def __init__(self):
Markdown.setPre("http://schema.org/")
self.typesCount = self.propsCount = self.namedCount = 0
self.openFile()
self.createDom()
self.setSkips()
self.getGraphs()
self.loadGraphs()
self.cleanup()
self.closeFile()
def __init__(self):
Markdown.setPre("http://schema.org/")
self.typesCount = self.propsCount = self.namedCount = 0
self.openFile()
self.createDom()
self.setSkips()
self.getGraphs()
self.loadGraphs()
self.cleanup()
self.closeFile()
def getCSVComment(self, term=None, graph=None):
query = '''select ?com where{
<%s> rdfs:comment ?com.
}''' % term
res = self.doQuery(graph, query)
ret = ', '.join([x.com for x in res])
#print "SUBTYPES of %s: '%s'" % (term,ret)
if self.markdown:
Markdown.setPre("http://schema.org/")
ret = Markdown.parse(ret)
Markdown.setPre()
return ret
def getCSVComment(self,term=None,graph=None):
query='''select ?com where{
<%s> rdfs:comment ?com.
}''' % term
res = self.doQuery(graph,query)
ret = ', '.join([x.com for x in res])
#print "SUBTYPES of %s: '%s'" % (term,ret)
if self.markdown:
Markdown.setPre(api.SdoConfig.vocabUri())
ret = Markdown.parse(ret)
Markdown.setPre()
return ret
def getCSVComment(self, term=None, graph=None):
query = '''select ?com where{
<%s> rdfs:comment ?com.
}''' % term
res = self.doQuery(graph, query)
ret = ', '.join([x.com for x in res])
#print "SUBTYPES of %s: '%s'" % (term,ret)
if self.markdown:
Markdown.setPre(api.SdoConfig.vocabUri())
ret = Markdown.parse(ret)
Markdown.setPre()
return ret.replace('\n', '')
def MdComments(self,g):#Process Markdown
Markdown.setPre("http://schema.org/")
for s,p,o in list(g.triples( (None, RDFS.comment, None) )):
no = Markdown.parse(o) #g.remove((s,p,o))
g.set((s,p,Literal(no)))
Markdown.setPre()
def buildSingleTermGraph(node, excludeAttic=True, markdown=True):
g = rdflib.Graph()
g.bind('owl', 'http://www.w3.org/2002/07/owl#')
g.bind('rdfa', 'http://www.w3.org/ns/rdfa#')
g.bind('dct', 'http://purl.org/dc/terms/')
g.bind('schema', 'http://schema.org/')
full = "http://schema.org/" + node
#n = URIRef(full)
n = SCHEMA.term(node)
n = n
full = str(n)
q = queryGraph()
ret = None
#log.info("NAME %s %s"% (n,full))
atts = None
try:
RDFLIBLOCK.acquire()
atts = list(
q.triples((n, SCHEMA.isPartOf, URIRef("http://attic.schema.org"))))
finally:
RDFLIBLOCK.release()
if len(atts):
#log.info("ATTIC TERM %s" % n)
excludeAttic = False
#Outgoing triples
try:
RDFLIBLOCK.acquire()
ret = list(q.triples((n, None, None)))
finally:
RDFLIBLOCK.release()
for (s, p, o) in ret:
#log.info("adding %s %s %s" % (s,p,o))
g.add((s, p, o))
#Incoming triples
ret = list(q.triples((None, None, n)))
for (s, p, o) in ret:
#log.info("adding %s %s %s" % (s,p,o))
g.add((s, p, o))
#super classes
query = '''select * where {
?term (^rdfs:subClassOf*) <%s>.
?term rdfs:subClassOf ?super.
?super ?pred ?obj.
}''' % n
ret = rdfQueryStore(query, q)
for row in ret:
#log.info("adding %s %s %s" % (row.term,RDFS.subClassOf,row.super))
g.add((row.term, RDFS.subClassOf, row.super))
g.add((row.super, row.pred, row.obj))
#poperties with superclasses in domain
query = '''select * where{
?term (^rdfs:subClassOf*) <%s>.
?prop <http://schema.org/domainIncludes> ?term.
?prop ?pred ?obj.
}
''' % n
ret = rdfQueryStore(query, q)
for row in ret:
g.add((row.prop, SCHEMA.domainIncludes, row.term))
g.add((row.prop, row.pred, row.obj))
#super properties
query = '''select * where {
?term (^rdfs:subPropertyOf*) <%s>.
?term rdfs:subPropertyOf ?super.
?super ?pred ?obj.
}''' % n
ret = rdfQueryStore(query, q)
for row in ret:
#log.info("adding %s %s %s" % (row.term,RDFS.subPropertyOf,row.super))
g.add((row.term, RDFS.subPropertyOf, row.super))
g.add((row.super, row.pred, row.obj))
#Enumeration for an enumeration value
query = '''select * where {
<%s> a ?type.
