~tfardet/NNGT

ref: f2dc429c7770e3a8e5f7a0c1c107c05bfe9458d2 NNGT/nngt/core/nx_graph.py -rwxr-xr-x 29.3 KiB
f2dc429cTanguy Fardet Bugfix - Spectral radius, NEST support 4 months ago
                                                                                
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
#-*- coding:utf-8 -*-
#
# core/nx_graph.py
#
# This file is part of the NNGT project, a graph-library for standardized and
# and reproducible graph analysis: generate and analyze networks with your
# favorite graph library (graph-tool/igraph/networkx) on any platform, without
# any change to your code.
# Copyright (C) 2015-2021 Tanguy Fardet
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.

""" Networkx subclassing """

from collections import OrderedDict, deque
from copy import deepcopy
from itertools import chain
import logging
import sys

import numpy as np
import scipy.sparse as ssp

import nngt
from nngt.lib import InvalidArgument, BWEIGHT, nonstring_container, is_integer
from nngt.lib.connect_tools import (_cleanup_edges, _set_dist_new_edges,
                                    _set_default_edge_attributes)
from nngt.lib.graph_helpers import (_get_dtype, _get_nx_weights,
                                    _post_del_update)
from nngt.lib.converters import _np_dtype, _to_np_array
from nngt.lib.logger import _log_message
from .graph_interface import GraphInterface, BaseProperty


logger = logging.getLogger(__name__)


# ---------- #
# Properties #
# ---------- #

class _NxNProperty(BaseProperty):

    '''
    Class for generic interactions with nodes properties (networkx)
    '''

    def __getitem__(self, name):
        g = self.parent()._graph

        lst = [g.nodes[i][name] for i in range(g.number_of_nodes())]

        dtype = _np_dtype(super(_NxNProperty, self).__getitem__(name))

        return _to_np_array(lst, dtype=dtype)

    def __setitem__(self, name, value):
        g    = self.parent()._graph
        size = g.number_of_nodes()

        if name in self:
            if len(value) == size:
                for i in range(size):
                    g.nodes[i][name] = value[i]
            else:
                raise ValueError("A list or a np.array with one entry per "
                                 "node in the graph is required")
        else:
            raise InvalidArgument("Attribute does not exist yet, use "
                                  "set_attribute to create it.")

    def new_attribute(self, name, value_type, values=None, val=None):
        g = self.parent()._graph

        if val is None:
            if value_type in ("int", "integer"):
                val = int(0)
            elif value_type in ("double", "float"):
                val = np.NaN
            elif value_type == "string":
                val = ""
            else:
                val = None
                value_type = "object"

        if values is None:
            values = [deepcopy(val)
                      for _ in range(g.number_of_nodes())]

        # store name and value type in the dict
        super(_NxNProperty, self).__setitem__(name, value_type)
        # store the real values in the attribute
        self[name] = values
        self._num_values_set[name] = len(values)

    def set_attribute(self, name, values, nodes=None):
        '''
        Set the node attribute.

        Parameters
        ----------
        name : str
            Name of the node attribute.
        values : array, size N
            Values that should be set.
        nodes : array-like, optional (default: all nodes)
            Nodes for which the value of the property should be set. If `nodes`
            is not None, it must be an array of size N.
        '''
        g = self.parent()._graph

        num_nodes = g.number_of_nodes()
        num_n = len(nodes) if nodes is not None else num_nodes

        if num_n == num_nodes:
            self[name] = values
        else:
            if num_n != len(values):
                raise ValueError("`nodes` and `values` must have the same "
                                 "size; got respectively " + str(num_n) + \
                                 " and " + str(len(values)) + " entries.")
            else:
                for n, val in zip(nodes, values):
                    g.nodes[n][name] = val
        self._num_values_set[name] = num_nodes


class _NxEProperty(BaseProperty):

