#!/usr/bin/python

"""
Sentence: A class for constructing and deconstructing text for NIPyLL lexicons.

Sentences can be constructed from strings, or built by in-place addition of
strings.  A built sentence can be iterated over, word by word.

Includes self-test.

You are free to use, modify or distribute this program under the terms of the
GNU General Public License:
    
    http://www.gnu.org/copyleft/gpl.html

Author: Chris Reece <car@jessies.org>

See also: http://www.jessies.org/~car/projects/nipyll/
"""

import re

class Sentence:
    
    nonWordsRE = re.compile('\W+')
    words = []
        
    def __init__(self, other = None):
        self.words = []
        if other == None:
            return
        self += other
    
    def __del__(self):
        pass
    
    def __iadd__(self, other):
        if other.__class__ == list:
            string = ' '.join(other)
        elif other.__class__ == str:
            string = other
        else:
            string = str(other)
        string = string.lower()
        string = Sentence.nonWordsRE.sub(' ', string)
        newWords = string.split()
        self.words += newWords
        return self
            
    def __getitem__(self, i):
        return self.words[i]
    
    def __str__(self):
        return ' '.join(self.words)
    
    def __repr__(self):
        return 'words = ' + str(self.words)

if __name__ == '__main__':
    for i in xrange(3):
        print 'SELF-TEST LOOP', i
        s = Sentence('This is Sentence\'s test sentence')
        print '__class__\n\t',
        print s.__class__
        print '__str__\n\t',
        print str(s)
        print '__repr__\n\t',
        print repr(s)
        print '__iadd__\n\t',
        s += 'plus'
        print s, '\n\t',
        s += ['more', 'and']
        print s, '\n\t',
        s += 'more words'
        print s
        print '__getitem__\n\t',
        for w in s:
            print w, '\n\t',
        print
        del s