1from django_elasticsearch_dsl import Document, fields 

2from django_elasticsearch_dsl.registries import registry 

3 

4from discuss_data.core.models import KeywordTagged, LanguageTagged 

5from discuss_data.dddatasets.models import ( 

6 Category, 

7 DataSet, 

8 DataSetCreator, 

9 CollectionMethodsTagged, 

10 AnalysisMethodsTagged, 

11 DisciplinesTagged, 

12) 

13 

14from discuss_data.ddusers.models import Affiliation, Country, User 

15 

16 

17@registry.register_document 

18class DataSetDocument(Document): 

19 countries = fields.ObjectField(properties={"name": fields.TextField()}) 

20 published_categories = fields.ObjectField(properties={"name": fields.TextField()}) 

21 published_main_category = fields.ObjectField( 

22 properties={"name": fields.TextField()} 

23 ) 

24 owner = fields.ObjectField( 

25 properties={"first_name": fields.TextField(), "last_name": fields.TextField()} 

26 ) 

27 dataset_creator = fields.ObjectField(properties={"name": fields.TextField()}) 

28 keywords = fields.ObjectField(properties={"name": fields.TextField()}) 

29 languages_of_data = fields.ObjectField(properties={"name": fields.TextField()}) 

30 methods_of_data_collection = fields.ObjectField( 

31 properties={"name": fields.TextField()} 

32 ) 

33 methods_of_data_analysis = fields.ObjectField( 

34 properties={"name": fields.TextField()} 

35 ) 

36 disciplines = fields.ObjectField(properties={"name": fields.TextField()}) 

37 # BleachField() instances are not recognized by django-elasticsearch-dsl 

38 # and need to be accessed as model attributes and added as a TextField() 

39 description = fields.TextField(attr="description") 

40 related_dataset_text = fields.TextField(attr="related_dataset_text") 

41 sources_of_data = fields.TextField(attr="sources_of_data") 

42 funding = fields.TextField(attr="funding") 

43 institutional_affiliation = fields.TextField(attr="institutional_affiliation") 

44 

45 class Index: 

46 # Name of the Elasticsearch index 

47 name = "datasets" 

48 # See Elasticsearch Indices API reference for available settings 

49 settings = {"number_of_shards": 1, "number_of_replicas": 0} 

50 

51 class Django: 

52 model = DataSet # The model associated with this Document 

53 

54 # The fields of the model you want to be indexed in Elasticsearch 

55 fields = [ 

56 "title", 

57 "subtitle", 

58 "version", 

59 "time_period_text", 

60 "time_period_from", 

61 "time_period_to", 

62 ] 

63 related_models = [ 

64 Country, 

65 Category, 

66 User, 

67 DataSetCreator, 

68 KeywordTagged, 

69 CollectionMethodsTagged, 

70 AnalysisMethodsTagged, 

71 DisciplinesTagged, 

72 ] 

73 

74 def get_instances_from_related(self, related_instance): 

75 """ If related_models is set, define how to retrieve the instance(s) from the related model. 

76 The related_models option should be used with caution because it can lead in the index 

77 to the updating of a lot of items. 

78 """ 

79 if isinstance(related_instance, Country): 

80 instance = related_instance.dataset_country.all() 

81 elif isinstance(related_instance, User): 

82 instance = related_instance.get_published_datasets() 

83 elif isinstance(related_instance, Category): 

84 instance = related_instance.get_published_datasets_category_all() 

85 elif isinstance(related_instance, KeywordTagged): 

86 instance = DataSet.objects.filter(keywords__name=related_instance.name) 

87 elif isinstance(related_instance, CollectionMethodsTagged): 

88 instance = DataSet.objects.filter( 

89 methods_of_data_collection__name=related_instance.name 

90 ) 

91 elif isinstance(related_instance, AnalysisMethodsTagged): 

92 instance = DataSet.objects.filter( 

93 methods_of_data_analysis__name=related_instance.name 

94 ) 

95 elif isinstance(related_instance, DisciplinesTagged): 

96 instance = DataSet.objects.filter(disciplines__name=related_instance.name) 

97 elif isinstance(related_instance, LanguageTagged): 

98 instance = DataSet.objects.filter( 

99 languages_of_data__name=related_instance.name 

100 ) 

101 elif isinstance(related_instance, DataSetCreator): 

102 instance = related_instance.dataset 

103 

104 return instance 

105 

106 def get_queryset(self): 

107 """ Mandatory for excluding instances from index by filtering by attributes. 