    ''' Class for generic interactions with edge properties (networkx)  '''

    def __getitem__(self, name):
        g = self.parent()._graph

        dtype = _np_dtype(super().__getitem__(name))
        eprop = np.empty(g.number_of_edges(), dtype=dtype)

        edges = list(g.edges(data=name))

        if len(edges):
            eids = np.asarray(list(g.edges(data="eid")))[:, 2]

            for i, eid in enumerate(np.argsort(eids)):
                eprop[i] = edges[eid][2]

        return eprop

    def __setitem__(self, name, value):
        g = self.parent()._graph

        if name in self:
            size = g.number_of_edges()

            if len(value) == size:
                edges = np.asarray(list(g.edges(data="eid")))

                if size:
                    order = np.argsort(edges[:, 2])

                    for i, idx in enumerate(order):
                        s, t, _ = edges[idx]
                        g.edges[s, t][name] = value[i]
            else:
                raise ValueError(
                    "A list or a np.array with one entry per edge in the "
                    "graph is required. For attribute "
                    "'{}', got {} entries vs {} edges.".format(
                        name, len(value), size))
        else:
            raise InvalidArgument("Attribute does not exist yet, use "
                                  "set_attribute to create it.")

    def get_eattr(self, edges, name=None):
        g = self.parent()._graph

        if nonstring_container(edges[0]):
            # many edges
            name = self.keys() if name is None else [name]

            eprop = {k: [] for k in self.keys()}

            for edge in edges:
                data = g.get_edge_data(*edge)

                if data is None:
                    raise ValueError(
                        "Edge {} does not exist.".format(edge))

                [eprop[k].append(data[k]) for k in name]

            for k, v in eprop.items():
                dtype    = _np_dtype(super().__getitem__(k))
                eprop[k] = _to_np_array(v, dtype)

            if len(name) == 1:
                return eprop[name[0]]

            return eprop

        # single edge
        data = deepcopy(g.get_edge_data(*edges))

        if not data:
            raise ValueError("Edge {} does not exist.".format(edges))

        if name is None:
            del data["eid"]

            return data

        return data[name]

    def new_attribute(self, name, value_type, values=None, val=None):
        g = self.parent()._graph

        if val is None:
            if value_type in ("int", "integer"):
                val = int(0)
            elif value_type in ("double", "float"):
                val = np.NaN
            elif value_type == "string":
                val = ""
            else:
                val = None
                value_type = "object"

        if values is None:
            values = [deepcopy(val)
                      for _ in range(g.number_of_edges())]

        # store name and value type in the dict
        super(_NxEProperty, self).__setitem__(name, value_type)

        # store the real values in the attribute
        self[name] = values
        self._num_values_set[name] = len(values)

    def set_attribute(self, name, values, edges=None, last_edges=False):
        '''
        Set the edge property.

        Parameters
        ----------
        name : str
            Name of the edge property.
        values : array
            Values that should be set.
        edges : array-like, optional (default: None)
            Edges for which the value of the property should be set. If `edges`
            is not None, it must be an array of shape `(len(values), 2)`.
        '''
        graph = self.parent()

        g = graph._graph

        num_edges = g.number_of_edges()
        num_e = len(edges) if edges is not None else num_edges

        if num_e != len(values):
            raise ValueError("`edges` and `values` must have the same "
                             "size; got respectively " + str(num_e) + \
                             " and " + str(len(values)) + " entries.")

        if edges is None:
            self[name] = values
        else:
            order = range(num_e)

            if not last_edges:
                get_eid = graph.edge_id

                eids  = [get_eid(e) for e in edges]
                order = np.argsort(np.argsort(eids))

            for i, e in zip(order, edges):
                edict = g[e[0]][e[1]]

                edict[name]  = values[i]

                g.add_edge(e[0], e[1], **edict)

        if num_e:
            self._num_values_set[name] = num_edges


# ----- #
# Graph #
# ----- #

class _NxGraph(GraphInterface):

    '''
    Subclass of networkx Graph
    '''

    _nattr_class = _NxNProperty
    _eattr_class = _NxEProperty

    #-------------------------------------------------------------------------#
    # Class properties

    di_value = {"string": "", "double": 0., "int": int(0)}

    #-------------------------------------------------------------------------#
    # Constructor and instance properties

    def __init__(self, nodes=0, copy_graph=None, directed=True, weighted=False,
                 **kwargs):
        self._nattr = _NxNProperty(self)
        self._eattr = _NxEProperty(self)

        self._max_eid = 0

        g = copy_graph.graph if copy_graph is not None else None

        if g is not None:
            if not directed and g.is_directed():
                g = g.to_undirected()
            elif directed and not g.is_directed():
                g = g.to_directed()

            self._from_library_graph(g, copy=True)

            self._max_eid = copy_graph._max_eid
        else:
            nx = nngt._config["library"]

            self._graph = nx.DiGraph() if directed else nx.Graph()

            if nodes:
                self._graph.add_nodes_from(range(nodes))

