logo

searx

My custom branche(s) on searx, a meta-search engine
commit: 82bf9c6b290423b5b7b300fc3af68bbfa32f3d4f
parent: cf8e2246010534b4cb617fe96d198f23d7d038dd
Author: Noémi Ványi <kvch@users.noreply.github.com>
Date:   Fri,  4 Jan 2019 22:10:38 +0100

Merge pull request #1464 from miicha/bugfix_startpage

restore startpage search results

Diffstat:

Msearx/engines/startpage.py7++++---
Mtests/unit/engines/test_startpage.py22+++++++++++-----------
2 files changed, 15 insertions(+), 14 deletions(-)

diff --git a/searx/engines/startpage.py b/searx/engines/startpage.py @@ -32,8 +32,9 @@ search_url = base_url + 'do/search' # specific xpath variables # ads xpath //div[@id="results"]/div[@id="sponsored"]//div[@class="result"] # not ads: div[@class="result"] are the direct childs of div[@id="results"] -results_xpath = '//div[@class="result"]' +results_xpath = '//li[contains(@class, "search-result") and contains(@class, "search-item")]' link_xpath = './/h3/a' +content_xpath = './p[@class="search-item__body"]' # do search-request @@ -79,8 +80,8 @@ def response(resp): title = extract_text(link) - if result.xpath('./p[@class="desc clk"]'): - content = extract_text(result.xpath('./p[@class="desc clk"]')) + if result.xpath(content_xpath): + content = extract_text(result.xpath(content_xpath)) else: content = '' diff --git a/tests/unit/engines/test_startpage.py b/tests/unit/engines/test_startpage.py @@ -31,14 +31,14 @@ class TestStartpageEngine(SearxTestCase): self.assertEqual(startpage.response(response), []) html = """ - <div class='result' style=' *width : auto; *margin-right : 10%;'> + <li class="search-result search-item"> <h3> <a href='http://this.should.be.the.link/' id='title_2' name='title_2' > This should be the title </a> <span id='title_stars_2' name='title_stars_2'> </span> </h3> - <p class='desc clk'> + <p class="search-item__body"> This should be the content. </p> <p> @@ -56,7 +56,7 @@ class TestStartpageEngine(SearxTestCase): Mis en surbrillance </A> </p> - </div> + </li> """ response = mock.Mock(text=html.encode('utf-8')) results = startpage.response(response) @@ -67,14 +67,14 @@ class TestStartpageEngine(SearxTestCase): self.assertEqual(results[0]['content'], 'This should be the content.') html = """ - <div class='result' style=' *width : auto; *margin-right : 10%;'> + <li class="search-result search-item"> <h3> <a href='http://www.google.com/aclk?sa=l&ai=C' id='title_2' name='title_2' > This should be the title </a> <span id='title_stars_2' name='title_stars_2'> </span> </h3> - <p class='desc clk'> + <p class="search-item__body"> This should be the content. </p> <p> @@ -92,20 +92,20 @@ class TestStartpageEngine(SearxTestCase): Mis en surbrillance </A> </p> - </div> - <div class='result' style=' *width : auto; *margin-right : 10%;'> + </li> + <li class="search-result search-item"> <h3> <span id='title_stars_2' name='title_stars_2'> </span> </h3> - <p class='desc clk'> + <p class="search-item__body"> This should be the content. </p> <p> <span class='url'>www.speed<b>test</b>.net/fr/ </span> </p> - </div> - <div class='result' style=' *width : auto; *margin-right : 10%;'> + </li> + <li class="search-result search-item"> <h3> <a href='http://this.should.be.the.link/' id='title_2' name='title_2' > This should be the title @@ -127,7 +127,7 @@ class TestStartpageEngine(SearxTestCase): Mis en surbrillance </A> </p> - </div> + </li> """ response = mock.Mock(text=html.encode('utf-8')) results = startpage.response(response)