2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								## Deviantart (Images)
							 | 
						
					
						
							
								
									
										
										
										
											2014-12-07 16:37:56 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								#
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								# @website     https://www.deviantart.com/
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								# @provide-api yes (https://www.deviantart.com/developers/) (RSS)
							 | 
						
					
						
							
								
									
										
										
										
											2014-12-07 16:37:56 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								#
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								# @using-api   no (TODO, rewrite to api)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								# @results     HTML
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								# @stable      no (HTML can change)
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 17:13:44 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								# @parse       url, title, thumbnail, img_src
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								#
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								# @todo        rewrite to api
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								from urllib import urlencode
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								from urlparse import urljoin
							 | 
						
					
						
							
								
									
										
										
										
											2014-02-05 20:24:31 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								from lxml import html
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								# engine dependent config
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								categories = ['images']
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								paging = True
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								# search-url
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								base_url = 'https://www.deviantart.com/'
							 | 
						
					
						
							
								
									
										
										
										
											2014-01-30 00:09:47 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								search_url = base_url+'search?offset={offset}&{query}'
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								# do search-request
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								def request(query, params):
							 | 
						
					
						
							
								
									
										
										
										
											2014-01-30 00:09:47 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    offset = (params['pageno'] - 1) * 24
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2014-01-30 00:09:47 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    params['url'] = search_url.format(offset=offset,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								                                      query=urlencode({'q': query}))
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    return params
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								# get response from search-request
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								def response(resp):
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    results = []
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    # return empty array if a redirection code is returned
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    if resp.status_code == 302:
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								        return []
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    dom = html.fromstring(resp.text)
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    # parse results
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    for result in dom.xpath('//div[contains(@class, "tt-a tt-fh")]'):
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								        link = result.xpath('.//a[contains(@class, "thumb")]')[0]
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								        url = urljoin(base_url, link.attrib.get('href'))
							 | 
						
					
						
							
								
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								        title_links = result.xpath('.//span[@class="details"]//a[contains(@class, "t")]')  # noqa
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								        title = ''.join(title_links[0].xpath('.//text()'))
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-22 23:34:45 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								        img_src = link.xpath('.//img')[0].attrib['src']
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								        # append result
							 | 
						
					
						
							
								
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								        results.append({'url': url,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								                        'title': title,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								                        'img_src': img_src,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								                        'template': 'images.html'})
							 | 
						
					
						
							
								
									
										
										
										
											2014-09-02 16:48:18 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    # return results
							 | 
						
					
						
							
								
									
										
										
										
											2013-10-20 11:12:10 +02:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    return results
							 |