1
0
mirror of https://github.com/searxng/searxng.git synced 2024-11-05 04:40:11 +01:00
searxng/searx/webapp.py
Alexandre Flament 2babf59adc [fix] pyright repported errors
The errors make pyright usage useless since a new error won't be seen [1].

[1] https://github.com/searxng/searxng/pull/1569

```
  searx/compat.py:11:27 - error: Expression of type "Type[cached_property[_T@cached_property]]" cannot be assigned to declared type "Type[cached_property]"
    "Type[cached_property[_T@cached_property]]" is incompatible with "Type[cached_property]"
    Type "Type[cached_property[_T@cached_property]]" cannot be assigned to type "Type[cached_property]" (reportGeneralTypeIssues)
  searx/utils.py:69:36 - error: Expression of type "None" cannot be assigned to parameter of type "str"
    Type "None" cannot be assigned to type "str" (reportGeneralTypeIssues)
  searx/utils.py:573:85 - error: Expression of type "None" cannot be assigned to parameter of type "int"
    Type "None" cannot be assigned to type "int" (reportGeneralTypeIssues)
  searx/webapp.py:1306:22 - error: Argument of type "str" cannot be assigned to parameter "__a" of type "BytesPath" in function "join"
    Type "str" cannot be assigned to type "BytesPath"
      "str" is incompatible with "bytes"
      "str" is incompatible with protocol "PathLike[bytes]"
        "__fspath__" is not present (reportGeneralTypeIssues)
  searx/webapp.py:1306:68 - error: Argument of type "Literal['themes']" cannot be assigned to parameter "paths" of type "BytesPath" in function "join"
    Type "Literal['themes']" cannot be assigned to type "BytesPath"
      "Literal['themes']" is incompatible with "bytes"
      "Literal['themes']" is incompatible with protocol "PathLike[bytes]"
        "__fspath__" is not present (reportGeneralTypeIssues)
  searx/webapp.py:1306:78 - error: Argument of type "str | Any | None" cannot be assigned to parameter "paths" of type "BytesPath" in function "join"
    Type "str | Any | None" cannot be assigned to type "BytesPath"
      Type "str" cannot be assigned to type "BytesPath"
        "str" is incompatible with "bytes"
        "str" is incompatible with protocol "PathLike[bytes]"
          "__fspath__" is not present (reportGeneralTypeIssues)
  searx/webapp.py:1306:85 - error: Argument of type "Literal['img']" cannot be assigned to parameter "paths" of type "BytesPath" in function "join"
    Type "Literal['img']" cannot be assigned to type "BytesPath"
      "Literal['img']" is incompatible with "bytes"
      "Literal['img']" is incompatible with protocol "PathLike[bytes]"
        "__fspath__" is not present (reportGeneralTypeIssues)
  searx/engines/mongodb.py:8:6 - warning: Import "pymongo" could not be resolved (reportMissingImports)
  searx/engines/mysql_server.py:9:8 - warning: Import "mysql.connector" could not be resolved (reportMissingImports)
  searx/engines/postgresql.py:9:8 - warning: Import "psycopg2" could not be resolved from source (reportMissingModuleSource)
  searx/engines/xpath.py:187:28 - warning: "categories" is not defined (reportUndefinedVariable)
  searx/search/__init__.py:184:82 - warning: "flask" is not defined (reportUndefinedVariable)
  searx/search/checker/background.py:19:26 - error: Type of "schedule" is partially unknown
    Type of "schedule" is "(delay: Any, func: Any, *args: Any) -> Literal[True]" (reportUnknownVariableType)
  searx/shared/__init__.py:8:12 - warning: Import "uwsgi" could not be resolved (reportMissingImports)
  searx/shared/shared_uwsgi.py:5:8 - warning: Import "uwsgi" could not be resolved (reportMissingImports)
```
2022-07-30 18:04:44 +02:00

1417 lines
51 KiB
Python
Executable File
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

#!/usr/bin/env python
# SPDX-License-Identifier: AGPL-3.0-or-later
# lint: pylint
# pyright: basic
"""WebbApp
"""
import hashlib
import hmac
import json
import os
import sys
import base64
from datetime import datetime, timedelta
from timeit import default_timer
from html import escape
from io import StringIO
import typing
from typing import List, Dict, Iterable
import urllib
import urllib.parse
from urllib.parse import urlencode, unquote
import httpx
from pygments import highlight
from pygments.lexers import get_lexer_by_name
from pygments.formatters import HtmlFormatter # pylint: disable=no-name-in-module
import flask
from flask import (
Flask,
render_template,
url_for,
make_response,
redirect,
send_from_directory,
)
from flask.wrappers import Response
from flask.json import jsonify
from flask_babel import (
Babel,
gettext,
format_date,
format_decimal,
)
from searx import (
logger,
get_setting,
settings,
searx_debug,
)
from searx import infopage
from searx.data import ENGINE_DESCRIPTIONS
from searx.results import Timing, UnresponsiveEngine
from searx.settings_defaults import OUTPUT_FORMATS
from searx.settings_loader import get_default_settings_path
from searx.exceptions import SearxParameterException
from searx.engines import (
OTHER_CATEGORY,
categories,
engines,
engine_shortcuts,
)
from searx.webutils import (
UnicodeWriter,
highlight_content,
get_static_files,
get_result_templates,
get_themes,
prettify_url,
new_hmac,
is_hmac_of,
is_flask_run_cmdline,
group_engines_in_tab,
)
from searx.webadapter import (
get_search_query_from_webapp,
get_selected_categories,
)
from searx.utils import (
html_to_text,
gen_useragent,
dict_subset,
match_language,
)
from searx.version import VERSION_STRING, GIT_URL, GIT_BRANCH
from searx.query import RawTextQuery
from searx.plugins import Plugin, plugins, initialize as plugin_initialize
from searx.plugins.oa_doi_rewrite import get_doi_resolver
from searx.preferences import (
Preferences,
ValidationException,
)
from searx.answerers import (
answerers,
ask,
)
from searx.metrics import (
get_engines_stats,
get_engine_errors,
get_reliabilities,
histogram,
counter,
)
from searx.flaskfix import patch_application
from searx.locales import (
LOCALE_NAMES,
RTL_LOCALES,
localeselector,
locales_initialize,
)
# renaming names from searx imports ...
from searx.autocomplete import search_autocomplete, backends as autocomplete_backends
from searx.languages import language_codes as languages
from searx.search import SearchWithPlugins, initialize as search_initialize
from searx.network import stream as http_stream, set_context_network_name
from searx.search.checker import get_result as checker_get_result
logger = logger.getChild('webapp')
# check secret_key
if not searx_debug and settings['server']['secret_key'] == 'ultrasecretkey':
logger.error('server.secret_key is not changed. Please use something else instead of ultrasecretkey.')
