open62541/tools/pyUANamespace/ua_nodeset.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.

###
### Author:  Chris Iatrou (ichrispa@core-vector.net)
### Version: rev 13
###
### This program was created for educational purposes and has been
### contributed to the open62541 project by the author. All licensing
### terms for this source is inherited by the terms and conditions
### specified for by the open62541 project (see the projects readme
### file for more information on the MPLv2 terms and restrictions).
###
### This program is not meant to be used in a production environment. The
### author is not liable for any complications arising due to the use of
### this program.
###

from __future__ import print_function
import sys
from struct import pack as structpack
from collections import deque
from time import struct_time, strftime, strptime, mktime
import logging; logger = logging.getLogger(__name__)

from ua_builtin_types import *;
from ua_node_types import *;
from ua_constants import *;

def getNextElementNode(xmlvalue):
  if xmlvalue == None:
    return None
  xmlvalue = xmlvalue.nextSibling
  while not xmlvalue == None and not xmlvalue.nodeType == xmlvalue.ELEMENT_NODE:
    xmlvalue = xmlvalue.nextSibling
  return xmlvalue

class NodeSet():
  """ Class holding and managing a set of OPCUA nodes.

      This class handles parsing XML description of namespaces, instantiating
      nodes, linking references, graphing the namespace and compiling a binary
      representation.

      Note that nodes assigned to this class are not restricted to having a
      single namespace ID. This class represents the entire physical address
      space of the binary representation and all nodes that are to be included
      in that segment of memory.
  """
  nodes = []
  nodeids = {}
  aliases = {}
  __linkLater__ = []
  __binaryIndirectPointers__ = []
  name = ""
  knownNodeTypes = ""
  namespaceIdentifiers = {} # list of 'int':'string' giving different namespace an array-mapable name

  def __init__(self, name):
    self.nodes = []
    self.knownNodeTypes = ['variable', 'object', 'method', 'referencetype', \
                           'objecttype', 'variabletype', 'methodtype', \
                           'datatype', 'referencetype', 'aliases']
    self.name = name
    self.nodeids = {}
    self.aliases = {}
    self.namespaceIdentifiers = {}
    self.__binaryIndirectPointers__ = []

  def addNamespace(self, numericId, stringURL):
    self.namespaceIdentifiers[numericId] = stringURL

  def linkLater(self, pointer):
    """ Called by nodes or references who have parsed an XML reference to a
        node represented by a string.

        No return value

        XML String representations of references have the form 'i=xy' or
        'ns=1;s="This unique Node"'. Since during the parsing of this attribute
        only a subset of nodes are known/parsed, this reference string cannot be
        linked when encountered.

        References register themselves with the namespace to have their target
        attribute (string) parsed by linkOpenPointers() when all nodes are
        created, so that target can be dereferenced an point to an actual node.
    """
    self.__linkLater__.append(pointer)

  def getUnlinkedPointers(self):
    """ Return the list of references registered for linking during the next call
        of linkOpenPointers()
    """
    return self.__linkLater__

  def unlinkedItemCount(self):
    """ Returns the number of unlinked references that will be processed during
        the next call of linkOpenPointers()
    """
    return len(self.__linkLater__)

  def buildAliasList(self, xmlelement):
    """ Parses the <Alias> XML Element present in must XML NodeSet definitions.

        No return value

        Contents the Alias element are stored in a dictionary for further
        dereferencing during pointer linkage (see linkOpenPointer()).
    """
    if not xmlelement.tagName == "Aliases":
      logger.error("XMLElement passed is not an Aliaslist")
      return
    for al in xmlelement.childNodes:
      if al.nodeType == al.ELEMENT_NODE:
        if al.hasAttribute("Alias"):
          aliasst = al.getAttribute("Alias")
          if sys.version_info[0] < 3:
            aliasnd = unicode(al.firstChild.data)
          else:
            aliasnd = al.firstChild.data
          if not aliasst in self.aliases:
            self.aliases[aliasst] = aliasnd
            logger.debug("Added new alias \"" + str(aliasst) + "\" == \"" + str(aliasnd) + "\"")
          else:
            if self.aliases[aliasst] != aliasnd:
              logger.error("Alias definitions for " + aliasst + " differ. Have " + self.aliases[aliasst] + " but XML defines " + aliasnd + ". Keeping current definition.")

  def getNodeByBrowseName(self, idstring):
    """ Returns the first node in the nodelist whose browseName matches idstring.
    """
    return next((n for n in self.nodes if idstring==str(n.browseName())), None)

  def getNodeByIDString(self, idstring):
    """ Returns the first node in the nodelist whose id string representation
        matches idstring.
    """
    return next((n for n in self.nodes if idstring==str(n.id())), None)

  def createNode(self, ndtype, xmlelement):
    """ createNode is instantiates a node described by xmlelement, its type being
        defined by the string ndtype.

