Source Code for Module TEES.Utils.InteractionXML.RecalculateIds

  1  import sys, os 
  2  sys.path.append(os.path.dirname(os.path.abspath(__file__))+"/../..") 
  3  try: 
  4      import xml.etree.cElementTree as ET 
  5  except ImportError: 
  6      import cElementTree as ET 
  7  import Utils.ElementTreeUtils as ETUtils 
  8   
  9 -def recalculateIds(input, output=None, onlyWithinSentence=False, docIndexStart=0): 
 10      print >> sys.stderr, "##### Recalculate hierarchical interaction XML ids #####" 
 11      print >> sys.stderr, "Loading corpus", input 
 12      corpusTree = ETUtils.ETFromObj(input) 
 13      print >> sys.stderr, "Corpus file loaded" 
 14      corpusRoot = corpusTree.getroot() 
 15   
 16      # Rebuild hierarchical ids 
 17      print >> sys.stderr, "Recalculating interaction xml ids" 
 18      corpusName = corpusRoot.attrib["source"] 
 19      documents = corpusRoot.findall("document") 
 20      # Recalculate ids for documents, sentences and entities 
 21      entDictionary = {} 
 22      docIndex = docIndexStart 
 23      for document in documents: 
 24          if not onlyWithinSentence: 
 25              document.attrib["id"] = corpusName + ".d" + str(docIndex) 
 26          sentIndex = 0 
 27          sentences = document.findall("sentence") 
 28          for sentence in sentences: 
 29              if not onlyWithinSentence: 
 30                  sentence.attrib["id"] = corpusName + ".d" + str(docIndex) + ".s" + str(sentIndex) 
 31              entIndex = 0 
 32              entities = sentence.findall("entity") 
 33              for entity in entities: 
 34                  if not onlyWithinSentence: 
 35                      entNewId = corpusName + ".d" + str(docIndex) + ".s" + str(sentIndex) + ".e" + str(entIndex) 
 36                  else: 
 37                      entNewId = sentence.attrib["id"] + ".e" + str(entIndex) 
 38                  assert not entDictionary.has_key(entity.attrib["id"]),entity.get("id") 
 39                  entDictionary[entity.attrib["id"]] = entNewId 
 40                  entity.attrib["id"] = entNewId 
 41                  entIndex += 1 
 42              sentIndex += 1 
 43          docIndex += 1 
 44      # Recalculate ids for pairs and interactions 
 45      docIndex = docIndexStart 
 46      for document in documents: 
 47          sentences = document.findall("sentence") 
 48          sentIndex = 0 
 49          for sentence in sentences: 
 50              interactions = sentence.findall("interaction") 
 51              intIndex = 0 
 52              for interaction in interactions: 
 53                  if onlyWithinSentence: 
 54                      interaction.attrib["id"] = sentence.attrib["id"] + ".i" + str(intIndex) 
 55                  else: 
 56                      interaction.attrib["id"] = corpusName + ".d" + str(docIndex) + ".s" + str(sentIndex) + ".i" + str(intIndex) 
 57                  if interaction.attrib["e1"] in entDictionary: 
 58                      interaction.attrib["e1"] = entDictionary[interaction.attrib["e1"]] 
 59                  if interaction.attrib["e2"] in entDictionary: 
 60                      interaction.attrib["e2"] = entDictionary[interaction.attrib["e2"]] 
 61                  intIndex += 1 
 62              pairs = sentence.findall("pair") 
 63              pairIndex = 0 
 64              for pair in pairs: 
 65                  if onlyWithinSentence: 
 66                      pair.attrib["id"] = sentence.attrib["id"] + ".p" + str(pairIndex) 
 67                  else: 
 68                      pair.attrib["id"] = corpusName + ".d" + str(docIndex) + ".s" + str(sentIndex) + ".p" + str(pairIndex) 
 69                  pair.attrib["e1"] = entDictionary[pair.attrib["e1"]] 
 70                  pair.attrib["e2"] = entDictionary[pair.attrib["e2"]] 
 71                  pairIndex += 1 
 72              sentIndex += 1 
 73          docIndex += 1 
 74       
 75      if output != None: 
 76          print >> sys.stderr, "Writing output to", output 
 77          ETUtils.write(corpusRoot, output) 
 78      return corpusTree 
 79   
 80  if __name__=="__main__": 
 81      import sys 
 82       
 83      from optparse import OptionParser 
 84      # Import Psyco if available 
 85      try: 
 86          import psyco 
 87          psyco.full() 
 88          print >> sys.stderr, "Found Psyco, using" 
 89      except ImportError: 
 90          print >> sys.stderr, "Psyco not installed" 
 91   
 92      defaultCorpusFilename = "BioInfer.xml" 
 93      defaultOutputName = "BioInfer.xml" 
 94      optparser = OptionParser(usage="%prog [options]\nPath generator.") 
 95      optparser.add_option("-i", "--input", default=defaultCorpusFilename, dest="input", help="Corpus in interaction xml format", metavar="FILE") 
 96      optparser.add_option("-o", "--output", default=defaultOutputName, dest="output", help="Output file in interaction xml format.") 
 97      optparser.add_option("-s", "--sentence", action="store_true", default=False, dest="sentence", help="Only recalculate within a sentence element.") 
 98      optparser.add_option("-d", "--docIndexStart", type="int", default=0, dest="docIndexStart", help="Start document indexing from.") 
 99      (options, args) = optparser.parse_args() 
100       
101      if options.input == None: 
102          print >> sys.stderr, "Error, input file not defined." 
103          optparser.print_help() 
104          sys.exit(1) 
105      if options.output == None: 
106          print >> sys.stderr, "Error, output file not defined." 
107          optparser.print_help() 
108          sys.exit(1) 
109       
110      recalculateIds(options.input, options.output, options.sentence, options.docIndexStart) 
111