From 4ec810749b45c02462a9be98090374b759db8151 Mon Sep 17 00:00:00 2001 From: Alexandre Flament Date: Mon, 7 Dec 2020 15:46:00 +0100 Subject: [PATCH] [fix] 1x engine --- searx/engines/www1x.py | 24 ++++++++++-------------- searx/settings.yml | 1 + 2 files changed, 11 insertions(+), 14 deletions(-) diff --git a/searx/engines/www1x.py b/searx/engines/www1x.py index 8d691c85..b8f111a5 100644 --- a/searx/engines/www1x.py +++ b/searx/engines/www1x.py @@ -7,12 +7,12 @@ @using-api no @results HTML @stable no (HTML can change) - @parse url, title, thumbnail, img_src, content + @parse url, title, thumbnail """ -from lxml import html +from lxml import html, etree from urllib.parse import urlencode, urljoin -from searx.utils import extract_text +from searx.utils import extract_text, eval_xpath_list, eval_xpath_getindex # engine dependent config categories = ['images'] @@ -21,6 +21,7 @@ paging = False # search-url base_url = 'https://1x.com' search_url = base_url + '/backend/search.php?{query}' +gallery_url = 'https://gallery.1x.com/' # do search-request @@ -33,23 +34,18 @@ def request(query, params): # get response from search-request def response(resp): results = [] - - dom = html.fromstring(resp.text) - for res in dom.xpath('//div[@class="List-item MainListing"]'): - # processed start and end of link - link = res.xpath('//a')[0] - + xmldom = etree.fromstring(resp.content) + xmlsearchresult = eval_xpath_getindex(xmldom, '//searchresult', 0) + dom = html.fragment_fromstring(xmlsearchresult.text, create_parent='div') + for link in eval_xpath_list(dom, '/div/table/tr/td/div[2]//a'): url = urljoin(base_url, link.attrib.get('href')) title = extract_text(link) - - thumbnail_src = urljoin(base_url, res.xpath('.//img')[0].attrib['src']) - # TODO: get image with higher resolution - img_src = thumbnail_src + thumbnail_src = urljoin(gallery_url, eval_xpath_getindex(link, './/img', 0).attrib['src']) # append result results.append({'url': url, 'title': title, - 'img_src': img_src, + 'img_src': thumbnail_src, 'content': '', 'thumbnail_src': thumbnail_src, 'template': 'images.html'}) diff --git a/searx/settings.yml b/searx/settings.yml index 04b658e0..f68770cd 100644 --- a/searx/settings.yml +++ b/searx/settings.yml @@ -289,6 +289,7 @@ engines: - name : 1x engine : www1x shortcut : 1x + timeout : 3.0 disabled : True - name : fdroid