import os
from strOp import strExt
from collections import deque
from tblOp import tblConcatclass FileNode:def __init__(self, _fileName_s='',_brothers=None,_sons=[],_isDir_b=False,_parent= None):self.fileName_s = _fileName_sself.bro = _brothersself.sons = _sonsself.isDir_b = _isDir_bself.parent = _parentdef addNodeUnderPathUnrecur(root, _path_s):''' inputs: root -> the root of directory tree. It must give the root of the d_path_s -> add the sons under the path of _path_s. if _path_s is equal to 'D:\\CS_DATA\\' then all the file under it is added as sons of the node named 'CS_DATA'outputs:Add all the files under _path_s as its sons. The input must give the root of directory'''node = searchNodeFromGivenFilePath(root, _path_s)filesUnderPath = os.listdir(_path_s)lenOfFilesUnderPath = len(filesUnderPath)for i in range(lenOfFilesUnderPath):if len(node.sons) == 0:newNode = FileNode(filesUnderPath[i], None, [], os.path.isdir(_path_s+filesUnderPath[i]), node)node.sons.append(newNode)else:newNode = FileNode(filesUnderPath[i], None, [], os.path.isdir(_path_s+filesUnderPath[i]), node)node.sons[len(node.sons)-1].bro = newNodenode.sons.append(newNode)#isSameName(node, newNode) file system will ensure that no the same name files exist.def searchNodeFromGivenFilePath(root, _path_s):''' inputs: root -> Must give the root of directory. Meaning the absolute path of a node._path_s -> The absolute path of a node. Examples: 'D:\\CS_DATA\\'output:Search the directory tree from root to find the node whose fileName_s is equal to 'CS_DATA'.So, you must give the absolute path. Whether 'D:\\CS_DATA\\' or 'D:\\CS_DATA' would be fine.'''if _path_s[-1] != '\\':_path_s += '\\'folderStructure = _path_s.split('\\')if root.bro != None:print 'input root is not root of file tree'returnif folderStructure[0] != root.fileName_s:print 'the head of input path is not same as root'returnstack = []stack.append(root)for i in range(1,len(folderStructure)-1):if len(stack) == 0:print 'stack is empty'breaknode = stack.pop()flag = 0for j in node.sons:if folderStructure[i] == j.fileName_s:stack.append(j)flag = 1if flag == 0:print 'can not find the folder %s' % folderStructure[i]return Nonenode = stack.pop()return nodedef addNodeAsSonFromGivenNode(root, _sonPath_s):''' inputs:root -> The root of the directory. Which directory that you want to add the node._sonPath_s -> The absolute path of added node. Examples: 'D:\\CS_DATA\\tree\\' means add the node named 'tree' to its parent 'CS_DATA'outputs:The directory tree with added node.'''if _sonPath_s[-1] != '\\':_sonPath_s += '\\'fileStructure = _sonPath_s.split('\\')lenOfFileStructure = len(fileStructure)if lenOfFileStructure <= 2:print 'These is not son in the input path %s' % _sonPath_sreturn_sonFileName_s = fileStructure[-2]_parentPath_s = ''for i in range(len(fileStructure)-2):_parentPath_s = _parentPath_s + fileStructure[i] + '\\'_addNodeAsSonFromGivenNode(root, _parentPath_s, _sonFileName_s)def _addNodeAsSonFromGivenNode(root, _parentPath_s, _sonFileName_s):''' inputs:root -> The root of directory tree._parentPath_s -> The absolute path of parent_sonFileName_s -> the filename of added nodeoutputs:This function is a auxiliary function of addNodeAsSonFromGivenNode'''if _parentPath_s[-1] != '\\':_parentPath_s += '\\'parentNode = searchNodeFromGivenFilePath(root, _parentPath_s)if parentNode == None:print 'can not find the parent folder %s' % _parentPath_sreturn Noneif len(parentNode.sons) == 0:newNode = FileNode(_sonFileName_s, None, [], os.path.isdir(_parentPath_s+_sonFileName_s), parentNode)if isSameName(parentNode, newNode):returnparentNode.sons.append(newNode)else:newNode = FileNode(_sonFileName_s, None, [], os.path.isdir(_parentPath_s+_sonFileName_s), parentNode)if isSameName(parentNode, newNode):returnparentNode.sons[len(parentNode.sons)-1].bro = newNodeparentNode.sons.append(newNode)def isSameName(parentNode, sonNode):''' inputs:parentNode -> The parent node.sonNode -> the son node.outputs:If sonNode is already in parentNode.sons then return True.'''for node in parentNode.sons:if node.fileName_s == sonNode.fileName_s:print 'has same node %s\\%s -> %s' % (parentNode.fileName_s, node.fileName_s, sonNode.fileName_s)return Truereturn Falsedef addNodeUnderPathRecur(root, _path_s):''' inputs:root -> The root of directory._path_s -> The absolute path wanted to be added. Examples: 'D:\\CS_DATA\\'outputs:1. Add all the file nodes under _path_s recursively. 