verb-reorder.py

'''
Created on Sep 28, 2013

@author: priyank
'''
import sys
import os
from MyParser import MyParser

parser=MyParser()
parser.add_argument('-i', '--input', help='Input file (with path)', required=True)
parser.add_argument('-b', '--bag', help='verb bag file (with path)', required=True)
args = parser.parse_args()

inputfile = args.input
bagfile = args.bag

if not os.path.isfile(inputfile):
    print  " Input file", inputfile ,"does not exist."
    sys.exit(0);
    

if not os.path.isfile(bagfile):
    print  "verb Bag file", bagfile ,"does not exist."
    sys.exit(0);
    
#reading input file 
f = open(inputfile)
inputdata = f.readlines()
f.close()


f = open(bagfile)
bagdata = f.readlines()
bagdata = [x.strip() for x in bagdata]
f.close()

"""function to search the feature with root in bag file data"""
def searchinbag(ftr, bagdata, root):
    flg = False
    for i in range(len(bagdata)):
        ftr_str = ftr.strip()[1:-1].split(' ')[1].split('=')[1]
        ftr_data = ftr_str.split(',')
        if bagdata[i].split(':')[0].strip() == root:
            bg_ftr = bagdata[i].split(':')[1].strip().split(',')
            #print ftr_data[0][1:], bg_ftr[0]
            if bg_ftr[0] == ftr_data[0][1:] and bg_ftr[1] == ftr_data[1] and bg_ftr[2] == ftr_data[2] and bg_ftr[3] == ftr_data[3] and bg_ftr[4] == ftr_data[4] and bg_ftr[5] == ftr_data[6]:
                flg = True
                return flg
    return flg


def searchinBagWithStructure(oldftr, bagdata, root):
    flg = False
    newftr = ''
    for i in range(len(bagdata)):
        baglinedata = bagdata[i].strip().split('\t')
        
        if baglinedata[1] == root:
            bagftr = baglinedata[2]
            if oldftr.strip() == bagftr:
                newftr = baglinedata[3].strip()
                flg = True
                return flg, newftr
        
    return flg, newftr
            
"""        
        ftr_str = ftr.strip()[1:-1].split(' ')[1].split('=')[1]
        ftr_data = ftr_str.split(',')
        if bagdata[i].split(':')[0].strip() == root:
            bg_ftr = bagdata[i].split(':')[1].strip().split(',')
            #print ftr_data[0][1:], bg_ftr[0]
            if bg_ftr[0] == ftr_data[0][1:] and bg_ftr[1] == ftr_data[1] and bg_ftr[2] == ftr_data[2] and bg_ftr[3] == ftr_data[3] and bg_ftr[4] == ftr_data[4] and bg_ftr[5] == ftr_data[6]:
                flg = True
                return flg
    return flg"""


for line in inputdata:
    
    if len(line.strip().split('\t')) == 4:
        
        features = line.strip().split('\t')[3].split('|')
        if len(features) == 1:
            print line.rstrip()
            
        elif len(features) == 2:
            flag_for_noOfFeatures = True
            if features[0][1:-1].split(' ')[1].split('=')[1].split(',')[1] == 'v':
                f1 = features[0][1:-1].split(' ')[1].split('=')[1].split(',')
                f2 = features[1][1:-1].split(' ')[1].split('=')[1].split(',')
                if f1[1] == 'v' and f1[6] == '0' and f1[2] == 'any' and f1[3] == 'any' and f1[4] == 'any':
                    if f2[1] == 'v' and f2[2] == 'm' or f2[3] == 'sg' and f2[4] == 'any' and f2[6] == 'yA':
                        line = line.split('\t')
                        features = reversed(features)
                        features = '|'.join(features)
                        print line[0]+'\t'+line[1]+'\t'+line[2]+'\t'+features 
                        flag_for_noOfFeatures = False
                    else:
                        
                        flag_for_noOfFeatures = True
                else:
                    
                    flag_for_noOfFeatures = True
            
            else:
                flag_for_noOfFeatures = True
                
            if flag_for_noOfFeatures:
                
                flg = False
                flaag = False
                truefeature = ''
                ii = 0
                root = line.split('\t')[1].strip()
                newftr = ''
                if line.split('\t')[2].strip() == 'VM' or line.split('\t')[2].strip() == 'VAUX':
                    flg, newftr = searchinBagWithStructure(line.strip().split('\t')[3], bagdata, root)

                if flg:
                    line = line.split('\t')
                    line[3] = newftr
                    newline = '\t'.join(line)
                    print newline.rstrip()
                
                else:
                    print line.rstrip()           
     
            
        elif len(features) == 0:
            print line.rstrip()
            
        elif len(features) > 2:
            
            flg = False
            flaag = False
            truefeature = ''
            ii = 0
            root = line.split('\t')[1].strip()
            newftr = ''
            if line.split('\t')[2].strip() == 'VM' or line.split('\t')[2].strip() == 'VAUX':
                flg, newftr = searchinBagWithStructure(line.strip().split('\t')[3], bagdata, root)

            if flg:
                line = line.split('\t')
                line[3] = newftr
                newline = '\t'.join(line)
                print newline.rstrip()
                
            else:
                print line.rstrip()           

    else:
        print line.rstrip()