14. Processing XML Documents

Consider the following XML document:

<tribu>
  <enseignant>
    <personne sexe="M">
      <nom>dupont</nom>
      <prenom>jean</prenom>
      <age>28</age>
      ceci est un commentaire
    </personne>
    <section>27</section>
  </enseignant>
  <etudiant>
    <personne sexe="F">
      <nom>martin</nom>
      <prenom>charline</prenom>
      <age>22</age>
    </personne>
    <formation>dess IAIE</formation>
  </etudiant>
</tribu>

We analyze this document to produce the following console output:

 tribu
  enseignant
   (personne,(sexe,M) )
    nom
     [dupont]
    /nom
    prenom
     [jean]
    /prenom
    age
     [28]
    /age
   /personne
   section
    [27]
   /section
  /enseignant
  etudiant
   (personne,(sexe,F) )
    nom
     [martin]
    /nom
    prenom
     [charline]
    /prenom
    age
     [22]
    /age
   /personne
   formation
    [dess IAIE]
   /formation
  /etudiant
 /tribu

We need to know how to recognize:

• a start tag such as <training>;
• an end tag such as </teacher>;
• a start tag with attributes such as <person gender="F">;
• the body of a tag, such as "martin" in <name>martin</name>.

A program that parses XML code is called an XML parser. Two modules provide the functionality needed to parse XML code: xml.sax and xml.sax.handler.

The [xml.sax] module provides us with an XML parser using the following statement:

xml_parser=xml.sax.make_parser()

This parser parses the XML text sequentially. It calls user-defined methods on events:

• the startElement method on a start tag;
• the endElement method on a closing tag;
• the characters method on the body of a tag.

We need to tell the parser which class implements these methods:

xml_parser.setContentHandler(XmlHandler())

We pass an instance of a class that implements the startElement, endElement, and characters methods to the parser’s setContentHandler method. The class used is a subclass of the xml.sax.handler.ContentHandler class. The preceding methods are called with parameters:

def startElement(self,name,attributs):

• name is the name of the start tag. attributes is the dictionary of the tag's attributes. Thus, for the tag <person sex="M">, we will have name="person" and attributes={'sex':'M'}

def endElement(self,name):

• name is the name of the end tag. Thus, for the </student> tag, we will have name='student'.

def characters(self,data):

• data is the body of the tag. Thus, if the tag is

<nom>
    dupont
</nom>

we will have data='\r\n dupont\r\n '. Generally, we will remove the whitespace preceding and following the data.

Now that this is explained, we can move on to the script for parsing an XML document:

---

The program (xml_sax_01)

---

#    -*- coding=utf-8 -*-

import xml.sax, xml.sax.handler,re

#    management class XML
class XmlHandler(xml.sax.handler.ContentHandler):

    #  function called when a start tag is encountered
    def startElement(self,name,attributs):
        global depth
        #    a sequence of spaces (indentation)
        print " " * depth,
        #    attributes
        precisions=""
        for (attrib,valeur) in attributs.items():
            precisions+="(%s,%s) " % (attrib,valeur)
        #  displays the tag name and any attributes
        if precisions :
            print "(%s,%s)" % (name,precisions)
        else :
            print name
        #    an extra level of tree structure
        depth+=1
        #  is it a data tag?
        global balisesDonnees,baliseDeDonnees
        if balisesDonnees.has_key(name.lower()):
            baliseDeDonnees=1

    #  the function called when an end tag is encountered
    def endElement(self,name):
        #    end of tag
        #    indentation level
        global depth
        depth-=1
        #    a sequence of spaces (indentation)
        print " " * depth,
        #    tag name
        print "/%s" % (name)

    #    data display function
    def characters(self,data):
        #    data
        global baliseDeDonnees

        #  is the current tag a data tag?
        if not baliseDeDonnees :
            return
        #    indentation level
        global depth
        #    a sequence of spaces (indentation)
        print " " * depth,
        #  data are displayed
        match=re.match(r"^\s*(.*)\s*$",data)
        if match:
            print "[%s]" % (match.groups()[0])
        #    end of data tag
        baliseDeDonnees=False

#    ------------------------------------------- main  
#  the program
#    data
file="data.xml"       #  the xml file
depth=0               #  indentation level=depth in tree structure
balisesDonnees={"nom":1,"prenom":1,"age":1,"section":1,"formation":1}
baliseDeDonnees=True     #    to true, indicates a data tag

#    create an xml text analysis object
xml_parser=xml.sax.make_parser()
#  the tag manager
xml_parser.setContentHandler(XmlHandler())
#    xml file processing
xml_parser.parse(file)

Notes:

• The script uses the function library from the xml.sax and xml.sax.handler modules (line 3);
• line 62: the parsed XML file;
• line 68: the XML parser;
• Line 70: The handler for events emitted by the parser will be an instance of the XmlHandler class;
• line 72: the parsing of the XML document begins;
• line 6: the class implementing the startElement, endElement, and characters methods. It is derived from the xml.sax.handler.ContentHandler class, which implements methods used by the parser;
• line 9: the startElement method;
• line 30: the endElement method;
• line 41: the characters method.

The results are those presented at the beginning of this paragraph.