D-Net

from elasticsearch import *

from elasticsearch_dsl import *

from os import path

import os

from typing import List

from swagger_server.models import Scholix, ObjectType, DSType, IdentifierType, RelationshipType, ObjectType, \

    ObjectTypeObjectType, CreatorType

def get_property():

    f = open(path.join(os.path.dirname(os.path.realpath(__file__)), '../../../api.properties'))

    p = {}

    for line in f:

        if not line.startswith("#"):

            data = line.strip().split("=")

            p[data[0].strip()] = data[1].strip()

    return p

class DLIESConnector(object):

    def __init__(self, index_host):

        props = get_property()

        index_name = props['api.index']

        self.index_host = [x.strip() for x in props['es_index'].split(',')]

        print "start with index " + index_name

        print index_host

        self.client = Elasticsearch(hosts=self.index_host)

        self.index_name = index_name

    def create_pid_pidType_query(self, pidType, pid):

        args = {'source.identifier.schema': pidType}

        args2 = {'source.identifier.identifier': pid}

        return Q('nested', path='source.identifier', query=Q('bool', must=[Q('match', **args), Q('match', **args2)]))

    def create_source_pid_query(self, value):

        args = {'source.identifier.identifier': value}

        return Q('nested', path='source.identifier', query=Q('bool', must=[Q('match', **args)]))

    def create_typology_query(self, value):

        args = {'target.objectType': value}

        return Q('nested', path='target', query=Q('bool', must=[Q('match', **args)]))

    def create_dataSource_query(self, value):

        args = {'linkprovider.name': value}

        return Q('nested', path='linkprovider', query=Q('bool', must=[Q('match', **args)]))

    def create_publisher_query(self, value):

        args = {'target.publisher.name': value}

        q = Q('nested', path='target.publisher', query=Q('bool', must=[Q('match', **args)]))

        return Q('nested', path='target', query=q)

    def list_datasources(self):

        search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query()

        search_object.aggs.bucket('all_datasources', 'nested', path='linkprovider').bucket('all_names', 'terms',

                                                                                           field='linkprovider.name',

                                                                                           size=100)

        response = search_object.execute()

        return [item.key for item in response.aggs.all_datasources.all_names.buckets]

    def convertLinkProvider(self, lp):

        result = []

        for item in lp:

            current_identifiers = []

            for ids in item['identifiers']:

                current_identifiers.append(IdentifierType(identifier=ids['identifier'], schema=ids['schema']))

            tmp = DSType(name=item['name'],

                         identifiers=[IdentifierType(x['identifier'], x['schema']) for x in item['identifiers']])

            result.append(tmp)

        return result

    def convertObject(self, inputObj):

        result = ObjectType()

        result.identifiers = []

        for item in inputObj.identifier:

            result.identifiers.append(IdentifierType(item.identifier, item.schema))

        result.identifiers.append(IdentifierType(inputObj.dnetIdentifier, 'dnetIdentifier'))

        result.object_provider = self.convertLinkProvider([x.provider for x in inputObj.collectedFrom])

        result.object_type = ObjectTypeObjectType(inputObj.objectType, '')

        if 'title' in inputObj:

            result.title = inputObj.title.replace(""","")

        if 'creator' in inputObj:

            result.creators = [CreatorType(x.name) for x in inputObj.creator]

        if 'publisher' in inputObj:

            publishers = []

            for p in inputObj.publisher:

                publishers.append(dict(name=p.name))

            result.publisher = publishers

        return result

    def convertScholix(self, response):

        result = []

        for item in response.hits:

            s = Scholix()

            s.relationship = RelationshipType(item.relationship.name, item.relationship.schema, '')

            s.link_provider = self.convertLinkProvider(item['linkprovider'])

            s.source = self.convertObject(item.source)

            s.target = self.convertObject(item.target)

            result.append(s)

        return result

    def realtionToPid(self, pid, pidType=None, datasource=None, typology=None, page=0):

        if pidType:

            query = self.create_pid_pidType_query(pidType.lower(), pid.lower())

        else:

            query = self.create_source_pid_query(pid.lower())

        filters = []

        if datasource and len(datasource):

            filters.append(self.create_dataSource_query(datasource))

        if typology and len(typology):

            filters.append(self.create_typology_query(typology))

        search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query(query)

        if len(filters):

            search_object = search_object.filter(Q('bool', must=filters))

        return self.convertScholix(search_object[page:page + 10].execute())

    def realtionToTypology(self, typology, page=0):

        search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query(

            self.create_typology_query(typology))

        return self.convertScholix(search_object[page:page + 10].execute())

    def realtionFromDatasource(self, datasource, page=0):

        search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query(

            self.create_dataSource_query(datasource))

        return self.convertScholix(search_object[page:page + 10].execute())

    def realtionFromPublisher(self, publisher, page=0):

        search_object = Search(using=self.client, index=self.index_name).doc_type('scholix').query(

            self.create_publisher_query(publisher))

        return self.convertScholix(search_object[page:page + 10].execute())