mirror of
https://github.com/searx/searx
synced 2025-09-03 07:56:09 +00:00
Onesearch engine without pagination
This commit is contained in:
56
searx/engines/onesearch.py
Normal file
56
searx/engines/onesearch.py
Normal file
@@ -0,0 +1,56 @@
|
|||||||
|
|
||||||
|
"""Onesearch
|
||||||
|
"""
|
||||||
|
|
||||||
|
from lxml.html import fromstring
|
||||||
|
|
||||||
|
import re
|
||||||
|
|
||||||
|
from searx.utils import (
|
||||||
|
eval_xpath,
|
||||||
|
extract_text,
|
||||||
|
)
|
||||||
|
|
||||||
|
from urllib.parse import unquote
|
||||||
|
|
||||||
|
# about
|
||||||
|
about = {
|
||||||
|
"website": 'https://www.onesearch.com/',
|
||||||
|
"wikidata_id": None,
|
||||||
|
"use_official_api": False,
|
||||||
|
"require_api_key": False,
|
||||||
|
"results": 'HTML',
|
||||||
|
}
|
||||||
|
|
||||||
|
# engine dependent config
|
||||||
|
categories = ['general']
|
||||||
|
|
||||||
|
# search-url
|
||||||
|
URL = 'https://www.onesearch.com/yhs/search;?p=%s'
|
||||||
|
|
||||||
|
def request(query, params):
|
||||||
|
params['url'] = URL % query
|
||||||
|
return params
|
||||||
|
|
||||||
|
|
||||||
|
# get response from search-request
|
||||||
|
def response(resp):
|
||||||
|
|
||||||
|
results = []
|
||||||
|
doc = fromstring(resp.text)
|
||||||
|
|
||||||
|
titles_tags = eval_xpath(doc, '//div[contains(@class, "algo")]//h3[contains(@class, "title")]')
|
||||||
|
contents = eval_xpath(doc, '//div[contains(@class, "algo")]/div[contains(@class, "compText")]/p')
|
||||||
|
onesearch_urls = eval_xpath(doc, '//div[contains(@class, "algo")]//h3[contains(@class, "title")]/a/@href')
|
||||||
|
|
||||||
|
for title_tag, content, onesearch_url in zip(titles_tags, contents, onesearch_urls):
|
||||||
|
print(f"{title_tag.text_content()} ---> {onesearch_url}")
|
||||||
|
matches = re.search(r'RU=(.*?)\/', onesearch_url)
|
||||||
|
results.append({
|
||||||
|
'title': title_tag.text_content(),
|
||||||
|
'content': extract_text(content),
|
||||||
|
'url': unquote(matches.group(1)),
|
||||||
|
})
|
||||||
|
|
||||||
|
return results
|
||||||
|
|
@@ -1623,12 +1623,8 @@ engines:
|
|||||||
|
|
||||||
- name: onesearch
|
- name: onesearch
|
||||||
shortcut: onesearch
|
shortcut: onesearch
|
||||||
engine: xpath
|
engine: onesearch
|
||||||
paging: false
|
paging: false
|
||||||
search_url: https://www.onesearch.com/yhs/search;?p={query}
|
|
||||||
url_xpath: //div[contains(@class, "algo")]//h3[contains(@class, "title")]/a/@href
|
|
||||||
title_xpath: //div[contains(@class, "algo")]//h3[contains(@class, "title")]
|
|
||||||
content_xpath: //div[contains(@class, "algo")]/div[contains(@class, "compText")]/p//text()
|
|
||||||
categories: general
|
categories: general
|
||||||
about:
|
about:
|
||||||
website: https://www.onesearch.com/
|
website: https://www.onesearch.com/
|
||||||
|
Reference in New Issue
Block a user