ebbf333 · searx · git.iko.soy

git.iko.soy/searx

[fix] bing images changes

commitebbf33343a19089ea7b121ae3001dc528a4d7bf3 authorAdam Tauber <asciimoo@gmail.com> dateMon 1 Feb 2016, 21:40 (11 years ago) parent37035b7

2 files changed+15−21

compare with master

unifiedsplit

searx/engines/bing_images.py+8−8 tests/unit/engines/test_bing_images.py+7−13

searx/engines/bing_images.py+8−8

@@ -17,7 +17,7 @@

1717

1818from urllib import urlencode

1919from lxml import html

20−from yaml import load

20+from json import loads

2121import re

2222

2323# engine dependent config

@@ -36,6 +36,9 @@ safesearch_types = {2: 'STRICT',

3636 0: 'OFF'}

3737

3838

39+_quote_keys_regex = re.compile('({|,)([a-z][a-z0-9]*):(")', re.I | re.U)

40+

41+

3942# do search-request

4043def request(query, params):

4144 offset = (params['pageno'] - 1) * 10 + 1

@@ -65,22 +68,19 @@ def response(resp):

6568

6669 dom = html.fromstring(resp.text)

6770

68− # init regex for yaml-parsing

69− p = re.compile('({|,)([a-z]+):(")')

70−

7171 # parse results

7272 for result in dom.xpath('//div[@class="dg_u"]'):

7373 link = result.xpath('./a')[0]

7474

75− # parse yaml-data (it is required to add a space, to make it parsable)

76− yaml_data = load(p.sub(r'\1\2: \3', link.attrib.get('m')))

75+ # parse json-data (it is required to add a space, to make it parsable)

76+ json_data = loads(_quote_keys_regex.sub(r'\1"\2": \3', link.attrib.get('m')))

7777

7878 title = link.attrib.get('t1')

7979 ihk = link.attrib.get('ihk')

8080

8181 # url = 'http://' + link.attrib.get('t3')

82− url = yaml_data.get('surl')

83− img_src = yaml_data.get('imgurl')

82+ url = json_data.get('surl')

83+ img_src = json_data.get('imgurl')

8484

8585 # append result

8686 results.append({'template': 'images.html',

tests/unit/engines/test_bing_images.py+7−13

@@ -39,16 +39,10 @@ class TestBingImagesEngine(SearxTestCase):

3939

4040 html = """

4141 <div class="dg_u" style="width:178px;height:144px;left:17px;top:0px">

42− <a href="#" ihk="HN.608003696942779811"

43− m="{ns:"images",k:"5045",

44−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

45−surl:"http://www.page.url/",imgurl:"http://test.url/Test%20Query.jpg",

46−oh:"238",tft:"0",oi:"http://www.image.url/Images/Test%20Query.jpg"}"

47− mid="59EB92C317974F34517A1CCAEBEF76A578E08DEE" onclick="return false;"

48− t1="Test Query" t2="650 x 517 · 31 kB · jpeg" t3="www.short.url" h="ID=images,5045.1">

49− <img src="https://tse4.mm.bing.net/th?id=HN.608003696942779811&o=4&pid=1.7"

50− style="height:144px;" width="178" height="144"/>

42+ <a href="/images/search?q=south&view=detailv2&&id=7E92863981CCFB89FBDD55205C742DFDA3290CF6&selectedIndex=9&ccid=vzvIfv5u&simid=608055786735667000&thid=OIP.Mbf3bc87efe6e0e476be8cc34bf6cd80eH0" ihk="OIP.Mbf3bc87efe6e0e476be8cc34bf6cd80eH0" t1="South Carolina" t2="747 x 589 · 29 kB · 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">

43+ <img class="img_hid" src2="https://tse4.mm.bing.net/th?id=OIP.Mbf3bc87efe6e0e476be8cc34bf6cd80eH0&w=210&h=154&c=7&rs=1&qlt=90&o=4&pid=1.1" style="width:210px;height:154px;" width="210" height="154">

5144 </a>

45+

5246 </div>

5347 """

5448 html = html.replace('\r\n', '').replace('\n', '').replace('\r', '')

@@ -56,11 +50,11 @@ oh:"238",tft:"0",oi:"http://www.image.url/Images/Test%2

5650 results = bing_images.response(response)

5751 self.assertEqual(type(results), list)

5852 self.assertEqual(len(results), 1)

59− self.assertEqual(results[0]['title'], 'Test Query')

60− self.assertEqual(results[0]['url'], 'http://www.page.url/')

53+ self.assertEqual(results[0]['title'], 'South Carolina')

54+ self.assertEqual(results[0]['url'], 'http://www.digital-topo-maps.com/county-map/south-carolina.shtml')

6155 self.assertEqual(results[0]['content'], '')

62− self.assertEqual(results[0]['thumbnail_src'], 'https://www.bing.com/th?id=HN.608003696942779811')

63− self.assertEqual(results[0]['img_src'], 'http://test.url/Test%20Query.jpg')

56+ self.assertEqual(results[0]['thumbnail_src'], 'https://www.bing.com/th?id=OIP.Mbf3bc87efe6e0e476be8cc34bf6cd80eH0')

57+ self.assertEqual(results[0]['img_src'], 'http://www.digital-topo-maps.com/county-map/south-carolina-county-map.gif')

6458

6559 html = """

6660 <a href="#" ihk="HN.608003696942779811"