Revision 46775
Added by Sandro La Bruzzo about 7 years ago
| DLIESConnector.py | ||
|---|---|---|
| 1 | 1 |
from elasticsearch import * |
| 2 | 2 |
from elasticsearch_dsl import * |
| 3 | 3 |
from typing import List |
| 4 |
from swagger_server.models import Scholix, ObjectType, DSType, IdentifierType, RelationshipType, ObjectType,ObjectTypeObjectType, CreatorType |
|
| 4 |
from swagger_server.models import Scholix, ObjectType, DSType, IdentifierType, RelationshipType, ObjectType, \ |
|
| 5 |
ObjectTypeObjectType, CreatorType |
|
| 5 | 6 |
|
| 6 | 7 |
|
| 7 |
|
|
| 8 | 8 |
class DLIESConnector(object): |
| 9 | 9 |
def __init__(self, index_host, index_name='dli'): |
| 10 | 10 |
self.index_host = index_host |
| ... | ... | |
| 12 | 12 |
self.index_name = index_name |
| 13 | 13 |
|
| 14 | 14 |
def create_pid_pidType_query(self, pidType, pid): |
| 15 |
args = {'target.identifier.schema': pidType}
|
|
| 16 |
args2 = {'target.identifier.identifier': pid}
|
|
| 17 |
return Q('nested', path='target.identifier', query=Q('bool', must=[Q('match', **args), Q('match', **args2)]))
|
|
| 15 |
args = {'source.identifier.schema': pidType}
|
|
| 16 |
args2 = {'source.identifier.identifier': pid}
|
|
| 17 |
return Q('nested', path='source.identifier', query=Q('bool', must=[Q('match', **args), Q('match', **args2)]))
|
|
| 18 | 18 |
|
| 19 |
def create_target_pid_query(self, value):
|
|
| 20 |
args = {'target.identifier.identifier': value}
|
|
| 21 |
return Q('nested', path='target.identifier', query=Q('bool', must=[Q('match', **args)]))
|
|
| 19 |
def create_source_pid_query(self, value):
|
|
| 20 |
args = {'source.identifier.identifier': value}
|
|
| 21 |
return Q('nested', path='source.identifier', query=Q('bool', must=[Q('match', **args)]))
|
|
| 22 | 22 |
|
| 23 | 23 |
def create_typology_query(self, value): |
| 24 | 24 |
args = {'target.objectType': value}
|
| ... | ... | |
| 30 | 30 |
|
| 31 | 31 |
def list_datasources(self): |
| 32 | 32 |
search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query()
|
| 33 |
search_object.aggs.bucket('all_datasources', 'nested', path='linkprovider').bucket('all_names', 'terms',field='linkprovider.name', size=100)
|
|
| 33 |
search_object.aggs.bucket('all_datasources', 'nested', path='linkprovider').bucket('all_names', 'terms',
|
|
| 34 |
field='linkprovider.name', |
|
| 35 |
size=100) |
|
| 34 | 36 |
|
| 35 | 37 |
response = search_object.execute() |
| 36 |
|
|
| 37 | 38 |
return [item.key for item in response.aggs.all_datasources.all_names.buckets] |
| 38 | 39 |
|
| 39 |
|
|
| 40 | 40 |
def convertLinkProvider(self, lp): |
| 41 | 41 |
result = [] |
| 42 | 42 |
for item in lp: |
| 43 |
current_identifiers =[] |
|
| 44 |
for ids in item['identifiers']:
|
|
| 43 |
current_identifiers = []
|
|
| 44 |
for ids in item['identifiers']: |
|
| 45 | 45 |
current_identifiers.append(IdentifierType(identifier=ids['identifier'], schema=ids['schema'])) |
| 46 |
tmp = DSType(name=item['name'], identifiers=[IdentifierType(x['identifier'], x['schema']) for x in item['identifiers']] ) |
|
| 46 |
tmp = DSType(name=item['name'], |
|
| 47 |
identifiers=[IdentifierType(x['identifier'], x['schema']) for x in item['identifiers']]) |
|
| 47 | 48 |
result.append(tmp) |
| 48 | 49 |
return result |
| 49 | 50 |
|
| 50 | 51 |
def convertObject(self, inputObj): |
| 51 | 52 |
result = ObjectType() |
| 52 |
result.identifiers= [] |
|
| 53 |
result.identifiers = []
|
|
| 53 | 54 |
for item in inputObj.identifier: |
| 54 |
result.identifiers.append(IdentifierType(item.identifier,item.schema)) |
|
| 55 |
result.identifiers.append(IdentifierType(item.identifier, item.schema))
|
|
| 55 | 56 |
result.identifiers.append(IdentifierType(inputObj.dnetIdentifier, 'dnetIdentifier')) |
| 56 |
