FreeOpcUa / opcua-asyncio

asyncua/server/address_space.py

import asyncio
import pickle
import shelve
import logging
import collections
from datetime import datetime
from concurrent.futures import ThreadPoolExecutor
from functools import partial

from asyncua import ua
from .users import User, UserRole

_logger = logging.getLogger(__name__)


class AttributeValue(object):

    def __init__(self, value):
        self.value = value
        self.value_callback = None
        self.datachange_callbacks = {}

    def __str__(self):
        return f"AttributeValue({self.value})"

    __repr__ = __str__


class NodeData:

    def __init__(self, nodeid):
        self.nodeid = nodeid
        self.attributes = {}
        self.references = []
        self.call = None

    def __str__(self):
        return f"NodeData(id:{self.nodeid}, attrs:{self.attributes}, refs:{self.references})"

    __repr__ = __str__


class AttributeService:

    def __init__(self, aspace: "AddressSpace"):
        self.logger = logging.getLogger(__name__)
        self._aspace: "AddressSpace" = aspace

    def read(self, params):
        # self.logger.debug("read %s", params)
        res = []
        for readvalue in params.NodesToRead:
            res.append(self._aspace.read_attribute_value(readvalue.NodeId, readvalue.AttributeId))
        return res

    async def write(self, params, user=User(role=UserRole.Admin)):
        # self.logger.debug("write %s as user %s", params, user)
        res = []
        for writevalue in params.NodesToWrite:
            if user.role != UserRole.Admin:
                if writevalue.AttributeId != ua.AttributeIds.Value:
                    res.append(ua.StatusCode(ua.StatusCodes.BadUserAccessDenied))
                    continue
                al = self._aspace.read_attribute_value(writevalue.NodeId, ua.AttributeIds.AccessLevel)
                ual = self._aspace.read_attribute_value(writevalue.NodeId, ua.AttributeIds.UserAccessLevel)
                if not al.StatusCode.is_good() or not ua.ua_binary.test_bit(
                        al.Value.Value, ua.AccessLevel.CurrentWrite) or not \
                        ua.ua_binary.test_bit(ual.Value.Value, ua.AccessLevel.CurrentWrite):
                    res.append(ua.StatusCode(ua.StatusCodes.BadUserAccessDenied))
                    continue
            res.append(
                await self._aspace.write_attribute_value(writevalue.NodeId, writevalue.AttributeId, writevalue.Value))
        return res


class ViewService(object):

    def __init__(self, aspace: "AddressSpace"):
        self.logger = logging.getLogger(__name__)
        self._aspace: "AddressSpace" = aspace

    def browse(self, params):
        # self.logger.debug("browse %s", params)
        res = []
        for desc in params.NodesToBrowse:
            res.append(self._browse(desc))
        return res

    def _browse(self, desc):
        res = ua.BrowseResult()
        if desc.NodeId not in self._aspace:
            res.StatusCode = ua.StatusCode(ua.StatusCodes.BadNodeIdInvalid)
            return res
        node = self._aspace[desc.NodeId]
        for ref in node.references:
            if not self._is_suitable_ref(desc, ref):
                continue
            res.References.append(ref)
        return res

    def _is_suitable_ref(self, desc, ref):
        if not self._suitable_direction(desc.BrowseDirection, ref.IsForward):
            # self.logger.debug("%s is not suitable due to direction", ref)
            return False
        if not self._suitable_reftype(desc.ReferenceTypeId, ref.ReferenceTypeId, desc.IncludeSubtypes):
            # self.logger.debug("%s is not suitable due to type", ref)
            return False
        if desc.NodeClassMask and ((desc.NodeClassMask & ref.NodeClass) == 0):
            # self.logger.debug("%s is not suitable due to class", ref)
            return False
        # self.logger.debug("%s is a suitable ref for desc %s", ref, desc)
        return True

