zaclys/searxng
1
0
Fork 1
mirror of https://github.com/searxng/searxng synced 2024-01-01 18:24:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'master' of https://github.com/asciimoo/searx into filtron

Browse Source
This commit is contained in:
Markus Heiser 2020-04-22 14:24:59 +02:00
commit 1fdd8cd3c0
6 changed files with 29 additions and 128 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
searx/data/useragents.json
View File

@ -1,11 +1,11 @@
{
"versions": [
"74.0",
"73.0.1",
"73.0"
"75.0",
"74.0.1",
"74.0"
],
"os": [
"Windows NT 10; WOW64",
"Windows NT 10.0; WOW64",
"X11; Linux x86_64"
],
"ua": "Mozilla/5.0 ({os}; rv:{version}) Gecko/20100101 Firefox/{version}"

96
searx/engines/faroo.py
View File

@ -1,96 +0,0 @@
"""
Faroo (Web, News)
@website http://www.faroo.com
@provide-api yes (http://www.faroo.com/hp/api/api.html), require API-key
@using-api no
@results JSON
@stable yes
@parse url, title, content, publishedDate, img_src
"""
from json import loads
import datetime
from searx.utils import searx_useragent
from searx.url_utils import urlencode
# engine dependent config
categories = ['general', 'news']
paging = True
language_support = True
number_of_results = 10
# search-url
url = 'http://www.faroo.com/'
search_url = url + 'instant.json?{query}'\
'&start={offset}'\
'&length={number_of_results}'\
'&l={language}'\
'&src={categorie}'\
'&i=false'\
'&c=false'
search_category = {'general': 'web',
'news': 'news'}
# do search-request
def request(query, params):
offset = (params['pageno'] - 1) * number_of_results + 1
categorie = search_category.get(params['category'], 'web')
if params['language'] == 'all':
language = 'en'
else:
language = params['language'].split('-')[0]
# if language is not supported, put it in english
if language != 'en' and\
language != 'de' and\
language != 'zh':
language = 'en'
params['url'] = search_url.format(offset=offset,
number_of_results=number_of_results,
query=urlencode({'q': query}),
language=language,
categorie=categorie)
params['headers']['Referer'] = url
return params
# get response from search-request
def response(resp):
# HTTP-Code 429: rate limit exceeded
if resp.status_code == 429:
raise Exception("rate limit has been exceeded!")
results = []
search_res = loads(resp.text)
# return empty array if there are no results
if not search_res.get('results', {}):
return []
# parse results
for result in search_res['results']:
publishedDate = None
result_json = {'url': result['url'], 'title': result['title'],
'content': result['kwic']}
if result['news']:
result_json['publishedDate'] = \
datetime.datetime.fromtimestamp(result['date'] / 1000.0)
# append image result if image url is set
if result['iurl']:
result_json['template'] = 'videos.html'
result_json['thumbnail'] = result['iurl']
results.append(result_json)
# return results
return results

4
searx/results.py
View File

@ -345,8 +345,8 @@ class ResultContainer(object):
return 0
return resultnum_sum / len(self._number_of_results)
def add_unresponsive_engine(self, engine_error):
self.unresponsive_engines.add(engine_error)
def add_unresponsive_engine(self, engine_name, error_type, error_message=None):
self.unresponsive_engines.add((engine_name, error_type, error_message))
def add_timing(self, engine_name, engine_time, page_load_time):
self.timings.append({

