logo

searx

Unnamed repository; edit this file 'description' to name the repository.
commit: ebbf33343a19089ea7b121ae3001dc528a4d7bf3
parent: 37035b7a40ec76ab582908827d2d25158f10301e
Author: Adam Tauber <asciimoo@gmail.com>
Date:   Mon,  1 Feb 2016 21:40:28 +0100

[fix] bing images changes

Diffstat:

searx/engines/bing_images.py | 16++++++++--------
tests/unit/engines/test_bing_images.py | 20+++++++-------------
2 files changed, 15 insertions(+), 21 deletions(-)

diff --git a/searx/engines/bing_images.py b/searx/engines/bing_images.py @@ -17,7 +17,7 @@ from urllib import urlencode from lxml import html -from yaml import load +from json import loads import re # engine dependent config @@ -36,6 +36,9 @@ safesearch_types = {2: 'STRICT', 0: 'OFF'} +_quote_keys_regex = re.compile('({|,)([a-z][a-z0-9]*):(")', re.I | re.U) + + # do search-request def request(query, params): offset = (params['pageno'] - 1) * 10 + 1 @@ -65,22 +68,19 @@ def response(resp): dom = html.fromstring(resp.text) - # init regex for yaml-parsing - p = re.compile('({|,)([a-z]+):(")') - # parse results for result in dom.xpath('//div[@class="dg_u"]'): link = result.xpath('./a')[0] - # parse yaml-data (it is required to add a space, to make it parsable) - yaml_data = load(p.sub(r'\1\2: \3', link.attrib.get('m'))) + # parse json-data (it is required to add a space, to make it parsable) + json_data = loads(_quote_keys_regex.sub(r'\1"\2": \3', link.attrib.get('m'))) title = link.attrib.get('t1') ihk = link.attrib.get('ihk') # url = 'http://' + link.attrib.get('t3') - url = yaml_data.get('surl') - img_src = yaml_data.get('imgurl') + url = json_data.get('surl') + img_src = json_data.get('imgurl') # append result results.append({'template': 'images.html', diff --git a/tests/unit/engines/test_bing_images.py b/tests/unit/engines/test_bing_images.py @@ -39,16 +39,10 @@ class TestBingImagesEngine(SearxTestCase): html = """ <div class="dg_u" style="width:178px;height:144px;left:17px;top:0px"> - <a href="#" ihk="HN.608003696942779811" - m="{ns:&quot;images&quot;,k:&quot;5045&quot;, -mid:&quot;659EB92C317974F34517A1CCAEBEF76A578E08DEE&quot;, -surl:&quot;http://www.page.url/&quot;,imgurl:&quot;http://test.url/Test%20Query.jpg&quot;, -oh:&quot;238&quot;,tft:&quot;0&quot;,oi:&quot;http://www.image.url/Images/Test%20Query.jpg&quot;}" - mid="59EB92C317974F34517A1CCAEBEF76A578E08DEE" onclick="return false;" - t1="Test Query" t2="650 x 517 ยท 31 kB ยท jpeg" t3="www.short.url" h="ID=images,5045.1"> - <img src="https://tse4.mm.bing.net/th?id=HN.608003696942779811&amp;o=4&amp;pid=1.7" - style="height:144px;" width="178" height="144"/> + <a href="/images/search?q=south&amp;view=detailv2&amp;&amp;id=7E92863981CCFB89FBDD55205C742DFDA3290CF6&amp;selectedIndex=9&amp;ccid=vzvIfv5u&amp;simid=608055786735667000&amp;thid=OIP.Mbf3bc87efe6e0e476be8cc34bf6cd80eH0" ihk="OIP.Mbf3bc87efe6e0e476be8cc34bf6cd80eH0" t1="South Carolina" t2="747 x 589 &#183; 29 kB &#183; gif" t3="www.digital-topo-maps.com/county-map/south-carolina.shtml" hh="236" hw="300" m='{ns:"images",k:"5117",mid:"7E92863981CCFB89FBDD55205C742DFDA3290CF6",md5:"bf3bc87efe6e0e476be8cc34bf6cd80e",surl:"http://www.digital-topo-maps.com/county-map/south-carolina.shtml",imgurl:"http://www.digital-topo-maps.com/county-map/south-carolina-county-map.gif",tid:"OIP.Mbf3bc87efe6e0e476be8cc34bf6cd80eH0",ow:"480",docid:"608055786735667000",oh:"378",tft:"45"}' mid="7E92863981CCFB89FBDD55205C742DFDA3290CF6" h="ID=images,5117.1"> + <img class="img_hid" src2="https://tse4.mm.bing.net/th?id=OIP.Mbf3bc87efe6e0e476be8cc34bf6cd80eH0&amp;w=210&amp;h=154&amp;c=7&amp;rs=1&amp;qlt=90&amp;o=4&amp;pid=1.1" style="width:210px;height:154px;" width="210" height="154"> </a> + </div> """ html = html.replace('\r\n', '').replace('\n', '').replace('\r', '') @@ -56,11 +50,11 @@ oh:&quot;238&quot;,tft:&quot;0&quot;,oi:&quot;http://www.image.url/Images/Test%2 results = bing_images.response(response) self.assertEqual(type(results), list) self.assertEqual(len(results), 1) - self.assertEqual(results[0]['title'], 'Test Query') - self.assertEqual(results[0]['url'], 'http://www.page.url/') + self.assertEqual(results[0]['title'], 'South Carolina') + self.assertEqual(results[0]['url'], 'http://www.digital-topo-maps.com/county-map/south-carolina.shtml') self.assertEqual(results[0]['content'], '') - self.assertEqual(results[0]['thumbnail_src'], 'https://www.bing.com/th?id=HN.608003696942779811') - self.assertEqual(results[0]['img_src'], 'http://test.url/Test%20Query.jpg') + self.assertEqual(results[0]['thumbnail_src'], 'https://www.bing.com/th?id=OIP.Mbf3bc87efe6e0e476be8cc34bf6cd80eH0') + self.assertEqual(results[0]['img_src'], 'http://www.digital-topo-maps.com/county-map/south-carolina-county-map.gif') html = """ <a href="#" ihk="HN.608003696942779811"