    def _suitable_reftype(self, ref1, ref2, subtypes):
        """
        """
        if ref1 == ua.NodeId(ua.ObjectIds.Null):
            # If ReferenceTypeId is not specified in the BrowseDescription,
            # all References are returned and includeSubtypes is ignored.
            return True
        if not subtypes and ref2.Identifier == ua.ObjectIds.HasSubtype:
            return False
        if ref1.Identifier == ref2.Identifier:
            return True
        oktypes = self._get_sub_ref(ref1)
        if not subtypes and ua.NodeId(ua.ObjectIds.HasSubtype) in oktypes:
            oktypes.remove(ua.NodeId(ua.ObjectIds.HasSubtype))
        return ref2 in oktypes

    def _get_sub_ref(self, ref):
        res = []
        nodedata = self._aspace[ref]
        if nodedata is not None:
            for ref in nodedata.references:
                if ref.ReferenceTypeId.Identifier == ua.ObjectIds.HasSubtype and ref.IsForward:
                    res.append(ref.NodeId)
                    res += self._get_sub_ref(ref.NodeId)
        return res

    def _suitable_direction(self, direction, isforward):
        if direction == ua.BrowseDirection.Both:
            return True
        if direction == ua.BrowseDirection.Forward and isforward:
            return True
        if direction == ua.BrowseDirection.Inverse and not isforward:
            return True
        return False

    def translate_browsepaths_to_nodeids(self, browsepaths):
        # self.logger.debug("translate browsepath: %s", browsepaths)
        results = []
        for path in browsepaths:
            results.append(self._translate_browsepath_to_nodeid(path))
        return results

    def _translate_browsepath_to_nodeid(self, path):
        # self.logger.debug("looking at path: %s", path)
        res = ua.BrowsePathResult()
        if not path.RelativePath.Elements[-1].TargetName:
            # OPC UA Part 4: Services, 5.8.4 TranslateBrowsePathsToNodeIds
            # it's unclear if this the check should also handle empty strings
            res.StatusCode = ua.StatusCode(ua.StatusCodes.BadBrowseNameInvalid)
            return res
        if path.StartingNode not in self._aspace:
            res.StatusCode = ua.StatusCode(ua.StatusCodes.BadNodeIdInvalid)
            return res
        current = path.StartingNode
        for el in path.RelativePath.Elements:
            nodeid = self._find_element_in_node(el, current)
            if not nodeid:
                res.StatusCode = ua.StatusCode(ua.StatusCodes.BadNoMatch)
                return res
            current = nodeid
        target = ua.BrowsePathTarget()
        target.TargetId = current
        target.RemainingPathIndex = 4294967295
        res.Targets = [target]
        return res

    def _find_element_in_node(self, el, nodeid):
        nodedata = self._aspace[nodeid]
        for ref in nodedata.references:
            if ref.BrowseName != el.TargetName:
                continue
            if ref.IsForward == el.IsInverse:
                continue
            if not el.IncludeSubtypes and ref.ReferenceTypeId != el.ReferenceTypeId:
                continue
            elif el.IncludeSubtypes and ref.ReferenceTypeId != el.ReferenceTypeId:
                if ref.ReferenceTypeId not in self._get_sub_ref(el.ReferenceTypeId):
                    continue
            return ref.NodeId
        self.logger.info("element %s was not found in node %s", el, nodeid)
        return None


class NodeManagementService:

    def __init__(self, aspace: "AddressSpace"):
        self.logger = logging.getLogger(__name__)
        self._aspace: "AddressSpace" = aspace

    def add_nodes(self, addnodeitems, user=User(role=UserRole.Admin)):
        results = []
        for item in addnodeitems:
            results.append(self._add_node(item, user))
        return results

    def try_add_nodes(self, addnodeitems, user=User(role=UserRole.Admin), check=True):
        for item in addnodeitems:
            ret = self._add_node(item, user, check=check)
            if not ret.StatusCode.is_good():
                yield item