    #-------------------------------------------------------------------------#
    # Graph manipulation

    def edge_id(self, edge):
        '''
        Return the ID a given edge or a list of edges in the graph.
        Raises an error if the edge is not in the graph or if one of the
        vertices in the edge is nonexistent.

        Parameters
        ----------
        edge : 2-tuple or array of edges
            Edge descriptor (source, target).

        Returns
        -------
        index : int or array of ints
            Index of the given `edge`.
        '''
        g = self._graph

        if is_integer(edge[0]):
            return g[edge[0]][edge[1]]["eid"]
        elif nonstring_container(edge[0]):
            return [g[e[0]][e[1]]["eid"] for e in edge]

        raise AttributeError("`edge` must be either a 2-tuple of ints or "
                             "an array of 2-tuples of ints.")

    def has_edge(self, edge):
        '''
        Whether `edge` is present in the graph.

        .. versionadded:: 2.0
        '''
        return self._graph.has_edge(*edge)

    @property
    def edges_array(self):
        '''
        Edges of the graph, sorted by order of creation, as an array of
        2-tuple.
        '''
        g     = self._graph
        edges = np.full((self._max_eid, 2), -1)

        # fast iteration using list comprehension
        # could also be done with deque and map (deque forces lazy map to run)
        # deque(map(lambda x: _gen_edges(edges, x), g.edges(data="eid")))
        [_gen_edges(edges, x) for x in g.edges(data="eid")]

        if self._max_eid > g.number_of_edges():
            return edges[edges[:, 0] > -1]

        return edges

    def _get_edges(self, source_node=None, target_node=None):
        '''
        Called by Graph.get_edges if source_node and target_node are not both
        integers.
        '''
        nx = nngt._config["library"]

        g = self._graph

        target_node = \
            [target_node] if is_integer(target_node) else target_node

        if source_node is not None:
            source_node = \
                [source_node] if is_integer(source_node) else source_node

            if target_node is None:

                if g.is_directed():
                    return list(g.out_edges(source_node))

                return [
                    e if e[0] <= e[1] else e[::-1] for e in g.edges(source_node)
                ]

            res_iter = nx.edge_boundary(g, source_node, target_node)

            if g.is_directed():
                return list(res_iter)

            return [e if e[0] <= e[1] else e[::-1] for e in res_iter]

        if target_node is None:
            # return all edges
            return list(g.edges)

        if g.is_directed():
            return list(g.in_edges(target_node))

        return [
            e if e[0] <= e[1] else e[::-1] for e in g.edges(target_node)
        ]

    def new_node(self, n=1, neuron_type=1, attributes=None, value_types=None,
                 positions=None, groups=None):
        '''
        Adding a node to the graph, with optional properties.

        Parameters
        ----------
        n : int, optional (default: 1)
            Number of nodes to add.
        neuron_type : int, optional (default: 1)
            Type of neuron (1 for excitatory, -1 for inhibitory)
        attributes : dict, optional (default: None)
            Dictionary containing the attributes of the nodes.
        value_types : dict, optional (default: None)
            Dict of the `attributes` types, necessary only if the `attributes`
            do not exist yet.
        positions : array of shape (n, 2), optional (default: None)
            Positions of the neurons. Valid only for
            :class:`~nngt.SpatialGraph` or :class:`~nngt.SpatialNetwork`.
        groups : str, int, or list, optional (default: None)
            :class:`~nngt.core.NeuralGroup` to which the neurons belong. Valid
            only for :class:`~nngt.Network` or :class:`~nngt.SpatialNetwork`.