        No return value

        If the xmlelement is an <Alias>, the contents will be parsed and stored
        for later dereferencing during pointer linking (see linkOpenPointers).

        Recognized types are:
        * UAVariable
        * UAObject
        * UAMethod
        * UAView
        * UAVariableType
        * UAObjectType
        * UAMethodType
        * UAReferenceType
        * UADataType

        For every recognized type, an appropriate node class is added to the node
        list of the namespace. The NodeId of the given node is created and parsing
        of the node attributes and elements is delegated to the parseXML() and
        parseXMLSubType() functions of the instantiated class.

        If the NodeID attribute is non-unique in the node list, the creation is
        deferred and an error is logged.
    """
    if not isinstance(xmlelement, dom.Element):
      logger.error( "Error: Can not create node from invalid XMLElement")
      return

    # An ID is mandatory for everything but aliases!
    id = None
    for idname in ['NodeId', 'NodeID', 'nodeid']:
      if xmlelement.hasAttribute(idname):
        id = xmlelement.getAttribute(idname)
    if ndtype == 'aliases':
      self.buildAliasList(xmlelement)
      return
    elif id == None:
      logger.info( "Error: XMLElement has no id, node will not be created!")
      return
    else:
      id = opcua_node_id_t(id)

    if str(id) in self.nodeids:
      # Normal behavior: Do not allow duplicates, first one wins
      #logger.error( "XMLElement with duplicate ID " + str(id) + " found, node will not be created!")
      #return
      # Open62541 behavior for header generation: Replace the duplicate with the new node
      logger.info( "XMLElement with duplicate ID " + str(id) + " found, node will be replaced!")
      nd = self.getNodeByIDString(str(id))
      self.nodes.remove(nd)
      self.nodeids.pop(str(nd.id()))

    node = None
    if (ndtype == 'variable'):
      node = opcua_node_variable_t(id, self)
    elif (ndtype == 'object'):
      node = opcua_node_object_t(id, self)
    elif (ndtype == 'method'):
      node = opcua_node_method_t(id, self)
    elif (ndtype == 'objecttype'):
      node = opcua_node_objectType_t(id, self)
    elif (ndtype == 'variabletype'):
      node = opcua_node_variableType_t(id, self)
    elif (ndtype == 'methodtype'):
      node = opcua_node_methodType_t(id, self)
    elif (ndtype == 'datatype'):
      node = opcua_node_dataType_t(id, self)
    elif (ndtype == 'referencetype'):
      node = opcua_node_referenceType_t(id, self)
    else:
      logger.error( "No node constructor for type " + ndtype)

    if node != None:
      node.parseXML(xmlelement)

    self.nodes.append(node)
    self.nodeids[str(node.id())] = node

  def removeNodeById(self, nodeId):
    nd = self.getNodeByIDString(nodeId)
    if nd == None:
      return False

    logger.debug("Removing nodeId " + str(nodeId))
    self.nodes.remove(nd)
    if nd.getInverseReferences() != None:
      for ref in nd.getInverseReferences():
        src = ref.target();
        src.removeReferenceToNode(nd)

    return True

  def registerBinaryIndirectPointer(self, node):
    """ Appends a node to the list of nodes that should be contained in the
        first 765 bytes (255 pointer slots a 3 bytes) in the binary
        representation (indirect referencing space).

        This function is reserved for references and dataType pointers.
    """
    if not node in self.__binaryIndirectPointers__:
      self.__binaryIndirectPointers__.append(node)
    return self.__binaryIndirectPointers__.index(node)

  def getBinaryIndirectPointerIndex(self, node):
    """ Returns the slot/index of a pointer in the indirect referencing space
        (first 765 Bytes) of the binary representation.
    """
    if not node in self.__binaryIndirectPointers__:
      return -1
    return self.__binaryIndirectPointers__.index(node)


  def parseXML(self, xmldoc):
    """ Reads an XML Namespace definition and instantiates node.

        No return value

        parseXML open the file xmldoc using xml.dom.minidom and searches for
        the first UANodeSet Element. For every Element encountered, createNode
        is called to instantiate a node of the appropriate type.
    """
    typedict = {}
    UANodeSet = dom.parse(xmldoc).getElementsByTagName("UANodeSet")
    if len(UANodeSet) == 0:
      logger.error( "Error: No NodeSets found")
      return
    if len(UANodeSet) != 1:
      logger.error( "Error: Found more than 1 Nodeset in XML File")

    UANodeSet = UANodeSet[0]
    for nd in UANodeSet.childNodes:
      if nd.nodeType != nd.ELEMENT_NODE:
        continue

      ndType = nd.tagName.lower()
      if ndType[:2] == "ua":
        ndType = ndType[2:]
      elif not ndType in self.knownNodeTypes:
        logger.warn("XML Element or NodeType " + ndType + " is unknown and will be ignored")
        continue

      if not ndType in typedict:
        typedict[ndType] = 1
      else:
        typedict[ndType] = typedict[ndType] + 1

      self.createNode(ndType, nd)
    logger.debug("Currently " + str(len(self.nodes)) + " nodes in address space. Type distribution for this run was: " + str(typedict))

  def linkOpenPointers(self):
    """ Substitutes symbolic NodeIds in references for actual node instances.