    def _add_node(self, item, user, check=True):
        # self.logger.debug("Adding node %s %s", item.RequestedNewNodeId, item.BrowseName)
        result = ua.AddNodesResult()

        if not user.role == UserRole.Admin:
            result.StatusCode = ua.StatusCode(ua.StatusCodes.BadUserAccessDenied)
            return result

        if item.RequestedNewNodeId.has_null_identifier():
            # If Identifier of requested NodeId is null we generate a new NodeId using
            # the namespace of the nodeid, this is an extention of the spec to allow
            # to requests the server to generate a new nodeid in a specified namespace
            # self.logger.debug("RequestedNewNodeId has null identifier, generating Identifier")
            item.RequestedNewNodeId = self._aspace.generate_nodeid(item.RequestedNewNodeId.NamespaceIndex)
        else:
            if item.RequestedNewNodeId in self._aspace:
                self.logger.warning("AddNodesItem: Requested NodeId %s already exists", item.RequestedNewNodeId)
                result.StatusCode = ua.StatusCode(ua.StatusCodes.BadNodeIdExists)
                return result

        if item.ParentNodeId.is_null():
            # self.logger.info("add_node: while adding node %s, requested parent node is null %s %s",
            # item.RequestedNewNodeId, item.ParentNodeId, item.ParentNodeId.is_null())
            if check:
                result.StatusCode = ua.StatusCode(ua.StatusCodes.BadParentNodeIdInvalid)
                return result

        parentdata = self._aspace.get(item.ParentNodeId)
        if parentdata is None and not item.ParentNodeId.is_null():
            self.logger.info("add_node: while adding node %s, requested parent node %s does not exists",
                             item.RequestedNewNodeId, item.ParentNodeId)
            result.StatusCode = ua.StatusCode(ua.StatusCodes.BadParentNodeIdInvalid)
            return result

        if item.ParentNodeId in self._aspace:
            for ref in self._aspace[item.ParentNodeId].references:
                # Check if the Parent has a "HasChild" Reference (or subtype of it) with the Node
                if ref.ReferenceTypeId.Identifier in [ua.ObjectIds.HasChild, ua.ObjectIds.HasComponent,
                                           ua.ObjectIds.HasProperty, ua.ObjectIds.HasSubtype,
                                           ua.ObjectIds.HasOrderedComponent] and ref.IsForward:
                    if item.BrowseName.Name == ref.BrowseName.Name:
                        self.logger.warning(f"AddNodesItem: Requested Browsename {item.BrowseName.Name}"
                                            f" already exists in Parent Node. ParentID:{item.ParentNodeId} --- "
                                            f"ItemId:{item.RequestedNewNodeId}")
                        result.StatusCode = ua.StatusCode(ua.StatusCodes.BadBrowseNameDuplicated)
                        return result

        nodedata = NodeData(item.RequestedNewNodeId)

        self._add_node_attributes(nodedata, item, add_timestamps=check)

        # now add our node to db
        self._aspace[nodedata.nodeid] = nodedata

        if parentdata is not None:
            self._add_ref_from_parent(nodedata, item, parentdata)
            self._add_ref_to_parent(nodedata, item, parentdata)

        # add type definition
        if item.TypeDefinition != ua.NodeId():
            self._add_type_definition(nodedata, item)

        result.StatusCode = ua.StatusCode()
        result.AddedNodeId = nodedata.nodeid

        return result

    def _add_node_attributes(self, nodedata, item, add_timestamps):
        # add common attrs
        nodedata.attributes[ua.AttributeIds.NodeId] = AttributeValue(
            ua.DataValue(ua.Variant(nodedata.nodeid, ua.VariantType.NodeId))
        )
        nodedata.attributes[ua.AttributeIds.BrowseName] = AttributeValue(
            ua.DataValue(ua.Variant(item.BrowseName, ua.VariantType.QualifiedName))
        )
        nodedata.attributes[ua.AttributeIds.NodeClass] = AttributeValue(
            ua.DataValue(ua.Variant(item.NodeClass, ua.VariantType.Int32))
        )
        # add requested attrs
        self._add_nodeattributes(item.NodeAttributes, nodedata, add_timestamps)

