39 mediaunit_elts = description.getElementsByTagName('mediaUnit') |
39 mediaunit_elts = description.getElementsByTagName('mediaUnit') |
40 for mediaunit in mediaunit_elts: |
40 for mediaunit in mediaunit_elts: |
41 has_native_content_elts = mediaunit.getElementsByTagName('wlr:hasNativeContent') |
41 has_native_content_elts = mediaunit.getElementsByTagName('wlr:hasNativeContent') |
42 for has_native_content in has_native_content_elts: |
42 for has_native_content in has_native_content_elts: |
43 a_url_image = has_native_content.childNodes[0].data |
43 a_url_image = has_native_content.childNodes[0].data |
44 print 'a_url_image' |
|
45 print a_url_image |
|
46 break |
44 break |
47 break |
45 break |
48 |
46 |
49 wls_score_elts = description.getElementsByTagName(self._ns + ':hasScore') |
47 wls_score_elts = description.getElementsByTagName(self._ns + ':hasScore') |
50 atitle = '' |
48 atitle = '' |
51 for title in dc_title_elts: |
49 for title in dc_title_elts: |
52 atitle = title.childNodes[0].data |
50 atitle = title.childNodes[0].data |
53 a_score = '0' |
51 a_score = '0' |
54 for wls_score in wls_score_elts: |
52 for wls_score in wls_score_elts: |
55 a_score = wls_score.childNodes[0].data |
53 a_score = wls_score.childNodes[0].data |
56 print 'a_score' |
54 |
57 print a_score |
55 |
58 if atitle != '': |
56 if atitle != '': |
59 has_description_elts = description.getElementsByTagName(self._ns + ':hasDescription') |
57 has_description_elts = description.getElementsByTagName(self._ns + ':hasDescription') |
60 cluster_abstract = '' |
58 cluster_abstract = '' |
61 for has_description in has_description_elts: |
59 for has_description in has_description_elts: |
62 cluster_abstract = has_description.childNodes[0].data |
60 cluster_abstract = has_description.childNodes[0].data |
63 concept_list.append({'about':about,'title': atitle, 'abstract':cluster_abstract, 'url_image':a_url_image, 'score':a_score}) |
61 concept_list.append({'about':about,'title': atitle, 'abstract':cluster_abstract, 'url_image':a_url_image, 'score':a_score}) |
64 logger.info('concept_list') |
|
65 logger.info(concept_list) |
|
66 return concept_list |
62 return concept_list |
67 |
63 |
68 def get_documents_concepts_(self, dom): |
64 def get_documents_concepts_(self, dom): |
69 concepts_with_documents_list = [] |
65 concepts_with_documents_list = [] |
70 for node in dom.getElementsByTagName('resultSet'): |
66 for node in dom.getElementsByTagName('resultSet'): |
85 mediaunit_elts = description.getElementsByTagName('mediaUnit') |
81 mediaunit_elts = description.getElementsByTagName('mediaUnit') |
86 for mediaunit in mediaunit_elts: |
82 for mediaunit in mediaunit_elts: |
87 has_native_content_elts = mediaunit.getElementsByTagName('wlr:hasNativeContent') |
83 has_native_content_elts = mediaunit.getElementsByTagName('wlr:hasNativeContent') |
88 for has_native_content in has_native_content_elts: |
84 for has_native_content in has_native_content_elts: |
89 img_internal_path = has_native_content.childNodes[0].data |
85 img_internal_path = has_native_content.childNodes[0].data |
90 print 'img_internal_path' |
|
91 print img_internal_path |
|
92 break |
86 break |
93 break |
87 break |
94 |
88 |
95 dc_identifier_elts = description.getElementsByTagName('dc:identifier') |
89 dc_identifier_elts = description.getElementsByTagName('dc:identifier') |
96 document_id = dc_identifier_elts[0].childNodes[0].data |
90 document_id = dc_identifier_elts[0].childNodes[0].data |
118 img_internal_path = has_exposed_content_elts[0].childNodes[0].data''' |
112 img_internal_path = has_exposed_content_elts[0].childNodes[0].data''' |
119 |
113 |
120 concept_documents_list.append({'id':document_id, 'score':score, 'rank':rank, 'isLinkedTo':isLinkedTo, 'image_path':img_internal_path, 'abstract':abstract}) |
114 concept_documents_list.append({'id':document_id, 'score':score, 'rank':rank, 'isLinkedTo':isLinkedTo, 'image_path':img_internal_path, 'abstract':abstract}) |
121 |
115 |
122 concepts_with_documents_list.append(concept_documents_list) |
116 concepts_with_documents_list.append(concept_documents_list) |
123 logger.info('concepts_with_documents_list') |
|
124 logger.info(concepts_with_documents_list) |
|
125 return concepts_with_documents_list |
117 return concepts_with_documents_list |
126 |
118 |
127 def get_document_ids_concepts_(self, dom, list_links_concepts): |
119 def get_document_ids_concepts_(self, dom, list_links_concepts): |
128 document_ids_concepts_list = [] |
120 document_ids_concepts_list = [] |
129 for list_links_of_a_concept in list_links_concepts: |
121 for list_links_of_a_concept in list_links_concepts: |