sys.exit(1)
# about static
logger.debug('static directory is %s', settings['ui']['static_path'])
static_files = get_static_files(settings['ui']['static_path'])
# about templates
logger.debug('templates directory is %s', settings['ui']['templates_path'])
default_theme = settings['ui']['default_theme']
templates_path = settings['ui']['templates_path']
themes = get_themes(templates_path)
result_templates = get_result_templates(templates_path)
STATS_SORT_PARAMETERS = {
'name': (False, 'name', ''),
'score': (True, 'score', 0),
'result_count': (True, 'result_count', 0),
'time': (False, 'total', 0),
'reliability': (False, 'reliability', 100),
}
# Flask app
app = Flask(__name__, static_folder=settings['ui']['static_path'], template_folder=templates_path)
app.jinja_env.trim_blocks = True
app.jinja_env.lstrip_blocks = True
app.jinja_env.add_extension('jinja2.ext.loopcontrols') # pylint: disable=no-member
app.jinja_env.filters['group_engines_in_tab'] = group_engines_in_tab # pylint: disable=no-member
app.secret_key = settings['server']['secret_key']
babel = Babel(app)
timeout_text = gettext('timeout')
parsing_error_text = gettext('parsing error')
http_protocol_error_text = gettext('HTTP protocol error')
network_error_text = gettext('network error')
exception_classname_to_text = {
None: gettext('unexpected crash'),
'timeout': timeout_text,
'asyncio.TimeoutError': timeout_text,
'httpx.TimeoutException': timeout_text,
'httpx.ConnectTimeout': timeout_text,
'httpx.ReadTimeout': timeout_text,
'httpx.WriteTimeout': timeout_text,
'httpx.HTTPStatusError': gettext('HTTP error'),
'httpx.ConnectError': gettext("HTTP connection error"),
'httpx.RemoteProtocolError': http_protocol_error_text,
'httpx.LocalProtocolError': http_protocol_error_text,
'httpx.ProtocolError': http_protocol_error_text,
'httpx.ReadError': network_error_text,
'httpx.WriteError': network_error_text,
'httpx.ProxyError': gettext("proxy error"),
'searx.exceptions.SearxEngineCaptchaException': gettext("CAPTCHA"),
'searx.exceptions.SearxEngineTooManyRequestsException': gettext("too many requests"),
'searx.exceptions.SearxEngineAccessDeniedException': gettext("access denied"),
'searx.exceptions.SearxEngineAPIException': gettext("server API error"),
'searx.exceptions.SearxEngineXPathException': parsing_error_text,
'KeyError': parsing_error_text,
'json.decoder.JSONDecodeError': parsing_error_text,
'lxml.etree.ParserError': parsing_error_text,
}
class ExtendedRequest(flask.Request):
"""This class is never initialized and only used for type checking."""
preferences: Preferences
errors: List[str]
user_plugins: List[Plugin]
form: Dict[str, str]
start_time: float
render_time: float
timings: List[Timing]
request = typing.cast(ExtendedRequest, flask.request)
@babel.localeselector
def get_locale():
locale = localeselector()
logger.debug("%s uses locale `%s`", urllib.parse.quote(request.url), locale)
return locale
def _get_browser_language(req, lang_list):
for lang in req.headers.get("Accept-Language", "en").split(","):
if ';' in lang:
lang = lang.split(';')[0]
if '-' in lang:
lang_parts = lang.split('-')
lang = "{}-{}".format(lang_parts[0], lang_parts[-1].upper())
locale = match_language(lang, lang_list, fallback=None)
if locale is not None:
return locale
return 'en'
def _get_locale_rfc5646(locale):
"""Get locale name for <html lang="...">
Chrom* browsers don't detect the language when there is a subtag (ie a territory).
For example "zh-TW" is detected but not "zh-Hant-TW".
This function returns a locale without the subtag.
"""
parts = locale.split('-')
return parts[0].lower() + '-' + parts[-1].upper()
# code-highlighter
@app.template_filter('code_highlighter')
def code_highlighter(codelines, language=None):
if not language:
language = 'text'
try:
# find lexer by programing language
lexer = get_lexer_by_name(language, stripall=True)
except Exception as e: # pylint: disable=broad-except
logger.exception(e, exc_info=True)
# if lexer is not found, using default one
lexer = get_lexer_by_name('text', stripall=True)
html_code = ''
tmp_code = ''
last_line = None
line_code_start = None
# parse lines
for line, code in codelines:
if not last_line:
line_code_start = line
# new codeblock is detected
if last_line is not None and last_line + 1 != line:
# highlight last codepart
formatter = HtmlFormatter(linenos='inline', linenostart=line_code_start, cssclass="code-highlight")
html_code = html_code + highlight(tmp_code, lexer, formatter)
# reset conditions for next codepart
tmp_code = ''
line_code_start = line
# add codepart
tmp_code += code + '\n'
# update line
last_line = line
# highlight last codepart
formatter = HtmlFormatter(linenos='inline', linenostart=line_code_start, cssclass="code-highlight")
html_code = html_code + highlight(tmp_code, lexer, formatter)
return html_code
def get_result_template(theme_name: str, template_name: str):
themed_path = theme_name + '/result_templates/' + template_name
if themed_path in result_templates:
return themed_path
return 'result_templates/' + template_name
def custom_url_for(endpoint: str, **values):
suffix = ""
if endpoint == 'static' and values.get('filename'):
file_hash = static_files.get(values['filename'])
if not file_hash:
# try file in the current theme
theme_name = request.preferences.get_value('theme')
filename_with_theme = "themes/{}/{}".format(theme_name, values['filename'])
file_hash = static_files.get(filename_with_theme)
if file_hash:
values['filename'] = filename_with_theme
if get_setting('ui.static_use_hash') and file_hash:
suffix = "?" + file_hash
if endpoint == 'info' and 'locale' not in values:
locale = request.preferences.get_value('locale')
if _INFO_PAGES.get_page(values['pagename'], locale) is None:
locale = _INFO_PAGES.locale_default
values['locale'] = locale
return url_for(endpoint, **values) + suffix
def proxify(url: str):
if url.startswith('//'):
url = 'https:' + url
if not settings.get('result_proxy'):
return url
url_params = dict(mortyurl=url)
if settings['result_proxy'].get('key'):
url_params['mortyhash'] = hmac.new(settings['result_proxy']['key'], url.encode(), hashlib.sha256).hexdigest()
return '{0}?{1}'.format(settings['result_proxy']['url'], urlencode(url_params))
def image_proxify(url: str):
if url.startswith('//'):
url = 'https:' + url
if not request.preferences.get_value('image_proxy'):
return url
if url.startswith('data:image/'):