    def _add_unique_reference(self, nodedata, desc):
        for r in nodedata.references:
            if r.ReferenceTypeId == desc.ReferenceTypeId and r.NodeId == desc.NodeId:
                if r.IsForward != desc.IsForward:
                    self.logger.error("Cannot add conflicting reference %s ", str(desc))
                    return ua.StatusCode(ua.StatusCodes.BadReferenceNotAllowed)
                break  # ref already exists
        else:
            nodedata.references.append(desc)
        return ua.StatusCode()

    def _add_ref_from_parent(self, nodedata, item, parentdata):
        desc = ua.ReferenceDescription()
        desc.ReferenceTypeId = item.ReferenceTypeId
        desc.NodeId = nodedata.nodeid
        desc.NodeClass = item.NodeClass
        desc.BrowseName = item.BrowseName
        desc.DisplayName = item.NodeAttributes.DisplayName
        desc.TypeDefinition = item.TypeDefinition
        desc.IsForward = True
        self._add_unique_reference(parentdata, desc)

    def _add_ref_to_parent(self, nodedata, item, parentdata):
        addref = ua.AddReferencesItem()
        addref.ReferenceTypeId = item.ReferenceTypeId
        addref.SourceNodeId = nodedata.nodeid
        addref.TargetNodeId = item.ParentNodeId
        addref.TargetNodeClass = parentdata.attributes[ua.AttributeIds.NodeClass].value.Value.Value
        addref.IsForward = False
        self._add_reference_no_check(nodedata, addref)

    def _add_type_definition(self, nodedata, item):
        addref = ua.AddReferencesItem()
        addref.SourceNodeId = nodedata.nodeid
        addref.IsForward = True
        addref.ReferenceTypeId = ua.NodeId(ua.ObjectIds.HasTypeDefinition)
        addref.TargetNodeId = item.TypeDefinition
        addref.TargetNodeClass = ua.NodeClass.DataType
        self._add_reference_no_check(nodedata, addref)

    def delete_nodes(self, deletenodeitems, user=User(role=UserRole.Admin)):
        results = []
        for item in deletenodeitems.NodesToDelete:
            results.append(self._delete_node(item, user))
        return results

    def _delete_node(self, item, user):
        if user.role != UserRole.Admin:
            return ua.StatusCode(ua.StatusCodes.BadUserAccessDenied)

        if item.NodeId not in self._aspace:
            self.logger.warning("DeleteNodesItem: NodeId %s does not exists", item.NodeId)
            return ua.StatusCode(ua.StatusCodes.BadNodeIdUnknown)

        if item.DeleteTargetReferences:
            for elem in self._aspace.keys():
                for rdesc in self._aspace[elem].references[:]:
                    if rdesc.NodeId == item.NodeId:
                        self._aspace[elem].references.remove(rdesc)

        self._delete_node_callbacks(self._aspace[item.NodeId])

        del (self._aspace[item.NodeId])

        return ua.StatusCode()

    def _delete_node_callbacks(self, nodedata):
        if ua.AttributeIds.Value in nodedata.attributes:
            for handle, callback in list(nodedata.attributes[ua.AttributeIds.Value].datachange_callbacks.items()):
                try:
                    callback(handle, None, ua.StatusCode(ua.StatusCodes.BadNodeIdUnknown))
                    self._aspace.delete_datachange_callback(handle)
                except Exception as ex:
                    self.logger.exception("Error calling delete node callback callback %s, %s, %s", nodedata,
                                          ua.AttributeIds.Value, ex)

    def add_references(self, refs, user=User(role=UserRole.Admin)):
        result = []
        for ref in refs:
            result.append(self._add_reference(ref, user))
        return result

