def project_uris_by_title(self, user_graph, user_uri):
projectsByTitle = defaultdict(list)
bind_namespaces(user_graph)
for row in user_graph.query("""
SELECT DISTINCT ?project ?title WHERE {
?user ore:aggregates ?project .
OPTIONAL {?project dc:title ?title .}
}
""",
initNs=ns,
initBindings={'user': URIRef(user_uri)}):
project_uri = uris.uri('semantic_store_projects', uri=row[0])
project_graph = Graph(store=rdfstore(), identifier=project_uri)
project_resource = Resource(project_graph, URIRef(row[0]))
titles = list(project_resource.objects(predicate=NS.dc['title']))
if len(titles) == 0 and row[1]:
# The project graph doesn't have a title triple, but the user graph does, so use that
projectsByTitle[unicode(row[1])].append(row[0])
else:
# Use the project graph's title triples (preferred)
for title in titles:
projectsByTitle[unicode(title)].append(row[0])
return projectsByTitle
def __init__(self, request, graph, *args, **kwargs):
mimetypes = list(accept_mimetypes(request.META['HTTP_ACCEPT']))
if settings.DEBUG and mimetypes[0].strip().lower().endswith('html'):
format = NegotiatedGraphResponse.default_format
kwargs['mimetype'] = '%s/%s' % (
NegotiatedGraphResponse.default_type,
NegotiatedGraphResponse.default_format)
else:
for mimetype in mimetypes:
format = mimetype[mimetype.rfind('/') + 1:].strip().lower()
if format in RDFLIB_SERIALIZER_FORMATS:
kwargs['mimetype'] = '%s/%s' % (
NegotiatedGraphResponse.default_type,
NegotiatedGraphResponse.default_format)
break
else:
format = NegotiatedGraphResponse.default_format
kwargs['mimetype'] = '%s/%s' % (
NegotiatedGraphResponse.default_type,
NegotiatedGraphResponse.default_format)
super(NegotiatedGraphResponse, self).__init__(self, *args, **kwargs)
bind_namespaces(graph)
self.content = graph.serialize(format=format)
def import_old_data(request):
everything_graph = Graph()
bind_namespaces(everything_graph)
# Either gather post data (must be one project/user graph at a time)
if request.method == 'POST':
logger.debug('!!!!!!!!!!!!!!! views.py - import_old_data')
parse_request_into_graph(request, everything_graph)
add_all_users(everything_graph)
# Create each user's default project
# Due to the structure of the data when exported from the old system, this also
# add each annotation to the project as an aggregated resource
create_project(everything_graph)
# or serialize from a folder, where each file is one project/user graph
else:
i = 0
for file_name in listdir("output/"):
if file_name.startswith('.'):
continue
try:
everything_graph.parse("output/" + file_name,
format=guess_format(file_name)
or 'turtle')
except Exception as e:
print "Failed to decode file '%s' with error message '%s'" % (
file_name, e.args[-1])
else:
add_all_users(everything_graph)
create_project(everything_graph)
return HttpResponse("I finished migrating data without errors.")
def delete_triples_from_project(request, uri):
"""Deletes the triples in a graph provided by a request object from the project graph.
Returns an HttpResponse of all the triples which were successfully removed from the graph."""
if request.user.is_authenticated():
if permissions.has_permission_over(uri, user=request.user, permission=NS.perm.mayUpdate):
removed = Graph()
bind_namespaces(removed)
try:
g = parse_request_into_graph(request)
except (ParserError, SyntaxError) as e:
return HttpResponse(status=400, content="Unable to parse serialization.\n%s" % e)
project_g = get_project_graph(uri)
project_metadata_g = get_project_metadata_graph(uri)
for t in g:
if t in project_g:
project_g.remove(t)
removed.add(t)
project_metadata_g.remove(t)
return NegotiatedGraphResponse(request, removed)
else:
return HttpResponseForbidden('User "%s" does not have update permissions over project "%s"' % (request.user.username, uri))
else:
return HttpResponse(status=401)
def add_all_users(graph):
bind_namespaces(graph)
query = graph.query("""SELECT ?user ?email
WHERE {
?user perm:hasPermissionOver ?project .
