def generate_predicate_URI( self, name, context ) :
"""
Generate a full URI for a predicate, using the type, the vocabulary, etc.
For details of this entry, see Section 4.4
@param name: name of the property, ie, what appears in @itemprop
@param context: an instance of an evaluation context
@type context: L{Evaluation_Context}
"""
if debug: print "name: %s, %s" % (name,context)
# Step 1: absolute URI-s are fine, take them as they are
if is_absolute_URI(name) : return name
# Step 2: if type is none, that this is just used as a fragment
# if not context.current_type :
if context.current_type == None and context.current_vocabulary == None :
return generate_URI( self.base, name )
#if context.current_type == None :
# return generate_URI( self.base, name )
# Step 3: set the scheme
try :
if context.current_vocabulary in registry and "propertyURI" in registry[context.current_vocabulary] :
scheme = registry[context.current_vocabulary]["propertyURI"]
else :
scheme = PropertySchemes.vocabulary
except :
# This is when the structure of the registry is broken
scheme = PropertySchemes.vocabulary
name = fragment_escape( name )
if scheme == PropertySchemes.contextual :
# Step 5.1
s = context.current_name
# s = context.current_type
if s != None and s.startswith("http://www.w3.org/ns/md?type=") :
# Step 5.2
return s + '.' + name
else :
# Step 5.3
return "http://www.w3.org/ns/md?type=" + fragment_escape(context.current_type) + "&prop=" + name
else :
# Step 4
if context.current_vocabulary[-1] == '#' or context.current_vocabulary[-1] == '/' :
return context.current_vocabulary + name
else :
return context.current_vocabulary + '#' + name
def generate_predicate_URI( self, name, context ) :
"""
Generate a full URI for a predicate, using the type, the vocabulary, etc.
For details of this entry, see Section 4.4
@param name: name of the property, ie, what appears in @itemprop
@param context: an instance of an evaluation context
@type context: L{Evaluation_Context}
"""
if debug: print( "name: %s, %s" % (name,context) )
# Step 1: absolute URI-s are fine, take them as they are
if is_absolute_URI(name) : return name
# Step 2: if type is none, that this is just used as a fragment
# if not context.current_type :
if context.current_type == None and context.current_vocabulary == None :
if self.base[-1] == '#' :
b = self.base[:-1]
else :
b = self.base
return b + '#' + fragment_escape(name)
#if context.current_type == None :
# return generate_URI( self.base, name )
# Step 3: set the scheme
try :
if context.current_vocabulary in registry and "propertyURI" in registry[context.current_vocabulary] :
scheme = registry[context.current_vocabulary]["propertyURI"]
else :
scheme = PropertySchemes.vocabulary
except :
# This is when the structure of the registry is broken
scheme = PropertySchemes.vocabulary
name = fragment_escape( name )
if scheme == PropertySchemes.contextual :
# Step 5.1
s = context.current_name
# s = context.current_type
if s != None and s.startswith("http://www.w3.org/ns/md?type=") :
# Step 5.2
expandedURI = s + '.' + name
else :
# Step 5.3
expandedURI = "http://www.w3.org/ns/md?type=" + fragment_escape(context.current_type) + "&prop=" + name
else :
# Step 4
if context.current_vocabulary[-1] == '#' or context.current_vocabulary[-1] == '/' :
expandedURI = context.current_vocabulary + name
else :
expandedURI = context.current_vocabulary + '#' + name
# see if there are subproperty/equivalentproperty relations
try :
vocab_mapping = registry[context.current_vocabulary]["properties"][name]
# if we got that far, we may have some mappings
expandedURIRef = URIRef(expandedURI)
try :
subpr = vocab_mapping["subPropertyOf"]
if subpr != None :
if isinstance(subpr,list) :
for p in subpr :
self.graph.add( (expandedURIRef, ns_rdfs["subPropertyOf"], URIRef(p)) )
else :
self.graph.add( (expandedURIRef, ns_rdfs["subPropertyOf"], URIRef(subpr)) )
except :
# Ok, no sub property
pass
try :
subpr = vocab_mapping["equivalentProperty"]
if subpr != None :
if isinstance(subpr,list) :
for p in subpr :
self.graph.add( (expandedURIRef, ns_owl["equivalentProperty"], URIRef(p)) )
else :
self.graph.add( (expandedURIRef, ns_owl["equivalentProperty"], URIRef(subpr)) )
except :
# Ok, no sub property
pass
except :
# no harm done, no extra vocabulary term
pass
return expandedURI
def generate_triples( self, item, context ) :
"""
Generate the triples for a specific item. See the W3C Note for the details.