    def try_add_references(self, refs, user=User(role=UserRole.Admin)):
        for ref in refs:
            if not self._add_reference(ref, user).is_good():
                yield ref

    def _add_reference(self, addref, user):
        sourcedata = self._aspace.get(addref.SourceNodeId)
        if sourcedata is None:
            return ua.StatusCode(ua.StatusCodes.BadSourceNodeIdInvalid)
        if addref.TargetNodeId not in self._aspace:
            return ua.StatusCode(ua.StatusCodes.BadTargetNodeIdInvalid)
        if user.role != UserRole.Admin:
            return ua.StatusCode(ua.StatusCodes.BadUserAccessDenied)
        return self._add_reference_no_check(sourcedata, addref)

    def _add_reference_no_check(self, sourcedata, addref):
        rdesc = ua.ReferenceDescription()
        rdesc.ReferenceTypeId = addref.ReferenceTypeId
        rdesc.IsForward = addref.IsForward
        rdesc.NodeId = addref.TargetNodeId
        if addref.TargetNodeClass == ua.NodeClass.Unspecified:
            rdesc.NodeClass = self._aspace.read_attribute_value(
                addref.TargetNodeId, ua.AttributeIds.NodeClass).Value.Value
        else:
            rdesc.NodeClass = addref.TargetNodeClass
        bname = self._aspace.read_attribute_value(addref.TargetNodeId, ua.AttributeIds.BrowseName).Value.Value
        if bname:
            rdesc.BrowseName = bname
        dname = self._aspace.read_attribute_value(addref.TargetNodeId, ua.AttributeIds.DisplayName).Value.Value
        if dname:
            rdesc.DisplayName = dname
        return self._add_unique_reference(sourcedata, rdesc)

    def delete_references(self, refs, user=User(role=UserRole.Admin)):
        result = []
        for ref in refs:
            result.append(self._delete_reference(ref, user))
        return result

    def _delete_unique_reference(self, item, invert=False):
        if invert:
            source, target, forward = item.TargetNodeId, item.SourceNodeId, not item.IsForward
        else:
            source, target, forward = item.SourceNodeId, item.TargetNodeId, item.IsForward
        for rdesc in self._aspace[source].references:
            if rdesc.NodeId == target and rdesc.ReferenceTypeId == item.ReferenceTypeId:
                if rdesc.IsForward == forward:
                    self._aspace[source].references.remove(rdesc)
                    return ua.StatusCode()
        return ua.StatusCode(ua.StatusCodes.BadNotFound)

    def _delete_reference(self, item, user):
        if item.SourceNodeId not in self._aspace:
            return ua.StatusCode(ua.StatusCodes.BadSourceNodeIdInvalid)
        if item.TargetNodeId not in self._aspace:
            return ua.StatusCode(ua.StatusCodes.BadTargetNodeIdInvalid)
        if item.ReferenceTypeId not in self._aspace:
            return ua.StatusCode(ua.StatusCodes.BadReferenceTypeIdInvalid)
        if user.role != UserRole.Admin:
            return ua.StatusCode(ua.StatusCodes.BadUserAccessDenied)

        if item.DeleteBidirectional:
            self._delete_unique_reference(item, True)
        return self._delete_unique_reference(item)

    def _add_node_attr(self, item, nodedata, name, vtype=None, add_timestamps=False):
        if item.SpecifiedAttributes & getattr(ua.NodeAttributesMask, name):

The variable getattr does not seem to be defined.

