1
0
mirror of https://github.com/searx/searx synced 2024-12-05 08:30:55 +01:00

gentoo engine test

This commit is contained in:
Alice Ferrazzi 2018-08-01 17:48:38 +09:00 committed by Alice Ferrazzi
parent b8978b4bbf
commit b7886bb15c
2 changed files with 126 additions and 0 deletions

122
searx/engines/gentoo.py Normal file
View File

@ -0,0 +1,122 @@
# -*- coding: utf-8 -*-
"""
Gentoo Wiki
@website https://wiki.gentoo.org
@provide-api no (Mediawiki provides API, but Arch Wiki blocks access to it
@using-api no
@results HTML
@stable no (HTML can change)
@parse url, title
"""
from lxml import html
from searx.engines.xpath import extract_text
from searx.url_utils import urlencode, urljoin
# engine dependent config
categories = ['it']
language_support = True
paging = True
base_url = 'https://wiki.gentoo.org'
# xpath queries
xpath_results = '//ul[@class="mw-search-results"]/li'
xpath_link = './/div[@class="mw-search-result-heading"]/a'
# cut 'en' from 'en-US', 'de' from 'de-CH', and so on
def locale_to_lang_code(locale):
if locale.find('-') >= 0:
locale = locale.split('-')[0]
return locale
# wikis for some languages were moved off from the main site, we need to make
# requests to correct URLs to be able to get results in those languages
lang_urls = {
'en': {
'base': 'https://wiki.gentoo.org',
'search': '/index.php?title=Special:Search&offset={offset}&{query}'
}
}
# get base & search URLs for selected language
def get_lang_urls(language):
if language in lang_urls:
return lang_urls[language]
return lang_urls['en']
# Language names to build search requests for
# those languages which are hosted on the main site.
main_langs = {
'ar': 'العربية',
'bg': 'Български',
'cs': 'Česky',
'da': 'Dansk',
'el': 'Ελληνικά',
'es': 'Español',
'he': 'עברית',
'hr': 'Hrvatski',
'hu': 'Magyar',
'it': 'Italiano',
'ko': '한국어',
'lt': 'Lietuviškai',
'nl': 'Nederlands',
'pl': 'Polski',
'pt': 'Português',
'ru': 'Русский',
'sl': 'Slovenský',
'th': 'ไทย',
'uk': 'Українська',
'zh': '简体中文'
}
supported_languages = dict(lang_urls, **main_langs)
# do search-request
def request(query, params):
# translate the locale (e.g. 'en-US') to language code ('en')
language = locale_to_lang_code(params['language'])
# if our language is hosted on the main site, we need to add its name
# to the query in order to narrow the results to that language
if language in main_langs:
query += b' (' + main_langs[language] + b')'
# prepare the request parameters
query = urlencode({'search': query})
offset = (params['pageno'] - 1) * 20
# get request URLs for our language of choice
urls = get_lang_urls(language)
search_url = urls['base'] + urls['search']
params['url'] = search_url.format(query=query, offset=offset)
return params
# get response from search-request
def response(resp):
# get the base URL for the language in which request was made
language = locale_to_lang_code(resp.search_params['language'])
base_url = get_lang_urls(language)['base']
results = []
dom = html.fromstring(resp.text)
# parse results
for result in dom.xpath(xpath_results):
link = result.xpath(xpath_link)[0]
href = urljoin(base_url, link.attrib.get('href'))
title = extract_text(link)
results.append({'url': href,
'title': title})
return results

View File

@ -273,6 +273,10 @@ engines:
timeout : 3.0
disabled: True
- name : gentoo
engine : gentoo
shortcut : ge
- name : gitlab
engine : json_engine
paging : True