@param item: the DOM Node for the specific item
@type item: DOM Node
@param context: an instance of an evaluation context
@type context: L{Evaluation_Context}
@return: a URIRef or a BNode for the (RDF) subject
"""
# Step 1,2: if the subject has to be set, store it in memory
subject = context.get_memory( item )
if subject == None :
# nop, there is no subject set. If there is a valid @itemid, that carries it
if item.hasAttribute("itemid") and is_absolute_URI( item.getAttribute("itemid") ):
subject = URIRef( item.getAttribute("itemid").strip() )
else :
subject = BNode()
context.set_memory( item, subject )
# Step 3: set the type triples if any
types = []
if item.hasAttribute("itemtype") :
types = item.getAttribute("itemtype").strip().split()
for t in types :
if is_absolute_URI( t ) :
self.graph.add( (subject, ns_rdf["type"], URIRef(t)) )
# Step 4, 5 and 6 to set the typing variable
if len(types) == 0 :
itype = None
else :
if is_absolute_URI(types[0]) :
itype = types[0]
context.current_name = None
elif context.current_type != None :
itype = context.current_type
else :
itype = None
# Step 7, 8, 9: Check the registry for possible keys and set the vocab
vocab = None
if itype != None :
for key in list(registry.keys()) :
if itype.startswith(key) :
# There is a predefined vocabulary for this type...
vocab = key
# Step 7: Issue an rdfa usesVocabulary triple
self.graph.add( (URIRef(self.base), RDFA_VOCAB, URIRef(vocab)))
self.vocabularies_used = True
break
# The registry has not set the vocabulary; has to be extracted from the type
if vocab == None :
parsed = urlsplit(itype)
if parsed.fragment != "" :
vocab = urlunsplit( (parsed.scheme,parsed.netloc,parsed.path,parsed.query,"") ) + '#'
elif parsed.path == "" and parsed.query == "" :
vocab = itype
if vocab[-1] != '/' : vocab += '/'
else :
vocab = itype.rsplit('/',1)[0] + '/'
# Step 9: update vocab in the context
if vocab != None :
context.current_vocabulary = vocab
elif item.hasAttribute("itemtype") :
context.current_vocabulary = None
# Step 10: set up a property list; this will be used to generate triples later.
# each entry in the dictionary is an array of RDF objects
property_list = {}
# Step 11: Get the item properties and run a cycle on those
for prop in self.get_item_properties(item) :
for name in prop.getAttribute("itemprop").strip().split() :
# 11.1.1. set a new context
new_context = context.new_copy(itype)
# 11.1.2, generate the URI for the property name, that will be the predicate
# Also update the context
new_context.current_name = predicate = self.generate_predicate_URI( name,new_context )
# 11.1.3, generate the property value. The extra flag signals that the value is a new item
# Note that 10.1.4 step is done in the method itself, ie, a recursion may occur there
# if a new item is hit (in which case the return value is a RDF resource chaining to a subject)
value = self.get_property_value( prop, new_context )
# 11.1.5, store all the values
if predicate in property_list :
property_list[predicate].append(value)
else :
property_list[predicate] = [ value ]
# step 12: generate the triples
for property in list(property_list.keys()) :
self.generate_property_values( subject, URIRef(property), property_list[property], context )
# Step 13: return the subject to the caller
return subject