result.object_provider= self.convertLinkProvider([x.provider for x in inputObj.collectedFrom]) |
|
| 57 |
result.objectType = ObjectTypeObjectType(inputObj.objectType,'')
|
|
| 57 |
result.object_provider = self.convertLinkProvider([x.provider for x in inputObj.collectedFrom])
|
|
| 58 |
result.object_type = ObjectTypeObjectType(inputObj.objectType, '')
|
|
| 58 | 59 |
if 'title' in inputObj: |
| 59 | 60 |
result.title = inputObj.title |
| 60 | 61 |
if 'creator' in inputObj: |
| 61 | 62 |
result.creators = [CreatorType(x.name) for x in inputObj.creator] |
| 62 | 63 |
if 'publisher' in inputObj: |
| 63 | 64 |
if len(inputObj.publisher.name) > 0: |
| 64 |
result.publisher=[DSType(inputObj.publisher.name)]
|
|
| 65 |
result.publisher = [DSType(inputObj.publisher.name)]
|
|
| 65 | 66 |
return result |
| 66 | 67 |
|
| 67 |
def convertScholix(self,response): |
|
| 68 |
def convertScholix(self, response):
|
|
| 68 | 69 |
result = [] |
| 69 | 70 |
for item in response.hits: |
| 70 |
s = Scholix()
|
|
| 71 |
s.relationship= RelationshipType(item.relationship.name,item.relationship.schema,'')
|
|
| 71 |
s = Scholix() |
|
| 72 |
s.relationship = RelationshipType(item.relationship.name, item.relationship.schema, '')
|
|
| 72 | 73 |
s.link_provider = self.convertLinkProvider(item['linkprovider']) |
| 73 |
s.source= self.convertObject(item.source) |
|
| 74 |
s.target= self.convertObject(item.target) |
|
| 74 |
s.source = self.convertObject(item.source)
|
|
| 75 |
s.target = self.convertObject(item.target)
|
|
| 75 | 76 |
result.append(s) |
| 76 | 77 |
return result |
| 77 | 78 |
|
| 78 |
def realtionToPid(self, pid, pidType=None, page=0): |
|
| 79 |
def realtionToPid(self, pid, pidType=None, datasource=None, typology = None, page=0):
|
|
| 79 | 80 |
if pidType: |
| 80 |
query = self.create_pid_pidType_query( pidType, pid)
|
|
| 81 |
query = self.create_pid_pidType_query(pidType, pid) |
|
| 81 | 82 |
else: |
| 82 |
query = self.create_target_pid_query(pid) |
|
| 83 |
query = self.create_source_pid_query(pid) |
|
| 84 |
filters = [] |
|
| 85 |
if datasource and len(datasource): |
|
| 86 |
filters.append(self.create_dataSource_query(datasource)) |
|
| 87 |
if typology and len(typology): |
|
| 88 |
filters.append(self.create_typology_query(typology)) |
|
| 89 |
search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query(query)
|
|
| 83 | 90 |
|
| 84 |
search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query(query)
|
|
| 85 |
return self.convertScholix(search_object[page:page+10].execute()) |
|
| 91 |
if len(filters): |
|
| 92 |
search_object = search_object.filter(Q('bool', must=filters))
|
|
| 93 |
return self.convertScholix(search_object[page:page + 10].execute()) |
|
| 86 | 94 |
|
| 87 | 95 |
def realtionToTypology(self, typology, page=0): |
| 88 |
search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query(self.create_typology_query(typology))
|
|
| 89 |
return self.convertScholix(search_object[page:page+10].execute()) |
|
| 96 |
search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query(
|
|
| 97 |
self.create_typology_query(typology)) |
|
| 98 |
return self.convertScholix(search_object[page:page + 10].execute()) |
|
| 90 | 99 |
|
| 91 | 100 |
def realtionFromDatasource(self, datasource, page=0): |
| 92 |
search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query(self.create_dataSource_query(datasource))
|
|
| 93 |
return self.convertScholix(search_object[page:page+10].execute()) |
|
| 101 |
search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query(
|
|
| 102 |
self.create_dataSource_query(datasource)) |
|
| 103 |
return self.convertScholix(search_object[page:page + 10].execute()) |
|
Also available in: Unified diff
uploaded a new version of dli_api