# 50 is an arbitrary number to get only the beginning of the image.
partial_base64 = url[len('data:image/') : 50].split(';')
if (
len(partial_base64) == 2
and partial_base64[0] in ['gif', 'png', 'jpeg', 'pjpeg', 'webp', 'tiff', 'bmp']
and partial_base64[1].startswith('base64,')
):
return url
return None
if settings.get('result_proxy'):
return proxify(url)
h = new_hmac(settings['server']['secret_key'], url.encode())
return '{0}?{1}'.format(url_for('image_proxy'), urlencode(dict(url=url.encode(), h=h)))
def get_translations():
return {
# when there is autocompletion
'no_item_found': gettext('No item found'),
# /preferences: the source of the engine description (wikipedata, wikidata, website)
'Source': gettext('Source'),
# infinite scroll
'error_loading_next_page': gettext('Error loading the next page'),
}
def _get_enable_categories(all_categories: Iterable[str]):
disabled_engines = request.preferences.engines.get_disabled()
enabled_categories = set(
# pylint: disable=consider-using-dict-items
category
for engine_name in engines
for category in engines[engine_name].categories
if (engine_name, category) not in disabled_engines
)
return [x for x in all_categories if x in enabled_categories]
def get_pretty_url(parsed_url: urllib.parse.ParseResult):
path = parsed_url.path
path = path[:-1] if len(path) > 0 and path[-1] == '/' else path
path = unquote(path.replace("/", " "))
return [parsed_url.scheme + "://" + parsed_url.netloc, path]
def get_client_settings():
req_pref = request.preferences
return {
'autocomplete_provider': req_pref.get_value('autocomplete'),
'autocomplete_min': get_setting('search.autocomplete_min'),
'http_method': req_pref.get_value('method'),
'infinite_scroll': req_pref.get_value('infinite_scroll'),
'translations': get_translations(),
'search_on_category_select': req_pref.plugins.choices['searx.plugins.search_on_category_select'],
'hotkeys': req_pref.plugins.choices['searx.plugins.vim_hotkeys'],
'theme_static_path': custom_url_for('static', filename='themes/simple'),
}
def render(template_name: str, **kwargs):
kwargs['client_settings'] = str(
base64.b64encode(
bytes(
json.dumps(get_client_settings()),
encoding='utf-8',
)
),
encoding='utf-8',
)
# values from the HTTP requests
kwargs['endpoint'] = 'results' if 'q' in kwargs else request.endpoint
kwargs['cookies'] = request.cookies
kwargs['errors'] = request.errors
# values from the preferences
kwargs['preferences'] = request.preferences
kwargs['autocomplete'] = request.preferences.get_value('autocomplete')
kwargs['infinite_scroll'] = request.preferences.get_value('infinite_scroll')
kwargs['results_on_new_tab'] = request.preferences.get_value('results_on_new_tab')
kwargs['advanced_search'] = request.preferences.get_value('advanced_search')
kwargs['query_in_title'] = request.preferences.get_value('query_in_title')
kwargs['safesearch'] = str(request.preferences.get_value('safesearch'))
kwargs['theme'] = request.preferences.get_value('theme')
kwargs['method'] = request.preferences.get_value('method')
kwargs['categories_as_tabs'] = list(settings['categories_as_tabs'].keys())
kwargs['categories'] = _get_enable_categories(categories.keys())
kwargs['OTHER_CATEGORY'] = OTHER_CATEGORY
# i18n
kwargs['language_codes'] = [l for l in languages if l[0] in settings['search']['languages']]
locale = request.preferences.get_value('locale')
kwargs['locale_rfc5646'] = _get_locale_rfc5646(locale)
if locale in RTL_LOCALES and 'rtl' not in kwargs:
kwargs['rtl'] = True
if 'current_language' not in kwargs:
kwargs['current_language'] = match_language(
request.preferences.get_value('language'), settings['search']['languages']
)
# values from settings
kwargs['search_formats'] = [x for x in settings['search']['formats'] if x != 'html']
kwargs['instance_name'] = get_setting('general.instance_name')
kwargs['searx_version'] = VERSION_STRING
kwargs['searx_git_url'] = GIT_URL
kwargs['get_setting'] = get_setting
kwargs['get_pretty_url'] = get_pretty_url
# values from settings: donation_url
donation_url = get_setting('general.donation_url')
if donation_url is True:
donation_url = custom_url_for('info', pagename='donate')
kwargs['donation_url'] = donation_url
# helpers to create links to other pages
kwargs['url_for'] = custom_url_for # override url_for function in templates
kwargs['image_proxify'] = image_proxify
kwargs['proxify'] = proxify if settings.get('result_proxy', {}).get('url') else None
kwargs['proxify_results'] = settings.get('result_proxy', {}).get('proxify_results', True)
kwargs['get_result_template'] = get_result_template
kwargs['opensearch_url'] = (
url_for('opensearch')
+ '?'