?type ?pred ?obj.
FILTER NOT EXISTS{?type a rdfs:class}.
}''' % n
ret = rdfQueryStore(query, q)
for row in ret:
#log.info("adding %s %s %s" % (row.type,row.pred,row.obj))
g.add((row.type, row.pred, row.obj))
if excludeAttic: #Remove triples referencing terms part of http://attic.schema.org
trips = list(g.triples((None, None, None)))
try:
RDFLIBLOCK.acquire()
for (s, p, o) in trips:
atts = list(
q.triples((s, SCHEMA.isPartOf,
URIRef("http://attic.schema.org"))))
if isinstance(o, URIRef):
atts.extend(
q.triples((o, SCHEMA.isPartOf,
URIRef("http://attic.schema.org"))))
for (rs, rp, ro) in atts:
#log.info("Removing %s" % rs)
g.remove((rs, None, None))
g.remove((None, None, rs))
finally:
RDFLIBLOCK.release()
if markdown:
try:
RDFLIBLOCK.acquire()
trips = list(g.triples((None, RDFS.comment, None)))
Markdown.setPre("http://schema.org/")
for (s, p, com) in trips:
mcom = Markdown.parse(com)
g.remove((s, p, com))
g.add((s, p, Literal(mcom)))
finally:
RDFLIBLOCK.release()
Markdown.setPre()
return g
def buildSingleTermGraph(node,excludeAttic=True,markdown=True):
g = rdflib.Graph()
g.bind('owl', 'http://www.w3.org/2002/07/owl#')
g.bind('rdfa', 'http://www.w3.org/ns/rdfa#')
g.bind('dct', 'http://purl.org/dc/terms/')
g.bind('schema', 'http://schema.org/')
full = "http://schema.org/" + node
#n = URIRef(full)
n = SCHEMA.term(node)
n = n
full = str(n)
q = queryGraph()
ret = None
#log.info("NAME %s %s"% (n,full))
atts = None
try:
RDFLIBLOCK.acquire()
atts = list(q.triples((n,SCHEMA.isPartOf,URIRef("http://attic.schema.org"))))
finally:
RDFLIBLOCK.release()
if len(atts):
#log.info("ATTIC TERM %s" % n)
excludeAttic = False
#Outgoing triples
try:
RDFLIBLOCK.acquire()
ret = list(q.triples((n,None,None)))
finally:
RDFLIBLOCK.release()
for (s,p,o) in ret:
#log.info("adding %s %s %s" % (s,p,o))
g.add((s,p,o))
#Incoming triples
ret = list(q.triples((None,None,n)))
for (s,p,o) in ret:
#log.info("adding %s %s %s" % (s,p,o))
g.add((s,p,o))
#super classes
query='''select * where {
?term (^rdfs:subClassOf*) <%s>.
?term rdfs:subClassOf ?super.
?super ?pred ?obj.
}''' % n
ret = rdfQueryStore(query,q)
for row in ret:
#log.info("adding %s %s %s" % (row.term,RDFS.subClassOf,row.super))
g.add((row.term,RDFS.subClassOf,row.super))
g.add((row.super,row.pred,row.obj))
#poperties with superclasses in domain
query='''select * where{
?term (^rdfs:subClassOf*) <%s>.
?prop <http://schema.org/domainIncludes> ?term.
?prop ?pred ?obj.
}
''' % n
ret = rdfQueryStore(query,q)
for row in ret:
g.add((row.prop,SCHEMA.domainIncludes,row.term))
g.add((row.prop,row.pred,row.obj))
#super properties
query='''select * where {
?term (^rdfs:subPropertyOf*) <%s>.
?term rdfs:subPropertyOf ?super.
?super ?pred ?obj.
}''' % n
ret = rdfQueryStore(query,q)
for row in ret:
#log.info("adding %s %s %s" % (row.term,RDFS.subPropertyOf,row.super))
g.add((row.term,RDFS.subPropertyOf,row.super))
g.add((row.super,row.pred,row.obj))
#Enumeration for an enumeration value
query='''select * where {
<%s> a ?type.
?type ?pred ?obj.
FILTER NOT EXISTS{?type a rdfs:class}.
