source: main/trunk/model-sites-dev/twso/PreProcessing/namefinder1.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()
        new_lines.append(line)

    return new_lines

#default split for a name: JESSICA TURNER --> TURNER, JESSICA
#names such as de Rooy, Ana must be entered manually
def splitName(name):
    name = name.split()
    if len(name) >= 2:
        return name[1] + ", " + name[0]
    else:
        return name[0]

#See if the name is in the array or not - used to avoid duplicate occurences of names
def checkForName(doc, n):

    if not doc:
        return False
    else:

        for name in doc:
            if n==name:
                return True
                break
            else:
                return False

#add the name to the document array in the correct format
#soloists should appear soloist Turner, Jessica followed by Turner, Jessica on a new line as they are also considered a player
#conductors should appear conductor Turner, Jessica and NOT Turner, Jessica on a separate line as they are not a player
#players should appear on a newline as Turner, Jessica
def appendName(doc, name, role):

    if (role == "soloist") | (role == "soloists"):
        doc.append("soloist " + name)
        doc.append(name)
    elif (role == "conductor") | (role == "conductors"):
        doc.append("conductor " + name)
    else:
        doc.append(name)

#This adds the player to the doc array, it is called after a role has been found
def addPlayer(doc, iterator, role):

    name = iterator.next()
    notfin = True
    
    while notfin == True:
        if not name:
            break
        else:
            print "NAME FOUND: " + name
            name = splitName(name)
            print "NAME TO BE ADDED: " + name
            if checkForName(doc, name) == False:
                #prompting user to select the name or not 
                #helps to prevent any mistakes if a role is found but the next line is not necessarily a name
                #Y = Yes - using this option will add the name to the document array
                #N = No - using this option will stop you from finding names
                #ENTER NAME = Enter the correct format of the name - for use with names such as van der Goes, Natalie
                #which will not be picked up correctly at the moment
                #Hit enter to continue will skip over this line and retrieve next line
                var = raw_input("Is this a name? Y|N|E|Hit any other key to continue...\n")
                var = var.upper()
                if var=="Y":
                    appendName(doc, name, role)
                elif var=="E":
                    var = raw_input("Enter name: ")
                    appendName(doc, var, role)
                elif var=="N":
                    notfin = False
            else:
                print "The name " + name + " is already in the document."

            try:
                name = iterator.next()
            except StopIteration:
                break

#check to see if role can be found in document
def findRole(roles, lines):

    nameDoc_lines = []
    #this is used so that if splitting needs to be done for program to detect role it can figure out whether to split it or not
    specialcases = ['1st violin', '2nd violin', '1st violins', '2nd violins']

    iterator = lines.__iter__()
    for line in iterator:
        
        if (not line):
            try:
                line = iterator.next()
            except StopIteration:
                return nameDoc_lines

        print line

        for case in specialcases:
            if line.lower() == case:
                line = line.split()
                line = line[1]
                break

        for role in roles:
            if role == line.lower():
                print "LINE: " + line
                print "MATCH FOUND " + role.upper()
                addPlayer(nameDoc_lines, iterator, role)
                break
    return nameDoc_lines

#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(orig_file, nameDoc):
    start_of_file = orig_file.split(".")
    new_file_name = start_of_file[0] + "_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 nameDoc:
        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():
    roles = openFile("roles")
    lines = openFile(sys.argv[1])

    nameDoc = findRole(roles, lines)
    
    writeToFile(sys.argv[1], nameDoc)

main()