        No return value

        References that have registered themselves with linkLater() to have
        their symbolic NodeId targets ("ns=2;i=32") substituted for an actual
        node will be iterated by this function. For each reference encountered
        in the list of unlinked/open references, the target string will be
        evaluated and searched for in the node list of this namespace. If found,
        the target attribute of the reference will be substituted for the
        found node.

        If a reference fails to get linked, it will remain in the list of
        unlinked references. The individual items in this list can be
        retrieved using getUnlinkedPointers().
    """
    linked = []

    logger.debug( str(self.unlinkedItemCount()) + " pointers need to get linked.")
    for l in self.__linkLater__:
      targetLinked = False
      if not l.target() == None and not isinstance(l.target(), opcua_node_t):
        if isinstance(l.target(),str) or isinstance(l.target(),unicode):
          # If is not a node ID, it should be an alias. Try replacing it
          # with a proper node ID
          if l.target() in self.aliases:
            l.target(self.aliases[l.target()])
          # If the link is a node ID, try to find it hopening that no ass has
          # defined more than one kind of id for that sucker
          if l.target()[:2] == "i=" or l.target()[:2] == "g=" or \
             l.target()[:2] == "b=" or l.target()[:2] == "s=" or \
             l.target()[:3] == "ns=" :
            tgt = self.getNodeByIDString(str(l.target()))
            if tgt == None:
              logger.error("Failed to link pointer to target (node not found) " + l.target())
            else:
              l.target(tgt)
              targetLinked = True
          else:
            logger.error("Failed to link pointer to target (target not Alias or Node) " + l.target())
        else:
          logger.error("Failed to link pointer to target (don't know dummy type + " + str(type(l.target())) + " +) " + str(l.target()))
      else:
        logger.error("Pointer has null target: " + str(l))


      referenceLinked = False
      if not l.referenceType() == None:
        if l.referenceType() in self.aliases:
          l.referenceType(self.aliases[l.referenceType()])
        tgt = self.getNodeByIDString(str(l.referenceType()))
        if tgt == None:
          logger.error("Failed to link reference type to target (node not found) " + l.referenceType())
        else:
          l.referenceType(tgt)
          referenceLinked = True
      else:
        referenceLinked = True

      if referenceLinked == True and targetLinked == True:
        linked.append(l)

    # References marked as "not forward" must be inverted (removed from source
    # node, assigned to target node and relinked)
    logger.warn("Inverting reference direction for all references with isForward==False attribute (is this correct!?)")
    for n in self.nodes:
      for r in n.getReferences():
        if r.isForward() == False:
          tgt = r.target()
          if isinstance(tgt, opcua_node_t):
            nref = opcua_referencePointer_t(n, parentNode=tgt)
            nref.referenceType(r.referenceType())
            tgt.addReference(nref)

    # Create inverse references for all nodes
    logger.debug("Updating all referencedBy fields in nodes for inverse lookups.")
    for n in self.nodes:
      n.updateInverseReferences()

    for l in linked:
      self.__linkLater__.remove(l)

    if len(self.__linkLater__) != 0:
      logger.warn(str(len(self.__linkLater__)) + " could not be linked.")

  def sanitize(self):
    remove = []
    logger.debug("Sanitizing nodes and references...")
    for n in self.nodes:
      if n.sanitize() == False:
        remove.append(n)
    if not len(remove) == 0:
      logger.warn(str(len(remove)) + " nodes will be removed because they failed sanitation.")
    # FIXME: Some variable ns=0 nodes fail because they don't have DataType fields...
    #        How should this be handles!?
    logger.warn("Not actually removing nodes... it's unclear if this is valid or not")

  def getRoot(self):
    """ Returns the first node instance with the browseName "Root".
    """
    return self.getNodeByBrowseName("Root")

  def buildEncodingRules(self):
    """ Calls buildEncoding() for all DataType nodes (opcua_node_dataType_t).

