def save_context(self, row_num, man, parent_list):
""" Save context information, will also add new context fields
as needed
"""
use_parents = False
context_uri = ''
if isinstance(parent_list, list):
if len(parent_list) > 0:
context_uri = URImanagement.make_oc_uri(parent_list[0], 'subjects')
use_parents = parent_list[::-1]
# save a record of the context URI
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 13
cell.record = context_uri
cell.save()
cell = None
if use_parents is not False:
pindex = 0
for parent_uuid in use_parents:
pindex += 1
context_label = self.deref_entity_label(parent_uuid)
field_num = self.get_add_context_field_number(pindex)
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = field_num
cell.record = context_label
cell.save()
cell = None
def save_authorship(self, row_num, man):
""" Saves authorship information """
authors = ''
auth = Authorship()
found = auth.get_authors(man.uuid,
man.project_uuid)
if found:
# save counts of different dc-terms:creator for use as table metadata
for auth_id in auth.creators:
if auth_id not in self.dc_creator_ids:
self.dc_creator_ids[auth_id] = 0
self.dc_creator_ids[auth_id] += 1
# save counts of different dc-terms:contributor for use as table metadata
for auth_id in auth.contributors:
if auth_id not in self.dc_contributor_ids:
self.dc_contributor_ids[auth_id] = 0
self.dc_contributor_ids[auth_id] += 1
all_author_ids = auth.creators + auth.contributors
all_authors = []
for auth_id in all_author_ids:
author = self.deref_entity_label(auth_id)
all_authors.append(author)
authors = '; '.join(all_authors)
# save Authors
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 7
cell.record = authors
cell.save()
cell = None
def add_source_cells(self, uuid, row_num, item_data):
""" Adds source data records for an assertion """
predicate_values = LastUpdatedOrderedDict()
project_uuid = item_data[0].project_uuid
for assertion in item_data:
predicate_uuid = assertion.predicate_uuid
object_uuid = assertion.object_uuid
if assertion.object_type == 'xsd:string':
try:
oc_str = OCstring.objects.get(uuid=object_uuid)
obj_val = oc_str.content
except OCstring.DoesNotExist:
obj_val = ''
elif assertion.object_type in ['xsd:integer', 'xsd:double']:
# numeric value
obj_val = str(assertion.data_num)
elif assertion.object_type == 'xsd:date':
obj_val = str(assertion.data_date)
else:
obj_val = str(self.deref_entity_label(object_uuid))
if predicate_uuid not in predicate_values:
# make a list, since some predicates are multi-valued
predicate_values[predicate_uuid] = []
predicate_values[predicate_uuid].append(obj_val)
for predicate_uuid, val_list in predicate_values.items():
field_num = self.get_add_predicate_field_number(predicate_uuid)
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = uuid
cell.project_uuid = project_uuid
cell.row_num = row_num
cell.field_num = field_num
cell.record = self.multi_source_value_delim.join(val_list) # semi-colon delim for multivalued predicates
cell.save()
cell = None
def save_authorship(self, row_num, man):
""" Saves authorship information """
authors = ''
auth = Authorship()
found = auth.get_authors(man.uuid, man.project_uuid)
if found:
# save counts of different dc-terms:creator for use as table metadata
for auth_id in auth.creators:
if auth_id not in self.dc_creator_ids:
self.dc_creator_ids[auth_id] = 0
self.dc_creator_ids[auth_id] += 1
# save counts of different dc-terms:contributor for use as table metadata
for auth_id in auth.contributors:
if auth_id not in self.dc_contributor_ids:
self.dc_contributor_ids[auth_id] = 0
self.dc_contributor_ids[auth_id] += 1
all_author_ids = auth.creators + auth.contributors
all_authors = []
for auth_id in all_author_ids:
author = self.deref_entity_label(auth_id)
all_authors.append(author)
authors = '; '.join(all_authors)
# save Authors
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 7
cell.record = authors
cell.save()
cell = None
def save_default_chrono(self, row_num, man, event_meta):
""" Saves earliest / latest times for an item """
