def add_position(person_uri, position):
"""
Given a person_uri and a position dictionary containing the attributes
of a position, generate the RDF necessary to create the position,
associate it with the person and assign its attributes.
"""
from vivofoundation import assert_resource_property
from vivofoundation import assert_data_property
from vivofoundation import add_dti
from vivofoundation import get_vivo_uri
ardf = ""
position_uri = get_vivo_uri()
dti = {
'start': position.get('start_date', None),
'end': position.get('end_date', None)
}
[add, dti_uri] = add_dti(dti)
ardf = ardf + add
ardf = ardf + assert_resource_property(position_uri, 'rdf:type',
position['position_type'])
ardf = ardf + assert_resource_property(position_uri, 'rdfs:label',
position['position_label'])
ardf = ardf + assert_resource_property(position_uri,
'vivo:dateTimeInterval', dti_uri)
ardf = ardf + assert_resource_property(position_uri, 'vivo:relates',
person_uri)
ardf = ardf + assert_resource_property(position_uri, 'vivo:relates',
position['position_orguri'])
return [ardf, position_uri]
def add_position(person_uri, position):
"""
Given a person_uri and a position dictionary containing the attributes
of a position, generate the RDF necessary to create the position,
associate it with the person and assign its attributes.
"""
from vivofoundation import assert_resource_property
from vivofoundation import assert_data_property
from vivofoundation import add_dti
from vivofoundation import get_vivo_uri
ardf = ""
position_uri = get_vivo_uri()
dti = {'start' : position.get('start_date',None),
'end': position.get('end_date',None)}
[add, dti_uri] = add_dti(dti)
ardf = ardf + add
ardf = ardf + assert_resource_property(position_uri,
'rdf:type', position['position_type'])
ardf = ardf + assert_resource_property(position_uri,
'rdfs:label', position['position_label'])
ardf = ardf + assert_resource_property(position_uri,
'vivo:dateTimeInterval', dti_uri)
ardf = ardf + assert_resource_property(position_uri,
'vivo:relates', person_uri)
ardf = ardf + assert_resource_property(position_uri,
'vivo:relates', position['position_orguri'])
return [ardf, position_uri]
"""
test_assert_resource_property.py -- Given a URI, a vivo resource property
predicate and a resource URI, generate RDF for asserting that the URI has
the resource property URI. The tests demonsrate that that the function does
not check for a valid predicate, nor a valid resource URI.
Version 0.1 MC 2013-12-27
-- Initial version
Version 0.2 MC 2014-08-30
-- PEP 8, support for vivofoundation
"""
__author__ = "Michael Conlon"
__copyright__ = "Copyright 2014, University of Florida"
__license__ = "BSD 3-Clause license"
__version__ = "0.2"
from vivofoundation import assert_resource_property
from datetime import datetime
print datetime.now(), "Start"
print assert_resource_property("http://a.b", "http://c.d", "http://e.f")
print assert_resource_property("http://a.b", "vivo:authorInAuthorship",
"http://c.d")
print assert_resource_property("http://a.b", "vivo:authorInAuthorlist",
"42")
print datetime.now(), "Finish"
def add_person(person):
"""
Add a person to VIVO. The person structure may have any number of
elements. These elements may represent direct assertions (label,
ufid, homeDept), vcard assertions (contact info, name parts),
and/or position assertions (title, tye, dept, start, end dates)
"""
from vivofoundation import assert_data_property
from vivofoundation import assert_resource_property
