Source Code for Module dataprocessor.output.taraxuwriter

 1  """ 
 2  @author: Eleftherios Avramidis 
 3  """ 
 4   
 5  from xml.dom import minidom 
 6  from sentence.parallelsentence import ParallelSentence 
 7  from xml.sax.saxutils import escape 
 8  from xmlwriter import XmlWriter 
 9   
10   
11 -class TaraXUWriter(XmlWriter): 
12      """ 
13      classdocs 
14      """ 
15   
16   
17 -    def __init__(self, data): 
18          """ 
19          Constructor 
20          """ 
21          if isinstance (data , minidom.Document): 
22              self.object_xml = data 
23          elif isinstance(data, list): 
24              self.object_xml = None 
25              self.convert_to_xml(data) 
26          else: 
27              self.object_xml = None 
28              self.convert_to_xml(data.get_parallelsentences()) 
29               
30           
31 -    def convert_to_xml(self, parallelsentences): 
32          """ 
33          Creates an XML for the document an populates that with the (parallel) sentences of the given object. 
34          Resulting XML object gets stored as a variable. 
35          @param parallelsentences: a list of ParallelSentence objects  
36          """ 
37          doc_xml = minidom.Document( ) 
38          jcml = doc_xml.createElement("doc") 
39           
40          i=0 
41           
42           
43          for ps in parallelsentences: 
44               
45              parallelsentence_xml = doc_xml.createElement("sentence") 
46               
47              #add attributes of parallel sentence 
48              for attribute_key in ps.get_attributes().keys(): 
49                  parallelsentence_xml.setAttribute( attribute_key , ps.get_attribute( attribute_key ) ) 
50               
51              #add source as a child of parallel sentence 
52              src_xml = self._create_xml_sentence(doc_xml, ps.get_source(), "source") 
53              parallelsentence_xml.appendChild( src_xml ) 
54   
55              #add translations 
56              for tgt in ps.get_translations(): 
57                  tgt_xml = self._create_xml_sentence(doc_xml, tgt, "target") 
58                  parallelsentence_xml.appendChild( tgt_xml ) 
59   
60              #add reference as a child of parallel sentence 
61              if ps.get_reference(): 
62                  ref_xml = self._create_xml_sentence(doc_xml, ps.get_reference(), "reference") 
63                  parallelsentence_xml.appendChild( ref_xml ) 
64   
65              #append the newly populated parallel sentence to the document 
66              jcml.appendChild(parallelsentence_xml) 
67               
68              #print ">", i 
69              i +=1 
70               
71          doc_xml.appendChild(jcml) 
72          self.object_xml = doc_xml 
73           
74           
75 -    def write_to_file(self, filename): 
76          file_object = open(filename, 'w') 
77          file_object.write(self.object_xml.toprettyxml("\t","\n","utf-8")) 
78          file_object.close()   
79              
80           
81           
82 -    def _create_xml_sentence(self, doc_xml, sentence, tag): 
83          """ 
84          Helper function that fetches the text and the attributes of a sentence 
85          and wraps them up into a minidom XML sentenceect 
86          """ 
87           
88          sentence_xml = doc_xml.createElement(tag) 
89   
90          for attribute_key in sentence.get_attributes().keys(): 
91              sentence_xml.setAttribute(attribute_key, escape(sentence.get_attribute(attribute_key)))        
92          textnode = escape(sentence.get_string().strip())      
93          sentence_xml.appendChild(doc_xml.createTextNode(textnode)) 
94           
95          return sentence_xml 
96