+ urlencode(
{
'method': request.preferences.get_value('method'),
'autocomplete': request.preferences.get_value('autocomplete'),
}
)
)
# scripts from plugins
kwargs['scripts'] = set()
for plugin in request.user_plugins:
for script in plugin.js_dependencies:
kwargs['scripts'].add(script)
# styles from plugins
kwargs['styles'] = set()
for plugin in request.user_plugins:
for css in plugin.css_dependencies:
kwargs['styles'].add(css)
start_time = default_timer()
result = render_template('{}/{}'.format(kwargs['theme'], template_name), **kwargs)
request.render_time += default_timer() - start_time # pylint: disable=assigning-non-slot
return result
@app.before_request
def pre_request():
request.start_time = default_timer() # pylint: disable=assigning-non-slot
request.render_time = 0 # pylint: disable=assigning-non-slot
request.timings = [] # pylint: disable=assigning-non-slot
request.errors = [] # pylint: disable=assigning-non-slot
preferences = Preferences(themes, list(categories.keys()), engines, plugins) # pylint: disable=redefined-outer-name
user_agent = request.headers.get('User-Agent', '').lower()
if 'webkit' in user_agent and 'android' in user_agent:
preferences.key_value_settings['method'].value = 'GET'
request.preferences = preferences # pylint: disable=assigning-non-slot
try:
preferences.parse_dict(request.cookies)
except Exception as e: # pylint: disable=broad-except
logger.exception(e, exc_info=True)
request.errors.append(gettext('Invalid settings, please edit your preferences'))
# merge GET, POST vars
# request.form
request.form = dict(request.form.items()) # pylint: disable=assigning-non-slot
for k, v in request.args.items():
if k not in request.form:
request.form[k] = v
if request.form.get('preferences'):
preferences.parse_encoded_data(request.form['preferences'])
else:
try:
preferences.parse_dict(request.form)
except Exception as e: # pylint: disable=broad-except
logger.exception(e, exc_info=True)
request.errors.append(gettext('Invalid settings'))
# language is defined neither in settings nor in preferences
# use browser headers
if not preferences.get_value("language"):
language = _get_browser_language(request, settings['search']['languages'])
preferences.parse_dict({"language": language})
logger.debug('set language %s (from browser)', preferences.get_value("language"))
# locale is defined neither in settings nor in preferences
# use browser headers
if not preferences.get_value("locale"):
locale = _get_browser_language(request, LOCALE_NAMES.keys())
preferences.parse_dict({"locale": locale})
logger.debug('set locale %s (from browser)', preferences.get_value("locale"))
# request.user_plugins
request.user_plugins = [] # pylint: disable=assigning-non-slot
allowed_plugins = preferences.plugins.get_enabled()
disabled_plugins = preferences.plugins.get_disabled()
for plugin in plugins:
if (plugin.default_on and plugin.id not in disabled_plugins) or plugin.id in allowed_plugins:
request.user_plugins.append(plugin)
@app.after_request
def add_default_headers(response: flask.Response):
# set default http headers
for header, value in settings['server']['default_http_headers'].items():
if header in response.headers:
continue
response.headers[header] = value
return response
@app.after_request
def post_request(response: flask.Response):
total_time = default_timer() - request.start_time
timings_all = [
'total;dur=' + str(round(total_time * 1000, 3)),
'render;dur=' + str(round(request.render_time * 1000, 3)),
]
if len(request.timings) > 0:
timings = sorted(request.timings, key=lambda t: t.total)
timings_total = [
'total_' + str(i) + '_' + t.engine + ';dur=' + str(round(t.total * 1000, 3)) for i, t in enumerate(timings)
]
timings_load = [
'load_' + str(i) + '_' + t.engine + ';dur=' + str(round(t.load * 1000, 3))
for i, t in enumerate(timings)
if t.load
]
timings_all = timings_all + timings_total + timings_load
response.headers.add('Server-Timing', ', '.join(timings_all))
return response
def index_error(output_format: str, error_message: str):
if output_format == 'json':
return Response(json.dumps({'error': error_message}), mimetype='application/json')
if output_format == 'csv':
response = Response('', mimetype='application/csv')
cont_disp = 'attachment;Filename=searx.csv'
response.headers.add('Content-Disposition', cont_disp)
return response
if output_format == 'rss':
response_rss = render(
'opensearch_response_rss.xml',
results=[],
q=request.form['q'] if 'q' in request.form else '',
number_of_results=0,
error_message=error_message,
)
return Response(response_rss, mimetype='text/xml')
# html
request.errors.append(gettext('search error'))
return render(
# fmt: off
'index.html',
selected_categories=get_selected_categories(request.preferences, request.form),
# fmt: on
)
@app.route('/', methods=['GET', 'POST'])
def index():
"""Render index page."""
# redirect to search if there's a query in the request
if request.form.get('q'):
query = ('?' + request.query_string.decode()) if request.query_string else ''
return redirect(url_for('search') + query, 308)
return render(
# fmt: off
'index.html',
selected_categories=get_selected_categories(request.preferences, request.form),
current_locale = request.preferences.get_value("locale"),
# fmt: on
)
@app.route('/healthz', methods=['GET'])
def health():
return Response('OK', mimetype='text/plain')
@app.route('/search', methods=['GET', 'POST'])
def search():
"""Search query in q and return results.
Supported outputs: html, json, csv, rss.