?user foaf:mbox ?email .
?user rdf:type foaf:Agent
}""",
initNs=ns)
for q in query:
username = ""
# Remove username from customized uri
s = q[0].split("/")[-1]
username = s.split("'),")[0]
email = q[1].split("mailto:")[-1]
try:
u = User.objects.create_user(username, email,
DEFAULT_PASSWORD_STRING)
except IntegrityError:
transaction.rollback_unless_managed()
print "User '%s' already existed in the database, and was not created." % (
username)
else:
u.save()
print "User '%s' was created successfully." % (username)
def delete_triples_from_project(request, uri):
"""Deletes the triples in a graph provided by a request object from the project graph.
Returns an HttpResponse of all the triples which were successfully removed from the graph."""
if request.user.is_authenticated():
if permissions.has_permission_over(uri, user=request.user, permission=NS.perm.mayUpdate):
removed = Graph()
bind_namespaces(removed)
try:
logger.debug('!!!!!!!!!!!!!!! project_texts.py - delete_triples_from_project')
g = parse_request_into_graph(request)
except (ParserError, SyntaxError) as e:
return HttpResponse(status=400, content="Unable to parse serialization.\n%s" % e)
project_g = get_project_graph(uri)
project_metadata_g = get_project_metadata_graph(uri)
for t in g:
if t in project_g:
project_g.remove(t)
removed.add(t)
project_metadata_g.remove(t)
return NegotiatedGraphResponse(request, removed)
else:
return HttpResponseForbidden('User "%s" does not have update permissions over project "%s"' % (request.user.username, uri))
else:
return HttpResponse(status=401)
def read_all_users(request):
g = Graph()
bind_namespaces(g)
for u in User.objects.filter():
g += user_metadata_graph(user=u)
return NegotiatedGraphResponse(request, g)
def import_old_data(request):
everything_graph = Graph()
bind_namespaces(everything_graph)
# Either gather post data (must be one project/user graph at a time)
if request.method == 'POST':
logger.debug('!!!!!!!!!!!!!!! views.py - import_old_data')
parse_request_into_graph(request, everything_graph)
add_all_users(everything_graph)
# Create each user's default project
# Due to the structure of the data when exported from the old system, this also
# add each annotation to the project as an aggregated resource
create_project(everything_graph)
# or serialize from a folder, where each file is one project/user graph
else:
i = 0
for file_name in listdir("output/"):
if file_name.startswith('.'):
continue
try:
everything_graph.parse("output/" + file_name, format=guess_format(file_name) or 'turtle')
except Exception as e:
print "Failed to decode file '%s' with error message '%s'"%(file_name, e.args[-1])
else:
add_all_users(everything_graph)
create_project(everything_graph)
return HttpResponse("I finished migrating data without errors.")
def add_all_users(graph):
bind_namespaces(graph)
query = graph.query("""SELECT ?user ?email
WHERE {
?user perm:hasPermissionOver ?project .
?user foaf:mbox ?email .
?user rdf:type foaf:Agent
}""", initNs = ns)
for q in query:
username = ""
# Remove username from customized uri
s = q[0].split("/")[-1]
username = s.split("'),")[0]
email = q[1].split("mailto:")[-1]
try:
u = User.objects.create_user(username, email, DEFAULT_PASSWORD_STRING)
except IntegrityError:
transaction.rollback_unless_managed()
print "User '%s' already existed in the database, and was not created."%(username)
else:
u.save()
print "User '%s' was created successfully."%(username)
def aggregated_uris_urls(uri, g):
bind_namespaces(g)
query = """SELECT DISTINCT ?resource_uri ?resource_url
WHERE {
?uri ore:aggregates ?resource_uri .
