| 
									
										
										
										
											2014-01-05 22:10:46 +01:00
										 |  |  | from urllib import urlencode | 
					
						
							|  |  |  | from HTMLParser import HTMLParser | 
					
						
							| 
									
										
										
										
											2014-01-06 22:15:46 +01:00
										 |  |  | from lxml import html | 
					
						
							| 
									
										
										
										
											2014-02-05 20:24:31 +01:00
										 |  |  | from xpath import extract_text | 
					
						
							| 
									
										
										
										
											2014-01-05 22:10:46 +01:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2014-01-06 22:15:46 +01:00
										 |  |  | base_url = 'http://vimeo.com' | 
					
						
							|  |  |  | search_url = base_url + '/search?{query}' | 
					
						
							| 
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 |  |  | url_xpath = None | 
					
						
							| 
									
										
										
										
											2014-01-11 11:14:46 +01:00
										 |  |  | content_xpath = None | 
					
						
							| 
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 |  |  | title_xpath = None | 
					
						
							| 
									
										
										
										
											2014-01-11 11:14:46 +01:00
										 |  |  | results_xpath = '' | 
					
						
							| 
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 |  |  | content_tpl = '<a href="{0}">  <img src="{2}"/> </a>' | 
					
						
							| 
									
										
										
										
											2014-01-05 22:10:46 +01:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 |  |  | # the cookie set by vimeo contains all the following values, | 
					
						
							|  |  |  | # but only __utma seems to be requiered | 
					
						
							| 
									
										
										
										
											2014-01-11 11:14:46 +01:00
										 |  |  | cookie = { | 
					
						
							| 
									
										
										
										
											2014-01-06 22:15:46 +01:00
										 |  |  |     #'vuid':'918282893.1027205400' | 
					
						
							|  |  |  |     # 'ab_bs':'%7B%223%22%3A279%7D' | 
					
						
							| 
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 |  |  |      '__utma': '00000000.000#0000000.0000000000.0000000000.0000000000.0' | 
					
						
							| 
									
										
										
										
											2014-01-06 22:15:46 +01:00
										 |  |  |     # '__utmb':'18302654.1.10.1388942090' | 
					
						
							|  |  |  |     #, '__utmc':'18302654' | 
					
						
							| 
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 |  |  |     #, '__utmz':'18#302654.1388942090.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none)'  # noqa | 
					
						
							| 
									
										
										
										
											2014-01-06 22:15:46 +01:00
										 |  |  |     #, '__utml':'search' | 
					
						
							| 
									
										
										
										
											2014-01-05 22:10:46 +01:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2014-01-05 22:10:46 +01:00
										 |  |  | def request(query, params): | 
					
						
							| 
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 |  |  |     params['url'] = search_url.format(query=urlencode({'q': query})) | 
					
						
							| 
									
										
										
										
											2014-01-11 11:14:46 +01:00
										 |  |  |     params['cookies'] = cookie | 
					
						
							| 
									
										
										
										
											2014-01-05 22:10:46 +01:00
										 |  |  |     return params | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2014-01-05 22:10:46 +01:00
										 |  |  | def response(resp): | 
					
						
							|  |  |  |     results = [] | 
					
						
							|  |  |  |     dom = html.fromstring(resp.text) | 
					
						
							| 
									
										
										
										
											2014-01-11 11:14:46 +01:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2014-01-06 22:15:46 +01:00
										 |  |  |     p = HTMLParser() | 
					
						
							| 
									
										
										
										
											2014-01-05 22:10:46 +01:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2014-01-06 22:15:46 +01:00
										 |  |  |     for result in dom.xpath(results_xpath): | 
					
						
							|  |  |  |         url = base_url + result.xpath(url_xpath)[0] | 
					
						
							|  |  |  |         title = p.unescape(extract_text(result.xpath(title_xpath))) | 
					
						
							| 
									
										
										
										
											2014-01-12 18:31:57 +01:00
										 |  |  |         thumbnail = extract_text(result.xpath(content_xpath)[0]) | 
					
						
							| 
									
										
										
										
											2014-01-20 02:31:20 +01:00
										 |  |  |         results.append({'url': url, | 
					
						
							|  |  |  |                         'title': title, | 
					
						
							|  |  |  |                         'content': content_tpl.format(url, title, thumbnail), | 
					
						
							|  |  |  |                         'template': 'videos.html', | 
					
						
							|  |  |  |                         'thumbnail': thumbnail}) | 
					
						
							| 
									
										
										
										
											2014-01-05 22:10:46 +01:00
										 |  |  |     return results |