}''' % n
ret = rdfQueryStore(query,q)
for row in ret:
#log.info("adding %s %s %s" % (row.type,row.pred,row.obj))
g.add((row.type,row.pred,row.obj))
if excludeAttic: #Remove triples referencing terms part of http://attic.schema.org
trips = list(g.triples((None,None,None)))
try:
RDFLIBLOCK.acquire()
for (s,p,o) in trips:
atts = list(q.triples((s,SCHEMA.isPartOf,URIRef("http://attic.schema.org"))))
if isinstance(o, URIRef):
atts.extend(q.triples((o,SCHEMA.isPartOf,URIRef("http://attic.schema.org"))))
for (rs,rp,ro) in atts:
#log.info("Removing %s" % rs)
g.remove((rs,None,None))
g.remove((None,None,rs))
finally:
RDFLIBLOCK.release()
if markdown:
try:
RDFLIBLOCK.acquire()
trips = list(g.triples((None,RDFS.comment,None)))
Markdown.setPre("http://schema.org/")
for (s,p,com) in trips:
mcom = Markdown.parse(com)
g.remove((s,p,com))
g.add((s,p,Literal(mcom)))
finally:
RDFLIBLOCK.release()
Markdown.setPre()
return g
def buildSingleTermGraph(node, excludeAttic=True, markdown=True):
q = queryGraph()
g = rdflib.Graph()
ns = q.namespaces()
for n in ns:
prefix, path = n
namespaceAdd(g, prefix=prefix, path=path)
namespaceAdd(g, api.SdoConfig.prefix(), api.SdoConfig.vocabUri())
full = "%s%s" % (api.SdoConfig.vocabUri(), node)
n = URIRef(full)
full = str(n)
ret = None
#log.info("NAME %s %s"% (n,full))
atts = None
attic = api.SdoConfig.atticUri()
if attic:
with RDFLIBLOCK:
atts = list(q.triples((n, SCHEMA.isPartOf, URIRef(attic))))
if len(atts):
#log.info("ATTIC TERM %s" % n)
excludeAttic = False
#Outgoing triples
with RDFLIBLOCK:
ret = list(q.triples((n, None, None)))
for (s, p, o) in ret:
#log.info("adding %s %s %s" % (s,p,o))
g.add((s, p, o))
#Incoming triples
with RDFLIBLOCK:
ret = list(q.triples((None, None, n)))
for (s, p, o) in ret:
#log.info("adding %s %s %s" % (s,p,o))
g.add((s, p, o))
#super classes
query = '''select * where {
?term (^rdfs:subClassOf*) <%s>.
?term rdfs:subClassOf ?super.
OPTIONAL {
?super ?pred ?obj.
FILTER (strstarts(str(?super),'%s'))
}
}
''' % (n, api.SdoConfig.vocabUri())
#log.info("Query: %s" % query)
ret = rdfQueryStore(query, q)
for row in ret:
#log.info("adding %s %s %s" % (row.term,RDFS.subClassOf,row.super))
g.add((row.term, RDFS.subClassOf, row.super))
pred = row.pred
obj = row.obj
if pred and obj:
g.add((row.super, row.pred, row.obj))
#poperties with superclasses in domain
query = '''select * where{
?term (^rdfs:subClassOf*) <%s>.
?prop <http://schema.org/domainIncludes> ?term.
OPTIONAL {
?prop ?pred ?obj.
FILTER (strstarts(str(?prop),'%s'))
}
}
''' % (n, api.SdoConfig.vocabUri())
#log.info("Query: %s" % query)
ret = rdfQueryStore(query, q)
for row in ret:
g.add((row.prop, SCHEMA.domainIncludes, row.term))
pred = row.pred
obj = row.obj
if pred and obj:
g.add((row.prop, row.pred, row.obj))
#super properties
query = '''select * where {
?term (^rdfs:subPropertyOf*) <%s>.
?term rdfs:subPropertyOf ?super.
OPTIONAL {
?super ?pred ?obj.
FILTER (strstarts(str(?super),'%s'))
}
}
''' % (n, api.SdoConfig.vocabUri())
#log.info("Query: %s" % query)
ret = rdfQueryStore(query, q)
for row in ret:
#log.info("adding %s %s %s" % (row.term,RDFS.subPropertyOf,row.super))
g.add((row.term, RDFS.subPropertyOf, row.super))
pred = row.pred
obj = row.obj
if pred and obj:
g.add((row.super, row.pred, row.obj))
#Enumeration for an enumeration value
query = '''select * where {
<%s> a ?type.
?type ?pred ?obj.
FILTER NOT EXISTS{?type a rdfs:class}.