from vivofoundation import untag_predicate
from vivofoundation import get_vivo_uri
ardf = ""
person_uri = get_vivo_uri()
# Add direct assertions
person_type = person['person_type']
ardf = ardf + assert_resource_property(person_uri, 'rdf:type', person_type)
ardf = ardf + assert_resource_property(person_uri, 'rdf:type',
untag_predicate('ufv:UFEntity'))
ardf = ardf + assert_resource_property(
person_uri, 'rdf:type', untag_predicate('ufv:UFCurrentEntity'))
direct_data_preds = {
'ufid': 'ufv:ufid',
'privacy_flag': 'ufv:privacyFlag',
'display_name': 'rdfs:label',
'gatorlink': 'ufv:gatorlink'
}
direct_resource_preds = {'homedept_uri': 'ufv:homeDept'}
for key in direct_data_preds:
if key in person:
pred = direct_data_preds[key]
val = person[key]
ardf = ardf + assert_data_property(person_uri, pred, val)
for key in direct_resource_preds:
if key in person:
pred = direct_resource_preds[key]
val = person[key]
ardf = ardf + assert_resource_property(person_uri, pred, val)
# Add Vcard Assertions
vcard = {}
for key in [
'last_name',
'first_name',
'middle_name',
'primary_email',
'name_prefix',
'name_suffix',
'fax',
'phone',
'preferred_title',
]:
if key in person.keys():
vcard[key] = person[key]
[add, vcard_uri] = add_vcard(person_uri, vcard)
ardf = ardf + add
# Add Position Assertions
position = {}
for key in [
'start_date', 'position_label', 'end_date', 'position_orguri',
'position_type'
]:
if key in person.keys():
position[key] = person[key]
[add, position_uri] = add_position(person_uri, position)
ardf = ardf + add
return [ardf, person_uri]
def update_vcard(vivo_vcard, source_vcard):
"""
Given a vivo vcard and a source vccard, generate the add and sub rdf
necesary to update vivo the the values ion the source
"""
from vivofoundation import update_entity
from vivofoundation import update_data_property
from vivofoundation import get_vivo_uri
from vivofoundation import assert_data_property
from vivofoundation import assert_resource_property
from vivofoundation import untag_predicate
ardf = ""
srdf = ""
# Update the name entity
name_keys = {
'given_name': {
'predicate': 'vcard:givenName',
'action': 'literal'
},
'family_name': {
'predicate': 'vcard:familyName',
'action': 'literal'
},
'additional_name': {
'predicate': 'vcard:additionalName',
'action': 'literal'
},
'honorific_prefix': {
'predicate': 'vcard:honorificPrefix',
'action': 'literal'
},
'honorific_suffix': {
'predicate': 'vcard:honorificSuffix',
'action': 'literal'
}
}
# Update name entity
if 'name' in source_vcard and 'name' not in vivo_vcard:
name_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(name_uri, 'rdf:type',
untag_predicate('vcard:Name'))
ardf = ardf + assert_resource_property(vcard['vcard_uri'],
'vcard:hasName', name_uri)
vivo_vcard['name_uri'] = name_uri
vivo_vcard['name'] = {}
if 'name' in source_vcard:
vivo_vcard['name']['uri'] = vivo_vcard['name_uri']
[add, sub] = update_entity(vivo_vcard['name'], source_vcard['name'],
name_keys)
ardf = ardf + add
srdf = srdf + sub
# Update title
if 'title' in source_vcard and 'title' not in vivo_vcard:
title_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(title_uri, 'rdf:type',
untag_predicate('vcard:Title'))
ardf = ardf + assert_resource_property(vivo_vcard['vcard_uri'],
'vcard:hasTitle', title_uri)
vivo_vcard['title_uri'] = title_uri
vivo_vcard['title'] = None
if 'title' in source_vcard:
[add, sub] = update_data_property(vivo_vcard['title_uri'],
'vcard:title', vivo_vcard['title'],
source_vcard['title'])