        Returns
        -------
        The node or a list of the nodes created.
        '''
        g = self._graph
        num_nodes = g.number_of_nodes()

        new_nodes = list(range(num_nodes, num_nodes + n))

        for v in new_nodes:
            g.add_node(v)

        attributes = {} if attributes is None else deepcopy(attributes)

        if attributes:
            for k, v in attributes.items():
                if k not in self._nattr:
                    self._nattr.new_attribute(k, value_types[k], val=v)
                else:
                    v = v if nonstring_container(v) else [v]
                    self._nattr.set_attribute(k, v, nodes=new_nodes)

        # set default values for all attributes that were not set
        for k in self.node_attributes:
            if k not in attributes:
                dtype = self.get_attribute_type(k)
                filler = [None for _ in new_nodes]

                # change for strings, doubles and ints
                if dtype == "string":
                    filler = ["" for _ in new_nodes]
                elif dtype == "double":
                    filler = [np.NaN for _ in new_nodes]
                elif dtype == "int":
                    filler = [0 for _ in new_nodes]

                self._nattr.set_attribute(k, filler, nodes=new_nodes)

        if self.is_spatial():
            old_pos      = self._pos
            self._pos    = np.full((self.node_nb(), 2), np.NaN)
            num_existing = len(old_pos) if old_pos is not None else 0

            if num_existing != 0:
                self._pos[:num_existing] = old_pos

        if positions is not None and len(positions):
            assert self.is_spatial(), \
                "`positions` argument requires a SpatialGraph/SpatialNetwork."
            self._pos[new_nodes] = positions

        if groups is not None:
            assert self.is_network(), \
                "`positions` argument requires a Network/SpatialNetwork."
            if nonstring_container(groups):
                assert len(groups) == n, "One group per neuron required."
                for g, node in zip(groups, new_nodes):
                    self.population.add_to_group(g, node)
            else:
                self.population.add_to_group(groups, new_nodes)

        if len(new_nodes) == 1:
            return new_nodes[0]

        return new_nodes

    def delete_nodes(self, nodes):
        '''
        Remove nodes (and associated edges) from the graph.
        '''
        g = self._graph

        if nonstring_container(nodes):
            for n in nodes:
                g.remove_node(n)
        else:
            g.remove_node(nodes)

        # relabel nodes from zero
        nx = nngt._config["library"]

        nx.relabel_nodes(g, {n: i for i, n in enumerate(g.nodes)}, copy=False)

        # update attributes
        for key in self._nattr:
            self._nattr._num_values_set[key] = self.node_nb()

        for key in self._eattr:
            self._eattr._num_values_set[key] = self.edge_nb()

        # check spatial and structure properties
        _post_del_update(self, nodes)

    def new_edge(self, source, target, attributes=None, ignore=False,
                 self_loop=False):
        '''
        Adding a connection to the graph, with optional properties.

        .. versionchanged :: 2.0
            Added `self_loop` argument to enable adding self-loops.

        Parameters
        ----------
        source : :class:`int/node`
            Source node.
        target : :class:`int/node`
            Target node.
        attributes : :class:`dict`, optional (default: ``{}``)
            Dictionary containing optional edge properties. If the graph is
            weighted, defaults to ``{"weight": 1.}``, the unit weight for the
            connection (synaptic strength in NEST).
        ignore : bool, optional (default: False)
            If set to True, ignore attempts to add an existing edge and accept
            self-loops; otherwise an error is raised.
        self_loop : bool, optional (default: False)
            Whether to allow self-loops or not.

        Returns
        -------
        The new connection or None if nothing was added.
        '''
        g = self._graph

        attributes = {} if attributes is None else deepcopy(attributes)

        # check that nodes exist
        num_nodes = g.number_of_nodes()

        if source >= num_nodes or target >= num_nodes:
            raise InvalidArgument("`source` or `target` does not exist.")

        # set default values for attributes that were not passed
        _set_default_edge_attributes(self, attributes, num_edges=1)

        if g.has_edge(source, target):
            if not ignore:
                raise InvalidArgument("Trying to add existing edge.")

            _log_message(logger, "INFO",
                         "Existing edge {} ignored.".format((source, target)))
        else:
            if source == target:
                if not ignore and not self_loop:
                    raise InvalidArgument("Trying to add a self-loop.")
                elif ignore:
                    _log_message(logger, "INFO",
                                 "Self-loop on {} ignored.".format(source))

                    return None

            for attr in attributes:
                if "_corr" in attr:
                    raise NotImplementedError("Correlated attributes are not "
                                              "available with networkx.")

            if self.is_weighted() and "weight" not in attributes:
                attributes["weight"] = 1.