            dv = ua.DataValue(ua.Variant(getattr(item, name), vtype))
            if add_timestamps:
                # dv.ServerTimestamp = datetime.utcnow()  # Disabled until someone explains us it should be there
                dv.SourceTimestamp = datetime.utcnow()
            nodedata.attributes[getattr(ua.AttributeIds, name)] = AttributeValue(dv)

    def _add_nodeattributes(self, item, nodedata, add_timestamps):
        self._add_node_attr(item, nodedata, "AccessLevel", ua.VariantType.Byte)
        self._add_node_attr(item, nodedata, "ArrayDimensions", ua.VariantType.UInt32)
        self._add_node_attr(item, nodedata, "BrowseName", ua.VariantType.QualifiedName)
        self._add_node_attr(item, nodedata, "ContainsNoLoops", ua.VariantType.Boolean)
        self._add_node_attr(item, nodedata, "DataType", ua.VariantType.NodeId)
        self._add_node_attr(item, nodedata, "Description", ua.VariantType.LocalizedText)
        self._add_node_attr(item, nodedata, "DisplayName", ua.VariantType.LocalizedText)
        self._add_node_attr(item, nodedata, "EventNotifier", ua.VariantType.Byte)
        self._add_node_attr(item, nodedata, "Executable", ua.VariantType.Boolean)
        self._add_node_attr(item, nodedata, "Historizing", ua.VariantType.Boolean)
        self._add_node_attr(item, nodedata, "InverseName", ua.VariantType.LocalizedText)
        self._add_node_attr(item, nodedata, "IsAbstract", ua.VariantType.Boolean)
        self._add_node_attr(item, nodedata, "MinimumSamplingInterval", ua.VariantType.Double)
        self._add_node_attr(item, nodedata, "NodeClass", ua.VariantType.Int32)
        self._add_node_attr(item, nodedata, "NodeId", ua.VariantType.NodeId)
        self._add_node_attr(item, nodedata, "Symmetric", ua.VariantType.Boolean)
        self._add_node_attr(item, nodedata, "UserAccessLevel", ua.VariantType.Byte)
        self._add_node_attr(item, nodedata, "UserExecutable", ua.VariantType.Boolean)
        self._add_node_attr(item, nodedata, "UserWriteMask", ua.VariantType.Byte)
        self._add_node_attr(item, nodedata, "ValueRank", ua.VariantType.Int32)
        self._add_node_attr(item, nodedata, "WriteMask", ua.VariantType.UInt32)
        self._add_node_attr(item, nodedata, "UserWriteMask", ua.VariantType.UInt32)
        self._add_node_attr(item, nodedata, "DataTypeDefinition", ua.VariantType.ExtensionObject)
        self._add_node_attr(item, nodedata, "Value", add_timestamps=add_timestamps)


class MethodService:

    def __init__(self, aspace: "AddressSpace"):
        self.logger = logging.getLogger(__name__)
        self._aspace: "AddressSpace" = aspace
        self._pool = ThreadPoolExecutor()

    def stop(self):
        self._pool.shutdown()

    async def call(self, methods):
        results = []
        for method in methods:
            res = await self._call(method)
            results.append(res)
        return results

    async def _call(self, method):
        self.logger.info("Calling: %s", method)
        res = ua.CallMethodResult()
        if method.ObjectId not in self._aspace or method.MethodId not in self._aspace:
            res.StatusCode = ua.StatusCode(ua.StatusCodes.BadNodeIdInvalid)
        else:
            node = self._aspace[method.MethodId]
            if node.call is None:
                res.StatusCode = ua.StatusCode(ua.StatusCodes.BadNothingToDo)
            else:
                try:
                    result = await self._run_method(node.call, method.ObjectId, *method.InputArguments)
                except Exception:
                    self.logger.exception("Error executing method call %s, an exception was raised: ", method)
                    res.StatusCode = ua.StatusCode(ua.StatusCodes.BadUnexpectedError)
                else:
                    if isinstance(result, ua.CallMethodResult):
                        res = result
                    elif isinstance(result, ua.StatusCode):
                        res.StatusCode = result
                    else:
                        res.OutputArguments = result
                    while len(res.InputArgumentResults) < len(method.InputArguments):
                        res.InputArgumentResults.append(ua.StatusCode())
        return res