2. The _path_s must exist in root.Unsafe:1. Some system directory can not be added recursively. Examples: 'D:\\System Volume Information'2. I do not make the judgment between files whether have same name when adding.3. So, this function must use in the premise of operation system ensuring the rule for us.'''if _path_s[-1] != '\\':_path_s = _path_s + '\\'fileStructure = _path_s.split('\\')if fileStructure[0] == root.fileName_s and len(fileStructure) == 2:print '_path_s can not be the root'returnreturnNode = currentNode = searchNodeFromGivenFilePath(root, _path_s)if currentNode == None:print 'can not find the path'returnqueue = deque([])fileName_sl = os.listdir(_path_s)for fileName_s in fileName_sl:file_s = _path_s + fileName_snewNode = FileNode(fileName_s, None, [], os.path.isdir(file_s), currentNode)queue.append(newNode)while(len(queue) != 0):newNode = queue.popleft()currentNode = newNode.parentlenOfSonsCurrentNode = len(currentNode.sons)if lenOfSonsCurrentNode == 0:currentNode.sons.append(newNode)else:currentNode.sons[lenOfSonsCurrentNode-1].bro = newNodecurrentNode.sons.append(newNode)if newNode.isDir_b == True:fullPathOfNewNode = getFullPathOfNode(newNode)subFileName_sl = os.listdir(fullPathOfNewNode)for subFileName_s in subFileName_sl:subNewNode = FileNode(subFileName_s, None, [], os.path.isdir(fullPathOfNewNode+subFileName_s), newNode)queue.append(subNewNode)return returnNode       def printBrosOfGivenNode(root, _path_s):''' inputs:root -> The root of the directory._path_s -> Examples: 'D:\\CS_DATA' , 'D:\\CS_DATA\\'outputs:print out the bros of 'CS_DATA' for 'D:\\CS_DATA'print out the sons of 'CS_DATA' for 'D:\\CS_DATA\\''''if _path_s[-1] != '\\':node = searchNodeFromGivenFilePath(root, _path_s)if node == None:print 'can not find the node'parentOfNode = node.parentheadOfSons = parentOfNode.sons[0]printStr = headOfSons.fileName_s + ','while(headOfSons.bro != None):headOfSons = headOfSons.broprintStr = printStr + headOfSons.fileName_s + ','else:node = searchNodeFromGivenFilePath(root, _path_s)if node == None:print 'can not find the node'printStr = ''if len(node.sons) == 0:print 'its sons is empty'else:for son in node.sons:printStr = printStr + son.fileName_s + ','print printStr[:-1]def crtFileTreeFromPath(_path_s):''' inputs:_path_s -> Examples: 'D:\\sketchDataset\\' outputs:This function will create the root node by 'D:',and then, call addNodeUnderPathUnrecur to add files under 'D:\\',and then, again call addNodeUnderPathUnrecur to add files under 'D:\\sketchDataset\\'This process is a loop until the last separator of _path_s.'''if _path_s[-1] != '\\':_path_s += '\\'fileStructure = _path_s.split('\\')lenOfFileStructure = len(fileStructure)root = FileNode(_fileName_s=fileStructure[0], _isDir_b=os.path.isdir(fileStructure[0]))fileStr = root.fileName_s + '\\'addNodeUnderPathUnrecur(root, fileStr)for i in range(1, lenOfFileStructure-1):file_s = fileStructure[i]fileStr = fileStr + file_s + '\\'addNodeUnderPathUnrecur(root, fileStr)return rootdef searchLeafNodeUnderGivenNode(root, _path_s):''' inputs:root -> For the given directory tree._path_s -> The absolute path of node that wanted to search all the leafs under it.outputs:Return all the leafs under the given _path_s.Leaf is the file whose has not sons and it is not a directory'''node = searchNodeFromGivenFilePath(root, _path_s)leafs = []if node == None:print 'can not find the node in searchLeafNodeUnderGivenNode'returnqueue = deque([])queue.append(node)while(len(queue) != 0):currentNode = queue.popleft()if len(currentNode.sons) == 0 and (currentNode.isDir_b == False):leafs.append(currentNode)else:for son in currentNode.sons:queue.append(son)return leafs        def getFullPathOfNode(givenNode):''' find the full(absolute) path of the input node.'''tmpNode = givenNodefullPathOfNode = tmpNode.fileName_s + '\\'while(tmpNode.parent != None):tmpNode = tmpNode.parentfullPathOfNode = tmpNode.fileName_s + '\\' + fullPathOfNodereturn fullPathOfNode

if __name__ == '__main__':root = crtFileTreeFromPath('D:\\sketchDataset\\')categroyNode = addNodeUnderPathRecur(root, 'D:\\sketchDataset\\category\\')leafs = searchLeafNodeUnderGivenNode(root, 'D:\\sketchDataset\\category\\')containModel_t = {}for i in range(len(leafs)):if leafs[i].parent.fileName_s not in containModel_t:containModel_t[leafs[i].parent.fileName_s] = []containModel_t[leafs[i].parent.fileName_s].append(strExt.extractModelIdWithSuffix(leafs[i].fileName_s, suffix_s='.off'))else:containModel_t[leafs[i].parent.fileName_s].append(strExt.extractModelIdWithSuffix(leafs[i].fileName_s, suffix_s='.off'))categroyNode = addNodeUnderPathRecur(root, 'D:\\sketchDataset\\all_categorized_sketches\\')sketchToCate_t = {}for son in categroyNode.sons:sketchNodes = son.sonsfor sketchNode in sketchNodes:sketchName = strExt.extractSketchNameWithSuffix(sketchNode.fileName_s, suffix_s='.txt')if sketchName not in sketchToCate_t:sketchToCate_t[sketchName] = son.fileName_swanted = tblConcat.concatTableByKey_ValAndVal_Vals(sketchToCate_t, containModel_t)print wanted

{'s165.txt': ['m1646.off', 'm1647.off', 'm1648.off', 'm1649.off', 'm1650.off', 'm1651.off', 'm1652.off', 'm1653.off', 'm1654.off', 'm1655.off', 'm1656.off', 'm1657.off', 'm1658.off', 'm1659.off', 'm1660.off', 'm1661.off', 'm1662.off', 'm1663.off', 'm1664.off', 'm1665.off'] ......}