ardf = ardf + add
srdf = srdf + sub
# Update phone. For now, assert a phone. We can't seem to tell which
# phone is to be "updated". If VIVO has telephones a and b, and the
# source says the phone number is c, what is the appropriate operation?
# We will just pick a phone and update it to c. If ther person has zero
# or one phones, everything is fine. These are the most likely cases.
# And in the above two phone case, we have a 50-50 chance of doing the
# right thing. So perhaps 1% of people will be effected adversely by
# the code that follows.
if 'phone' in source_vcard and source_vcard['phone'] is not None:
if 'telephones' not in vivo_vcard or vivo_vcard['telephones'] == []:
telephone_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(
vivo_vcard['vcard_uri'], 'vcard:hasTelephone', telephone_uri)
ardf = ardf + assert_resource_property(
telephone_uri, 'rdf:type', untag_predicate('vcard:telephone'))
telephone_value = None
else:
for telephone in vivo_vcard['telephones']:
if telephone['telephone_type'] == 'Telephone':
telephone_uri = telephone['telephone_uri']
telephone_value = telephone['telephone_number']
continue
[add,
sub] = update_data_property(telephone_uri, 'vcard:telephone',
telephone_value, source_vcard['phone'])
ardf = ardf + add
srdf = srdf + sub
# Analogous processing with analogous comments for a fax number
if 'fax' in source_vcard and source_vcard['fax'] is not None:
if 'telephones' not in vivo_vcard or vivo_vcard['telephones'] == []:
telephone_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(
vivo_vcard['vcard_uri'], 'vcard:hasTelephone', telephone_uri)
ardf = ardf + assert_resource_property(
telephone_uri, 'rdf:type', untag_predicate('vcard:Fax'))
telephone_value = None
else:
for telephone in vivo_vcard['telephones']:
if telephone['telephone_type'] == 'Fax':
telephone_uri = telephone['telephone_uri']
telephone_value = telephone['telephone_number']
continue
[add, sub] = update_data_property(telephone_uri, 'vcard:telephone',
telephone_value, source_vcard['fax'])
ardf = ardf + add
srdf = srdf + sub
# Analogous processing with analogous comments for an email address
if 'primary_email' in source_vcard and \
source_vcard['primary_email'] is not None:
if 'email_addresses' not in vivo_vcard or \
vivo_vcard['email_addresses'] == []:
email_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(vivo_vcard['vcard_uri'],
'vcard:hasEmail', email_uri)
ardf = ardf + assert_resource_property(
email_uri, 'rdf:type', untag_predicate('vcard:Email'))
email_value = None
else:
email_uri = email_addresses[0]['email_uri']
email_value = email_address[0]['email_address']
[add,
sub] = update_data_property(email_uri, 'vcard:email', email_value,
source_vcard['primary_email'])
ardf = ardf + add
srdf = srdf + sub
return [ardf, srdf]
def add_vcard(person_uri, vcard):
"""
Given a person_uri and a vcard dictionary of items on the vcard,
generate ther RDF necessary to create the vcard, associate it with
the person, and associate attributes to the vcard.
The person_uri will be associated to the vcard and the vcard may have
any number of single entry entities to references. The single_entry
table controls the processing of these entities.
The name entity is a special case. All values are attrbuted to the name
entity.
The single_entry table contains some additional keys for future use
Both the name table and the single entry table are easily extensible to
handle additional name attributes and additional single entry entities
respectively.
"""
from vivofoundation import assert_resource_property
from vivofoundation import assert_data_property
from vivofoundation import get_vivo_uri
from vivofoundation import untag_predicate
single_entry = {
'primary_email': {
'resource': 'vcard:hasEmail',
'type': 'vcard:Email',
'pred': 'vcard:email'
},
'email': {
'resource': 'vcard:hasEmail',
'type': 'vcard:Email',
'pred': 'vcard:email'
},
'fax': {
'resource': 'vcard:hasTelephone',
'type': 'vcard:Fax',
'pred': 'vcard:telephone'
},
'telephone': {
'resource': 'vcard:hasTelephone',
'type': 'vcard:Telephone',
'pred': 'vcard:telephone'
},
'preferred_title': {
'resource': 'vcard:hasTitle',
'type': 'vcard:Title',
'pred': 'vcard:title'
},
'title': {
'resource': 'vcard:hasTitle',
