Merge pull request #178 from NeuralEnsemble/feat/path-util

Towards semi-automated path generation

neuroml/nml/generatedssupersuper.py

@@ -553,3 +553,61 @@ def _check_arg_list(self, **kwargs):
                 print(err)
                 self.info()
                 raise ValueError(err)
+
+    @classmethod
+    def get_class_hierarchy(cls):
+        """Get the class hierarchy for a component classs.
+
+        Reference: https://stackoverflow.com/a/75161393/375067
+
+        See the methods in neuroml.utils to use this generated hierarchy.
+
+        :returns: nested single key dictionaries where the key of each
+            dictionary is the root node of that subtree, and keys are its
+            immediate descendents
+
+        """
+        # classes that don't have any members, like ZeroOrNone, which is an Enum
+        schema = sys.modules[cls.__module__]
+        try:
+            allmembers = cls._get_members()
+        except AttributeError:
+            return {cls.__name__: []}
+
+        retlist = []
+        for member in allmembers:
+            if member is not None:
+                # is it a complex type, which will have a corresponding class?
+                member_class = getattr(schema, member.get_data_type(), None)
+                # if it isn't a class, so a simple type, just added it with an
+                # empty list
+                if member_class is None:
+                    retlist.append({member.get_name(): []})
+                else:
+                    # if it is a class, see if it has a hierarchy
+                    try:
+                        retlist.append(member_class.get_class_hierarchy())
+                    except AttributeError:
+                        retlist.append({member_class.__name__: []})
+
+        return {cls.__name__: retlist}
+
+    @classmethod
+    def get_nml2_class_hierarchy(cls):
+        """Return the NeuroML class hierarchy.
+
+        The root here is NeuroMLDocument.
+        This is useful in calculating paths to different components to aid in
+        construction of relative paths.
+
+        This caches the value as a class variable so that it is not
+        re-calculated when used multiple times.
+        """
+        # if hierarchy exists, return
+        try:
+            return cls.__nml_hier
+        # first run
+        except AttributeError:
+            schema = sys.modules[cls.__module__]
+            cls.__nml_hier = schema.NeuroMLDocument.get_class_hierarchy()
+        return cls.__nml_hier

neuroml/test/test_nml.py

@@ -12,7 +12,7 @@
 except ImportError:
     import unittest
 
-from neuroml.utils import component_factory
+from neuroml.utils import (component_factory, print_hierarchy)
 import neuroml
 
 
@@ -762,3 +762,18 @@ def test_morphinfo(self):
 
         cell.morphinfo(True)
         cell.biophysinfo()
+
+    def test_class_hierarchy(self):
+        """Test the class hierarchy getter and printer
+        """
+        hier = neuroml.Cell.get_class_hierarchy()
+        self.assertIsNotNone(hier)
+        print()
+        print_hierarchy(hier)
+
+        hier = neuroml.Morphology.get_class_hierarchy()
+        self.assertIsNotNone(hier)
+        print()
+        print(hier)
+        print()
+        print_hierarchy(hier)

neuroml/test/test_utils.py

@@ -7,10 +7,11 @@
 Copyright 2023 NeuroML contributors
 """
 
-import neuroml
-from neuroml.utils import component_factory
 import unittest
-import tempfile
+
+import neuroml
+from neuroml.utils import (component_factory, get_relative_component_path,
+                           print_hierarchy)
 
 
 class UtilsTestCase(unittest.TestCase):
@@ -42,3 +43,13 @@ def test_component_factory_should_fail(self):
             "IafCell",
             id="test_cell",
         )
+
+    def test_networkx_hier_graph(self):
+        """Test constructing a networkx graph of a hierarchy"""
+        hier = neuroml.NeuroMLDocument.get_class_hierarchy()
+        self.assertIsNotNone(hier)
+        print_hierarchy(hier)
+
+        path, graph = get_relative_component_path("Input", "Instance")
+        self.assertIsNotNone(graph)
+        self.assertEqual(path, "../Population/Instance")

neuroml/utils.py

@@ -3,12 +3,16 @@
 Utilities for checking generated code
 
 """
-import sys
 import inspect
+import os
+import sys
 import warnings
-from typing import Union, Any
+from typing import Any, Dict, Union, Optional, Type
+
+import networkx
 
 import neuroml.nml.nml as schema
+
 from . import loaders
 
 
@@ -228,6 +232,79 @@ def component_factory(
     return new_obj
 
 
+def print_hierarchy(tree, indent=4, current_ind=0):
+    """Print the hierarchy tree generated by get_class_hierarchy
+
+    Reference: https://stackoverflow.com/a/75161393/375067
+    """
+    for k, v in tree.items():
+        if current_ind:
+            before_dashes = current_ind - indent
+            print(' ' * before_dashes + '└' + '-'*(indent-1) + k)
+        else:
+            print(k)
+        for sub_tree in v:
+            print_hierarchy(sub_tree, indent=indent, current_ind=current_ind + indent)
+
+
+def get_hier_graph_networkx(graph: networkx.DiGraph, hier: Dict[str, Any]):
+    """Get a networkx graph of the NeuroML hierarchy
+
+    :param graph: graph object to populate
+    :param hier: component type hierarchy obtained from `get_class_hierarchy`
+        and `get_nml2_class_hierarchy` methods
+    :returns: None
+
+    """
+    for k, vs in hier.items():
+        for v in vs:
+            if type(v) is dict:
+                graph.add_edge(k, list(v.keys())[0])
+                get_hier_graph_networkx(graph, v)
+            else:
+                graph.add_edge(k, v)
+
+
+def get_relative_component_path(src: str, dest: str, root: Type =
+                                schema.NeuroMLDocument, graph:
+                                Optional[networkx.DiGraph] = None):
+    """Construct a path from src component to dest in a neuroml document.
+
+    Useful when referring to components in other components
+    Note that
+
+    :param src: source component
+    :param dest: destination component
+    :param root: root component of the hierarchy
+    :param graph: a networkx digraph of the NeuroML hierarchy if available
+        if not, one is constructed
+    :returns: calculated path and networkx digraph for future use
+    """
+    if graph is None:
+        graph = networkx.DiGraph()
+        get_hier_graph_networkx(graph, root.get_nml2_class_hierarchy())
+
+    p1 = (list(networkx.all_shortest_paths(graph, root.__name__, "Instance")))
+    p2 = (list(networkx.all_shortest_paths(graph, root.__name__, "Input")))
+
+    if len(p1) > 1 or len(p2) > 1:
+        print("Multiple paths found, cannot calculate recommended path")
+        print("Paths are:")
+        for p in p1 + p2:
+            print("/".join(p1[0]))
+    else:
+        p1s = "/".join(p1[0])
+        p2s = "/".join(p2[0])
+        print(f"Path1: {p1s}")
+        print(f"Path2: {p2s}")
+        # remove one "../" because we do not need to get to the common level
+        # here, unlike actual file system path traversal
+        path = os.path.relpath(p1s, p2s).replace("../", "", 1)
+        print("Relative path: " + path)
+
+    return (path, graph)
+
+
 def main():
     if len(sys.argv) != 2:
         print("Please specify the name of the NeuroML2 file...")