17
searx/search.py
View File

@ -127,11 +127,7 @@ def search_one_offline_request_safe(engine_name, query, request_params, result_c
logger.exception('engine {0} : invalid input : {1}'.format(engine_name, e))
except Exception as e:
record_offline_engine_stats_on_error(engine, result_container, start_time)
result_container.add_unresponsive_engine((
engine_name,
u'{0}: {1}'.format(gettext('unexpected crash'), e),
))
result_container.add_unresponsive_engine(engine_name, 'unexpected crash', str(e))
logger.exception('engine {0} : exception : {1}'.format(engine_name, e))
@ -186,24 +182,21 @@ def search_one_http_request_safe(engine_name, query, request_params, result_cont
engine.stats['errors'] += 1
if (issubclass(e.__class__, requests.exceptions.Timeout)):
result_container.add_unresponsive_engine((engine_name, gettext('timeout')))
result_container.add_unresponsive_engine(engine_name, 'timeout')
# requests timeout (connect or read)
logger.error("engine {0} : HTTP requests timeout"
"(search duration : {1} s, timeout: {2} s) : {3}"
.format(engine_name, engine_time, timeout_limit, e.__class__.__name__))
requests_exception = True
elif (issubclass(e.__class__, requests.exceptions.RequestException)):
result_container.add_unresponsive_engine((engine_name, gettext('request exception')))
result_container.add_unresponsive_engine(engine_name, 'request exception')
# other requests exception
logger.exception("engine {0} : requests exception"
"(search duration : {1} s, timeout: {2} s) : {3}"
.format(engine_name, engine_time, timeout_limit, e))
requests_exception = True
else:
result_container.add_unresponsive_engine((
engine_name,
u'{0}: {1}'.format(gettext('unexpected crash'), e),
))
result_container.add_unresponsive_engine(engine_name, 'unexpected crash', str(e))
# others errors
logger.exception('engine {0} : exception : {1}'.format(engine_name, e))
@ -238,7 +231,7 @@ def search_multiple_requests(requests, result_container, start_time, timeout_lim
remaining_time = max(0.0, timeout_limit - (time() - start_time))
th.join(remaining_time)
if th.isAlive():
result_container.add_unresponsive_engine((th._engine_name, gettext('timeout')))
result_container.add_unresponsive_engine(th._engine_name, 'timeout')
logger.warning('engine timeout: {0}'.format(th._engine_name))

5
searx/settings.yml
View File

@ -219,11 +219,6 @@ engines:
shortcut : et
disabled : True
- name : faroo
engine : faroo
shortcut : fa
disabled : True
- name : 1x
engine : www1x
shortcut : 1x

25
searx/webapp.py
View File

@ -56,6 +56,7 @@ from flask import (
from babel.support import Translations
import flask_babel
from flask_babel import Babel, gettext, format_date, format_decimal
from flask.ctx import has_request_context
from flask.json import jsonify
from searx import brand
from searx import settings, searx_dir, searx_debug
@ -165,13 +166,11 @@ _flask_babel_get_translations = flask_babel.get_translations
# monkey patch for flask_babel.get_translations
def _get_translations():
translation_locale = request.form.get('use-translation')
if translation_locale:
if has_request_context() and request.form.get('use-translation') == 'oc':
babel_ext = flask_babel.current_app.extensions['babel']
translation = Translations.load(next(babel_ext.translation_directories), 'oc')
else:
translation = _flask_babel_get_translations()
return translation
return Translations.load(next(babel_ext.translation_directories), 'oc')
return _flask_babel_get_translations()
flask_babel.get_translations = _get_translations
@ -627,7 +626,7 @@ def index():
'corrections': list(result_container.corrections),
'infoboxes': result_container.infoboxes,
'suggestions': list(result_container.suggestions),
'unresponsive_engines': list(result_container.unresponsive_engines)},
'unresponsive_engines': __get_translated_errors(result_container.unresponsive_engines)}, # noqa
default=lambda item: list(item) if isinstance(item, set) else item),
mimetype='application/json')
elif output_format == 'csv':
@ -695,7 +694,7 @@ def index():
corrections=correction_urls,
infoboxes=result_container.infoboxes,
paging=result_container.paging,
unresponsive_engines=result_container.unresponsive_engines,
unresponsive_engines=__get_translated_errors(result_container.unresponsive_engines),
current_language=match_language(search_query.lang,
LANGUAGE_CODES,
fallback=request.preferences.get_value("language")),
@ -706,6 +705,16 @@ def index():
)
def __get_translated_errors(unresponsive_engines):
translated_errors = []
for unresponsive_engine in unresponsive_engines:
error_msg = gettext(unresponsive_engine[1])
if unresponsive_engine[2]:
error_msg = "{} {}".format(error_msg, unresponsive_engine[2])
translated_errors.append((unresponsive_engine[0], error_msg))
return translated_errors
@app.route('/about', methods=['GET'])
def about():
"""Render about page"""