[Release] v0.1.3 (#288)

* 0.1.2 release

* oops

* more fixes on windows

* [Bugfix] fix download dir (#275)

* fix download dir

* add doc for the env var

* windows 7 -> 10

* doc update

* [Bugfix] Fix conversion from networkx (#286)

* fix from_nx when no edge id available

* add test cases

* more detailed tests

* more comments

* [Bugfix] Switch to sparse_coo_matrix for torch 1.0+ (#282)

* switch to sparse_coo_matrix for torch 1.0+

* fix bug when the version is 0.4.1.post2

* change to distutils

python/dgl/backend/pytorch/tensor.py

 from __future__ import absolute_import
 
+from distutils.version import LooseVersion
+
 import torch as th
 from torch.utils import dlpack
 
+TH_VERSION = LooseVersion(th.__version__)
+
 def data_type_dict():
     return {'float16' : th.float16,
             'float32' : th.float32,
@@ -19,7 +23,8 @@ def cpu():
 def tensor(data, dtype=None):
     return th.tensor(data, dtype=dtype)
 
-def sparse_matrix(data, index, shape, force_format=False):
+if TH_VERSION.version[0] == 0:
+    def sparse_matrix(data, index, shape, force_format=False):
         fmt = index[0]
         if fmt != 'coo':
             raise TypeError('Pytorch backend only supports COO format. But got %s.' % fmt)
@@ -27,6 +32,15 @@ def sparse_matrix(data, index, shape, force_format=False):
         spmat = th._sparse_coo_tensor_unsafe(index[1], data, shape)
         # No conversion is required.
         return spmat, None
+else:
+    # VERSION 1.0+
+    def sparse_matrix(data, index, shape, force_format=False):
+        fmt = index[0]
+        if fmt != 'coo':
+            raise TypeError('Pytorch backend only supports COO format. But got %s.' % fmt)
+        spmat = th.sparse_coo_tensor(index[1], data, shape)
+        # No conversion is required.
+        return spmat, None
 
 def sparse_matrix_indices(spmat):
     return ('coo', spmat._indices())

python/dgl/graph.py

@@ -1167,6 +1167,7 @@ class DGLGraph(object):
             else:
                 return F.tensor(lst)
         if node_attrs is not None:
+            # mapping from feature name to a list of tensors to be concatenated
             attr_dict = defaultdict(list)
             for nid in range(self.number_of_nodes()):
                 for attr in node_attrs:
@@ -1175,14 +1176,19 @@ class DGLGraph(object):
                 self._node_frame[attr] = _batcher(attr_dict[attr])
         if edge_attrs is not None:
             has_edge_id = 'id' in next(iter(nx_graph.edges(data=True)))[-1]
+            # mapping from feature name to a list of tensors to be concatenated
             attr_dict = defaultdict(lambda: [None] * self.number_of_edges())
+            # each defaultdict value is initialized to be a list of None
+            # None here serves as placeholder to be replaced by feature with
+            # corresponding edge id
             if has_edge_id:
                 for _, _, attrs in nx_graph.edges(data=True):
                     for key in edge_attrs:
                         attr_dict[key][attrs['id']] = attrs[key]
             else:
                 # XXX: assuming networkx iteration order is deterministic
-                for eid, (_, _, attr) in enumerate(nx_graph.edges(data=True)):
+                #      so the order is the same as graph_index.from_networkx
+                for eid, (_, _, attrs) in enumerate(nx_graph.edges(data=True)):
                     for key in edge_attrs:
                         attr_dict[key][eid] = attrs[key]
             for attr in edge_attrs:

tests/pytorch/test_basics.py

@@ -142,6 +142,8 @@ def test_nx_conversion():
     # check conversion between networkx and DGLGraph
 
     def _check_nx_feature(nxg, nf, ef):
+        # check node and edge feature of nxg
+        # this is used to check to_networkx
         num_nodes = len(nxg)
         num_edges = nxg.size()
         if num_nodes > 0:
@@ -185,18 +187,24 @@ def test_nx_conversion():
     assert nxg.size() == 4
     _check_nx_feature(nxg, {'n1': n1, 'n3': n3}, {'e1': e1, 'e2': e2})
 
-    # convert to DGLGraph
+    # convert to DGLGraph, nx graph has id in edge feature
     # use id feature to test non-tensor copy
     g.from_networkx(nxg, node_attrs=['n1'], edge_attrs=['e1', 'id'])
+    # check graph size
     assert g.number_of_nodes() == 5
     assert g.number_of_edges() == 4
-    assert U.allclose(g.get_n_repr()['n1'], n1)
-    assert U.allclose(g.get_e_repr()['e1'], e1)
+    # check number of features
+    # test with existing dglgraph (so existing features should be cleared)
+    assert len(g.ndata) == 1
+    assert len(g.edata) == 2
+    # check feature values
+    assert U.allclose(g.ndata['n1'], n1)
+    # with id in nx edge feature, e1 should follow original order
+    assert U.allclose(g.edata['e1'], e1)
     assert th.equal(g.get_e_repr()['id'], th.arange(4))
 
-    g.pop_e_repr('id')
-
-    # test modifying DGLGraph
+    # test conversion after modifying DGLGraph
+    g.pop_e_repr('id') # pop id so we don't need to provide id when adding edges
     new_n = th.randn(2, 3)
     new_e = th.randn(3, 5)
     g.add_nodes(2, data={'n1': new_n})
@@ -210,6 +218,28 @@ def test_nx_conversion():
     assert nxg.size() == 7
     _check_nx_feature(nxg, {'n1': n1}, {'e1': e1})
 
+    # now test convert from networkx without id in edge feature
+    # first pop id in edge feature
+    for _, _, attr in nxg.edges(data=True):
+        attr.pop('id')
+    # test with a new graph
+    g = DGLGraph(multigraph=True)
+    g.from_networkx(nxg, node_attrs=['n1'], edge_attrs=['e1'])
+    # check graph size
+    assert g.number_of_nodes() == 7
+    assert g.number_of_edges() == 7
+    # check number of features
+    assert len(g.ndata) == 1
+    assert len(g.edata) == 1
+    # check feature values
+    assert U.allclose(g.ndata['n1'], n1)
+    # edge feature order follows nxg.edges()
+    edge_feat = []
+    for _, _, attr in nxg.edges(data=True):
+        edge_feat.append(attr['e1'].unsqueeze(0))
+    edge_feat = th.cat(edge_feat, dim=0)
+    assert U.allclose(g.edata['e1'], edge_feat)
+
 def test_batch_send():
     g = generate_graph()
     def _fmsg(edges):