"""
# pylint: disable=too-many-locals, too-many-return-statements, too-many-branches
# pylint: disable=too-many-statements
# output_format
output_format = request.form.get('format', 'html')
if output_format not in OUTPUT_FORMATS:
output_format = 'html'
if output_format not in settings['search']['formats']:
flask.abort(403)
# check if there is query (not None and not an empty string)
if not request.form.get('q'):
if output_format == 'html':
return render(
# fmt: off
'index.html',
selected_categories=get_selected_categories(request.preferences, request.form),
# fmt: on
)
return index_error(output_format, 'No query'), 400
# search
search_query = None
raw_text_query = None
result_container = None
try:
search_query, raw_text_query, _, _ = get_search_query_from_webapp(request.preferences, request.form)
# search = Search(search_query) # without plugins
search = SearchWithPlugins(search_query, request.user_plugins, request) # pylint: disable=redefined-outer-name
result_container = search.search()
except SearxParameterException as e:
logger.exception('search error: SearxParameterException')
return index_error(output_format, e.message), 400
except Exception as e: # pylint: disable=broad-except
logger.exception(e, exc_info=True)
return index_error(output_format, gettext('search error')), 500
# results
results = result_container.get_ordered_results()
number_of_results = result_container.results_number()
if number_of_results < result_container.results_length():
number_of_results = 0
# checkin for a external bang
if result_container.redirect_url:
return redirect(result_container.redirect_url)
# Server-Timing header
request.timings = result_container.get_timings() # pylint: disable=assigning-non-slot
current_template = None
previous_result = None
# output
for result in results:
if output_format == 'html':
if 'content' in result and result['content']:
result['content'] = highlight_content(escape(result['content'][:1024]), search_query.query)
if 'title' in result and result['title']:
result['title'] = highlight_content(escape(result['title'] or ''), search_query.query)
else:
if result.get('content'):
result['content'] = html_to_text(result['content']).strip()
# removing html content and whitespace duplications
result['title'] = ' '.join(html_to_text(result['title']).strip().split())
if 'url' in result:
result['pretty_url'] = prettify_url(result['url'])
# TODO, check if timezone is calculated right # pylint: disable=fixme
if result.get('publishedDate'): # do not try to get a date from an empty string or a None type
try: # test if publishedDate >= 1900 (datetime module bug)
result['pubdate'] = result['publishedDate'].strftime('%Y-%m-%d %H:%M:%S%z')
except ValueError:
result['publishedDate'] = None
else:
if result['publishedDate'].replace(tzinfo=None) >= datetime.now() - timedelta(days=1):
timedifference = datetime.now() - result['publishedDate'].replace(tzinfo=None)
minutes = int((timedifference.seconds / 60) % 60)
hours = int(timedifference.seconds / 60 / 60)
if hours == 0:
result['publishedDate'] = gettext('{minutes} minute(s) ago').format(minutes=minutes)
else:
result['publishedDate'] = gettext('{hours} hour(s), {minutes} minute(s) ago').format(
hours=hours, minutes=minutes
)
else:
result['publishedDate'] = format_date(result['publishedDate'])
# set result['open_group'] = True when the template changes from the previous result
# set result['close_group'] = True when the template changes on the next result
if current_template != result.get('template'):
result['open_group'] = True
if previous_result:
previous_result['close_group'] = True # pylint: disable=unsupported-assignment-operation
current_template = result.get('template')
previous_result = result
if previous_result:
previous_result['close_group'] = True
if output_format == 'json':
x = {
'query': search_query.query,
'number_of_results': number_of_results,
'results': results,
'answers': list(result_container.answers),
'corrections': list(result_container.corrections),
'infoboxes': result_container.infoboxes,
'suggestions': list(result_container.suggestions),
'unresponsive_engines': __get_translated_errors(result_container.unresponsive_engines),
}
response = json.dumps(x, default=lambda item: list(item) if isinstance(item, set) else item)
return Response(response, mimetype='application/json')
if output_format == 'csv':
csv = UnicodeWriter(StringIO())
keys = ('title', 'url', 'content', 'host', 'engine', 'score', 'type')
csv.writerow(keys)
for row in results:
row['host'] = row['parsed_url'].netloc
row['type'] = 'result'
csv.writerow([row.get(key, '') for key in keys])
for a in result_container.answers:
row = {'title': a, 'type': 'answer'}
csv.writerow([row.get(key, '') for key in keys])
for a in result_container.suggestions:
row = {'title': a, 'type': 'suggestion'}
csv.writerow([row.get(key, '') for key in keys])
for a in result_container.corrections:
row = {'title': a, 'type': 'correction'}
csv.writerow([row.get(key, '') for key in keys])
csv.stream.seek(0)
response = Response(csv.stream.read(), mimetype='application/csv')
cont_disp = 'attachment;Filename=searx_-_{0}.csv'.format(search_query.query)
response.headers.add('Content-Disposition', cont_disp)
return response
if output_format == 'rss':
response_rss = render(
'opensearch_response_rss.xml',
results=results,
answers=result_container.answers,
corrections=result_container.corrections,
suggestions=result_container.suggestions,
q=request.form['q'],
number_of_results=number_of_results,
)
return Response(response_rss, mimetype='text/xml')
# HTML output format
# suggestions: use RawTextQuery to get the suggestion URLs with the same bang
suggestion_urls = list(
map(
lambda suggestion: {'url': raw_text_query.changeQuery(suggestion).getFullQuery(), 'title': suggestion},
result_container.suggestions,
)
)
correction_urls = list(
map(
lambda correction: {'url': raw_text_query.changeQuery(correction).getFullQuery(), 'title': correction},
result_container.corrections,
)
)
return render(
# fmt: off
'results.html',
results = results,
q=request.form['q'],
selected_categories = search_query.categories,
pageno = search_query.pageno,
time_range = search_query.time_range,
number_of_results = format_decimal(number_of_results),
suggestions = suggestion_urls,
answers = result_container.answers,
corrections = correction_urls,
infoboxes = result_container.infoboxes,
engine_data = result_container.engine_data,
paging = result_container.paging,
unresponsive_engines = __get_translated_errors(
result_container.unresponsive_engines
),
current_locale = request.preferences.get_value("locale"),
current_language = match_language(
search_query.lang,
settings['search']['languages'],
