logo

searx

My custom branche(s) on searx, a meta-search engine git clone https://hacktivis.me/git/searx.git
commit: 8f3b33de23090e6d11a5a29a239719bb6915d2ec
parent 4cea71e3bb34020dff0f3e28f5c1398c0b0d8278
Author: a01200356 <a01200356@itesm.mx>
Date:   Sun, 28 Feb 2016 02:05:52 -0600

[fix] remove unnecesary async calls in wolframalpha_noapi

setting async to false in the request did the job, lol.

Diffstat:

Msearx/engines/wolframalpha_noapi.py56++------------------------------------------------------
Mtests/unit/engines/test_wolframalpha_api.py32++++++++++++++++----------------
Mtests/unit/engines/test_wolframalpha_noapi.py72++++++++++++++++++++++++++++--------------------------------------------
3 files changed, 46 insertions(+), 114 deletions(-)

diff --git a/searx/engines/wolframalpha_noapi.py b/searx/engines/wolframalpha_noapi.py @@ -20,7 +20,7 @@ from searx.poolrequests import get as http_get url = 'https://www.wolframalpha.com/' search_url = url + 'input/json.jsp'\ - '?async=true'\ + '?async=false'\ '&banners=raw'\ '&debuggingdata=false'\ '&format=image,plaintext,imagemap,minput,moutput'\ @@ -38,14 +38,6 @@ referer_url = url + 'input/?{query}' token = {'value': '', 'last_updated': None} -# xpath variables -success_xpath = '/pod[attribute::error="false"]' -plaintext_xpath = './plaintext' -title_xpath = './@title' -image_xpath = './img' -img_src_xpath = './img/@src' -img_alt_xpath = './img/@alt' - # pods to display as image in infobox # this pods do return a plaintext, but they look better and are more useful as images image_pods = {'VisualRepresentation', @@ -79,42 +71,6 @@ def request(query, params): return params -# get additional pod -# NOTE: this makes an additional requests to server, so the response will take longer and might reach timeout -def get_async_pod(url): - try: - resp = http_get(url, timeout=2.0) - except: - return None - - if resp: - return parse_async_pod(resp) - - -def parse_async_pod(resp): - pod = {'subpods': []} - - resp_pod = XML(resp.content) - - if resp_pod.xpath(success_xpath): - for subpod in resp_pod: - new_subpod = {'title': subpod.xpath(title_xpath)[0]} - - plaintext = subpod.xpath(plaintext_xpath)[0].text - if plaintext: - new_subpod['plaintext'] = plaintext - else: - new_subpod['plaintext'] = '' - - if subpod.xpath(image_xpath): - new_subpod['img'] = {'src': subpod.xpath(img_src_xpath)[0], - 'alt': subpod.xpath(img_alt_xpath)[0]} - - pod['subpods'].append(new_subpod) - - return pod - - # get response from search-request def response(resp): results = [] @@ -132,15 +88,7 @@ def response(resp): pod_title = pod.get('title', '') if 'subpods' not in pod: - # comment this section if your requests always reach timeout - if pod['async']: - result = get_async_pod(pod['async']) - if result: - pod = result - else: - continue - else: - continue + continue if pod_id == 'Input' or not infobox_title: infobox_title = pod['subpods'][0]['plaintext'] diff --git a/tests/unit/engines/test_wolframalpha_api.py b/tests/unit/engines/test_wolframalpha_api.py @@ -87,16 +87,16 @@ class TestWolframAlphaAPIEngine(SearxTestCase): results = wolframalpha_api.response(response) self.assertEqual(type(results), list) self.assertEqual(len(results), 2) - self.assertIn('input_plaintext', results[0]['infobox']) + self.assertEqual('input_plaintext', results[0]['infobox']) self.assertEqual(len(results[0]['attributes']), 