    async def _run_method(self, func, parent, *args):
        if asyncio.iscoroutinefunction(func):
            return await func(parent, *args)
        p = partial(func, parent, *args)
        res = await asyncio.get_event_loop().run_in_executor(self._pool, p)
        return res


class AddressSpace:
    """
    The address space object stores all the nodes of the OPC-UA server and helper methods.
    The methods are thread safe
    """

    def __init__(self):
        self.logger = logging.getLogger(__name__)
        self._nodes = {}
        self._datachange_callback_counter = 200
        self._handle_to_attribute_map = {}
        self._default_idx = 2
        self._nodeid_counter = {0: 20000, 1: 2000}

    def __getitem__(self, nodeid):
        return self._nodes.__getitem__(nodeid)

    def get(self, nodeid):
        return self._nodes.get(nodeid, None)

    def __setitem__(self, nodeid, value):
        return self._nodes.__setitem__(nodeid, value)

    def __contains__(self, nodeid):
        return self._nodes.__contains__(nodeid)

    def __delitem__(self, nodeid):
        self._nodes.__delitem__(nodeid)

    def generate_nodeid(self, idx=None):
        if idx is None:
            idx = self._default_idx
        if idx in self._nodeid_counter:
            self._nodeid_counter[idx] += 1
        else:
            # get the biggest identifier number from the existed nodes in address space
            identifier_list = sorted([
                nodeid.Identifier for nodeid in self._nodes.keys()
                if nodeid.NamespaceIndex == idx and nodeid.NodeIdType in (
                    ua.NodeIdType.Numeric, ua.NodeIdType.TwoByte, ua.NodeIdType.FourByte
                )
            ])
            if identifier_list:
                self._nodeid_counter[idx] = identifier_list[-1]
            else:
                self._nodeid_counter[idx] = 1
        nodeid = ua.NodeId(self._nodeid_counter[idx], idx)
        while True:
            if nodeid in self._nodes:
                nodeid = self.generate_nodeid(idx)
            else:
                return nodeid

    def keys(self):
        return self._nodes.keys()

    def empty(self):
        """Delete all nodes in address space"""
        self._nodes = {}

    def dump(self, path):
        """
        Dump address space as binary to file; note that server must be stopped for this method to work
        DO NOT DUMP AN ADDRESS SPACE WHICH IS USING A SHELF (load_aspace_shelf), ONLY CACHED NODES WILL GET DUMPED!
        """
        # prepare nodes in address space for being serialized
        for nodeid, ndata in self._nodes.items():
            # if the node has a reference to a method call, remove it so the object can be serialized
            if ndata.call is not None:
                self._nodes[nodeid].call = None

        with open(path, 'wb') as f:
            pickle.dump(self._nodes, f, pickle.HIGHEST_PROTOCOL)

    def load(self, path):
        """
        Load address space from a binary file, overwriting everything in the current address space
        """
        with open(path, 'rb') as f:
            self._nodes = pickle.load(f)

    def make_aspace_shelf(self, path):
        """
        Make a shelf for containing the nodes from the standard address space; this is typically only done on first
        start of the server. Subsequent server starts will load the shelf, nodes are then moved to a cache
        by the LazyLoadingDict class when they are accessed. Saving data back to the shelf
        is currently NOT supported, it is only used for the default OPC UA standard address space

        Note: Intended for slow devices, such as Raspberry Pi, to greatly improve start up time
        """
        with shelve.open(path, 'n', protocol=pickle.HIGHEST_PROTOCOL) as s:
            for nodeid, ndata in self._nodes.items():
                s[nodeid.to_string()] = ndata

    def load_aspace_shelf(self, path):
        """
        Load the standard address space nodes from a python shelve via LazyLoadingDict as needed.
        The dump() method can no longer be used if the address space is being loaded from a shelf

        Note: Intended for slow devices, such as Raspberry Pi, to greatly improve start up time
        """
        raise NotImplementedError

        # ToDo: async friendly implementation - load all at once?
        class LazyLoadingDict(collections.MutableMapping):

The variable collections does not seem to be defined.