fallback=request.preferences.get_value("language")
),
timeout_limit = request.form.get('timeout_limit', None)
# fmt: on
)
def __get_translated_errors(unresponsive_engines: Iterable[UnresponsiveEngine]):
translated_errors = []
# make a copy unresponsive_engines to avoid "RuntimeError: Set changed size
# during iteration" it happens when an engine modifies the ResultContainer
# after the search_multiple_requests method has stopped waiting
for unresponsive_engine in unresponsive_engines:
error_user_text = exception_classname_to_text.get(unresponsive_engine.error_type)
if not error_user_text:
error_user_text = exception_classname_to_text[None]
error_msg = gettext(error_user_text)
if unresponsive_engine.suspended:
error_msg = gettext('Suspended') + ': ' + error_msg
translated_errors.append((unresponsive_engine.engine, error_msg))
return sorted(translated_errors, key=lambda e: e[0])
@app.route('/about', methods=['GET'])
def about():
"""Redirect to about page"""
# custom_url_for is going to add the locale
return redirect(custom_url_for('info', pagename='about'))
@app.route('/info/<locale>/<pagename>', methods=['GET'])
def info(pagename, locale):
"""Render page of online user documentation"""
page = _INFO_PAGES.get_page(pagename, locale)
if page is None:
flask.abort(404)
user_locale = request.preferences.get_value('locale')
return render(
'info.html',
all_pages=_INFO_PAGES.iter_pages(user_locale, fallback_to_default=True),
active_page=page,
active_pagename=pagename,
)
@app.route('/autocompleter', methods=['GET', 'POST'])
def autocompleter():
"""Return autocompleter results"""
# run autocompleter
results = []
# set blocked engines
disabled_engines = request.preferences.engines.get_disabled()
# parse query
raw_text_query = RawTextQuery(request.form.get('q', ''), disabled_engines)
sug_prefix = raw_text_query.getQuery()
# normal autocompletion results only appear if no inner results returned
# and there is a query part
if len(raw_text_query.autocomplete_list) == 0 and len(sug_prefix) > 0:
# get language from cookie
language = request.preferences.get_value('language')
if not language or language == 'all':
language = 'en'
else:
language = language.split('-')[0]
# run autocompletion
raw_results = search_autocomplete(request.preferences.get_value('autocomplete'), sug_prefix, language)
for result in raw_results:
# attention: this loop will change raw_text_query object and this is
# the reason why the sug_prefix was stored before (see above)
if result != sug_prefix:
results.append(raw_text_query.changeQuery(result).getFullQuery())
if len(raw_text_query.autocomplete_list) > 0:
for autocomplete_text in raw_text_query.autocomplete_list:
results.append(raw_text_query.get_autocomplete_full_query(autocomplete_text))
for answers in ask(raw_text_query):
for answer in answers:
results.append(str(answer['answer']))
if request.headers.get('X-Requested-With') == 'XMLHttpRequest':
# the suggestion request comes from the searx search form
suggestions = json.dumps(results)
mimetype = 'application/json'
else:
# the suggestion request comes from browser's URL bar
suggestions = json.dumps([sug_prefix, results])
mimetype = 'application/x-suggestions+json'
suggestions = escape(suggestions, False)
return Response(suggestions, mimetype=mimetype)
@app.route('/preferences', methods=['GET', 'POST'])
def preferences():
"""Render preferences page && save user preferences"""
# pylint: disable=too-many-locals, too-many-return-statements, too-many-branches
# pylint: disable=too-many-statements
# save preferences using the link the /preferences?preferences=...&save=1
if request.args.get('save') == '1':
resp = make_response(redirect(url_for('index', _external=True)))
return request.preferences.save(resp)
# save preferences
if request.method == 'POST':
resp = make_response(redirect(url_for('index', _external=True)))
try:
request.preferences.parse_form(request.form)
except ValidationException:
request.errors.append(gettext('Invalid settings, please edit your preferences'))
return resp
return request.preferences.save(resp)
# render preferences
image_proxy = request.preferences.get_value('image_proxy') # pylint: disable=redefined-outer-name
disabled_engines = request.preferences.engines.get_disabled()
allowed_plugins = request.preferences.plugins.get_enabled()
# stats for preferences page
filtered_engines = dict(filter(lambda kv: request.preferences.validate_token(kv[1]), engines.items()))
engines_by_category = {}
for c in categories: # pylint: disable=consider-using-dict-items
engines_by_category[c] = [e for e in categories[c] if e.name in filtered_engines]
# sort the engines alphabetically since the order in settings.yml is meaningless.
list.sort(engines_by_category[c], key=lambda e: e.name)
# get first element [0], the engine time,
# and then the second element [1] : the time (the first one is the label)
stats = {} # pylint: disable=redefined-outer-name
max_rate95 = 0
for _, e in filtered_engines.items():
h = histogram('engine', e.name, 'time', 'total')
median = round(h.percentage(50), 1) if h.count > 0 else None
rate80 = round(h.percentage(80), 1) if h.count > 0 else None
rate95 = round(h.percentage(95), 1) if h.count > 0 else None
max_rate95 = max(max_rate95, rate95 or 0)
result_count_sum = histogram('engine', e.name, 'result', 'count').sum
successful_count = counter('engine', e.name, 'search', 'count', 'successful')
result_count = int(result_count_sum / float(successful_count)) if successful_count else 0
stats[e.name] = {
'time': median,
'rate80': rate80,
'rate95': rate95,
'warn_timeout': e.timeout > settings['outgoing']['request_timeout'],
'supports_selected_language': _is_selected_language_supported(e, request.preferences),
'result_count': result_count,
}
# end of stats
# reliabilities
reliabilities = {}
engine_errors = get_engine_errors(filtered_engines)
checker_results = checker_get_result()
checker_results = (
checker_results['engines'] if checker_results['status'] == 'ok' and 'engines' in checker_results else {}
)
for _, e in filtered_engines.items():
checker_result = checker_results.get(e.name, {})
checker_success = checker_result.get('success', True)
errors = engine_errors.get(e.name) or []
if counter('engine', e.name, 'search', 'count', 'sent') == 0:
# no request
reliablity = None
elif checker_success and not errors:
reliablity = 100
elif 'simple' in checker_result.get('errors', {}):
# the basic (simple) test doesn't work: the engine is broken accoding to the checker
# even if there is no exception
reliablity = 0
else:
# pylint: disable=consider-using-generator
reliablity = 100 - sum([error['percentage'] for error in errors if not error.get('secondary')])