OPTIONAL { ?resource_url ore:describes ?resource_uri } .
OPTIONAL { ?resource_uri ore:isDescribedBy ?resource_url }
}"""
qres = g.query(query, initNs=ns, initBindings={'uri': URIRef(uri)})
return list(qres)
def find_resource(manifest_uri, g, pred, obj):
bind_namespaces(g)
query = """SELECT DISTINCT ?resource_uri ?resource_url
WHERE {
<%s> ore:aggregates ?resource_uri .
?resource_url ore:describes ?resource_uri .
?resource_uri %s "%s" .
}""" % (manifest_uri, pred, obj)
qres = g.query(query, initNs=ns)
return qres
def find_resource(manifest_uri, g, pred, obj):
bind_namespaces(g)
query = """SELECT DISTINCT ?resource_uri ?resource_url
WHERE {
<%s> ore:aggregates ?resource_uri .
?resource_url ore:describes ?resource_uri .
?resource_uri %s "%s" .
}""" % (manifest_uri, pred, obj)
qres = g.query(query, initNs = ns)
return qres
def aggregated_uris_urls(uri, g):
bind_namespaces(g)
query = """SELECT DISTINCT ?resource_uri ?resource_url
WHERE {
?uri ore:aggregates ?resource_uri .
OPTIONAL { ?resource_url ore:describes ?resource_uri } .
OPTIONAL { ?resource_uri ore:isDescribedBy ?resource_url }
}"""
qres = g.query(query, initNs=ns, initBindings={'uri': URIRef(uri)})
return list(qres)
def image_annotations(manifest_uri, g):
bind_namespaces(g)
query = """SELECT DISTINCT ?resource_uri ?resource_url
WHERE {
?manifest_uri ore:aggregates ?resource_uri .
?resource_uri rdf:type dms:ImageAnnotationList .
OPTIONAL { ?resource_url ore:describes ?resource_uri } .
OPTIONAL { ?resource_uri ore:isDescribedBy ?resource_url }
}"""
qres = g.query(query, initNs=ns, initBindings={'manifest_uri': URIRef(manifest_uri)})
return list(qres)
def image_annotations(manifest_uri, g):
bind_namespaces(g)
query = """SELECT DISTINCT ?resource_uri ?resource_url
WHERE {
?manifest_uri ore:aggregates ?resource_uri .
?resource_uri rdf:type dms:ImageAnnotationList .
OPTIONAL { ?resource_url ore:describes ?resource_uri } .
OPTIONAL { ?resource_uri ore:isDescribedBy ?resource_url }
}"""
qres = g.query(query,
initNs=ns,
initBindings={'manifest_uri': URIRef(manifest_uri)})
return list(qres)
def resource_urls(manifest_uri, g):
bind_namespaces(g)
query = """SELECT DISTINCT ?resource_url
WHERE {
?manifest_uri ore:aggregates ?resource_uri .
?resource_url ore:describes ?resource_uri
}"""
qres = g.query(query, initNs=ns, initBindings={'manifest_uri': URIRef(manifest_uri)})
if len(qres) == 0:
query = """SELECT DISTINCT ?resource_url
WHERE {
?manifest_uri ore:aggregates ?resource_uri .
