source: main/trunk/model-sites-dev/twso/PreProcessing/masternamecreator2.py@ 34227

import sys
import os

#open the file and return an array of the lines in the file
def openFile(filename):
    
    programme = open(filename)
    lines = programme.readlines()
    programme.close()

    new_lines = []

    for line in lines:
        line = line.strip()
                if line:
                new_lines.append(line)

    return new_lines

def checkForSpecial(name):
                
    name_parts = name.split()
    if len(name_parts) > 2:
        if (name_parts[0] == "conductor") | (name_parts[0] == "soloist"):
            i = 1
            name = ""
            while i < len(name_parts):
                name += name_parts[i] + " "
                i +=1
                                    
    return name.strip()

def removeDuplicates(lines):

    checked = []
    for name in lines:
        if name not in checked:
            checked.append(name)

    return checked

#write names out to file, takes orig_file as name so that new file can be created based on original file name
#takes doc array of names to write to file
def writeToFile(lines):

    new_file_name = "master_name_list.txt"
    
    if os.path.exists(new_file_name):
        new_file = file(new_file_name, "r+")
    else:
        new_file = file(new_file_name, "w+")

    for name in lines:
        new_file.write(name + "\n")
    
    print "Finished writing names to " + new_file_name

#main method which calls methods in order of execution required
def main():

        new_lines = []
    i = 1
    while i < len(sys.argv):
        lines = openFile(sys.argv[i])
        extras = openFile("master_name_list.txt")

        for extra in extras:
            new_lines.append(extra)

        for name in lines:
            new_name = checkForSpecial(name)
            new_lines.append(new_name)

        new_lines.sort()

        new_lines = removeDuplicates(new_lines)

        writeToFile(new_lines)
        i += 1

main()