hsh_triplify/triplify.py

# coding=utf-8
#SQL
from sqlalchemy import *
from sqlalchemy.ext.declarative import declarative_base
from sqlalchemy.orm import sessionmaker
from sqlalchemy.orm import relationship, backref

Base = declarative_base()

class HsHOrganizationalUnitType(Base):
	__tablename__ = 'organizational_unit_type'
	id = Column(Integer, primary_key=True)
	name = Column(String)

class HsHOrganizationalUnit(Base):
	__tablename__ = 'organizational_unit';
	id = Column(Integer, primary_key=True)
	name = Column(String)
	parent = Column(Integer, ForeignKey('organizational_unit.id'))
	acronym = Column(String)
	post_address = Column(Integer)
	organizational_unit_type = Column(Integer, ForeignKey('organizational_unit_type.id'))
	oParent = relationship("HsHOrganizationalUnit", remote_side=[id])

class HsHMembership(Base):
	__tablename__ = 'membership'
	id = Column(Integer, primary_key=True)
	person = Column(Integer, ForeignKey('person.id'))
	organizational_unit = Column(Integer, ForeignKey('organizational_unit.id'))
	start_date = Column(DateTime)
	end_date = Column(DateTime)
	active = Column(String)
	business_role = Column(Integer, ForeignKey('business_role.id'))
	mPerson = relationship("HsHPerson")
	mBusinessRole = relationship("HsHBusinessRole")
	mOrg = relationship("HsHOrganizationalUnit")

class HsHBusinessRole(Base):
	__tablename__ = 'business_role'
	id = Column(Integer, primary_key=True)
	name = Column(String)
	description = Column(String)

class HsHPerson(Base):
	__tablename__ = 'person'
	id = Column(Integer, primary_key=True)
	name = Column(String)
	firstname = Column(String)
	title_prefix = Column(String)
	title_suffix = Column(String)
	gender = Column(String)
	#birthday = Column(DateTime) #TODO: without timezone!
	memberships = relationship("HsHMembership")

class HsHTelephone(Base):
	__tablename__ = 'telephone'
	id = Column(Integer, primary_key=True)
	call_number = Column(String)
	description = Column(String)

#RDF
from rdflib import Namespace
from rdfalchemy import rdfSingle, rdfMultiple
from rdfalchemy.rdfSubject import rdfSubject

from rdflib import Literal, BNode, Namespace, URIRef
from rdflib import RDF, RDFS, Graph, OWL
from rdflib.namespace import XSD

foaf = Namespace('http://xmlns.com/foaf/0.1/')
vitro = Namespace('http://vitro.mannlib.cornell.edu/ns/vitro/0.7')
vivo = Namespace('http://vivoweb.org/ontology/core#')
vcard = Namespace('http://www.w3.org/2006/vcard/ns#')
obo = Namespace('http://purl.obolibrary.org/obo/')
hsh = Namespace('http://data.hs-hannover.de/ontology/hshOntologie#')

localPerson = Namespace('http://data.hs-hannover.de/individual/p')
localOrg = Namespace('http://data.hs-hannover.de/individual/o')
localMembership = Namespace('http://data.hs-hannover.de/individual/m')

def get_graph():
	"""Small little helper to create graph with namespaces ;-)"""
	g = rdfSubject.db
	g.bind('foaf', foaf)
	g.bind('vitro', vitro)
	g.bind('vivo', vivo)
	g.bind('vcard', vcard)
	g.bind('obo', obo)
	g.bind('localperson', localPerson)
	g.bind('localorg', localOrg)
	g.bind('localMembership', localMembership)
	g.bind('hsh', hsh)
	return g

class Thing(rdfSubject):
    rdf_type = OWL.Thing
    label = rdfSingle(RDFS.label)

class hshThing(Thing):
	rdf_type = hsh.hshLocal

class Person(hshThing):
    rdf_type = foaf.Person
    firstname = rdfSingle(vcard.givenName)
    name = rdfSingle(vcard.familyName)
    hasContactInfo = rdfSingle(obo.ARG_2000028)
    label = rdfSingle(RDFS.label)
    associatedOe = rdfSingle(vivo.relatedBy)

class FacultyMember(Person):
	rdf_type = vivo.FacultyMember

class NonFacultyAcademic(Person):
	rdf_type = vivo.NonFacultyAcademic

class NonAcademic(Person):
	rdf_type = vivo.NonAcademic

class ContactInfo(Thing):
	rdf_type = vcard.Individual
	hasTitle = rdfSingle(vcard.hasTitle,range_type=vcard.Title)
	contactInformationFor = rdfSingle(obo.ARG_2000029,range_type=foaf.Person)

class Title(Thing):
	rdf_type = vcard.Title
	title = rdfSingle(vcard.title,range_type=XSD.String)

