#!/usr/bin/env python
'''Keeps GDep available for processing multiple sentences/texts.

Get GDep from: http://www.cs.cmu.edu/~sagae/parser/gdep

The advantage of this module is that GDep is readily avalaiable once it's loaded.
So you don't have to lose time reading in the modules for each different call to GDep.  

USAGE
    >>> parser = Parser('gdep-beta1')
    >>> print parser.parse('The The cat smiles .\\nShe is lovely .')
    1	The	The	B-NP	DT	O	4	NMOD
    2	The	The	I-NP	DT	O	4	NMOD
    3	cat	cat	I-NP	NN	O	4	NMOD
    4	smiles	smile	I-NP	NNS	O	0	ROOT
    5	.	.	O	.	O	4	P

    1	She	She	B-NP	PRP	O	2	SUB
    2	is	be	B-VP	VBZ	O	0	ROOT
    3	lovely	lovely	B-ADJP	JJ	O	2	PRD
    4	.	.	O	.	O	2	P

# License: GNU General Public License, see http://www.clips.ua.ac.be/~vincent/scripts/LICENSE.txt
'''
__date__='October 2010'
__version__='1.0'
__author__='Vincent Van Asch'
__url__ = 'http://www.clips.ua.ac.be/~vincent/software.html'

import os, sys, time, subprocess


class Parser(subprocess.Popen):
    '''
    A Gdep interface object
    
    Note: Creating a Parser takes 15secs startup time.
    '''
    def __init__(self, gdepdir):
        '''gdepdir: a directory where a working binary of Gdep resides.'''
        self.gdepdir = os.path.abspath(os.path.expanduser(gdepdir))           
        command = [os.path.join(self.gdepdir, 'gdep'), '-nt']

        # Dump STDERR
        null = open(os.devnull, 'w')
        
        # Save current dir and go to gdepdir
        dir = os.getcwd()
        os.chdir(self.gdepdir)    
        try:
            subprocess.Popen.__init__(self, command, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=null)
        finally:
            os.chdir(dir)
            
        time.sleep(15)
            
        
    def __repr__(self):
        return '<Parser object based on %s>' %self.gdepdir
        
    def write(self, string):
        '''Send a UTF8 encoded string to the parser.'''
        self.stdin.write(string.strip()+'\n')
        
    def readline(self):
        '''Read one output line from the parser.'''
        return self.stdout.readline()
        
    def parse(self, string):
        '''
        Takes a string and returns the parser's answer.
                
        string: tokenized sentences on newlines.
        '''
        # Send to parser
        string = string.strip()
        self.write(string)
        
        # Keep the number of tokens for checking the output
        length=len(string.split())

        # Collect output
        line=self.readline()
        output=''
        count=0
        while count < length+string.count('\n'):
            output+=line
            line = self.readline()
            count+=1

        return output.strip()



if __name__ == '__main__': print __doc__