earliest = ''
latest = ''
if event_meta is not False:
times = []
for event in event_meta:
times.append(event.start)
times.append(event.stop)
earliest = min(times)
latest = max(times)
if self.dates_bce_ce is False:
earliest = 1950 - earliest
latest = 1950 - latest
earliest = round(earliest, 0)
latest = round(latest, 0)
# save earliest
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 11
cell.record = str(earliest)
cell.save()
cell = None
# save latest
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 12
cell.record = str(latest)
cell.save()
cell = None
def save_default_chrono(self, row_num, man, event_meta):
""" Saves earliest / latest times for an item """
earliest = ''
latest = ''
if event_meta is not False:
times = []
for event in event_meta:
times.append(event.start)
times.append(event.stop)
earliest = min(times)
latest = max(times)
if self.dates_bce_ce is False:
earliest = 1950 - earliest
latest = 1950 - latest
earliest = round(earliest, 0)
latest = round(latest, 0)
# save earliest
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 11
cell.record = str(earliest)
cell.save()
cell = None
# save latest
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 12
cell.record = str(latest)
cell.save()
cell = None
def save_context(self, row_num, man, parent_list):
""" Save context information, will also add new context fields
as needed
"""
use_parents = False
context_uri = ''
if isinstance(parent_list, list):
if len(parent_list) > 0:
context_uri = URImanagement.make_oc_uri(
parent_list[0], 'subjects')
use_parents = parent_list[::-1]
# save a record of the context URI
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 13
cell.record = context_uri
cell.save()
cell = None
if use_parents is not False:
pindex = 0
for parent_uuid in use_parents:
pindex += 1
context_label = self.deref_entity_label(parent_uuid)
field_num = self.get_add_context_field_number(pindex)
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = field_num
cell.record = context_label
cell.save()
cell = None
def add_source_cells(self, uuid, row_num, item_data):
""" Adds source data records for an assertion """
predicate_values = LastUpdatedOrderedDict()
project_uuid = item_data[0].project_uuid
for assertion in item_data:
predicate_uuid = assertion.predicate_uuid
object_uuid = assertion.object_uuid
if assertion.object_type == 'xsd:string':
try:
oc_str = OCstring.objects.get(uuid=object_uuid)
obj_val = oc_str.content
except OCstring.DoesNotExist:
obj_val = ''
elif assertion.object_type in ['xsd:integer', 'xsd:double']:
# numeric value
obj_val = str(assertion.data_num)
elif assertion.object_type == 'xsd:date':
obj_val = str(assertion.data_date)
else:
obj_val = str(self.deref_entity_label(object_uuid))
if predicate_uuid not in predicate_values:
# make a list, since some predicates are multi-valued
predicate_values[predicate_uuid] = []
predicate_values[predicate_uuid].append(obj_val)
for predicate_uuid, val_list in predicate_values.items():
field_num = self.get_add_predicate_field_number(predicate_uuid)
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = uuid
cell.project_uuid = project_uuid
cell.row_num = row_num
cell.field_num = field_num
cell.record = self.multi_source_value_delim.join(
val_list) # semi-colon delim for multivalued predicates
cell.save()
cell = None
def save_default_geo(self, row_num, man, geo_meta):
""" Saves geo lat / lon data for an item """
latitude = ''
longitude = ''
note = 'Best available location data'
if geo_meta is not False:
for geo in geo_meta:
if geo.meta_type == 'oc-gen:discovey-location':
latitude = geo.latitude
longitude = geo.longitude
if geo.specificity < 0:
note = 'Location approximated '
note += 'as a security precaution (Zoom: ' + str(
abs(geo.specificity)) + ')'
break
# save Latitude
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 8
cell.record = str(latitude)
cell.save()
cell = None
# save Longitude
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 9
cell.record = str(longitude)
cell.save()
cell = None
