71de593 · searx · git.iko.soy

git.iko.soy/searx

[fix] bing images page change

commit71de593a5b9025e360da32813e9efeaeaa4992b2 authorAdam Tauber <asciimoo@gmail.com> dateSat 26 Mar 2016, 02:07 (11 years ago) parent6d55642

2 files changed+10−154

compare with master

unifiedsplit

searx/engines/bing_images.py+1−1 tests/unit/engines/test_bing_images.py+9−153

searx/engines/bing_images.py+1−1

@@ -69,7 +69,7 @@ def response(resp):

6969 dom = html.fromstring(resp.text)

7070

7171 # parse results

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

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

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

7474

7575 # parse json-data (it is required to add a space, to make it parsable)

tests/unit/engines/test_bing_images.py+9−153

@@ -38,12 +38,12 @@ class TestBingImagesEngine(SearxTestCase):

3838 self.assertEqual(bing_images.response(response), [])

3939

4040 html = """

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

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

4242 <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">

4343 <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">

4444 </a>

4545

46− </div>

46+ </div></div>

4747 """ # noqa

4848 html = html.replace('\r\n', '').replace('\n', '').replace('\r', '')

4949 response = mock.Mock(text=html)

@@ -78,7 +78,7 @@ class TestBingImagesEngine(SearxTestCase):

7878 self.assertEqual(len(results), 0)

7979

8080 html = """

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

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

8282 <a href="#" ihk="HN.608003696942779811"

8383 m="{ns:"images",k:"5045",

8484mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

@@ -89,8 +89,8 @@ oh:"238",tft:"0",oi:"http://www.image.url/Images/Test%2

8989 <img src="https://tse4.mm.bing.net/th?id=HN.608003696942779811&o=4&pid=1.7"

9090 style="height:144px;" width="178" height="144"/>

9191 </a>

92− </div>

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

92+ </div></div>

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

9494 <a href="#" ihk="HN.608003696942779811"

9595 m="{ns:"images",k:"5045",

9696mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

@@ -101,8 +101,8 @@ oh:"238",tft:"0",oi:"http://www.image.url/Images/Test%2

101101 <img src="https://tse4.mm.bing.net/th?id=HN.608003696942779811&o=4&pid=1.7"

102102 style="height:144px;" width="178" height="144"/>

103103 </a>

104− </div>

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

104+ </div></div>

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

106106 <a href="#" ihk="HN.608003696942779811"

107107 m="{ns:"images",k:"5045",

108108mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

@@ -113,154 +113,10 @@ oh:"238",tft:"0",oi:"http://www.image.url/Images/Test%2

113113 <img src="https://tse4.mm.bing.net/th?id=HN.608003696942779811&o=4&pid=1.7"

114114 style="height:144px;" width="178" height="144"/>

115115 </a>

116− </div>

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

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

119− m="{ns:"images",k:"5045",

120−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

127− </a>

128− </div>

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

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

131− m="{ns:"images",k:"5045",

132−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

139− </a>

140− </div>

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

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

143− m="{ns:"images",k:"5045",

144−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

151− </a>

152− </div>

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

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

155− m="{ns:"images",k:"5045",

156−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

163− </a>

164− </div>

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

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

167− m="{ns:"images",k:"5045",

168−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

175− </a>

176− </div>

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

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

179− m="{ns:"images",k:"5045",

180−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

187− </a>

188− </div>

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

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

191− m="{ns:"images",k:"5045",

192−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

199− </a>

200− </div>

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

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

203− m="{ns:"images",k:"5045",

204−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

211− </a>

212− </div>

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

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

215− m="{ns:"images",k:"5045",

216−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

223− </a>

224− </div>

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

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

227− m="{ns:"images",k:"5045",

228−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

235− </a>

236− </div>

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

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

239− m="{ns:"images",k:"5045",

240−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

247− </a>

248− </div>

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

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

251− m="{ns:"images",k:"5045",

252−mid:"659EB92C317974F34517A1CCAEBEF76A578E08DEE",

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

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

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

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

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

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

259− </a>

260− </div>

116+ </div></div>

261117 """

262118 html = html.replace('\r\n', '').replace('\n', '').replace('\r', '')

263119 response = mock.Mock(text=html)

264120 results = bing_images.response(response)

265121 self.assertEqual(type(results), list)

266− self.assertEqual(len(results), 10)

122+ self.assertEqual(len(results), 3)