def get_pubmed_load_config(self):
index_item = es_utils.get_info_for_index_id(ID_PUBMED)
pubmed_index = index_item['index']
pubmed_type = index_item['index_type']
load_config = LoadConfig()
load_config.root_directory = self.ct_load_config.root_directory
load_config.server = self.ct_load_config.server
load_config.index = pubmed_index
load_config.type = pubmed_type
load_config.data_extractor = PubmedDataExtractor()
load_config.data_mapper = PubmedDataMapper()
return load_config
def get_load_config(self):
load_config = LoadConfig()
load_config.root_directory = self.root_directory
load_config.process_count = psutil.cpu_count()
load_config.server = self.server
load_config.server_username = self.server_username
load_config.server_password = self.server_password
load_config.index = self.index
load_config.type = self.type
load_config.data_mapper = self.get_data_mapper()
load_config.data_extractor = self.get_data_extractor()
load_config.max_memory_percent = self.get_max_memory_percent()
return load_config
def create_load_config():
load_config = LoadConfig()
load_config.root_directory = ROOT_DIRECTORY
# load_config.data_source_name = 'extended_relations'
load_config.process_count = psutil.cpu_count()
load_config.server = LOCAL_SERVER
load_config.index = INDEX
load_config.type = TYPE
load_config.data_mapper = IRDBDataMapper()
load_config.data_extractor = IRDBDataExtractor()
# load_config.data_source_name = file_name.split('.')[0]
load_config.max_memory_percent = 75
return load_config
def get_load_config():
irdb_load_config = irdb_load_config_getter.get_load_config()
load_config = LoadConfig()
load_config.root_directory = irdb_load_config.root_directory
load_config.server = irdb_load_config.server
load_config.index = INDEX_MAPPING[ID_PUBMED]['index']
load_config.type = INDEX_MAPPING[ID_PUBMED]['type']
load_config.data_extractor = PubmedDataExtractor()
load_config.data_mapper = PubmedDataMapper()
# load_config.data_source_name = file_name.split('.')[0]
load_config.process_count = irdb_load_config.process_count
# load_config.log_level = LOG_LEVEL_TRACE
return load_config
def get_load_config():
load_config = LoadConfig()
load_config.root_directory = ROOT_DIRECTORY
load_config.server = SERVER
load_config.index = INDEX
load_config.type = TYPE
load_config.process_count = PROCESS_COUNT
load_config.bulk_data_size = BULK_DATA_SIZE
load_config.data_loader_batch_size = DATA_LOADER_BATCH_SIZE
load_config.data_source_batch_size = DATA_SOURCE_BATCH_SIZE
load_config.doc_fetch_batch_size = DOC_FETCH_BATCH_SIZE
# load_config.log_level = LOG_LEVEL_TRACE
load_config.data_extractor = CTDataExtractor()
load_config.data_mapper = CTDataMapper()
# load_config.data_source_name = file_name.split('.')[0]
load_config.max_memory_percent = 80
return load_config
citations = list(citations_set | added_citations_set)
if 'removed_citations' in update_history_item:
removed_citations = update_history_item[
'removed_citations']
citations = list(set(citations) - set(removed_citations))
return citations
load_config = LoadConfig()
load_config.root_directory = '/data/data_loading/pubmed_2019/pubmed2019/fix_citations'
# load_config.process_count = psutil.cpu_count()
load_config.server = 'http://localhost:9200'
load_config.server_username = ''
load_config.server_password = ''
load_config.index = "pubmed2019"
load_config.type = "article"
load_config.data_mapper = PubmedDataMapper()
load_config.data_extractor = PubmedDataExtractor()
load_config.max_memory_percent = 75
load_config.source = ""
load_config.append_relations = False
load_config.process_count = 4
fix_citations = FixCitations(load_config)