Source Code for Module TEES.Utils.InteractionXML.IDUtils

 1  import sys 
 2   
 3 -def getNextFreeId(elements): 
 4      highest = -1 
 5      for element in elements: 
 6          id = element.get("id") 
 7          assert id.find(".") != -1, id 
 8          lastPart = id.rsplit(".",1)[-1] 
 9          assert len(lastPart) > 1, id 
10          number = int(lastPart[1:]) 
11          if number > highest: 
12              highest = number 
13      if len(elements) - 1 > highest: 
14          highest = len(elements) - 1 
15      return highest + 1 
16   
17 -def getIdNumberType(idNumber): 
18      if idNumber[0] == "d": 
19          return "document" 
20      elif idNumber[0] == "s": 
21          return "sentence" 
22      elif idNumber[0] == "i": 
23          return "interaction" 
24      elif idNumber[0] == "p": 
25          return "pair" 
26      elif idNumber[0] == "e": 
27          return "entity" 
28      elif idNumber[0] == "a": 
29          return "path" 
30      else: 
31          sys.exit("Unknown type") 
32   
33 -def splitInteractionId(id): 
34      dict = {} 
35      splits = id.split(".") 
36      dict["corpus"] = splits[0] 
37      for i in splits[1:]: 
38          dict[getIdNumberType(i)] = int(i[1:]) 
39      return dict 
40   
41 -def buildInteractionId(id): 
42      idString = id["corpus"] 
43      if id.has_key("document"): 
44          idString += ".d" + str(id["document"]) 
45      if id.has_key("sentence"): 
46          idString += ".s" + str(id["sentence"]) 
47       
48      if id.has_key("entity"): 
49          idString += ".e" + str(id["e"]) 
50      if id.has_key("interaction"): 
51          idString += ".i" + str(id["interaction"]) 
52      if id.has_key("pair"): 
53          idString += ".p" + str(id["pair"]) 
54      if id.has_key("path"): 
55          idString += ".a" + str(id["path"]) 
56      return idString 
57       
58 -def sortInteractionIds(id1, id2): 
59      splits1 = id1.split(".") 
60      splits2 = id2.split(".") 
61      # start from split 1 since token 0 is the name of the dataset 
62      for i in range(1,len(splits1)): 
63          # each level's index starts with a letter (s for sentence etc.) 
64          number1 = int(splits1[i][1:])  
65          number2 = int(splits2[i][1:]) 
66       
67          if number1 > number2: 
68             return 1 
69          elif number1 < number2: 
70             return -1 
71      return 0 # last number1 == number 2 
72