From 1a82ed6f54a1e529485943d3f018b589d9b3aefc Mon Sep 17 00:00:00 2001 From: Adam Tauber Date: Tue, 8 Sep 2015 22:58:09 +0200 Subject: [PATCH 1/3] [mod] 500px rewrite --- searx/engines/www500px.py | 43 +++++++++++++++++++++++---------------- 1 file changed, 26 insertions(+), 17 deletions(-) diff --git a/searx/engines/www500px.py b/searx/engines/www500px.py index f1bc6c583..c2a50a696 100644 --- a/searx/engines/www500px.py +++ b/searx/engines/www500px.py @@ -12,12 +12,10 @@ @todo rewrite to api """ - +from json import loads from urllib import urlencode from urlparse import urljoin -from lxml import html -import re -from searx.engines.xpath import extract_text +from xml.sax.saxutils import escape # engine dependent config categories = ['images'] @@ -25,13 +23,27 @@ paging = True # search-url base_url = 'https://500px.com' -search_url = base_url + '/search?search?page={pageno}&type=photos&{query}' +search_url = 'https://api.500px.com/v1/photos/search?type=photos'\ + '&{query}'\ + '&image_size%5B%5D=4'\ + '&image_size%5B%5D=20'\ + '&image_size%5B%5D=21'\ + '&image_size%5B%5D=1080'\ + '&image_size%5B%5D=1600'\ + '&image_size%5B%5D=2048'\ + '&include_states=true'\ + '&formats=jpeg%2Clytro'\ + '&include_tags=true'\ + '&exclude_nude=true'\ + '&page={pageno}'\ + '&rpp=50'\ + '&sdk_key=b68e60cff4c929bedea36ca978830c5caca790c3' # do search-request def request(query, params): params['url'] = search_url.format(pageno=params['pageno'], - query=urlencode({'q': query})) + query=urlencode({'term': query})) return params @@ -40,19 +52,16 @@ def request(query, params): def response(resp): results = [] - dom = html.fromstring(resp.text) - regex = re.compile(r'3\.jpg.*$') + response_json = loads(resp.text) # parse results - for result in dom.xpath('//div[@class="photo"]'): - link = result.xpath('.//a')[0] - url = urljoin(base_url, link.attrib.get('href')) - title = extract_text(result.xpath('.//div[@class="title"]')) - thumbnail_src = link.xpath('.//img')[0].attrib.get('src') - # To have a bigger thumbnail, uncomment the next line - # thumbnail_src = regex.sub('4.jpg', thumbnail_src) - content = extract_text(result.xpath('.//div[@class="info"]')) - img_src = regex.sub('2048.jpg', thumbnail_src) + for result in response_json['photos']: + url = urljoin(base_url, result['url']) + title = escape(result['name']) + # last index is the biggest resolution + img_src = result['image_url'][-1] + thumbnail_src = result['image_url'][0] + content = escape(result['description'] or '') # append result results.append({'url': url, From 4676e1dcdee2b5039c8519b544e0211e718398d9 Mon Sep 17 00:00:00 2001 From: Adam Tauber Date: Sun, 11 Dec 2016 03:24:26 +0100 Subject: [PATCH 2/3] [fix] remove result escaping from engine --- searx/engines/www500px.py | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/searx/engines/www500px.py b/searx/engines/www500px.py index c2a50a696..546521ba3 100644 --- a/searx/engines/www500px.py +++ b/searx/engines/www500px.py @@ -15,7 +15,6 @@ from json import loads from urllib import urlencode from urlparse import urljoin -from xml.sax.saxutils import escape # engine dependent config categories = ['images'] @@ -57,11 +56,11 @@ def response(resp): # parse results for result in response_json['photos']: url = urljoin(base_url, result['url']) - title = escape(result['name']) + title = result['name'] # last index is the biggest resolution img_src = result['image_url'][-1] thumbnail_src = result['image_url'][0] - content = escape(result['description'] or '') + content = result['description'] or '' # append result results.append({'url': url, From 9b2bd6847f707e76d10e1ff8b80d68c56b4956f3 Mon Sep 17 00:00:00 2001 From: Adam Tauber Date: Sun, 11 Dec 2016 03:30:13 +0100 Subject: [PATCH 3/3] [fix] 500px unit tests --- tests/unit/engines/test_www500px.py | 63 ++++------------------------- 1 file changed, 7 insertions(+), 56 deletions(-) diff --git a/tests/unit/engines/test_www500px.py b/tests/unit/engines/test_www500px.py index 8df15b945..e50601da0 100644 --- a/tests/unit/engines/test_www500px.py +++ b/tests/unit/engines/test_www500px.py @@ -22,62 +22,13 @@ class TestWww500pxImagesEngine(SearxTestCase): self.assertRaises(AttributeError, www500px.response, '') self.assertRaises(AttributeError, www500px.response, '[]') - response = mock.Mock(text='') - self.assertEqual(www500px.response(response), []) - - html = """ -
- - - -
-
-
- - This is the title - -
-
- - This is the content - -
-
44.8
-
-
-
- """ - response = mock.Mock(text=html) + json = """ +{"current_page":1,"total_pages":1000,"total_items":862178,"photos":[{"id":64531569,"user_id":111147,"name":"Grand Canyon Afternoon","description":"Looking west on a very windy winter afternoon.","camera":"Canon EOS 5D Mark II","lens":"EF24-105mm f/4L IS USM","focal_length":"28","iso":"200","shutter_speed":"1/1250","aperture":"6.3","times_viewed":4809,"rating":48.5,"status":1,"created_at":"2014-03-22T03:44:46-04:00","category":8,"location":null,"latitude":36.0323916666667,"longitude":-111.85273,"taken_at":"2014-02-27T14:10:43-05:00","hi_res_uploaded":2,"for_sale":true,"width":5476,"height":3651,"votes_count":108,"favorites_count":35,"comments_count":5,"nsfw":false,"sales_count":0,"for_sale_date":null,"highest_rating":91.9,"highest_rating_date":"2014-03-22T22:34:54-04:00","license_type":0,"converted":31,"collections_count":10,"crop_version":0,"privacy":false,"profile":true,"image_url":["https://drscdn.500px.org/photo/64531569/w%3D70_h%3D70/449d50817f28d85395e23bbb415b3cdb?v=0","https://drscdn.500px.org/photo/64531569/q%3D50_w%3D140_h%3D140/3e3e123734a596644ede78105268bdb2?v=0","https://drscdn.500px.org/photo/64531569/q%3D80_h%3D300/2ce2f61714aebdca710967dfdc3efb04","https://drscdn.500px.org/photo/64531569/q%3D80_h%3D450/c8ec030441f2c68b9bd40a114903348a","https://drscdn.500px.org/photo/64531569/q%3D80_h%3D600/ab6562d0581b359679ecc8ef2e939396","https://drscdn.500px.org/photo/64531569/q%3D80_m%3D1000/bd7dbc54a505e041a8c9a70dfa434272","https://drscdn.500px.org/photo/64531569/q%3D80_m%3D1500/eb4d7f8f6a32d3e5c168c2cb55d29c12","https://drscdn.500px.org/photo/64531569/q%3D80_m%3D2000/d519f91b8a568e7357a8a7fa1aabbe74","https://drscdn.500px.org/photo/64531569/m%3D2048/4c52fb18cc2b2b6f91a0d04609786507","https://drscdn.500px.org/photo/64531569/m%3D900/fb620ae39569ab4a421e9170a94b1a0f","https://drscdn.500px.org/photo/64531569/m%3D900_s%3D1_k%3D1_a%3D1/02b95ce64db090c1f94f890960974612?v=0"],"images":[{"size":1,"url":"https://drscdn.500px.org/photo/64531569/w%3D70_h%3D70/449d50817f28d85395e23bbb415b3cdb?v=0","https_url":"https://drscdn.500px.org/photo/64531569/w%3D70_h%3D70/449d50817f28d85395e23bbb415b3cdb?v=0","format":"jpeg"},{"size":2,"url":"https://drscdn.500px.org/photo/64531569/q%3D50_w%3D140_h%3D140/3e3e123734a596644ede78105268bdb2?v=0","https_url":"https://drscdn.500px.org/photo/64531569/q%3D50_w%3D140_h%3D140/3e3e123734a596644ede78105268bdb2?v=0","format":"jpeg"},{"size":4,"url":"https://drscdn.500px.org/photo/64531569/m%3D900/fb620ae39569ab4a421e9170a94b1a0f","https_url":"https://drscdn.500px.org/photo/64531569/m%3D900/fb620ae39569ab4a421e9170a94b1a0f","format":"jpeg"},{"size":14,"url":"https://drscdn.500px.org/photo/64531569/m%3D900_s%3D1_k%3D1_a%3D1/02b95ce64db090c1f94f890960974612?v=0","https_url":"https://drscdn.500px.org/photo/64531569/m%3D900_s%3D1_k%3D1_a%3D1/02b95ce64db090c1f94f890960974612?v=0","format":"jpeg"},{"size":31,"url":"https://drscdn.500px.org/photo/64531569/q%3D80_h%3D450/c8ec030441f2c68b9bd40a114903348a","https_url":"https://drscdn.500px.org/photo/64531569/q%3D80_h%3D450/c8ec030441f2c68b9bd40a114903348a","format":"jpeg"},{"size":32,"url":"https://drscdn.500px.org/photo/64531569/q%3D80_h%3D300/2ce2f61714aebdca710967dfdc3efb04","https_url":"https://drscdn.500px.org/photo/64531569/q%3D80_h%3D300/2ce2f61714aebdca710967dfdc3efb04","format":"jpeg"},{"size":33,"url":"https://drscdn.500px.org/photo/64531569/q%3D80_h%3D600/ab6562d0581b359679ecc8ef2e939396","https_url":"https://drscdn.500px.org/photo/64531569/q%3D80_h%3D600/ab6562d0581b359679ecc8ef2e939396","format":"jpeg"},{"size":34,"url":"https://drscdn.500px.org/photo/64531569/q%3D80_m%3D1000/bd7dbc54a505e041a8c9a70dfa434272","https_url":"https://drscdn.500px.org/photo/64531569/q%3D80_m%3D1000/bd7dbc54a505e041a8c9a70dfa434272","format":"jpeg"},{"size":35,"url":"https://drscdn.500px.org/photo/64531569/q%3D80_m%3D1500/eb4d7f8f6a32d3e5c168c2cb55d29c12","https_url":"https://drscdn.500px.org/photo/64531569/q%3D80_m%3D1500/eb4d7f8f6a32d3e5c168c2cb55d29c12","format":"jpeg"},{"size":36,"url":"https://drscdn.500px.org/photo/64531569/q%3D80_m%3D2000/d519f91b8a568e7357a8a7fa1aabbe74","https_url":"https://drscdn.500px.org/photo/64531569/q%3D80_m%3D2000/d519f91b8a568e7357a8a7fa1aabbe74","format":"jpeg"},{"size":2048,"url":"https://drscdn.500px.org/photo/64531569/m%3D2048/4c52fb18cc2b2b6f91a0d04609786507","https_url":"https://drscdn.500px.org/photo/64531569/m%3D2048/4c52fb18cc2b2b6f91a0d04609786507","format":"jpeg"}],"url":"/photo/64531569/grand-canyon-afternoon-by-todd-hakala","positive_votes_count":108,"converted_bits":31,"tags":["landscape","river","arizona","canyon","grand","colorado","south","southwest","az","west","rim","CanonGetaway"],"watermark":false,"image_format":"jpeg","licensing_requested":false,"licensing_suggested":false,"is_free_photo":false,"user":{"id":111147,"username":"ToddHakala","firstname":"Todd","lastname":"Hakala","city":"Albuquerque","country":"US","usertype":0,"fullname":"Todd Hakala","userpic_url":"https://pacdn.500px.org/111147/ea167926a64ce9b32e44cbec61e3af4f75b762cb/1.jpg?2","userpic_https_url":"https://pacdn.500px.org/111147/ea167926a64ce9b32e44cbec61e3af4f75b762cb/1.jpg?2","cover_url":"https://pacdn.500px.org/111147/ea167926a64ce9b32e44cbec61e3af4f75b762cb/cover_2048.jpg?8","upgrade_status":1,"store_on":true,"affection":5217,"avatars":{"default":{"https":"https://pacdn.500px.org/111147/ea167926a64ce9b32e44cbec61e3af4f75b762cb/1.jpg?2"},"large":{"https":"https://pacdn.500px.org/111147/ea167926a64ce9b32e44cbec61e3af4f75b762cb/2.jpg?2"},"small":{"https":"https://pacdn.500px.org/111147/ea167926a64ce9b32e44cbec61e3af4f75b762cb/3.jpg?2"},"tiny":{"https":"https://pacdn.500px.org/111147/ea167926a64ce9b32e44cbec61e3af4f75b762cb/4.jpg?2"}},"followers_count":171}}]} + """ # noqa + response = mock.Mock(text=json) results = www500px.response(response) self.assertEqual(type(results), list) self.assertEqual(len(results), 1) - self.assertEqual(results[0]['title'], 'This is the title') - self.assertEqual(results[0]['url'], 'https://500px.com/this.should.be.the.url') - self.assertEqual(results[0]['content'], 'This is the content') - self.assertEqual(results[0]['thumbnail_src'], 'https://image.url/3.jpg?v=0') - self.assertEqual(results[0]['img_src'], 'https://image.url/2048.jpg') - - html = """ - - - -
-
-
- - This is the title - -
-
- - Oliver Turpin - -
-
44.8
-
-
- """ - response = mock.Mock(text=html) - results = www500px.response(response) - self.assertEqual(type(results), list) - self.assertEqual(len(results), 0) + self.assertEqual(results[0]['title'], u'Grand Canyon Afternoon') + self.assertEqual(results[0]['url'], 'https://500px.com/photo/64531569/grand-canyon-afternoon-by-todd-hakala') + self.assertEqual(results[0]['content'], u'Looking west on a very windy winter afternoon.')