reliabilities[e.name] = {
'reliablity': reliablity,
'errors': [],
'checker': checker_results.get(e.name, {}).get('errors', {}).keys(),
}
# keep the order of the list checker_results[e.name]['errors'] and deduplicate.
# the first element has the highest percentage rate.
reliabilities_errors = []
for error in errors:
error_user_text = None
if error.get('secondary') or 'exception_classname' not in error:
continue
error_user_text = exception_classname_to_text.get(error.get('exception_classname'))
if not error:
error_user_text = exception_classname_to_text[None]
if error_user_text not in reliabilities_errors:
reliabilities_errors.append(error_user_text)
reliabilities[e.name]['errors'] = reliabilities_errors
# supports
supports = {}
for _, e in filtered_engines.items():
supports_selected_language = _is_selected_language_supported(e, request.preferences)
safesearch = e.safesearch
time_range_support = e.time_range_support
for checker_test_name in checker_results.get(e.name, {}).get('errors', {}):
if supports_selected_language and checker_test_name.startswith('lang_'):
supports_selected_language = '?'
elif safesearch and checker_test_name == 'safesearch':
safesearch = '?'
elif time_range_support and checker_test_name == 'time_range':
time_range_support = '?'
supports[e.name] = {
'supports_selected_language': supports_selected_language,
'safesearch': safesearch,
'time_range_support': time_range_support,
}
return render(
# fmt: off
'preferences.html',
selected_categories = get_selected_categories(request.preferences, request.form),
locales = LOCALE_NAMES,
current_locale = request.preferences.get_value("locale"),
image_proxy = image_proxy,
engines_by_category = engines_by_category,
stats = stats,
max_rate95 = max_rate95,
reliabilities = reliabilities,
supports = supports,
answerers = [
{'info': a.self_info(), 'keywords': a.keywords}
for a in answerers
],
disabled_engines = disabled_engines,
autocomplete_backends = autocomplete_backends,
shortcuts = {y: x for x, y in engine_shortcuts.items()},
themes = themes,
plugins = plugins,
doi_resolvers = settings['doi_resolvers'],
current_doi_resolver = get_doi_resolver(request.preferences),
allowed_plugins = allowed_plugins,
preferences_url_params = request.preferences.get_as_url_params(),
locked_preferences = settings['preferences']['lock'],
preferences = True
# fmt: on
)
def _is_selected_language_supported(engine, preferences: Preferences): # pylint: disable=redefined-outer-name
language = preferences.get_value('language')
if language == 'all':
return True
x = match_language(
language, getattr(engine, 'supported_languages', []), getattr(engine, 'language_aliases', {}), None
)
return bool(x)
@app.route('/image_proxy', methods=['GET'])
def image_proxy():
# pylint: disable=too-many-return-statements, too-many-branches
url = request.args.get('url')
if not url:
return '', 400
if not is_hmac_of(settings['server']['secret_key'], url.encode(), request.args.get('h', '')):
return '', 400
maximum_size = 5 * 1024 * 1024
forward_resp = False
resp = None
try:
request_headers = {
'User-Agent': gen_useragent(),
'Accept': 'image/webp,*/*',
'Accept-Encoding': 'gzip, deflate',
'Sec-GPC': '1',
'DNT': '1',
}
set_context_network_name('image_proxy')
resp, stream = http_stream(method='GET', url=url, headers=request_headers, allow_redirects=True)
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 != 200:
logger.debug('image-proxy: wrong response code: %i', resp.status_code)
if resp.status_code >= 400:
return '', resp.status_code
return '', 400
if not resp.headers.get('Content-Type', '').startswith('image/') and not resp.headers.get(
'Content-Type', ''
).startswith('binary/octet-stream'):
logger.debug('image-proxy: wrong content-type: %s', resp.headers.get('Content-Type', ''))
return '', 400
forward_resp = True
except httpx.HTTPError:
logger.exception('HTTP error')
return '', 400
finally:
if resp and not forward_resp:
# the code is about to return an HTTP 400 error to the browser
# we make sure to close the response between searxng and the HTTP server
try:
resp.close()
except httpx.HTTPError:
logger.exception('HTTP error on closing')
def close_stream():
nonlocal resp, stream
try:
if resp:
resp.close()
del resp
del stream
except httpx.HTTPError as e:
logger.debug('Exception while closing response', e)
try:
headers = dict_subset(resp.headers, {'Content-Type', 'Content-Encoding', 'Content-Length', 'Length'})
response = Response(stream, mimetype=resp.headers['Content-Type'], headers=headers, direct_passthrough=True)
response.call_on_close(close_stream)
return response
except httpx.HTTPError:
close_stream()
return '', 400
@app.route('/engine_descriptions.json', methods=['GET'])
def engine_descriptions():
locale = get_locale().split('_')[0]
result = ENGINE_DESCRIPTIONS['en'].copy()
if locale != 'en':
for engine, description in ENGINE_DESCRIPTIONS.get(locale, {}).items():
result[engine] = description
for engine, description in result.items():
if len(description) == 2 and description[1] == 'ref':
ref_engine, ref_lang = description[0].split(':')
description = ENGINE_DESCRIPTIONS[ref_lang][ref_engine]
if isinstance(description, str):
description = [description, 'wikipedia']
result[engine] = description
# overwrite by about:description (from settings)
for engine_name, engine_mod in engines.items():
descr = getattr(engine_mod, 'about', {}).get('description', None)
if descr is not None:
result[engine_name] = [descr, "SearXNG config"]
return jsonify(result)
@app.route('/stats', methods=['GET'])
def stats():
"""Render engine statistics page."""
