Little Helper Script to convert CSV files for neo4j

txt_to_neoscv.py

# -*- coding: utf-8 -*-                                                                                               
"""                                                                                                                   
Created on Wed May 25 11:26:57 2016                                                                                   
                                                                                                                      
@author: julius                                                                                                       
"""                                                                                                                   
                                                                                                                      
                                                                                                                      
import getopt, sys                                                                                                    
                                                                                                                      
def getOptionValue(opts, name, is_long=False):                                                                        
    fmt = "-{}"                                                                                                       
    if is_long:                                                                                                       
        fmt = "-" + fmt                                                                                               
    o = [item for item in opts if item[0] == fmt.format(name)]                                                        
    if len(o) > 0:                                                                                                    
        return o.pop()[1]                                                                                             
    else:                                                                                                             
        return None                                                                                                   
   
# Sample usage 
#  python3 txt_to_neocsv.py -s data/location/wln/wln_label-type.txt -d data/location/wln/wln_label-type.txt.csv -w tab -h "locationid:ID(Location),title:string,locationlabel:string" --stringify=2
def main():                                                                                                           
    opts, args = getopt.getopt(sys.argv[1:], 's:d:h:w:', longopts=['source_file=', 'destination_file=', 'stringify='])
    print(opts)                                                                                                       
    s_source = getOptionValue(opts, 's')                                                                              
    s_destination = getOptionValue(opts, 'd')                                                                         
    s_header = getOptionValue(opts, 'h')                                                                              
    s_whitespace = getOptionValue(opts, 'w')                                                                          
    i_stringify = int(getOptionValue(opts, 'stringify', is_long=True)) - 1                                            
    if s_whitespace == "tab":                                                                                         
        s_whitespace = '\t'                                                                                           
    if s_whitespace == "ws":                                                                                          
        s_whitespace = ' '                                                                                            
    print(s_source)                                                                                                   
    print(s_destination)                                                                                              
    print(s_header)                                                                                                   
    print(i_stringify)                                                                                                
    f_src = open(s_source, 'r')                                                                                       
    f_dst = open(s_destination, 'w+')                                                                                 
    f_dst.writelines([s_header])                                                                                      
    next(f_src)                                                                                                       
    f_dst.write("\n")                                                                                                 
    for line in f_src:                                                                                                
        tmp = line.replace('"', '\\"').split(s_whitespace)                                                            
        if i_stringify >= 0 and len(tmp) > i_stringify:                                                               
            tmp[i_stringify] = '"' + tmp[i_stringify] + '"'                                                           
        tmp = ','.join(tmp)                                                                                           
        f_dst.write(tmp)                                                                                              
    print("Done")                                                                                                     
    f_dst.close()                                                                                                     
    f_src.close()                                                                                                     
                                                                                                                      
                                                                                                                      
if __name__ == '__main__':          
  main()