Source Code for Module TEES.Utils.InteractionXML.SplitMergedElements

  1  import sys, os, copy 
  2  extraPath = os.path.dirname(os.path.abspath(__file__))+"/../../" 
  3  sys.path.append(extraPath) 
  4  from Utils.ProgressCounter import ProgressCounter 
  5  try: 
  6      import xml.etree.cElementTree as ET 
  7  except ImportError: 
  8      import cElementTree as ET 
  9  import Utils.ElementTreeUtils as ETUtils 
 10  import IDUtils 
 11   
 12  # Splits merged types generated from overlapping entities/edges into their components 
 13 -def getElementTypes(element, separator="---"): 
 14      typeName = element.get("type") 
 15      if typeName.find(separator) != -1: 
 16          return typeName.split(separator) 
 17      else: 
 18          return [typeName] 
 19       
 20 -def splitMerged(sentence, elementName, countsByType): 
 21      elements = sentence.findall(elementName) 
 22      elementCount = len(elements) 
 23      newElements = [] 
 24      # split old elements and remove them 
 25      removeCount = 0 
 26      newIdCount = IDUtils.getNextFreeId(elements) 
 27      for element in elements: 
 28          types = getElementTypes(element) 
 29          if len(types) > 1: 
 30              for type in types: 
 31                  #newElement = copy.deepcopy(element) 
 32                  newElement = ET.Element(elementName) 
 33                  for k,v in element.attrib.iteritems(): 
 34                      newElement.set(k, v) 
 35                  newElement.set("type", type) 
 36                  idSplits = element.get("id").rsplit(".",1) 
 37                  newElement.set("id", idSplits[0] + "." + idSplits[1][0] + str(newIdCount) ) 
 38                  newIdCount += 1 
 39                  #print "new id", idSplits[0] + "." + idSplits[1][0] + str(elementCount) 
 40                  newElements.append(newElement) 
 41                  elementCount += 1 
 42              sentence.remove(element) 
 43              removeCount += 1 
 44      # insert the new elements 
 45      elements = sentence.findall(elementName) 
 46      if len(newElements) > 0: 
 47          insertPos = 0 
 48          if len(elements) > 0: 
 49              for element in sentence: 
 50                  if element == elements[-1]: 
 51                      break 
 52                  insertPos += 1 
 53          for newElement in newElements: 
 54              sentence.insert(insertPos, newElement) 
 55      # increment counts 
 56      if countsByType != None: 
 57          countsByType[elementName][0] += removeCount 
 58          countsByType[elementName][1] += len(newElements) 
 59               
 60  # Splits entities/edges with merged types into separate elements 
 61 -def processSentence(sentence, countsByType): 
 62      splitMerged(sentence, "entity", countsByType) 
 63      splitMerged(sentence, "interaction", countsByType) 
 64      splitMerged(sentence, "pair", countsByType) 
 65   
 66 -def splitMergedElements(inputFilename, outputFilename=None): 
 67      print >> sys.stderr, "##### Split elements with merged types #####" 
 68      print >> sys.stderr, "Loading corpus", inputFilename 
 69      corpusTree = ETUtils.ETFromObj(inputFilename) 
 70      corpusRoot = corpusTree.getroot() 
 71       
 72      documents = corpusRoot.findall("document") 
 73      counter = ProgressCounter(len(documents), "Documents") 
 74      countsByType = {"entity":[0,0], "interaction":[0,0], "pair":[0,0]} 
 75      for document in documents: 
 76          counter.update() 
 77          for sentence in document.findall("sentence"): 
 78              processSentence(sentence, countsByType) 
 79      print >> sys.stderr, "Results" 
 80      for k in sorted(countsByType.keys()): 
 81          print >> sys.stderr, "  " + k + ": removed", countsByType[k][0], "created", countsByType[k][1] 
 82       
 83      if outputFilename != None: 
 84          print >> sys.stderr, "Writing output to", outputFilename 
 85          ETUtils.write(corpusRoot, outputFilename) 
 86      return corpusTree 
 87   
 88  if __name__=="__main__": 
 89      import sys 
 90       
 91      from optparse import OptionParser 
 92      # Import Psyco if available 
 93      try: 
 94          import psyco 
 95          psyco.full() 
 96          print >> sys.stderr, "Found Psyco, using" 
 97      except ImportError: 
 98          print >> sys.stderr, "Psyco not installed" 
 99   
100      optparser = OptionParser(usage="%prog [options]\nPath generator.") 
101      optparser.add_option("-i", "--input", default=None, dest="input", help="Corpus in interaction xml format", metavar="FILE") 
102      optparser.add_option("-o", "--output", default=None, dest="output", help="Output file in interaction xml format.") 
103      (options, args) = optparser.parse_args() 
104       
105      if options.input == None: 
106          print >> sys.stderr, "Error, input file not defined." 
107          optparser.print_help() 
108          sys.exit(1) 
109      if options.output == None: 
110          print >> sys.stderr, "Error, output file not defined." 
111          optparser.print_help() 
112          sys.exit(1) 
113       
114      run(options.input, options.output) 
115