1
0
mirror of https://github.com/searxng/searxng.git synced 2024-11-20 03:10:10 +01:00
searxng/searx/engines/www1x.py
2015-02-01 13:52:43 +01:00

83 lines
2.0 KiB
Python

## 1x (Images)
#
# @website http://1x.com/
# @provide-api no
#
# @using-api no
# @results HTML
# @stable no (HTML can change)
# @parse url, title, thumbnail, img_src, content
from urllib import urlencode
from urlparse import urljoin
from lxml import html
import string
import re
# engine dependent config
categories = ['images']
paging = False
# search-url
base_url = 'http://1x.com'
search_url = base_url+'/backend/search.php?{query}'
# do search-request
def request(query, params):
params['url'] = search_url.format(query=urlencode({'q': query}))
return params
# get response from search-request
def response(resp):
results = []
# get links from result-text
regex = re.compile('(</a>|<a)')
results_parts = re.split(regex, resp.text)
cur_element = ''
# iterate over link parts
for result_part in results_parts:
# processed start and end of link
if result_part == '<a':
cur_element = result_part
continue
elif result_part != '</a>':
cur_element += result_part
continue
cur_element += result_part
# fix xml-error
cur_element = string.replace(cur_element, '"></a>', '"/></a>')
dom = html.fromstring(cur_element)
link = dom.xpath('//a')[0]
url = urljoin(base_url, link.attrib.get('href'))
title = link.attrib.get('title', '')
thumbnail_src = urljoin(base_url, link.xpath('.//img')[0].attrib['src'])
# TODO: get image with higher resolution
img_src = thumbnail_src
# check if url is showing to a photo
if '/photo/' not in url:
continue
# append result
results.append({'url': url,
'title': title,
'img_src': img_src,
'content': '',
'thumbnail_src': thumbnail_src,
'template': 'images.html'})
# return results
return results