?resource_uri ore:isDescribedBy ?resource_url
}"""
qres = g.query(query, initNs=ns, initBindings={'manifest_uri': URIRef(manifest_uri)})
return qres
def create_project_graph(host, user, title, project):
if not project:
project = uris.uuid()
elif not (type(project) == URIRef):
project = URIRef(project)
g = Graph()
bind_namespaces(g)
if not title:
title = "Default Project"
g.add((project, NS.rdf.type, NS.foaf.Project))
g.add((project, NS.rdf.type, NS.dm.Project))
g.add((project, NS.rdf['type'], NS.dcmitype['Collection']))
g.add((project, NS.rdf['type'], NS.ore['Aggregation']))
g.add((project, NS.dc['title'], Literal(title)))
g.add((project, NS.dcterms['created'], Literal(datetime.utcnow())))
return g
def create_project_graph(host, user, title, project):
if not project:
project = uris.uuid()
elif not (type(project) == URIRef):
project = URIRef(project)
g = Graph()
bind_namespaces(g)
if not title:
title = "Default Project"
g.add((project, NS.rdf.type, NS.foaf.Project))
g.add((project, NS.rdf.type, NS.dm.Project))
g.add((project, NS.rdf['type'], NS.dcmitype['Collection']))
g.add((project, NS.rdf['type'], NS.ore['Aggregation']))
g.add((project, NS.dc['title'], Literal(title)))
g.add((project, NS.dcterms['created'], Literal(datetime.utcnow())))
return g
def print_triples(triples):
"""Prints a turtle serialization of an iterable of triples"""
g = Graph()
bind_namespaces(g)
g += triples
serialization = g.serialize(format='turtle')
# Remove the @prefix section for easier readability
last_match = None
for match in re.finditer('^@prefix.+\.$', serialization, flags=re.MULTILINE):
last_match = match
if last_match:
print serialization[last_match.end() + 1:].strip('\n')
else:
print serialization
def print_triples(triples):
"""Prints a turtle serialization of an iterable of triples"""
g = Graph()
bind_namespaces(g)
g += triples
serialization = g.serialize(format='turtle')
# Remove the @prefix section for easier readability
last_match = None
for match in re.finditer('^@prefix.+\.$',
serialization,
flags=re.MULTILINE):
last_match = match
if last_match:
print serialization[last_match.end() + 1:].strip('\n')
else:
print serialization
def read_project_text(project_uri, text_uri):
# Correctly format project uri and get project graph
project_identifier = uris.uri('semantic_store_projects', uri=project_uri)
project_g = Graph(rdfstore(), identifier=project_identifier)
# Make text uri URIRef (so Graph will understand)
text_uri = URIRef(text_uri)
# Create an empty graph and bind namespaces
text_g = Graph()
bind_namespaces(text_g)
text_g += resource_annotation_subgraph(project_g, text_uri)
text_g += specific_resources_subgraph(project_g, text_uri, project_uri)
overwrite_text_graph_from_model(text_uri, project_uri, text_g)
# Return graph about text
return text_g
def resource_urls(manifest_uri, g):
bind_namespaces(g)
query = """SELECT DISTINCT ?resource_url
WHERE {
?manifest_uri ore:aggregates ?resource_uri .
?resource_url ore:describes ?resource_uri
}"""
qres = g.query(query,
initNs=ns,
initBindings={'manifest_uri': URIRef(manifest_uri)})
if len(qres) == 0:
query = """SELECT DISTINCT ?resource_url
WHERE {
?manifest_uri ore:aggregates ?resource_uri .
?resource_uri ore:isDescribedBy ?resource_url
}"""
qres = g.query(query,
initNs=ns,
initBindings={'manifest_uri': URIRef(manifest_uri)})
return qres
def resource_uris_urls_old(manifest_uri, g):
bind_namespaces(g)
query = """SELECT DISTINCT ?resource_uri ?resource_url
WHERE {
?manifest_uri ore:aggregates ?resource_uri .
?resource_url ore:describes ?resource_uri
}"""
qres = g.query(query, initNs=ns, initBindings={'manifest_uri': URIRef(manifest_uri)})
uris_urls = set(qres)
query = """SELECT DISTINCT ?resource_uri ?resource_url
WHERE {
?manifest_uri ore:aggregates ?resource_uri .