            # check distance
            _set_dist_new_edges(attributes, self, [(source, target)])

            g.add_edge(source, target)

            g[source][target]["eid"] = self._max_eid

            self._max_eid += 1

            # call parent function to set the attributes
            self._attr_new_edges([(source, target)], attributes=attributes)

        return (source, target)

    def new_edges(self, edge_list, attributes=None, check_duplicates=False,
                  check_self_loops=True, check_existing=True,
                  ignore_invalid=False):
        '''
        Add a list of edges to the graph.

        .. versionchanged:: 2.0
            Can perform all possible checks before adding new edges via the
            ``check_duplicates`` ``check_self_loops``, and ``check_existing``
            arguments.

        Parameters
        ----------
        edge_list : list of 2-tuples or np.array of shape (edge_nb, 2)
            List of the edges that should be added as tuples (source, target)
        attributes : :class:`dict`, optional (default: ``{}``)
            Dictionary containing optional edge properties. If the graph is
            weighted, defaults to ``{"weight": ones}``, where ``ones`` is an
            array the same length as the `edge_list` containing a unit weight
            for each connection (synaptic strength in NEST).
        check_duplicates : bool, optional (default: False)
            Check for duplicate edges within `edge_list`.
        check_self_loops : bool, optional (default: True)
            Check for self-loops.
        check_existing : bool, optional (default: True)
            Check whether some of the edges in `edge_list` already exist in the
            graph or exist multiple times in `edge_list` (also performs
            `check_duplicates`).
        ignore_invalid : bool, optional (default: False)
            Ignore invalid edges: they are not added to the graph and are
            silently dropped. Unless this is set to true, an error is raised
            whenever one of the three checks fails.

        .. warning::

            Setting `check_existing` to False will lead to undefined behavior
            if existing edges are provided! Only use it (for speedup) if you
            are sure that you are indeed only adding new edges.

        Returns
        -------
        Returns new edges only.
        '''
        g = self._graph

        attributes = {} if attributes is None else deepcopy(attributes)
        num_edges  = len(edge_list)

        # check that all nodes exist
        if num_edges:
            if np.max(edge_list) >= g.number_of_nodes():
                raise InvalidArgument("Some nodes do no exist.")

            for k, v in attributes.items():
                assert nonstring_container(v) and len(v) == num_edges, \
                    "One attribute per edge is required."

                if "_corr" in k:
                    raise NotImplementedError("Correlated attributes are not "
                                              "available with networkx.")

            # set default values for attributes that were not passed
            _set_default_edge_attributes(self, attributes, num_edges)

            # check edges
            new_attr = None

            if check_duplicates or check_self_loops or check_existing:
                edge_list, new_attr = _cleanup_edges(
                    self, edge_list, attributes, check_duplicates,
                    check_self_loops, check_existing, ignore_invalid)
            else:
                new_attr = attributes

            # create the edges
            initial_eid = self._max_eid

            num_added = len(edge_list)

            if num_added:
                arr_edges = np.zeros((num_added, 3), dtype=int)

                arr_edges[:, :2] = edge_list
                arr_edges[:, 2]  = np.arange(
                    initial_eid, initial_eid + num_added)

                # create the edges with an eid attribute
                g.add_weighted_edges_from(arr_edges, weight="eid")

                # check distance
                _set_dist_new_edges(new_attr, self, edge_list)

                # call parent function to set the attributes
                self._attr_new_edges(edge_list, attributes=new_attr)

                self._max_eid += num_added

        return edge_list

    def delete_edges(self, edges):
        ''' Remove a list of edges '''
        if len(edges):
            if nonstring_container(edges[0]):
                self._graph.remove_edges_from(edges)
            else:
                self._graph.remove_edge(*edges)

            for key in self._eattr:
                self._eattr._num_values_set[key] = self.edge_nb()

    def clear_all_edges(self):
        ''' Remove all edges from the graph '''
        g = self._graph
        g.remove_edges_from(tuple(g.edges()))
        self._eattr.clear()