3) - self.assertIn('Input', results[0]['attributes'][0]['label']) - self.assertIn('input_plaintext', results[0]['attributes'][0]['value']) - self.assertIn('Result', results[0]['attributes'][1]['label']) - self.assertIn('result_plaintext', results[0]['attributes'][1]['value']) - self.assertIn('Manipulatives illustration', results[0]['attributes'][2]['label']) - self.assertIn('illustration_img_src.gif', results[0]['attributes'][2]['image']['src']) - self.assertIn('illustration_img_alt', results[0]['attributes'][2]['image']['alt']) + self.assertEqual('Input', results[0]['attributes'][0]['label']) + self.assertEqual('input_plaintext', results[0]['attributes'][0]['value']) + self.assertEqual('Result', results[0]['attributes'][1]['label']) + self.assertEqual('result_plaintext', results[0]['attributes'][1]['value']) + self.assertEqual('Manipulatives illustration', results[0]['attributes'][2]['label']) + self.assertEqual('illustration_img_src.gif', results[0]['attributes'][2]['image']['src']) + self.assertEqual('illustration_img_alt', results[0]['attributes'][2]['image']['alt']) self.assertEqual(len(results[0]['urls']), 1) @@ -129,8 +129,8 @@ class TestWolframAlphaAPIEngine(SearxTestCase): <plaintext>integral_plaintext</plaintext> </subpod> </pod> - <pod title='Plot' - scanner='Plotter' + <pod title='Plot of the integral' + scanner='Integral' id='Plot' error='false' numsubpods='1'> @@ -147,14 +147,14 @@ class TestWolframAlphaAPIEngine(SearxTestCase): results = wolframalpha_api.response(response) self.assertEqual(type(results), list) self.assertEqual(len(results), 2) - self.assertIn('integral_plaintext', results[0]['infobox']) + self.assertEqual('integral_plaintext', results[0]['infobox']) self.assertEqual(len(results[0]['attributes']), 2) - self.assertIn('Indefinite integral', results[0]['attributes'][0]['label']) - self.assertIn('integral_plaintext', results[0]['attributes'][0]['value']) - self.assertIn('Plot', results[0]['attributes'][1]['label']) - self.assertIn('plot.gif', results[0]['attributes'][1]['image']['src']) - self.assertIn('plot_alt', results[0]['attributes'][1]['image']['alt']) + self.assertEqual('Indefinite integral', results[0]['attributes'][0]['label']) + self.assertEqual('integral_plaintext', results[0]['attributes'][0]['value']) + self.assertEqual('Plot of the integral', results[0]['attributes'][1]['label']) + self.assertEqual('plot.gif', results[0]['attributes'][1]['image']['src']) + self.assertEqual('plot_alt', results[0]['attributes'][1]['image']['alt']) self.assertEqual(len(results[0]['urls']), 1) diff --git a/tests/unit/engines/test_wolframalpha_noapi.py b/tests/unit/engines/test_wolframalpha_noapi.py @@ -124,16 +124,16 @@ class TestWolframAlphaNoAPIEngine(SearxTestCase): results = wolframalpha_noapi.response(response) self.assertEqual(type(results), list) self.assertEqual(len(results), 2) - self.assertIn('input_plaintext', results[0]['infobox']) + self.assertEqual('input_plaintext', results[0]['infobox']) self.assertEqual(len(results[0]['attributes']), 3) - self.assertIn('Input', results[0]['attributes'][0]['label']) - self.assertIn('input_plaintext', results[0]['attributes'][0]['value']) - self.assertIn('Result', results[0]['attributes'][1]['label']) - self.assertIn('result_plaintext', results[0]['attributes'][1]['value']) - self.assertIn('Manipulatives