'type': 'vcard:Title',
'pred': 'vcard:title'
}
}
name_table = {
'first_name': 'vcard:givenName',
'last_name': 'vcard:familyName',
'middle_name': 'vcard:additionalName',
'name_prefix': 'vcard:honoraryPrefix',
'name_suffix': 'vcard:honorarySuffix'
}
ardf = ""
vcard_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(vcard_uri, 'rdf:type',
untag_predicate('vcard:Individual'))
ardf = ardf + assert_resource_property(person_uri, 'obo:ARG2000028',
vcard_uri) # hasContactInfo
ardf = ardf + assert_resource_property(vcard_uri, 'obo:ARG2000029',
person_uri) # contactInfoOf
# Create the name entity and attach to vcard. For each key in the
# name_table, assert its value to the name entity
name_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(name_uri, 'rdf:type',
untag_predicate('vcard:Name'))
ardf = ardf + assert_resource_property(vcard_uri, 'vcard:hasName',
name_uri)
for key in vcard.keys():
if key in name_table:
pred = name_table[key]
val = vcard[key]
ardf = ardf + assert_data_property(name_uri, pred, val)
# Process single entry vcard bits of info:
# Go through the keys in the vcard. If it's a single entry key, then
# create it. Assign the data vaue and link the vcard to the single
# entry entity
for key in vcard.keys():
if key in single_entry:
val = vcard[key]
entry = single_entry[key]
entry_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(
entry_uri, 'rdf:type', untag_predicate(entry['type']))
ardf = ardf + assert_data_property(entry_uri, entry['pred'], val)
ardf = ardf + assert_resource_property(
vcard_uri, entry['resource'], entry_uri)
return [ardf, vcard_uri]
def add_person(person):
"""
Add a person to VIVO. The person structure may have any number of
elements. These elements may represent direct assertions (label,
ufid, homeDept), vcard assertions (contact info, name parts),
and/or position assertions (title, tye, dept, start, end dates)
"""
from vivofoundation import assert_data_property
from vivofoundation import assert_resource_property
from vivofoundation import untag_predicate
from vivofoundation import get_vivo_uri
ardf = ""
person_uri = get_vivo_uri()
# Add direct assertions
person_type = person['person_type']
ardf = ardf + assert_resource_property(person_uri, 'rdf:type', person_type)
ardf = ardf + assert_resource_property(person_uri, 'rdf:type',
untag_predicate('ufv:UFEntity'))
ardf = ardf + assert_resource_property(person_uri, 'rdf:type',
untag_predicate('ufv:UFCurrentEntity'))
direct_data_preds = {'ufid':'ufv:ufid',
'privacy_flag':'ufv:privacyFlag',
'display_name':'rdfs:label',
'gatorlink':'ufv:gatorlink'
}
direct_resource_preds = {'homedept_uri':'ufv:homeDept'}
for key in direct_data_preds:
if key in person:
pred = direct_data_preds[key]
val = person[key]
ardf = ardf + assert_data_property(person_uri, pred, val)
for key in direct_resource_preds:
if key in person:
pred = direct_resource_preds[key]
val = person[key]
ardf = ardf + assert_resource_property(person_uri, pred, val)
# Add Vcard Assertions
vcard = {}
for key in ['last_name', 'first_name', 'middle_name', 'primary_email',
'name_prefix', 'name_suffix', 'fax', 'phone', 'preferred_title',
]:
if key in person.keys():
vcard[key] = person[key]
[add, vcard_uri] = add_vcard(person_uri, vcard)
ardf = ardf + add
# Add Position Assertions
position = {}
for key in ['start_date', 'position_label', 'end_date', 'position_orguri',
'position_type']:
if key in person.keys():
position[key] = person[key]
[add, position_uri] = add_position(person_uri, position)
ardf = ardf + add
return [ardf, person_uri]
def update_vcard(vivo_vcard, source_vcard):
"""
Given a vivo vcard and a source vccard, generate the add and sub rdf
necesary to update vivo the the values ion the source
"""
from vivofoundation import update_entity
from vivofoundation import update_data_property
from vivofoundation import get_vivo_uri
from vivofoundation import assert_data_property
from vivofoundation import assert_resource_property
from vivofoundation import untag_predicate
ardf = ""
srdf = ""
# Update the name entity
name_keys = {
'given_name' : {'predicate':'vcard:givenName','action':'literal'},
'family_name' : {'predicate':'vcard:familyName',
'action':'literal'},
'additional_name' : {'predicate':'vcard:additionalName',
