import xml.dom.minidom,re
sre = re.compile('&')

class SensevalInst (object):

    def __init__(self,sense,position,inst_id,context):
        self.sense = sense
        self.position = position
        self.id = inst_id
        self.context = context

def parse_file_and_extract_events(xml_file,event_list,indent=0):
        g = open(xml_file,'r')
        g_doc = xml.dom.minidom.parse(g)
        g.close()
        extract_events(g_doc,event_list, indent,True)

def extract_events (node,event_list, indent=0, tagging=False):
    """
    C{node} is the root of an xml.dom.minidom tree for a senseval
    xml file.  Construct a list of events (feature-dict, class pairs).
    """
    while node:
        if node.nodeType == node.ELEMENT_NODE and node.nodeName=='senseval_instances':
            print True
            feature_dict = {}
            sis = node.getElementsByTagName('senseval_instance')
            for si in sis:
                sense = si.getAttribute('sense')
                position = int(si.getAttribute('position'))
                inst_id = si.getAttribute('id')
                sentence = si.firstChild.wholeText
                ##  Now make sentence a list of word tag pairs
                context = [wordtag.split('_') for wordtag in sentence.split()]
                si = SensevalInst(sense,position,inst_id,context)
                ## Now extract features from the list C{context}.
                ## C{context[position]} returns our word of interest.
                # print context
                event_list.append((si,sense))
        extract_events (node.firstChild,event_list,indent+4, tagging)
        node = node.nextSibling
        

def wsd_features (context,position,context_id,vocab):
    """
    Not implemented.
    
    Return a dictionary of the features extracted.
    """
    features = {}
    ### Code here!
    return features


def extract_vocab(event_list, n=300):
    """
    Return a dictionary of n most frequently occurring words in THESE
    contexts, excluding stop words.

    Shd probably compare stemmed forms.  Not done yet.

    @return vocab: vocab[{word}] returns a list of the context
         ids C{word} occurs in.
    """
    # Google's stoplist with most preps removed. "and" added, word added
    stopwords = [ 'I',    'a',    'an',    'are',    'as',    'and',
                  'be',    'com',   'how',  'is',    'it',    'of',    'or',
                  'that',    'the',  'this',    'was',    'what',
                  'when',   'where',    'who',    'will',    'with',    
                  'the',    'www']
    vocab = {}
    ## Code goes here.    
    return vocab

def print_event (fh,feature_dict,sense):
    print >> fh, 'BEGIN EVENT'
    print >> fh, sense
    for k in feature_dict:
        print >> fh, '%s\t%s' % (k,feature_dict[k])
    print >> fh, 'END EVENT'
    
    
if __name__ == '__main__':
    
    word = 'hard.pos'
    data_file = 'senseval-hard.xml'
    event_file = 'senseval-hard.evt'
    event_list = []
    parse_file_and_extract_events(data_file,event_list)
    ## Extract the 300 most frequent vocab words from the events
    ## Store as a dictionary.
    vocab = extract_vocab(event_list,300)
    e_fh = open(event_file,'w')
    for (s_inst, sense) in event_list:
        context = s_inst.context
        position = s_inst.position
        context_id = s_inst.id
        feature_dict = wsd_features(context,position,context_id,vocab)
        print_event(e_fh,feature_dict,sense)
    e_fh.close()