# save Note
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 10
cell.record = note
cell.save()
cell = None
def save_default_geo(self, row_num, man, geo_meta):
""" Saves geo lat / lon data for an item """
latitude = ''
longitude = ''
note = 'Best available location data'
if geo_meta is not False:
for geo in geo_meta:
if geo.meta_type == 'oc-gen:discovey-location':
latitude = geo.latitude
longitude = geo.longitude
if geo.specificity < 0:
note = 'Location approximated '
note += 'as a security precaution (Zoom: ' + str(abs(geo.specificity)) + ')'
break
# save Latitude
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 8
cell.record = str(latitude)
cell.save()
cell = None
# save Longitude
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 9
cell.record = str(longitude)
cell.save()
cell = None
# save Note
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 10
cell.record = note
cell.save()
cell = None
def save_basic_default_field_cells(self, row_num, man):
""" Saves the default fields that do not involve containment lookups """
# save URI
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 1
cell.record = URImanagement.make_oc_uri(man.uuid, man.item_type)
cell.save()
cell = None
# save label
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 2
cell.record = man.label
cell.save()
cell = None
# save project label
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 3
cell.record = self.deref_entity_label(man.project_uuid)
cell.save()
cell = None
# save project URI
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 4
cell.record = URImanagement.make_oc_uri(man.project_uuid, 'projects')
cell.save()
cell = None
# save item category / class
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 5
cell.record = self.deref_entity_label(man.class_uri)
cell.save()
cell = None
# last updated
if man.revised is datetime:
last_update = man.revised
else:
last_update = man.record_updated
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 6
cell.record = last_update.strftime('%Y-%m-%d')
cell.save()
cell = None
def add_ld_cells(self, uuid, row_num, item_data, pred_ld_equiv_uri):
""" Adds linked data records for an assertion """
if self.do_boolean_multiple_ld_fields(pred_ld_equiv_uri):
multi_ld_fields = True
else:
multi_ld_fields = False
obj_values = LastUpdatedOrderedDict()
obj_values['[URI]'] = []
obj_values['[Label]'] = []
obj_values['[Source]'] = []
project_uuid = item_data[0].project_uuid
for assertion in item_data:
object_uuid = assertion.object_uuid
if assertion.object_type == 'xsd:string':
try:
oc_str = OCstring.objects.get(uuid=object_uuid)
obj_label = oc_str.content
except OCstring.DoesNotExist:
obj_label = ''
else:
obj_label = self.deref_entity_label(object_uuid)
obj_label = str(obj_label)
if obj_label not in obj_values['[Source]']:
obj_values['[Source]'].append(obj_label)
obj_ld_found = False
if object_uuid in self.ld_object_equivs:
for obj_ld_equiv_uri in self.ld_object_equivs[object_uuid]:
obj_ld_found = True
if multi_ld_fields:
cell_value = self.boolean_multiple_ld_fields
field_num = self.get_add_ld_field_number('[Has]',
pred_ld_equiv_uri,
obj_ld_equiv_uri)
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = uuid
cell.project_uuid = project_uuid
cell.row_num = row_num
cell.field_num = field_num
cell.record = cell_value
cell.save()
cell = None
else:
# predicate not broken into seperate fields for different values
obj_equiv_label = self.deref_entity_label(obj_ld_equiv_uri)
if obj_equiv_label is False:
obj_equiv_label = obj_ld_equiv_uri
if obj_equiv_label not in obj_values['[Label]']:
obj_values['[Label]'].append(obj_equiv_label)
if obj_ld_equiv_uri not in obj_values['[URI]']:
obj_values['[URI]'].append(obj_ld_equiv_uri)
if obj_ld_found is False:
print('No linked data for object:' + object_uuid)
if multi_ld_fields is False:
# predicate not broken into seperate fields for different values
for field_type, value_list in obj_values.items():
if len(value_list) > 0:
try:
cell_value = '; '.join(value_list)
except:
# some messiness in the data, won't join into a string
cell_value = False