        No return value
    """
    stat = {True: 0, False: 0}
    for n in self.nodes:
      if isinstance(n, opcua_node_dataType_t):
        n.buildEncoding()
        stat[n.isEncodable()] = stat[n.isEncodable()] + 1
    logger.debug("Type definitions built/passed: " +  str(stat))

  def allocateVariables(self):
    for n in self.nodes:
      if isinstance(n, opcua_node_variable_t):
        n.allocateValue()

  def getSubTypesOf(self, tdNodes = None, currentNode = None, hasSubtypeRefNode = None):
    # If this is a toplevel call, collect the following information as defaults
    if tdNodes == None:
      tdNodes = []
    if currentNode == None:
      currentNode = self.getNodeByBrowseName("HasTypeDefinition")
      tdNodes.append(currentNode)
      if len(tdNodes) < 1:
        return []
    if hasSubtypeRefNode == None:
      hasSubtypeRefNode = self.getNodeByBrowseName("HasSubtype")
      if hasSubtypeRefNode == None:
        return tdNodes

    # collect all subtypes of this node
    for ref in currentNode.getReferences():
      if ref.isForward() and ref.referenceType().id() == hasSubtypeRefNode.id():
        tdNodes.append(ref.target())
        self.getTypeDefinitionNodes(tdNodes=tdNodes, currentNode = ref.target(), hasSubtypeRefNode=hasSubtypeRefNode)

    return tdNodes


  def printDotGraphWalk(self, depth=1, filename="out.dot", rootNode=None,
                        followInverse = False, excludeNodeIds=[]):
    """ Outputs a graphiz/dot description the nodes centered around rootNode.

        References beginning from rootNode will be followed for depth steps. If
        "followInverse = True" is passed, then inverse (not Forward) references
        will also be followed.

        Nodes can be excluded from the graph by passing a list of NodeIds as
        string representation using excludeNodeIds (ex ["i=53", "ns=2;i=453"]).

        Output is written into filename to be parsed by dot/neato/srfp...
    """
    iter = depth
    processed = []
    if rootNode == None or \
       not isinstance(rootNode, opcua_node_t) or \
       not rootNode in self.nodes:
      root = self.getRoot()
    else:
      root = rootNode

    file=open(filename, 'w+')

    if root == None:
      return

    file.write("digraph ns {\n")
    file.write(root.printDot())
    refs=[]
    if followInverse == True:
      refs = root.getReferences(); # + root.getInverseReferences()
    else:
      for ref in root.getReferences():
        if ref.isForward():
          refs.append(ref)
    while iter > 0:
      tmp = []
      for ref in refs:
        if isinstance(ref.target(), opcua_node_t):
          tgt = ref.target()
          if not str(tgt.id()) in excludeNodeIds:
            if not tgt in processed:
              file.write(tgt.printDot())
              processed.append(tgt)
              if ref.isForward() == False and followInverse == True:
                tmp = tmp + tgt.getReferences(); # + tgt.getInverseReferences()
              elif ref.isForward() == True :
                tmp = tmp + tgt.getReferences();
      refs = tmp
      iter = iter - 1

    file.write("}\n")
    file.close()

  def getSubTypesOf2(self, node):
    re = [node]
    for ref in node.getReferences():
      if isinstance(ref.target(), opcua_node_t):
        if ref.referenceType().displayName() == "HasSubtype" and ref.isForward():
          re = re + self.getSubTypesOf2(ref.target())
    return re

  def reorderNodesMinDependencies(self, printedExternally):
    #Kahn's algorithm
    #https://algocoding.wordpress.com/2015/04/05/topological-sorting-python/
    
    relevant_types = ["HierarchicalReferences", "HasComponent"]
    
    temp = []
    for t in relevant_types:
        temp = temp + self.getSubTypesOf2(self.getNodeByBrowseName(t))
    relevant_types = temp

    in_degree = { u : 0 for u in self.nodes }     # determine in-degree
    for u in self.nodes: # of each node
      if u not in printedExternally:
        for ref in u.getReferences():
         if isinstance(ref.target(), opcua_node_t):
           if(ref.referenceType() in relevant_types and ref.isForward()):
             in_degree[ref.target()] += 1

    Q = deque()                 # collect nodes with zero in-degree
    for u in in_degree:
      if in_degree[u] == 0:
        Q.appendleft(u)

    L = []     # list for order of nodes

    while Q:
      u = Q.pop()          # choose node of zero in-degree
      L.append(u)          # and 'remove' it from graph
      for ref in u.getReferences():
       if isinstance(ref.target(), opcua_node_t):
         if(ref.referenceType() in relevant_types and ref.isForward()):
           in_degree[ref.target()] -= 1
           if in_degree[ref.target()] == 0:
             Q.appendleft(ref.target())
    if len(L) == len(self.nodes):
        self.nodes = L
    else:                    # if there is a cycle,
        logger.error("Node graph is circular on the specified references")
        self.nodes = L + [x for x in self.nodes if x not in L]
    return