'action':'literal'},
'honorific_prefix' : {'predicate':'vcard:honorificPrefix',
'action':'literal'},
'honorific_suffix' : {'predicate':'vcard:honorificSuffix',
'action':'literal'}
}
# Update name entity
if 'name' in source_vcard and 'name' not in vivo_vcard:
name_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(name_uri, 'rdf:type',
untag_predicate('vcard:Name'))
ardf = ardf + assert_resource_property(vcard['vcard_uri'],
'vcard:hasName', name_uri)
vivo_vcard['name_uri'] = name_uri
vivo_vcard['name'] = {}
if 'name' in source_vcard:
vivo_vcard['name']['uri'] = vivo_vcard['name_uri']
[add, sub] = update_entity(vivo_vcard['name'],
source_vcard['name'], name_keys)
ardf = ardf + add
srdf = srdf + sub
# Update title
if 'title' in source_vcard and 'title' not in vivo_vcard:
title_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(title_uri, 'rdf:type',
untag_predicate('vcard:Title'))
ardf = ardf + assert_resource_property(vivo_vcard['vcard_uri'],
'vcard:hasTitle', title_uri)
vivo_vcard['title_uri'] = title_uri
vivo_vcard['title'] = None
if 'title' in source_vcard:
[add, sub] = update_data_property(vivo_vcard['title_uri'],
'vcard:title', vivo_vcard['title'], source_vcard['title'])
ardf = ardf + add
srdf = srdf + sub
# Update phone. For now, assert a phone. We can't seem to tell which
# phone is to be "updated". If VIVO has telephones a and b, and the
# source says the phone number is c, what is the appropriate operation?
# We will just pick a phone and update it to c. If ther person has zero
# or one phones, everything is fine. These are the most likely cases.
# And in the above two phone case, we have a 50-50 chance of doing the
# right thing. So perhaps 1% of people will be effected adversely by
# the code that follows.
if 'phone' in source_vcard and source_vcard['phone'] is not None:
if 'telephones' not in vivo_vcard or vivo_vcard['telephones'] == []:
telephone_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(vivo_vcard['vcard_uri'],
'vcard:hasTelephone', telephone_uri)
ardf = ardf + assert_resource_property(telephone_uri,
'rdf:type', untag_predicate('vcard:telephone'))
telephone_value = None
else:
for telephone in vivo_vcard['telephones']:
if telephone['telephone_type'] == 'Telephone':
telephone_uri = telephone['telephone_uri']
telephone_value = telephone['telephone_number']
continue
[add, sub] = update_data_property(telephone_uri,
'vcard:telephone', telephone_value, source_vcard['phone'])
ardf = ardf + add
srdf = srdf + sub
# Analogous processing with analogous comments for a fax number
if 'fax' in source_vcard and source_vcard['fax'] is not None:
if 'telephones' not in vivo_vcard or vivo_vcard['telephones'] == []:
telephone_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(vivo_vcard['vcard_uri'],
'vcard:hasTelephone', telephone_uri)
ardf = ardf + assert_resource_property(telephone_uri,
'rdf:type', untag_predicate('vcard:Fax'))
telephone_value = None
else:
for telephone in vivo_vcard['telephones']:
if telephone['telephone_type'] == 'Fax':
telephone_uri = telephone['telephone_uri']
telephone_value = telephone['telephone_number']
continue
[add, sub] = update_data_property(telephone_uri,
'vcard:telephone', telephone_value, source_vcard['fax'])
ardf = ardf + add
srdf = srdf + sub
# Analogous processing with analogous comments for an email address
if 'primary_email' in source_vcard and \
source_vcard['primary_email'] is not None:
if 'email_addresses' not in vivo_vcard or \
vivo_vcard['email_addresses'] == []:
email_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(vivo_vcard['vcard_uri'],
'vcard:hasEmail', email_uri)
ardf = ardf + assert_resource_property(email_uri,
'rdf:type', untag_predicate('vcard:Email'))
email_value = None
else:
email_uri = email_addresses[0]['email_uri']
email_value = email_address[0]['email_address']
[add, sub] = update_data_property(email_uri,
'vcard:email', email_value, source_vcard['primary_email'])
ardf = ardf + add
srdf = srdf + sub
return [ardf, srdf]
def add_vcard(person_uri, vcard):
"""
Given a person_uri and a vcard dictionary of items on the vcard,
generate ther RDF necessary to create the vcard, associate it with
the person, and associate attributes to the vcard.
