Merge pull request #691 from RDFLib/fix_609

Added __iadd__ method to collections, much faster for adding multiple…

rdflib/collection.py

@@ -1,7 +1,6 @@
 from rdflib.namespace import RDF
 from rdflib.term import BNode
 from rdflib.term import Literal
-from rdflib.graph import Graph
 from rdflib.py3compat import format_doctest_out
 
 __all__ = ['Collection']
@@ -43,8 +42,7 @@ class Collection(object):
     def __init__(self, graph, uri, seq=[]):
         self.graph = graph
         self.uri = uri or BNode()
-        for item in seq:
-            self.append(item)
+        self += seq
 
     def n3(self):
         """
@@ -184,6 +182,16 @@ def __iter__(self):
         """Iterator over items in Collections"""
         return self.graph.items(self.uri)
 
+    def _end(self):
+        # find end of list
+        container = self.uri
+        while True:
+            rest = self.graph.value(container, RDF.rest)
+            if rest == None or rest == RDF.nil:
+                return container
+            else:
+                container = rest
+
     def append(self, item):
         """
         >>> from rdflib.graph import Graph
@@ -192,29 +200,36 @@ def append(self, item):
         >>> c = Collection(g,listName,[Literal(1),Literal(2)])
         >>> links = [
         ...     list(g.subjects(object=i, predicate=RDF.first))[0] for i in c]
-        >>> len([i for i in links if (i,RDF.rest, RDF.nil) in g])
+        >>> len([i for i in links if (i, RDF.rest, RDF.nil) in g])
         1
 
         """
-        container = self.uri
-        graph = self.graph
-        # iterate to the end of the linked list
-        rest = graph.value(container, RDF.rest)
-        while rest:
-            if rest == RDF.nil:
-                # the end, append to the end of the linked list
-                node = BNode()
-                graph.set((container, RDF.rest, node))
-                container = node
-                break
-            else:
-                # move down one link
-                if container != self.uri:
-                    rest = graph.value(rest, RDF.rest)
-                if not rest == RDF.nil:
-                    container = rest
-        graph.add((container, RDF.first, item))
-        graph.add((container, RDF.rest, RDF.nil))
+
+        end = self._end()
+        if (end, RDF.first, None) in self.graph:
+            # append new node to the end of the linked list
+            node = BNode()
+            self.graph.set((end, RDF.rest, node))
+            end = node
+
+        self.graph.add((end, RDF.first, item))
+        self.graph.add((end, RDF.rest, RDF.nil))
+
+    def __iadd__(self, other):
+
+        end = self._end()
+        self.graph.remove((end, RDF.rest, None))
+
+        for item in other:
+            if (end, RDF.first, None) in self.graph:
+                nxt = BNode()
+                self.graph.add((end, RDF.rest, nxt))
+                end = nxt
+
+            self.graph.add((end, RDF.first, item))
+
+
+        self.graph.add((end, RDF.rest, RDF.nil))
 
     def clear(self):
         container = self.uri
@@ -233,6 +248,7 @@ def test():
 if __name__ == "__main__":
     test()
 
+    from rdflib import Graph
     g = Graph()
 
     c = Collection(g, BNode())

rdflib/graph.py

@@ -266,6 +266,7 @@
 from rdflib.parser import create_input_source
 from rdflib.namespace import NamespaceManager
 from rdflib.resource import Resource
+from rdflib.collection import Collection
 from rdflib import py3compat
 b = py3compat.b
 
@@ -1177,6 +1178,28 @@ def all_nodes(self):
         res.update(self.subjects())
         return res
 
+    def collection(self, identifier):
+        """Create a new ``Collection`` instance.
+
+        Parameters:
+
+        - ``identifier``: a URIRef or BNode instance.
+
+        Example::
+
+            >>> graph = Graph()
+            >>> uri = URIRef("http://example.org/resource")
+            >>> collection = graph.collection(uri)
+            >>> assert isinstance(collection, Collection)
+            >>> assert collection.uri is uri
+            >>> assert collection.graph is graph
+            >>> collection += [ Literal(1), Literal(2) ]
+        """
+
+        return Collection(self, identifier)
+
+
+
     def resource(self, identifier):
         """Create a new ``Resource`` instance.
 

test/test_issue223.py

@@ -11,9 +11,9 @@ def test_collection_with_duplicates():
     g = Graph().parse(data=ttl, format="turtle")
     for _,_,o in g.triples((URIRef("http://example.org/s"), URIRef("http://example.org/p"), None)):
         break
-    c = Collection(g, o)
+    c = g.collection(o)
     assert list(c) == list(URIRef("http://example.org/" + x) for x in ["a", "b", "a"])
     assert len(c) == 3
-    
+
 if __name__ == '__main__':
     test_collection_with_duplicates()