    #-------------------------------------------------------------------------#
    # Getters

    def node_nb(self):
        ''' Number of nodes in the graph '''
        return self._graph.number_of_nodes()

    def edge_nb(self):
        ''' Number of edges in the graph '''
        return self._graph.number_of_edges()

    def get_degrees(self, mode="total", nodes=None, weights=None):
        g = self._graph
        w = _get_nx_weights(self, weights)

        nodes  = range(g.number_of_nodes()) if nodes is None else nodes
        dtype  = int if weights in {False, None} else float
        di_deg = None

        if mode == 'total' or not self._graph.is_directed():
            di_deg = g.degree(nodes, weight=w)
        elif mode == 'in':
            di_deg = g.in_degree(nodes, weight=w)
        elif mode == 'out':
            di_deg = g.out_degree(nodes, weight=w)
        else:
            raise ValueError("Unknown `mode` '{}'".format(mode))

        if nonstring_container(nodes):
            return np.array([di_deg[i] for i in nodes], dtype=dtype)

        return di_deg

    def is_connected(self, mode="strong"):
        '''
        Return whether the graph is connected.

        Parameters
        ----------
        mode : str, optional (default: "strong")
            Whether to test connectedness with directed ("strong") or
            undirected ("weak") connections.
        '''
        g = self._graph

        if g.is_directed() and mode == "weak":
            g = g.to_undirected(as_view=True)

        try:
            import networkx as nx
            nx.diameter(g)
            return True
        except nx.exception.NetworkXError:
            return False

    def neighbours(self, node, mode="all"):
        '''
        Return the neighbours of `node`.

        Parameters
        ----------
        node : int
            Index of the node of interest.
        mode : string, optional (default: "all")
            Type of neighbours that will be returned: "all" returns all the
            neighbours regardless of directionality, "in" returns the
            in-neighbours (also called predecessors) and "out" retruns the
            out-neighbours (or successors).

        Returns
        -------
        neighbours : set
            The neighbours of `node`.
        '''
        g = self._graph

        # special case for undirected
        if not g.is_directed():
            return set(g.neighbors(node))

        if mode == "all":
            # for directed graphs, neighbors ~ successors
            return set(g.successors(node)).union(g.predecessors(node))
        elif mode == "in":
            return set(g.predecessors(node))
        elif mode == "out":
            return set(g.successors(node))

        raise ArgumentError('Invalid `mode` argument {}; possible values are '
                            '"all", "out" or "in".'.format(mode))

    def _from_library_graph(self, graph, copy=True):
        ''' Initialize `self._graph` from existing library object. '''
        import networkx as nx

        nodes = {n: i for i, n in enumerate(graph)}

        num_nodes = graph.number_of_nodes()
        num_edges = graph.number_of_edges()

        # check if nodes start from 0 and are continuous
        if set(nodes.keys()) != set(range(num_nodes)):
            # forced copy to restore nodes to [0, num_nodes[
            g = None

            if graph.is_directed():
                g = nx.DiGraph()
            else:
                g = nx.Graph()

            # add nodes
            for i, (n, attr) in enumerate(graph.nodes(data=True)):
                attr["id"] = n
                g.add_node(i, **attr)

            # add edges
            [g.add_edge(nodes[u], nodes[v], **attr)
             for u, v, attr in graph.edges(data=True)]

            # make edges ids
            def set_eid(e, eid):
                g.edges[e]["eid"] = eid

            [set_eid(e, i) for i, e in enumerate(g.edges)]

            self._max_eid = num_edges

            self._graph = graph = g
        else:
            # all good
            self._graph = graph.copy() if copy else graph

        # get attributes names and "types" and initialize them
        if num_nodes:
            for key, val in graph.nodes[0].items():
                super(type(self._nattr), self._nattr).__setitem__(
                    key, _get_dtype(val))

        if num_edges:
            e0 = next(iter(graph.edges))

            for key, val in graph.edges[e0].items():
                if key != "eid":
                    super(type(self._eattr), self._eattr).__setitem__(
                        key, _get_dtype(val))


# tool function to generate the edges_array

def _gen_edges(array, edata):
    source, target, eid = edata
    array[eid] = (source, target)