?resource_uri ore:isDescribedBy ?resource_url
}"""
qres = g.query(query, initNs=ns, initBindings={'manifest_uri': URIRef(manifest_uri)})
for i in qres:
uris_urls.add(i)
return uris_urls
def read_project_text(project_uri, text_uri):
# Correctly format project uri and get project graph
project_identifier = uris.uri('semantic_store_projects', uri=project_uri)
project_g = Graph(rdfstore(), identifier=project_identifier)
# Make text uri URIRef (so Graph will understand)
text_uri = URIRef(text_uri)
# Create an empty graph and bind namespaces
text_g = Graph()
bind_namespaces(text_g)
text_g += resource_annotation_subgraph(project_g, text_uri)
text_g += specific_resources_subgraph(project_g, text_uri, project_uri)
overwrite_text_graph_from_model(text_uri, project_uri, text_g)
# Return graph about text
return text_g
def page_attributes(g, page_uri, res_uri):
bind_namespaces(g)
qres = g.query("""SELECT DISTINCT ?res_title ?title ?width ?height ?image
WHERE {
{?res_uri dc:title ?res_title} UNION {?res_uri rdfs:label ?res_title} .
{?page_uri dc:title ?title} UNION {?page_uri rdfs:label ?title} .
?page_uri exif:width ?width .
?page_uri exif:height ?height .
?anno oa:hasTarget ?page_uri .
?anno oa:hasBody ?image .
?image rdf:type dcmitype:Image .
}""", initBindings={
'res_uri': URIRef(res_uri),
'page_uri': URIRef(page_uri)
})
if qres:
(res_title, page_title, width, height, image) = list(qres)[0]
return (unicode(res_title), unicode(page_title), int(width), int(height),
unicode(image))
else:
return None, None, None, None, None
def __init__(self, request, graph, *args, **kwargs):
mimetypes = list(accept_mimetypes(request.META['HTTP_ACCEPT']))
if settings.DEBUG and mimetypes[0].strip().lower().endswith('html'):
format = NegotiatedGraphResponse.default_format
kwargs['mimetype'] = '%s/%s' % (NegotiatedGraphResponse.default_type, NegotiatedGraphResponse.default_format)
else:
for mimetype in mimetypes:
format = mimetype[mimetype.rfind('/') + 1:].strip().lower()
if format in RDFLIB_SERIALIZER_FORMATS:
kwargs['mimetype'] = '%s/%s' % (NegotiatedGraphResponse.default_type, NegotiatedGraphResponse.default_format)
break
else:
format = NegotiatedGraphResponse.default_format
kwargs['mimetype'] = '%s/%s' % (NegotiatedGraphResponse.default_type, NegotiatedGraphResponse.default_format)
super(NegotiatedGraphResponse, self).__init__(self, *args, **kwargs)
bind_namespaces(graph)
self.content = graph.serialize(format=format)
def resources(request, uri, ext=None):
if request.user.is_authenticated():
perms = ProjectPermission.objects.filter(user=request.user)
else:
perms = []
uri = uri.rstrip('/')
store_g = Graph(store=rdfstore(), identifier=URIRef(uri))
g = Graph()
g += store_g
if len(g) > 0:
for i in perms:
anno_uri = settings.URI_MINT_BASE \
+ "/projects/" + i.identifier \
+ "/resources/" + uri \
+ "/annotations/"
anno_url = reverse('semantic_store_project_annotations',
kwargs={'project_uri': i.identifier}) \
+ "?uri=" + uri
g.add((URIRef(uri), NS.ore['aggregates'], URIRef(anno_uri)))
g.add(
(URIRef(anno_uri), NS.ore['isDescribedBy'], URIRef(anno_url)))
g.add((URIRef(anno_uri), NS.rdf['type'], NS.ore['Aggregation']))
g.add((URIRef(anno_uri), NS.rdf['type'], NS.rdf['List']))
g.add((URIRef(anno_uri), NS.rdf['type'], NS.dms['AnnotationList']))
return NegotiatedGraphResponse(request, g)
else:
main_graph_store = ConjunctiveGraph(store=rdfstore(),
identifier=default_identifier)
main_graph = Graph()
main_graph += main_graph_store
g = Graph()
bind_namespaces(g)
for t in main_graph.triples((URIRef(uri), None, None)):
g.add(t)
if len(g) > 0:
return NegotiatedGraphResponse(request, g)
else:
return HttpResponseNotFound()
def page_attributes(g, page_uri, res_uri):
bind_namespaces(g)
qres = g.query("""SELECT DISTINCT ?res_title ?title ?width ?height ?image
WHERE {
{?res_uri dc:title ?res_title} UNION {?res_uri rdfs:label ?res_title} .
