Source Code for Module TEES.Utils.InteractionXML.ListInteractors

 1  import sys, os 
 2  extraPath = os.path.dirname(os.path.abspath(__file__))+"/../.." 
 3  #IF LOCAL 
 4  extraPath = os.path.dirname(os.path.abspath(__file__))+"/../../JariSandbox/ComplexPPI/Source" 
 5  #ENDIF 
 6  sys.path.append(extraPath) 
 7  from Utils.ProgressCounter import ProgressCounter 
 8  try: 
 9      import xml.etree.cElementTree as ET 
10  except ImportError: 
11      import cElementTree as ET 
12  import Utils.ElementTreeUtils as ETUtils 
13   
14 -def processCorpus(input, attrs=["text"]): 
15      print attrs 
16      print >> sys.stderr, "Loading corpus file", input 
17      corpusRoot = ETUtils.ETFromObj(input).getroot() 
18       
19      documents = corpusRoot.findall("document") 
20      counter = ProgressCounter(len(documents), "Documents") 
21      countsByType = {} 
22      interactors = {} 
23      for document in documents: 
24          entDict = {} 
25          for entity in document.getiterator("entity"): 
26              entDict[entity.get("id")] = entity 
27          for interaction in document.getiterator("interaction"): 
28              e1 = entDict[interaction.get("e1")] 
29              e2 = entDict[interaction.get("e2")] 
30              # form identifier tuples 
31              e1Tuple = [] 
32              for attr in attrs: e1Tuple.append(e1.get(attr)) 
33              e1Tuple = tuple(e1Tuple) 
34              e2Tuple = [] 
35              for attr in attrs: e2Tuple.append(e2.get(attr)) 
36              e2Tuple = tuple(e2Tuple) 
37              interactors = [e1Tuple, e2Tuple] 
38              #interactors.sort() 
39              print interactors 
40              # add interactors 
41  #            if not interactors.has_key(e1): 
42  #                interactors[e1] = set() 
43  #            if not interactors.has_key(e2): 
44  #                interactors[e2] = set() 
45  #            interactors[e1].add(e2) 
46  #            interactors[e2].add(e1) 
47   
48  if __name__=="__main__": 
49      import sys 
50      print >> sys.stderr, "##### Split elements with merged types #####" 
51       
52      from optparse import OptionParser 
53      # Import Psyco if available 
54      try: 
55          import psyco 
56          psyco.full() 
57          print >> sys.stderr, "Found Psyco, using" 
58      except ImportError: 
59          print >> sys.stderr, "Psyco not installed" 
60   
61      optparser = OptionParser(usage="%prog [options]\nPath generator.") 
62      optparser.add_option("-i", "--input", default=None, dest="input", help="Corpus in interaction xml format", metavar="FILE") 
63      optparser.add_option("-a", "--attr", default=None, dest="attr", help="Output file in interaction xml format.") 
64      (options, args) = optparser.parse_args() 
65       
66      if options.input == None: 
67          print >> sys.stderr, "Error, input file not defined." 
68          optparser.print_help() 
69          sys.exit(1) 
70   
71      processCorpus(options.input, eval(options.attr)) 
72