Simple script to convert a CSV file to XML

csv2xml.py

#!/usr/bin/env python
# encoding: utf-8
"""
csv2xml.py

Created by Justin van Wees on 2011-04-18.
"""
import sys
import os
import string
import re
import csv
import libxml2


VERSION = '0.1 (2011-04-18)'

class ConvertToXML(object):
    def __init__(self, options, source_csv, dest_file=None, ):
        self.csv = self.parse_csv(filename=source_csv,  delimiter=options.delimiter,
                                    quotechar=options.quotechar)
        self.headers = self.parse_headers(self.csv[0])
        
        self.xml = self.create_xml(root_element=options.xml_root,
                                    record_element=options.xml_record, 
                                    headers=self.headers, csv=self.csv)
        
        if dest_file:
            self.save(dest_file, self.xml)
        else:
            print self.xml
        
    def parse_csv(self, filename, delimiter, quotechar):
        csv.register_dialect('custom', delimiter=delimiter, 
                                quotechar=quotechar)
        
        csv_file = open(filename, mode='r')
        csv_file = list(csv.reader(csv_file))
        return csv_file
    
    def parse_headers(self, headers):
        punct = set(string.punctuation)
        parsed_headers = []
                        
        for head in headers:
            # Strip punct
            head = ''.join(ch for ch in head if ch not in punct)
            
            # Strip whitespace at beginning and end of string, make lowercase
            head = head.strip().lower()
            
            # Replace space with underscores
            head = head.replace(' ', '_')
            
            parsed_headers.append(head)
        
        return parsed_headers
    
    def create_xml(self, root_element, record_element, headers, csv):
        doc = libxml2.newDoc(version='1.0')
        root = doc.newChild(None, root_element, None)
        
        for record in csv[1:]:
            this_record = root.newChild(None, record_element, None)
            
            for index, header in enumerate(headers):
                if len(record[index]) > 0:
                    this_record.newChild(None, header, record[index])
                else:
                    this_record.newChild(None, header, None)
        
        return doc.serialize(encoding='utf-8', format=1)
    
    def create_xml2(self, root_element, record_element, headers, csv):
        doc = Document()
        
        root = doc.createElement(root_element)
        doc.appendChild(root)
        
        for record in csv[1:]:
            this_record = doc.createElement(record_element)
            
            for index, header in enumerate(headers):
                this_item = doc.createElement(unicode(header, 'utf-8'))
                
                if len(record[index]) > 0:
                    this_item.appendChild(doc.createTextNode(unicode(record[index], 'utf-8')))
                
                this_record.appendChild(this_item)
                
            root.appendChild(this_record)
            
        print doc.toprettyxml(encoding="UTF-8")
    
    def save(self, filename, xml):
        xml_file = open(filename, 'w')
        xml_file.write(xml)
        xml_file.close()

if __name__ == '__main__':
    from optparse import OptionParser
    
    parser = OptionParser(version="%prog " + VERSION,
                            usage='%prog [options] SOURCE_CSV DEST_XML')
    parser.disable_interspersed_args()
    parser.add_option('-d', '--delimiter', dest='delimiter', type='str', default=',',
                        help="One-char string used to separate fields in the CSV file")
    parser.add_option('-q', '--quote-char', dest='quotechar', type='str',
                        default='"', help="One-char string used to quote fields that contain 'special' chars")
    parser.add_option('-r', '--root-element', dest="xml_root", type='str', default='root', 
                        help="Name of the root element")
    parser.add_option('-i', '--record-element', dest="xml_record", type='str',
                        default='record', help="Name of the record elements")
    
    
    (options, args) = parser.parse_args()
    if len(args)  > 1:
        xml = ConvertToXML(options, args[0], args[1])
    else:
        print ConvertToXML(options, args[0])