sort_order = request.args.get('sort', default='name', type=str)
selected_engine_name = request.args.get('engine', default=None, type=str)
filtered_engines = dict(filter(lambda kv: request.preferences.validate_token(kv[1]), engines.items()))
if selected_engine_name:
if selected_engine_name not in filtered_engines:
selected_engine_name = None
else:
filtered_engines = [selected_engine_name]
checker_results = checker_get_result()
checker_results = (
checker_results['engines'] if checker_results['status'] == 'ok' and 'engines' in checker_results else {}
)
engine_stats = get_engines_stats(filtered_engines)
engine_reliabilities = get_reliabilities(filtered_engines, checker_results)
if sort_order not in STATS_SORT_PARAMETERS:
sort_order = 'name'
reverse, key_name, default_value = STATS_SORT_PARAMETERS[sort_order]
def get_key(engine_stat):
reliability = engine_reliabilities.get(engine_stat['name'], {}).get('reliablity', 0)
reliability_order = 0 if reliability else 1
if key_name == 'reliability':
key = reliability
reliability_order = 0
else:
key = engine_stat.get(key_name) or default_value
if reverse:
reliability_order = 1 - reliability_order
return (reliability_order, key, engine_stat['name'])
engine_stats['time'] = sorted(engine_stats['time'], reverse=reverse, key=get_key)
return render(
# fmt: off
'stats.html',
sort_order = sort_order,
engine_stats = engine_stats,
engine_reliabilities = engine_reliabilities,
selected_engine_name = selected_engine_name,
searx_git_branch = GIT_BRANCH,
# fmt: on
)
@app.route('/stats/errors', methods=['GET'])
def stats_errors():
filtered_engines = dict(filter(lambda kv: request.preferences.validate_token(kv[1]), engines.items()))
result = get_engine_errors(filtered_engines)
return jsonify(result)
@app.route('/stats/checker', methods=['GET'])
def stats_checker():
result = checker_get_result()
return jsonify(result)
@app.route('/robots.txt', methods=['GET'])
def robots():
return Response(
"""User-agent: *
Allow: /info/en/about
Disallow: /stats
Disallow: /image_proxy
Disallow: /preferences
Disallow: /*?*q=*
""",
mimetype='text/plain',
)
@app.route('/opensearch.xml', methods=['GET'])
def opensearch():
method = 'post'
if request.preferences.get_value('method') == 'GET':
method = 'get'
# chrome/chromium only supports HTTP GET....
if request.headers.get('User-Agent', '').lower().find('webkit') >= 0:
method = 'get'
autocomplete = request.preferences.get_value('autocomplete')
ret = render('opensearch.xml', opensearch_method=method, autocomplete=autocomplete)
resp = Response(response=ret, status=200, mimetype="application/opensearchdescription+xml")
return resp
@app.route('/favicon.ico')
def favicon():
theme = request.preferences.get_value("theme")
return send_from_directory(
os.path.join(app.root_path, settings['ui']['static_path'], 'themes', theme, 'img'), # pyright: ignore
'favicon.png',
mimetype='image/vnd.microsoft.icon',
)
@app.route('/clear_cookies')
def clear_cookies():
resp = make_response(redirect(url_for('index', _external=True)))
for cookie_name in request.cookies:
resp.delete_cookie(cookie_name)
return resp
@app.route('/config')
def config():
"""Return configuration in JSON format."""
_engines = []
for name, engine in engines.items():
if not request.preferences.validate_token(engine):
continue
supported_languages = engine.supported_languages
if isinstance(engine.supported_languages, dict):
supported_languages = list(engine.supported_languages.keys())
_engines.append(
{
'name': name,
'categories': engine.categories,
'shortcut': engine.shortcut,
'enabled': not engine.disabled,
'paging': engine.paging,
'language_support': engine.language_support,
'supported_languages': supported_languages,
'safesearch': engine.safesearch,
'time_range_support': engine.time_range_support,
'timeout': engine.timeout,
}
)
_plugins = []
for _ in plugins:
_plugins.append({'name': _.name, 'enabled': _.default_on})
return jsonify(
{
'categories': list(categories.keys()),
'engines': _engines,
'plugins': _plugins,
'instance_name': settings['general']['instance_name'],
'locales': LOCALE_NAMES,
'default_locale': settings['ui']['default_locale'],
'autocomplete': settings['search']['autocomplete'],
'safe_search': settings['search']['safe_search'],
'default_theme': settings['ui']['default_theme'],
'version': VERSION_STRING,
'brand': {
'PRIVACYPOLICY_URL': get_setting('general.privacypolicy_url'),
'CONTACT_URL': get_setting('general.contact_url'),
'GIT_URL': GIT_URL,
'GIT_BRANCH': GIT_BRANCH,
'DOCS_URL': get_setting('brand.docs_url'),
},
'doi_resolvers': list(settings['doi_resolvers'].keys()),
'default_doi_resolver': settings['default_doi_resolver'],
}
)
@app.errorhandler(404)
def page_not_found(_e):
return render('404.html'), 404
# see https://flask.palletsprojects.com/en/1.1.x/cli/
# True if "FLASK_APP=searx/webapp.py FLASK_ENV=development flask run"
flask_run_development = (
os.environ.get("FLASK_APP") is not None and os.environ.get("FLASK_ENV") == 'development' and is_flask_run_cmdline()
)
# True if reload feature is activated of werkzeug, False otherwise (including uwsgi, etc..)
# __name__ != "__main__" if searx.webapp is imported (make test, make docs, uwsgi...)
# see run() at the end of this file : searx_debug activates the reload feature.
werkzeug_reloader = flask_run_development or (searx_debug and __name__ == "__main__")
# initialize the engines except on the first run of the werkzeug server.
if not werkzeug_reloader or (werkzeug_reloader and os.environ.get("WERKZEUG_RUN_MAIN") == "true"):
locales_initialize()
_INFO_PAGES = infopage.InfoPageSet()
plugin_initialize(app)
search_initialize(enable_checker=True, check_network=True, enable_metrics=settings['general']['enable_metrics'])
def run():
logger.debug('starting webserver on %s:%s', settings['server']['bind_address'], settings['server']['port'])
app.run(
debug=searx_debug,
use_debugger=searx_debug,
port=settings['server']['port'],
host=settings['server']['bind_address'],
threaded=True,
extra_files=[get_default_settings_path()],
)
application = app
patch_application(app)
if __name__ == "__main__":
run()