Source Code for Module dataprocessor.input.posteditingreader

 1  #!/usr/bin/python 
 2  # -*- coding: utf-8 -*- 
 3  ''' 
 4  Created on Jun 27, 2011 
 5   
 6  @author: jogin 
 7  ''' 
 8   
 9   
10  from xml.dom.minidom import parse 
11  from sentence.parallelsentence import ParallelSentence 
12  from sentence.sentence import SimpleSentence 
13  from xml.sax.saxutils import unescape 
14  from dataprocessor.input.genericreader import GenericReader 
15    
16   
17 -class PosteditingReader(GenericReader): 
18      """ 
19      Reader able to parse the ranking results from taraxu 1st evaluation round, as exported by cfedermann 
20      """ 
21   
22 -    def __init__(self, input_xml_filename, load = True): 
23          """ 
24          Constructor. Creates an XML object that handles ranking file data 
25          @param input_xml_filename: the name of XML file 
26          @type input_xml_filename: string 
27          @param load: by turning this option to false, the instance will be  
28                       initialized without loading everything into memory 
29          @type load: boolean  
30          """ 
31          self.input_filename = input_xml_filename 
32          self.loaded = load 
33          if load: 
34              self.load() 
35       
36 -    def load(self): 
37          """ 
38          Loads the data of the file into memory. It is useful if the Classes has 
39          been asked not to load the filename upon initialization 
40          """ 
41          self.xmlObject = parse(self.input_filename) 
42   
43 -    def unload(self): 
44          self.xmlObject.unlink() 
45           
46       
47   
48 -    def get_parallelsentences(self): 
49          """ 
50          This function parses a ranking xml file and returns a list of parallel 
51          sentence objects. 
52          @return ps_list: list of tuples in format (ranking-item_id, ParallelSentence) 
53          @type ps_list: list of tuples 
54          """ 
55          r_items = self.xmlObject.getElementsByTagName('editing-item') 
56          ps_list = [] 
57          for r_item in r_items: 
58              stc_id = r_item.getAttribute('sentence_id') 
59              id = r_item.getAttribute('id') 
60              src = '' 
61              tgt_list = [] 
62              for rank_child in r_item.childNodes: 
63                  if rank_child.nodeName == 'source': 
64                      src = SimpleSentence(unescape(rank_child.childNodes[0].nodeValue)) 
65                  elif rank_child.nodeName == 'post-edited': 
66                      tgt = SimpleSentence(unescape(rank_child.childNodes[0].nodeValue)) 
67                      tgt.add_attribute('system', 'post-edited') 
68                      tgt_list.append(tgt) 
69                  elif rank_child.nodeName != '#text':                     
70                      tgt = SimpleSentence(unescape(rank_child.childNodes[0].nodeValue)) 
71                      for attribute_name in rank_child.attributes.keys(): 
72                          attribute_value = rank_child.getAttribute(attribute_name) 
73                          tgt.add_attribute(attribute_name, attribute_value) 
74                      tgt.add_attribute('system', rank_child.getAttribute('name')) 
75                      tgt.add_attribute('origin','post-editing') 
76  #                   tgt.add_attribute('rank', rank_child.getAttribute('rank')) 
77                      tgt_list.append(tgt) 
78               
79               
80              ps_attributes = {} 
81              for attributeKey in r_item.attributes.keys(): 
82                  ps_attributes[attributeKey] = unescape(r_item.attributes[attributeKey].value)             
83                                   
84               
85              ps = ParallelSentence(src, tgt_list) 
86              ps.add_attributes({'sentence_id': stc_id}) 
87              ps.add_attributes({'id': id}) 
88              ps_list.append(ps) 
89          return ps_list 
90