SAX-based parser to extract blocks of elements as Python dictionaries

parser.py

from xml.sax import ContentHandler, parseString
from datetime import datetime

class Parser(ContentHandler):
    def __init__(self, block_name, attrs=None):
        ContentHandler.__init__(self)
        self._block_name = block_name
        self._name =  None
        self._blocks = list()
        self._current_block = None
        self._content = ""
        self._attrs = attrs

    def startElement(self, name, attrs):
        if self._block_name == name:
            self._current_block = list()
        else:
            self._name = name

        if self._attrs and self._name in self._attrs:
            for elem_name, attr_name in self._attrs.items():
                if elem_name == self._name:
                    self._current_block.append(("%s_%s" % (elem_name, attr_name), attrs[attr_name]))

    def characters(self, content):
        self._content += content

    def endElement(self, name):
        if self._name == name and len(self._content) > 0:
            self._current_block.append((self._name, self._content.strip()))
            self._content = ""
        if self._block_name == name:
            self._blocks.append(self._current_block)
            self._current_block = None

    def endDocument(self):
        result = list()
        for block in self._blocks:
            new_dict = {}
            for (name, value) in block:
                if name in new_dict:
                    if isinstance(new_dict[name], list):
                        new_dict[name].append(value)
                    else:
                        lst = [new_dict[name], value]
                        new_dict[name] = lst
                else:
                    new_dict[name] = value
            result.append(new_dict)
        self._blocks = result


def parse_date(date_string):
    date =  None
    if date_string is None or len(date_string) == 0:
        return None

    if date_string == "ongoing":
        now = datetime.now()
        end_date =  None
        try:
            end_date = datetime(year=now.year, month=now.month, day=now.day+7)
        except ValueError:
            if now.month == 12:
                end_date = datetime(year=now.year+1,  month=1, day=7)
            else:
                end_date = datetime(year=now.year, month=now.month+1, day=7)
        return end_date

    # try to split the string on a space
    split_string = date_string.split()

    # if there are two portions to the date the first is the 
    # date, the second is the time.
    if len(split_string) > 1:
        date_string = split_string[0]


    # split the string to get out the parts of the date
    year, month, day = date_string.split('-')
    # create a new datetime object
    date = datetime(int(year), int(month), int(day))
    return date