1
0
mirror of https://github.com/searxng/searxng.git synced 2024-11-05 12:50:11 +01:00
searxng/searx/engines/ebay.py

77 lines
2.3 KiB
Python
Raw Normal View History

# SPDX-License-Identifier: AGPL-3.0-or-later
"""
Ebay (Videos, Music, Files)
"""
2018-02-09 23:52:11 +01:00
from lxml import html
from searx.engines.xpath import extract_text
from urllib.parse import quote
# about
about = {
"website": 'https://www.ebay.com',
"wikidata_id": 'Q58024',
"official_api_documentation": 'https://developer.ebay.com/',
"use_official_api": False,
"require_api_key": False,
"results": 'HTML',
}
2018-02-09 23:52:11 +01:00
categories = ['shopping']
paging = True
# Set base_url in settings.yml in order to
# have the desired local TLD.
base_url = None
search_url = '/sch/i.html?_nkw={query}&_sacat={pageno}'
2018-02-09 23:52:11 +01:00
results_xpath = '//li[contains(@class, "s-item")]'
url_xpath = './/a[@class="s-item__link"]/@href'
title_xpath = './/h3[@class="s-item__title"]'
content_xpath = './/div[@span="SECONDARY_INFO"]'
price_xpath = './/div[contains(@class, "s-item__detail")]/span[@class="s-item__price"][1]/text()'
shipping_xpath = './/span[contains(@class, "s-item__shipping")]/text()'
source_country_xpath = './/span[contains(@class, "s-item__location")]/text()'
thumbnail_xpath = './/img[@class="s-item__image-img"]/@src'
def request(query, params):
params['url'] = f'{base_url}' + search_url.format(query=quote(query), pageno=params['pageno'])
2018-02-09 23:52:11 +01:00
return params
def response(resp):
results = []
dom = html.fromstring(resp.text)
results_dom = dom.xpath(results_xpath)
if not results_dom:
return []
for result_dom in results_dom:
url = extract_text(result_dom.xpath(url_xpath))
title = extract_text(result_dom.xpath(title_xpath))
content = extract_text(result_dom.xpath(content_xpath))
price = extract_text(result_dom.xpath(price_xpath))
shipping = extract_text(result_dom.xpath(shipping_xpath))
source_country = extract_text(result_dom.xpath(source_country_xpath))
thumbnail = extract_text(result_dom.xpath(thumbnail_xpath))
if title == "":
continue
results.append(
{
'url': url,
'title': title,
'content': content,
'price': price,
'shipping': shipping,
'source_country': source_country,
'thumbnail': thumbnail,
'template': 'products.html',
}
)
2018-02-09 23:52:11 +01:00
return results