''' Paths are enumerated threads through a particular graph. They are implemented as an optimized hierarchical graph using python primitives to save memory and processing time... Each "leaf" node may be tracked back for it's entire path by tracking back to parents. ''' def newPathNode(parent=None, **kwargs): ''' Create a new path node with the given properties ''' ret = (parent, [], kwargs) if parent is not None: parent[1].append(ret) return ret def getNodeParent(pnode): return pnode[0] def delPathNode(pnode): ''' Prune (remove) this node from the parent... ''' p = getNodeParent(pnode) if p is not None: p[1].remove(pnode) def getNodeIndex(pnode): p = getNodeParent(pnode) if p is not None: return p[1].index(pnode) return None def getNodeKids(pnode): ''' Return the list of children nodes for the specified node. Example: for knode in getNodeKids(pnode): ''' return pnode[1] def getRootNode(pnode): ''' Get the root node for the path tree which contains pnode. Example: root = getRootNode(branchnode) ''' ret = pnode while pnode[0] is not None: pnode = pnode[0] return pnode def getLeafNodes(pnode): ''' Get all the leaf nodes for the path tree which contains the pnode. Example: for leaf in getLeafNodes(root): ''' root = getRootNode(pnode) ret = [] todo = [root, ] while len(todo): x = todo.pop() if len(x[1]) == 0: ret.append(x) continue for n in x[1]: todo.append(n) return ret def getPathToNode(pnode): ''' Return a list of the path nodes which lead from the root node to the specified path node. ''' path = [] while pnode is not None: path.append(pnode) pnode = pnode[0] path.reverse() return path def getAllPaths(pnode): ''' Get a list of lists which has each path flattened out. Example: for path in getAllPaths(pnode): for node in path: doStuff() ''' leafs = getLeafNodes(pnode) paths = [] for leaf in leafs: path = getPathToNode(leaf) paths.append(path) return paths def getNodeProp(pnode, key, default=None): ''' Get a property from the given node, returning default in the case that the specified property is not present. Example: name = getNodeProp(pnode, 'name', 'Unknown') ''' return pnode[2].get(key, default) def getPathProp(pnode, key, default=None): ''' Retrieve the specified property by walking the give path backward until the property is found. Returns the specified default if the specified key is not found as a property of any node in the given path. Example: name = getPathProp(pnode, 'name', 'Unknown') ''' parent = pnode while parent is not None: parent, kids, props = parent x = props.get(key) if x is not None: return x return default def setNodeProp(pnode, key, value): ''' Set a spcified property on the given path node. Example: setNodeProp(pnode, 'name', 'woot') ''' pnode[2][key] = value def isPathLoop(pnode, key, value): ''' Assuming you have some identifier property (such as graph node id) being set on the path nodes, you may use this API to determine if the current path has a node with the specified key/value property. Example: if searchPathLoop(pnode, 'nid', 5): continue ''' parent = pnode while parent is not None: parent, kids, props = parent if props.get(key) == value: return True return False def getPathLoopCount(pnode, key, value): ''' Assuming that the key is unique, walk the current path and see how many times "key" has the specified value. This will be how many instances of a loop have been encountered. ''' count = 0 parent = pnode while parent is not None: parent, kids, props = parent if props.get(key) == value: count += 1 return count def trimPath(pnode): ''' Doing analysis on a path tree can be very memory consuming. If an analysis tool is powering through bajillions of nodes, the tree can consume all RAM. However, the analysis may be doable with some selective trimming of the path tree. Only trim when you are done with a path node, and make sure your algorithm uses newPathNode on all "children" before traversing into one of them. trimPath() will remove pnode from it's parent node, then check if the parent node has any other children nodes, and recursively remove path links to all unnecessary path nodes leading up to pnode. caveat: if you are not calling newPathNode (which adds a reference from it's parent) on all child nodes before traversing into any of them, it is possible that the current node could be trimmed because it looks like it doesn't have any children. ''' while True: p = getNodeParent(pnode) # if we don't have a parent, we're done if p is None: break # remove our burden from our parent if pnode in p[1]: p[1].remove(pnode) # if our parent still has kids living at home, we're done if len(p[1]): break # time to kill our parent pnode = p def reprPath(node, startFromRoot=True): r = '' if startFromRoot: node = getRootNode(node) snid = getNodeProp(node, 'nid') todo = [(snid, node, 1)] r += hex(snid) r += '\n' while len(todo): nid, pnode, indent = todo.pop() for tpnode in pnode[1]: tnid = getNodeProp(tpnode, 'nid') outstr = " |"*indent + "--" + hex(tnid) if tnid == snid: outstr += " <<<<-- our node!" r += outstr r += '\n' todo.append((tnid, tpnode, indent + 1)) return r