            """
            Special dict that only loads nodes as they are accessed. If a node is accessed it gets copied from the
            shelve to the cache dict. All user nodes are saved in the cache ONLY. Saving data back to the shelf
            is currently NOT supported
            """

            def __init__(self, source):
                self.source = source  # python shelf
                self.cache = {}  # internal dict

            def __getitem__(self, key):
                # try to get the item (node) from the cache, if it isn't there get it from the shelf
                try:
                    return self.cache[key]
                except KeyError:
                    node = self.cache[key] = self.source[key.to_string()]
                    return node

            def __setitem__(self, key, value):
                # add a new item to the cache; if this item is in the shelf it is not updated
                self.cache[key] = value

            def __contains__(self, key):
                return key in self.cache or key.to_string() in self.source

            def __delitem__(self, key):
                # only deleting items from the cache is allowed
                del self.cache[key]

            def __iter__(self):
                # only the cache can be iterated over
                return iter(self.cache.keys())

            def __len__(self):
                # only returns the length of items in the cache, not unaccessed items in the shelf
                return len(self.cache)

        self._nodes = LazyLoadingDict(shelve.open(path, "r"))

    def read_attribute_value(self, nodeid, attr):
        # self.logger.debug("get attr val: %s %s", nodeid, attr)
        if nodeid not in self._nodes:
            dv = ua.DataValue()
            dv.StatusCode = ua.StatusCode(ua.StatusCodes.BadNodeIdUnknown)
            return dv
        node = self._nodes[nodeid]
        if attr not in node.attributes:
            dv = ua.DataValue()
            dv.StatusCode = ua.StatusCode(ua.StatusCodes.BadAttributeIdInvalid)
            return dv
        attval = node.attributes[attr]
        if attval.value_callback:
            return attval.value_callback()
        return attval.value

    async def write_attribute_value(self, nodeid, attr, value):
        # self.logger.debug("set attr val: %s %s %s", nodeid, attr, value)
        node = self._nodes.get(nodeid, None)
        if node is None:
            return ua.StatusCode(ua.StatusCodes.BadNodeIdUnknown)
        attval = node.attributes.get(attr, None)
        if attval is None:
            return ua.StatusCode(ua.StatusCodes.BadAttributeIdInvalid)

        old = attval.value
        attval.value = value
        cbs = []
        if old.Value != value.Value:  # only send call callback when a value change has happend
            cbs = list(attval.datachange_callbacks.items())

        for k, v in cbs:
            try:
                await v(k, value)
            except Exception as ex:
                self.logger.exception("Error calling datachange callback %s, %s, %s", k, v, ex)

        return ua.StatusCode()

    def add_datachange_callback(self, nodeid, attr, callback):
        self.logger.debug("set attr callback: %s %s %s", nodeid, attr, callback)
        if nodeid not in self._nodes:
            return ua.StatusCode(ua.StatusCodes.BadNodeIdUnknown), 0
        node = self._nodes[nodeid]
        if attr not in node.attributes:
            return ua.StatusCode(ua.StatusCodes.BadAttributeIdInvalid), 0
        attval = node.attributes[attr]
        self._datachange_callback_counter += 1
        handle = self._datachange_callback_counter
        attval.datachange_callbacks[handle] = callback
        self._handle_to_attribute_map[handle] = (nodeid, attr)
        return ua.StatusCode(), handle

    def delete_datachange_callback(self, handle):
        if handle in self._handle_to_attribute_map:
            nodeid, attr = self._handle_to_attribute_map.pop(handle)
            self._nodes[nodeid].attributes[attr].datachange_callbacks.pop(handle)

    def add_method_callback(self, methodid, callback):
        node = self._nodes[methodid]
        node.call = callback