for val in value_list:
val = str(val)
if cell_value is False:
cell_value = val
else:
cell_value += '; ' + val
field_num = self.get_add_ld_field_number(field_type,
pred_ld_equiv_uri)
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = uuid
cell.project_uuid = project_uuid
cell.row_num = row_num
cell.field_num = field_num
cell.record = cell_value
cell.save()
cell = None
def save_basic_default_field_cells(self, row_num, man):
""" Saves the default fields that do not involve containment lookups """
# save URI
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 1
cell.record = URImanagement.make_oc_uri(man.uuid, man.item_type)
cell.save()
cell = None
# save label
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 2
cell.record = man.label
cell.save()
cell = None
# save project label
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 3
cell.record = self.deref_entity_label(man.project_uuid)
cell.save()
cell = None
# save project URI
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 4
cell.record = URImanagement.make_oc_uri(man.project_uuid, 'projects')
cell.save()
cell = None
# save item category / class
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 5
cell.record = self.deref_entity_label(man.class_uri)
cell.save()
cell = None
# last updated
if man.revised is datetime:
last_update = man.revised
else:
last_update = man.record_updated
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = man.uuid
cell.project_uuid = man.project_uuid
cell.row_num = row_num
cell.field_num = 6
cell.record = last_update.strftime('%Y-%m-%d')
cell.save()
cell = None
def add_ld_cells(self, uuid, row_num, item_data, pred_ld_equiv_uri):
""" Adds linked data records for an assertion """
if self.do_boolean_multiple_ld_fields(pred_ld_equiv_uri):
multi_ld_fields = True
else:
multi_ld_fields = False
obj_values = LastUpdatedOrderedDict()
obj_values['[URI]'] = []
obj_values['[Label]'] = []
obj_values['[Source]'] = []
project_uuid = item_data[0].project_uuid
for assertion in item_data:
object_uuid = assertion.object_uuid
if assertion.object_type == 'xsd:string':
try:
oc_str = OCstring.objects.get(uuid=object_uuid)
obj_label = oc_str.content
except OCstring.DoesNotExist:
obj_label = ''
else:
obj_label = self.deref_entity_label(object_uuid)
obj_label = str(obj_label)
if obj_label not in obj_values['[Source]']:
obj_values['[Source]'].append(obj_label)
obj_ld_found = False
if object_uuid in self.ld_object_equivs:
for obj_ld_equiv_uri in self.ld_object_equivs[object_uuid]:
obj_ld_found = True
if multi_ld_fields:
cell_value = self.boolean_multiple_ld_fields
field_num = self.get_add_ld_field_number(
'[Has]', pred_ld_equiv_uri, obj_ld_equiv_uri)
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = uuid
cell.project_uuid = project_uuid
cell.row_num = row_num
cell.field_num = field_num
cell.record = cell_value
cell.save()
cell = None
else:
# predicate not broken into seperate fields for different values
obj_equiv_label = self.deref_entity_label(
obj_ld_equiv_uri)
if obj_equiv_label is False:
obj_equiv_label = obj_ld_equiv_uri
if obj_equiv_label not in obj_values['[Label]']:
obj_values['[Label]'].append(obj_equiv_label)
if obj_ld_equiv_uri not in obj_values['[URI]']:
obj_values['[URI]'].append(obj_ld_equiv_uri)
if obj_ld_found is False:
print('No linked data for object:' + object_uuid)
if multi_ld_fields is False:
# predicate not broken into seperate fields for different values
for field_type, value_list in obj_values.items():
if len(value_list) > 0:
try:
cell_value = '; '.join(value_list)
except:
# some messiness in the data, won't join into a string
cell_value = False
for val in value_list:
val = str(val)
if cell_value is False:
cell_value = val
else:
cell_value += '; ' + val
field_num = self.get_add_ld_field_number(
field_type, pred_ld_equiv_uri)
cell = ExpCell()
cell.table_id = self.table_id
cell.uuid = uuid
cell.project_uuid = project_uuid
cell.row_num = row_num
cell.field_num = field_num
cell.record = cell_value
cell.save()
cell = None