{?page_uri dc:title ?title} UNION {?page_uri rdfs:label ?title} .
?page_uri exif:width ?width .
?page_uri exif:height ?height .
?anno oa:hasTarget ?page_uri .
?anno oa:hasBody ?image .
?image rdf:type dcmitype:Image .
}""",
initBindings={
'res_uri': URIRef(res_uri),
'page_uri': URIRef(page_uri)
})
if qres:
(res_title, page_title, width, height, image) = list(qres)[0]
return (unicode(res_title), unicode(page_title), int(width),
int(height), unicode(image))
else:
return None, None, None, None, None
def resources(request, uri, ext=None):
if request.user.is_authenticated():
perms = ProjectPermission.objects.filter(user=request.user)
else:
perms = []
uri = uri.rstrip('/')
store_g = Graph(store=rdfstore(), identifier=URIRef(uri))
g = Graph()
g += store_g
if len(g) > 0:
for i in perms:
anno_uri = settings.URI_MINT_BASE \
+ "/projects/" + i.identifier \
+ "/resources/" + uri \
+ "/annotations/"
anno_url = reverse('semantic_store_project_annotations',
kwargs={'project_uri': i.identifier}) \
+ "?uri=" + uri
g.add((URIRef(uri), NS.ore['aggregates'], URIRef(anno_uri)))
g.add((URIRef(anno_uri), NS.ore['isDescribedBy'], URIRef(anno_url)))
g.add((URIRef(anno_uri), NS.rdf['type'], NS.ore['Aggregation']))
g.add((URIRef(anno_uri), NS.rdf['type'], NS.rdf['List']))
g.add((URIRef(anno_uri), NS.rdf['type'], NS.dms['AnnotationList']))
return NegotiatedGraphResponse(request, g)
else:
main_graph_store = ConjunctiveGraph(store=rdfstore(),
identifier=default_identifier)
main_graph = Graph()
main_graph += main_graph_store
g = Graph()
bind_namespaces(g)
for t in main_graph.triples((URIRef(uri), None, None)):
g.add(t)
if len(g) > 0:
return NegotiatedGraphResponse(request, g)
else:
return HttpResponseNotFound()
def resource_uris_urls_old(manifest_uri, g):
bind_namespaces(g)
query = """SELECT DISTINCT ?resource_uri ?resource_url
WHERE {
?manifest_uri ore:aggregates ?resource_uri .
?resource_url ore:describes ?resource_uri
}"""
qres = g.query(query,
initNs=ns,
initBindings={'manifest_uri': URIRef(manifest_uri)})
uris_urls = set(qres)
query = """SELECT DISTINCT ?resource_uri ?resource_url
WHERE {
?manifest_uri ore:aggregates ?resource_uri .
?resource_uri ore:isDescribedBy ?resource_url
}"""
qres = g.query(query,
initNs=ns,
initBindings={'manifest_uri': URIRef(manifest_uri)})
for i in qres:
uris_urls.add(i)
return uris_urls
def graph():
g = Graph()
bind_namespaces(g)
return g
def graph():
g = Graph()
bind_namespaces(g)
return g
def serialization_iterator(project_uri, format):
yield ''
export_graph = project_export_graph(project_uri)
bind_namespaces(export_graph)
yield export_graph.serialize(format=format)
def serialization_iterator(project_uri, format):
yield ''
export_graph = project_export_graph(project_uri)
bind_namespaces(export_graph)
yield export_graph.serialize(format=format)