class Organization(hshThing):
	rdf_type = foaf.Organization
	label = rdfSingle(RDFS.label)
	parentOe = rdfSingle(obo.BFO_0000050)
	acronym = rdfSingle(vivo.abbreviation)

class OboMembership(Thing):
	rdf_type = obo.BFO_0000020

class Membership(OboMembership):
	rdf_type = vivo.Position
	label = rdfSingle(RDFS.label)
	relates = rdfMultiple(vivo.relates)
	dateTimeInterval = rdfSingle(vivo.dateTimeInterval)

class FacultyPosition(Membership):
	rdf_type = vivo.FacultyPosition

class NonFacultyPosition(Membership):
	rdf_type = vivo.NonFacultyPosition

class NonAcademicPosition(Membership):
	rdf_type = vivo.NonAcademicPosition

class OboDateTimeInterval(Thing):
	rdf_type = obo.BFO_0000038

class DateTimeInterval(OboDateTimeInterval):
	rdf_type = vivo.DateTimeInterval
	start = rdfSingle(vivo.start)
	end = rdfSingle(vivo.end)

class OboDateTimeValue(Thing):
	rdf_type = obo.BFO_0000148

class DateTimeValue(OboDateTimeValue):
	rdf_type = vivo.DateTimeValue
	dateTime = rdfSingle(vivo.dateTime)
	dateTimePrecision = rdfSingle(vivo.dateTimePrecision)

class IdSequence:
	"""Small Helper for easy sequences"""
	def __init__(self, start):
		self.num = start - 1
	def getNext(self):
		self.num += 1
		return self.num

def getBusinessRoleName(membership):
	"""Gender based mapping for business role names."""
	names = {
		"M": {
			1000000000: "Beschäftigter",
			1000000001: "Professor",
			1000000002: "Wissenschaftlicher Mitarbeiter",
			1000000003: "Auszubildender",
			1000000004: "Lehrkraft für besondere Aufgaben",
			1000000005: "Lehrbeauftragter",
			1000000006: "Praktikant",
			1000000007: "Wissenschaftliche Hilfskraft",
			1000000008: "Studentische Hilfskraft",
			1000000009: "Lehrkraft",
			1000000010: "Leiter"
		},
		"F": {
			1000000000: "Beschäftigte",
			1000000001: "Professorin",
			1000000002: "Wissenschaftliche Mitarbeiterin",
			1000000003: "Auszubildende",
			1000000004: "Lehrkraft für besondere Aufgaben",
			1000000005: "Lehrbeauftragte",
			1000000006: "Praktikantin",
			1000000007: "Wissenschaftliche Hilfskraft",
			1000000008: "Studentische Hilfskraft",
			1000000009: "Lehrkraft",
			1000000010: "Leiterin"
		},
		"?": {
			1000000000: "Beschäftigte(r)",
			1000000001: "ProfessorIn",
			1000000002: "Wissenschaftliche(r) MitarbeiterIn",
			1000000003: "Auszubildende(r)",
			1000000004: "Lehrkraft für besondere Aufgaben",
			1000000005: "Lehrbeauftragte(r)",
			1000000006: "PraktikantIn",
			1000000007: "Wissenschaftliche Hilfskraft",
			1000000008: "Studentische Hilfskraft",
			1000000009: "Lehrkraft",
			1000000010: "LeiterIn"
		}
	}
	gender = membership.mPerson.gender
	business_role = membership.mBusinessRole.id
	if gender == None:
		gender = "?"
	return names[gender][business_role]

def addPersonToGraph(sqlP, additionalIdSeq):
	"""Add a given HsHPerson to the graph."""
	combined_title = ''
	if sqlP.title_prefix != None:
		combined_title = sqlP.title_prefix
	if sqlP.title_suffix != None:
		if combined_title == '':
			combined_title = sqlP.title_suffix
		else:
			combined_title += ' ' + sqlP.title_suffix

	person_uri = URIRef("%s%s" % (localPerson, sqlP.id))
	hshThing(person_uri)
	rdfP = Person(person_uri) #WTF?!
	rdfP.firstname = sqlP.firstname;
	rdfP.name = sqlP.name;
	rdfP.label = "%s, %s" % (sqlP.name, sqlP.firstname)
	# If there is a title to add, do it.
	if(combined_title != ''):
		title_uri = URIRef("%s%s" % (localPerson, additionalIdSeq.getNext()))
		rdfTitle = Title(title_uri)
		rdfTitle.title = combined_title

		contact_info_uri = URIRef("%s%s" % (localPerson, additionalIdSeq.getNext()))
		rdfCi = ContactInfo(contact_info_uri)
		rdfCi.hasTitle = rdfTitle
		rdfCi.contactInformationFor = rdfP

		rdfP.hasContactInfo = rdfCi
	return person_uri

def processMembership(membership, additionalIdSeq):
	#create membership
	membershipUri = URIRef("%s%s" % (localMembership, membership.id))
	personUri = URIRef("%s%s" % (localPerson, membership.mPerson.id))
	organizationUri = URIRef("%s%s" % (localOrg, membership.mOrg.id))