The person_uri will be associated to the vcard and the vcard may have
any number of single entry entities to references. The single_entry
table controls the processing of these entities.
The name entity is a special case. All values are attrbuted to the name
entity.
The single_entry table contains some additional keys for future use
Both the name table and the single entry table are easily extensible to
handle additional name attributes and additional single entry entities
respectively.
"""
from vivofoundation import assert_resource_property
from vivofoundation import assert_data_property
from vivofoundation import get_vivo_uri
from vivofoundation import untag_predicate
single_entry = {
'primary_email': {'resource':'vcard:hasEmail','type':'vcard:Email',
'pred':'vcard:email'},
'email': {'resource':'vcard:hasEmail','type':'vcard:Email',
'pred':'vcard:email'},
'fax': {'resource':'vcard:hasTelephone','type':'vcard:Fax',
'pred':'vcard:telephone'},
'telephone': {'resource':'vcard:hasTelephone','type':'vcard:Telephone',
'pred':'vcard:telephone'},
'preferred_title': {'resource':'vcard:hasTitle','type':'vcard:Title',
'pred':'vcard:title'},
'title': {'resource':'vcard:hasTitle','type':'vcard:Title',
'pred':'vcard:title'}
}
name_table = {
'first_name' : 'vcard:givenName',
'last_name' : 'vcard:familyName',
'middle_name' : 'vcard:additionalName',
'name_prefix' : 'vcard:honoraryPrefix',
'name_suffix' : 'vcard:honorarySuffix'
}
ardf = ""
vcard_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(vcard_uri, 'rdf:type',
untag_predicate('vcard:Individual'))
ardf = ardf + assert_resource_property(person_uri, 'obo:ARG2000028',
vcard_uri) # hasContactInfo
ardf = ardf + assert_resource_property(vcard_uri, 'obo:ARG2000029',
person_uri) # contactInfoOf
# Create the name entity and attach to vcard. For each key in the
# name_table, assert its value to the name entity
name_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(name_uri, 'rdf:type',
untag_predicate('vcard:Name'))
ardf = ardf + assert_resource_property(vcard_uri, 'vcard:hasName',
name_uri)
for key in vcard.keys():
if key in name_table:
pred = name_table[key]
val = vcard[key]
ardf = ardf + assert_data_property(name_uri,
pred, val)
# Process single entry vcard bits of info:
# Go through the keys in the vcard. If it's a single entry key, then
# create it. Assign the data vaue and link the vcard to the single
# entry entity
for key in vcard.keys():
if key in single_entry:
val = vcard[key]
entry = single_entry[key]
entry_uri = get_vivo_uri()
ardf = ardf + assert_resource_property(entry_uri,
'rdf:type', untag_predicate(entry['type']))
ardf = ardf + assert_data_property(entry_uri,
entry['pred'], val)
ardf = ardf + assert_resource_property(vcard_uri,
entry['resource'], entry_uri)
return [ardf, vcard_uri]
position_data = read_csv('position_data.csv')
current_ufids = set()
for row_number, row in position_data.items():
current_ufids.add(row['UFID'])
print datetime.now(), "UF has ", len(current_ufids), "ufids on the pay list"
# Compare each ufid in VIVO to the ufids in the set of current ufids. If
# current, mark as current. If not current, subtract current
k = 0
current_uri = untag_predicate('ufVivo:UFCurrentEntity')
for ufid in vivo_ufids.keys():
k += 1
if k % 1000 == 0:
print datetime.now(), k
if ufid in current_ufids:
ardf += assert_resource_property(vivo_ufids[ufid], 'rdf:type', current_uri)
else:
srdf += assert_resource_property(vivo_ufids[ufid], 'rdf:type', current_uri)
ardf += rdf_footer()
srdf += rdf_footer()
add_file = open("current_add.rdf", "w")
sub_file = open("current_sub.rdf", "w")
print >>add_file, ardf
print >>sub_file, srdf
add_file.close()
sub_file.close()
print datetime.now(), "End"