||
if use_cache:
|
||||
params['data']['oc'] = "on"
|
||||
# pylint: enable=undefined-variable
|
||||
|
||||
if isinstance(country, str):
|
||||
args['country'] = country
|
||||
if params['time_range'] in time_range_dict:
|
||||
params['dateRestrict'] = time_range_dict[params['time_range']]
|
||||
else:
|
||||
params['dateRestrict'] = ''
|
||||
|
||||
args['lastUpdated'] = time_range_dict[params['time_range']]
|
||||
if params['pageno'] > 1:
|
||||
# Page 1 is n/a, Page 2 is 11, page 3 is 21, ...
|
||||
params['data']['start'] = ''.join([str(params['pageno'] - 1), "1"])
|
||||
args['page'] = params['pageno']
|
||||
|
||||
if params['headers'] is None:
|
||||
params['headers'] = {}
|
||||
params['url'] = f"{base_url}/search/__data.json?{urlencode(args)}"
|
||||
|
||||
assign_headers(params['headers'])
|
||||
return params
|
||||
|
||||
|
||||
def extract_result(dom_result: list[html.HtmlElement]):
|
||||
# Infoboxes sometimes appear in the beginning and will have a length of 0
|
||||
if len(dom_result) == 3:
|
||||
[a_elem, h3_elem, p_elem] = dom_result
|
||||
elif len(dom_result) == 4:
|
||||
[_, a_elem, h3_elem, p_elem] = dom_result
|
||||
else:
|
||||
return None
|
||||
|
||||
return {
|
||||
'url': extract_text(a_elem.text),
|
||||
'title': extract_text(h3_elem),
|
||||
'content': extract_text(p_elem),
|
||||
}
|
||||
|
||||
|
||||
def extract_results(search_results: html.HtmlElement):
|
||||
for search_result in search_results:
|
||||
dom_result = eval_xpath_list(search_result, 'div/div/*')
|
||||
result = extract_result(dom_result)
|
||||
if result is not None:
|
||||
yield result
|
||||
|
||||
|
||||
def response(resp: Response):
|
||||
"""Checks if connected to Mullvad VPN, then extracts the search results from
|
||||
the DOM resp: requests response object"""
|
||||
json_response = resp.json()
|
||||
|
||||
dom = html.fromstring(resp.text)
|
||||
if not is_vpn_connected(dom):
|
||||
raise SearxEngineResponseException('Not connected to Mullvad VPN')
|
||||
search_results = eval_xpath(dom.body, '//main/div[2]/div')
|
||||
return list(extract_results(search_results))
|
||||
nodes = json_response["nodes"]
|
||||
# 0: is None
|
||||
# 1: has "connected=True", not useful
|
||||
# 2: query results within "data"
|
||||
|
||||
data_nodes = nodes[2]["data"]
|
||||
# Instead of nested object structure, all objects are flattened into a
|
||||
# list. Rather, the first object in data_node provides indices into the
|
||||
# "data_nodes" to access each searchresult (which is an object of more
|
||||
# indices)
|
||||
#
|
||||
# Read the relative TypedDict definitions for details
|
||||
|
||||
def fetch_traits(engine_traits: EngineTraits):
|
||||
"""Fetch languages and regions from Mullvad-Leta
|
||||
query_meta_data: DataNodeQueryMetaDataIndices = data_nodes[0]
|
||||
|
||||
.. warning::
|
||||
query_items_indices = query_meta_data['items']
|
||||
|
||||
Fetching the engine traits also requires a Mullvad VPN connection. If
|
||||
not connected, then an error message will print and no traits will be
|
||||
updated.
|
||||
"""
|
||||
# pylint: disable=import-outside-toplevel
|
||||
# see https://github.com/searxng/searxng/issues/762
|
||||
from searx.network import post as http_post
|
||||
results = []
|
||||
for idx in data_nodes[query_items_indices]:
|
||||
query_item_indices: DataNodeResultIndices = data_nodes[idx]
|
||||
results.append(
|
||||
{
|
||||
'url': data_nodes[query_item_indices['link']],
|
||||
'title': data_nodes[query_item_indices['title']],
|
||||
'content': data_nodes[query_item_indices['snippet']],
|
||||
'favicon_url': base_url + "/" + data_nodes[query_item_indices['favicon']],
|
||||
}
|
||||
)
|
||||
|
||||
# pylint: enable=import-outside-toplevel
|
||||
resp = http_post(search_url, headers=assign_headers({}))
|
||||
if not isinstance(resp, Response):
|
||||
print("ERROR: failed to get response from mullvad-leta. Are you connected to the VPN?")
|
||||
return
|
||||
if not resp.ok:
|
||||
print("ERROR: response from mullvad-leta is not OK. Are you connected to the VPN?")
|
||||
return
|
||||
dom = html.fromstring(resp.text)
|
||||
if not is_vpn_connected(dom):
|
||||
print('ERROR: Not connected to Mullvad VPN')
|
||||
return
|
||||
# supported region codes
|
||||
options = eval_xpath_list(dom.body, '//main/div/form/div[2]/div/select[1]/option')
|
||||
if options is None or len(options) <= 0:
|
||||
print('ERROR: could not find any results. Are you connected to the VPN?')
|
||||
for x in options:
|
||||
eng_country = x.get("value")
|
||||
|
||||
sxng_locales = get_official_locales(eng_country, engine_traits.languages.keys(), regional=True)
|
||||
|
||||
if not sxng_locales:
|
||||
print(
|
||||
"ERROR: can't map from Mullvad-Leta country %s (%s) to a babel region."
|
||||
% (x.get('data-name'), eng_country)
|
||||
)
|
||||
continue
|
||||
|
||||
for sxng_locale in sxng_locales:
|
||||
engine_traits.regions[region_tag(sxng_locale)] = eng_country
|
||||
return results
|
||||
|
|
|
|||
|
|
@ -1398,11 +1398,9 @@ engines:
|
|||
require_api_key: false
|
||||
results: JSON
|
||||
|
||||
# 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]
|
||||
# shortcut: ml
|
||||
|
|
|
|||
Loading…
Reference in a new issue