	#dtiUri = URIRef("%s%s" % (localMembership, additionalIdSeq.getNext()))
	#dtsUri = URIRef("%s%s" % (localMembership, additionalIdSeq.getNext()))
	#dteUri = URIRef("%s%s" % (localMembership, additionalIdSeq.getNext()))

	rdfM = OboMembership(membershipUri)
	rdfM = Membership(membershipUri)
	rdfM.relates = [personUri, organizationUri]
	rdfM.label = getBusinessRoleName(membership)
	#rdfM.dateTimeInterval = dtiUri

	if membership.mBusinessRole.name == "ProfessorIn":
		rdfM = FacultyPosition(membershipUri)
	elif membership.mBusinessRole.name == "WiMi":
		rdfM = NonFacultyPosition(membershipUri)
	else:
		rdfM = NonAcademicPosition(membershipUri)

	#rdfDts = OboDateTimeValue(dtsUri)
	#rdfDts = DateTimeValue(dtsUri)
	#rdfDts.dateTime = membership.start_date
	#rdfDts.dateTimePrecision = vivo.yearPrecision

	#rdfDte = OboDateTimeValue(dteUri)
	#rdfDte = DateTimeValue(dteUri)
	#rdfDte.dateTime = membership.end_date
	#rdfDte.dateTimePrecision = vivo.yearPrecision

	#rdfDti = OboDateTimeInterval(dtiUri)
	#rdfDti = DateTimeInterval(dtiUri)
	#rdfDti.start = dtsUri
	#rdfDti.end = dteUri

def processPersons(session, additionalIdSeq):
	"""Fetch persons, create triples for them."""
	persons = session.query(HsHPerson)
	for count, sqlP in enumerate(persons):
		if sqlP.memberships == []:
			continue #SKIP PEOPLE WITHOUT ANY MEMBERSHIPS AT ALL!

		#create person
		personUri = addPersonToGraph(sqlP, additionalIdSeq)

		#used to determine whether person is non-academic type
		isNonAcademic = True

		#use memberships to determine person type
		for i, membership in enumerate(sqlP.memberships):
			if membership.active != 'Y':
				continue #SKIP INACTIVE MEMBERSHIPS!

			processMembership(membership, additionalIdSeq)

			#This is an active membership with an id and a business_role.
			if membership.mBusinessRole.name == 'ProfessorIn':
				isNonAcademic = False
				organization_uri = URIRef("%s%s" % (localOrg, membership.organizational_unit))
				rdfP = FacultyMember(personUri)
				rdfP.associatedOe = organization_uri
			elif membership.mBusinessRole.name == 'WiMi':
				isNonAcademic = False
				organization_uri = URIRef("%s%s" % (localOrg, membership.organizational_unit))
				rdfP = NonFacultyAcademic(personUri)
				rdfP.associatedOe = organization_uri

		#assign non-academic person type if neccessary
		if isNonAcademic:
			NonAcademic(personUri)

def processOrganizations(session, additionalIdSeq):
	"""Generate triples for organizational units."""
	organizations = session.query(HsHOrganizationalUnit)
	for count, sqlO in enumerate(organizations):
		organization_uri = URIRef("%s%s" % (localOrg, sqlO.id))
		rdfO = hshThing(organization_uri)
		rdfO = Organization(organization_uri)
		rdfO.label = sqlO.name
		if sqlO.acronym != None:
			rdfO.acronym = sqlO.acronym
		if sqlO.oParent != None:
			parent_organization_uri = URIRef("%s%s" % (localOrg, sqlO.oParent.id))
			rdfO.parentOe = parent_organization_uri

def createTriples():
	"""Fetch memberships, evaluate them, create persons,..."""
	engine = create_engine("postgresql://hshinfo:hshinfotest@141.71.2.152/hshinfo")
	session = sessionmaker(bind=engine)()

	g = get_graph()
	additionalIdSeq = IdSequence(2000000000) #don't care sequence

	processOrganizations(session, additionalIdSeq)
	processPersons(session, additionalIdSeq)

	#we're done.
	triples = g.serialize(format='n3')
	g.close()

	return triples

if __name__ == '__main__':
	writeToFile = True

	triples = createTriples()
	print(triples)

	if writeToFile == True:
		f = open('data.n3', 'wb')
		f.write(triples)
		f.close()