}''' % n
ret = rdfQueryStore(query, q)
for row in ret:
#log.info("adding %s %s %s" % (row.type,row.pred,row.obj))
g.add((row.type, row.pred, row.obj))
if excludeAttic: #Remove triples referencing terms part of http://attic.schema.org
trips = list(g.triples((None, None, None)))
with RDFLIBLOCK:
for (s, p, o) in trips:
atts = list(q.triples((s, SCHEMA.isPartOf, URIRef(attic))))
if isinstance(o, URIRef):
atts.extend(q.triples((o, SCHEMA.isPartOf, URIRef(attic))))
for (rs, rp, ro) in atts:
#log.info("Removing %s" % rs)
g.remove((rs, None, None))
g.remove((None, None, rs))
if markdown:
with RDFLIBLOCK:
trips = list(g.triples((None, RDFS.comment, None)))
Markdown.setPre(api.SdoConfig.vocabUri())
for (s, p, com) in trips:
mcom = Markdown.parse(com)
g.remove((s, p, com))
g.add((s, p, Literal(mcom)))
Markdown.setPre()
return g
def buildSingleTermGraph(node,excludeAttic=True,markdown=True):
q = queryGraph()
g = rdflib.Graph()
ns = q.namespaces()
for n in ns:
prefix, path = n
namespaceAdd(g,prefix=prefix,path=path)
namespaceAdd(g,api.SdoConfig.prefix(),api.SdoConfig.vocabUri())
full = "%s%s" % (api.SdoConfig.vocabUri(), node)
n = URIRef(full)
full = str(n)
ret = None
#log.info("NAME %s %s"% (n,full))
atts = None
attic = api.SdoConfig.atticUri()
if attic:
with RDFLIBLOCK:
atts = list(q.triples((n,SCHEMA.isPartOf,URIRef(attic))))
if len(atts):
#log.info("ATTIC TERM %s" % n)
excludeAttic = False
#Outgoing triples
with RDFLIBLOCK:
ret = list(q.triples((n,None,None)))
for (s,p,o) in ret:
#log.info("adding %s %s %s" % (s,p,o))
g.add((s,p,o))
#Incoming triples
with RDFLIBLOCK:
ret = list(q.triples((None,None,n)))
for (s,p,o) in ret:
#log.info("adding %s %s %s" % (s,p,o))
g.add((s,p,o))
#super classes
query='''select * where {
?term (^rdfs:subClassOf*) <%s>.
?term rdfs:subClassOf ?super.
OPTIONAL {
?super ?pred ?obj.
FILTER (strstarts(str(?super),'%s'))
}
}
''' % (n,api.SdoConfig.vocabUri())
#log.info("Query: %s" % query)
ret = rdfQueryStore(query,q)
for row in ret:
#log.info("adding %s %s %s" % (row.term,RDFS.subClassOf,row.super))
g.add((row.term,RDFS.subClassOf,row.super))
pred = row.pred
obj = row.obj
if pred and obj:
g.add((row.super,row.pred,row.obj))
#poperties with superclasses in domain
query='''select * where{
?term (^rdfs:subClassOf*) <%s>.
?prop <http://schema.org/domainIncludes> ?term.
OPTIONAL {
?prop ?pred ?obj.
FILTER (strstarts(str(?prop),'%s'))
}
}
''' % (n,api.SdoConfig.vocabUri())
#log.info("Query: %s" % query)
ret = rdfQueryStore(query,q)
for row in ret:
g.add((row.prop,SCHEMA.domainIncludes,row.term))
pred = row.pred
obj = row.obj
if pred and obj:
g.add((row.prop,row.pred,row.obj))
#super properties
query='''select * where {
?term (^rdfs:subPropertyOf*) <%s>.
?term rdfs:subPropertyOf ?super.
OPTIONAL {
?super ?pred ?obj.
FILTER (strstarts(str(?super),'%s'))
}
}
''' % (n,api.SdoConfig.vocabUri())
#log.info("Query: %s" % query)
ret = rdfQueryStore(query,q)
for row in ret:
#log.info("adding %s %s %s" % (row.term,RDFS.subPropertyOf,row.super))
g.add((row.term,RDFS.subPropertyOf,row.super))
pred = row.pred
obj = row.obj
if pred and obj:
g.add((row.super,row.pred,row.obj))
#Enumeration for an enumeration value
query='''select * where {
<%s> a ?type.
?type ?pred ?obj.
FILTER NOT EXISTS{?type a rdfs:class}.
}''' % n
ret = rdfQueryStore(query,q)
for row in ret:
#log.info("adding %s %s %s" % (row.type,row.pred,row.obj))
g.add((row.type,row.pred,row.obj))
if excludeAttic: #Remove triples referencing terms part of http://attic.schema.org
trips = list(g.triples((None,None,None)))
with RDFLIBLOCK:
for (s,p,o) in trips:
atts = list(q.triples((s,SCHEMA.isPartOf,URIRef(attic))))
if isinstance(o, URIRef):
atts.extend(q.triples((o,SCHEMA.isPartOf,URIRef(attic))))
for (rs,rp,ro) in atts:
#log.info("Removing %s" % rs)
g.remove((rs,None,None))
g.remove((None,None,rs))
if markdown:
with RDFLIBLOCK:
trips = list(g.triples((None,RDFS.comment,None)))
Markdown.setPre(api.SdoConfig.vocabUri())
for (s,p,com) in trips:
mcom = Markdown.parse(com)
g.remove((s,p,com))
g.add((s,p,Literal(mcom)))
Markdown.setPre()
return g