illustration', results[0]['attributes'][2]['label']) - self.assertIn('illustration_img_src.gif', results[0]['attributes'][2]['image']['src']) - self.assertIn('illustration_img_alt', results[0]['attributes'][2]['image']['alt']) + self.assertEqual('Input', results[0]['attributes'][0]['label']) + self.assertEqual('input_plaintext', results[0]['attributes'][0]['value']) + self.assertEqual('Result', results[0]['attributes'][1]['label']) + self.assertEqual('result_plaintext', results[0]['attributes'][1]['value']) + self.assertEqual('Manipulatives illustration', results[0]['attributes'][2]['label']) + self.assertEqual('illustration_img_src.gif', results[0]['attributes'][2]['image']['src']) + self.assertEqual('illustration_img_alt', results[0]['attributes'][2]['image']['alt']) self.assertEqual(len(results[0]['urls']), 1) @@ -184,8 +184,19 @@ class TestWolframAlphaNoAPIEngine(SearxTestCase): ], "id" : "Plot", "error" : false, - "numsubpods" : 0, - "async" : "invalid_async_url" + "numsubpods" : 1, + "subpods" : [ + { + "title" : "", + "img" : { + "src" : "plot.gif", + "alt" : "plot_alt", + "title" : "plot_title" + }, + "plaintext" : "", + "minput" : "plot_minput" + } + ] } ] }} @@ -194,11 +205,14 @@ class TestWolframAlphaNoAPIEngine(SearxTestCase): results = wolframalpha_noapi.response(response) self.assertEqual(type(results), list) self.assertEqual(len(results), 2) - self.assertIn('integral_plaintext', results[0]['infobox']) + self.assertEqual('integral_plaintext', results[0]['infobox']) - self.assertEqual(len(results[0]['attributes']), 1) - self.assertIn('Indefinite integral', results[0]['attributes'][0]['label']) - self.assertIn('integral_plaintext', results[0]['attributes'][0]['value']) + self.assertEqual(len(results[0]['attributes']), 2) + self.assertEqual('Indefinite integral', results[0]['attributes'][0]['label']) + self.assertEqual('integral_plaintext', results[0]['attributes'][0]['value']) + self.assertEqual('Plot of the integral', results[0]['attributes'][1]['label']) + self.assertEqual('plot.gif', results[0]['attributes'][1]['image']['src']) + self.assertEqual('plot_alt', results[0]['attributes'][1]['image']['alt']) self.assertEqual(len(results[0]['urls']), 1) @@ -206,33 +220,3 @@ class TestWolframAlphaNoAPIEngine(SearxTestCase): self.assertEqual('Wolfram|Alpha', results[0]['urls'][0]['title']) self.assertEqual(referer_url, results[1]['url']) self.assertEqual('Wolfram|Alpha', results[1]['title']) - - def test_parse_async_pod(self): - self.assertRaises(AttributeError, wolframalpha_noapi.parse_async_pod, None) - self.assertRaises(AttributeError, wolframalpha_noapi.parse_async_pod, []) - self.assertRaises(AttributeError, wolframalpha_noapi.parse_async_pod, '') - self.assertRaises(AttributeError, wolframalpha_noapi.parse_async_pod, '[]') - - # test plot - xml = '''<?xml version='1.0' encoding='UTF-8'?> - <pod title='Plot' - scanner='Plot' - id='Plot' - error='false' - numsubpods='1'> - <subpod title=''> - <img src='plot_img_src.gif' - alt='plot_img_alt' - title='plot_img_title' /> - <plaintext>plot_plaintext</plaintext> - <minput>plot_minput</minput> - </subpod> - </pod> - ''' - response = mock.Mock(content=xml) - pod = wolframalpha_noapi.parse_async_pod(response) - self.assertEqual(len(pod['subpods']), 1) - self.assertEqual('', pod['subpods'][0]['title']) - self.assertEqual('plot_plaintext', pod['subpods'][0]['plaintext']) - self.assertEqual('plot_img_src.gif', pod['subpods'][0]['img']['src']) - self.assertEqual('plot_img_alt', pod['subpods'][0]['img']['alt'])