108 Works only for build_index command. 

109 Also to improve performance we can select related in one sql request 

110 """ 

111 return ( 

112 super(DataSetDocument, self) 

113 .get_queryset() 

114 .filter(published="True") 

115 .select_related("owner",) 

116 ) 

117 

118 @staticmethod 

119 def filter_instance(instance): 

120 """ return instance only if published is True""" 

121 if isinstance(instance, DataSet): 

122 if instance.published: 

123 return [instance] 

124 else: 

125 return [] 

126 else: 

127 return instance 

128 

129 def update(self, instance, refresh=None, action="index", **kwargs): 

130 """ Apply exclude from index filter also for index update by signal (eg instance changes)""" 

131 return super().update(self.filter_instance(instance), refresh, action, **kwargs) 

132 

133 

134@registry.register_document 

135class UserDocument(Document): 

136 # ObjectField works for ForeignKey relations using related_name 

137 # NestedField problematic in Search 

138 countries = fields.ObjectField(properties={"name": fields.TextField()}) 

139 interests = fields.ObjectField(properties={"name": fields.TextField()}) 

140 

141 affiliation_user = fields.ObjectField( 

142 properties={ 

143 "position": fields.TextField(), 

144 "name_of_institution": fields.TextField(), 

145 "place_of_institution": fields.TextField(), 

146 "country_of_institution": fields.TextField(), 

147 } 

148 ) 

149 

150 class Index: 

151 # Name of the Elasticsearch index 

152 name = "users" 

153 # See Elasticsearch Indices API reference for available settings 

154 settings = {"number_of_shards": 1, "number_of_replicas": 0} 

155 

156 class Django: 

157 model = User # The model associated with this Document 

158 

159 # The fields of the model you want to be indexed in Elasticsearch 

160 fields = [ 

161 "first_name", 

162 "last_name", 

163 ] 

164 related_models = [ 

165 Country, 

166 KeywordTagged, 

167 Affiliation, 

168 ] 

169 

170 def get_instances_from_related(self, related_instance): 

171 """If related_models is set, define how to retrieve the instance(s) from the related model. 

172 The related_models option should be used with caution because it can lead in the index 

173 to the updating of a lot of items. 

174 """ 

175 if isinstance(related_instance, Country): 

176 return related_instance.user_country.all() 

177 elif isinstance(related_instance, KeywordTagged): 

178 return User.objects.filter(interests__name=related_instance.name) 

179 elif isinstance(related_instance, Affiliation): 

180 return related_instance.user 

181 

182 def get_queryset(self): 

183 """ Mandatory for excluding instances from index by filtering by attributes. 

184 Works only for build_index command. 

185 Also to improve performance we can select related in one sql request 

186 """ 

187 return ( 

188 super(UserDocument, self) 

189 .get_queryset() 

190 .exclude(profile_accessibility="HID") 

191 .exclude(is_active=False) 

192 .exclude(first_name="") 

193 .exclude(last_name="") 

194 .exclude(username="AnonymousUser") 

195 ) 

196 

197 @staticmethod 

198 def filter_instance(instance): 

199 """filter hidden and internally used user accounts""" 

200 if isinstance(instance, User): 

201 if ( 

202 instance.profile_accessibility == "HID" 

203 or not instance.is_active 

204 or instance.first_name == "" 

205 or instance.last_name == "" 

206 or instance.username == "AnonymousUser" 

207 ): 

208 return [] 

209 else: 

210 return [instance] 

211 else: 

212 return instance 

213 

214 def update(self, instance, refresh=None, action="index", **kwargs): 

215 """ Apply exclude from index filter also for index update by signal (eg instance changes)""" 

216 return super().update(self.filter_instance(instance), refresh, action, **kwargs)