# imdb.py
#
# CREATED: 2017.10.08  ABS  copied from process_svm
# CREATED: 2017.10.11  ABS  refactored
# CREATED: 2017.11.14  ABS  parms on input flags
#
# ToDo: [X] put parms on input flags
#       [_] make file len=0 mean read to EOF
#       [_] skip crew info
#       [_] scrape show URL from IMDB.com
#       [_] deal with non-Latin characters

import sys
import urllib
import subprocess
import argparse
parser = argparse.ArgumentParser()
parser.add_argument("-a", "--actors", help="length of actors file",
                    type=int,default=480)
parser.add_argument("--actors_filename", help="actors file name",
                    default='actors.txt')
parser.add_argument("-s", "--shows", help="length of shows file",
                    type=int,default=71)
parser.add_argument("--shows_filename", help="shows file name",
                    default='show_urls.txt')
parser.add_argument("-t", "--test", help="test arg parsing then exit",
                    action="store_true")
args = parser.parse_args()

if args.actors:
    len_actors_file = args.actors;
    if args.test:
        print 'TEST: len_actors_file: {0}'.format(len_actors_file)
else:
    if args.test:
        print("TEST: actors not set")

if args.shows:
    len_shows_file = args.shows;
    if args.test:
        print 'TEST: len_shows_file: {0}'.format(len_shows_file)
else:
    if args.test:
        print("TEST: shows not set")

if args.test:
    print 'TEST: actors filename: {0}'.format(args.actors_filename)
    print 'TEST: shows filename: {0}'.format(args.shows_filename)

def print_ruler(num_shows):
    print '<TR>'
    print '    <TD BGCOLOR=#c0c0c0><B><FONT SIZE=-1>actor \ show</FONT></B></TD>'
    for ns in range(num_shows):
        if ns < 11:
            print '    <TD BGCOLOR=#c0c0c0><FONT SIZE=-2><FONT COLOR=#c0c0c0>_</FONT>{0}</FONT></TD>'.format(ns + 1);
        else:
            print '    <TD BGCOLOR=#c0c0c0><FONT SIZE=-2>{0}</FONT></TD>'.format(ns + 1);
    print '    <TD BGCOLOR=#c0c0c0><FONT SIZE=-2>tot</FONT></TD>'
    print '</TR>'

actors_file = open(args.actors_filename, 'rU')
shows_file = open(args.shows_filename, 'rU')

if args.test:
    sys.exit()

#
# init vars
#
tot = 0;
actor_count_min = 1;
actor_count_max = 1;
show_count_min = 1;
show_count_max = 1;
#
# init arrays
#
actor_array = ["joe" for i in range(len_actors_file)] 
show_array = ["felix the cat" for i in range(len_shows_file)] 
actor_count_array = [0 for i in range(len_actors_file)] 
show_count_array = [0 for i in range(len_shows_file)] 
matrix = [[0 for j in range(len_shows_file)] for i in range(len_actors_file)] 
#
# loop through actors file and set actor_array
#
for actor_num in range (0, len_actors_file):
    #
    # read next line in actors file
    #
    actors_line = actors_file.readline()
    stripped_actors_line = actors_line.rstrip('\n')
   
    actor = stripped_actors_line; 
    actor_array[actor_num] = actor;
actors_file.close()
#
# loop through shows file
#
for show_num in range (0, len_shows_file):
    #
    # read next line in shows file
    #
    shows_line = shows_file.readline()
    stripped_shows_line = shows_line.rstrip('\n')
    parsed_show_line = stripped_shows_line.split(";")

    show_array[show_num] = parsed_show_line[0]
    link = parsed_show_line[1]

    f = urllib.urlopen(link)
    myfile = f.read()
    #
    # loop through actor_array
    #
    for actor_num in range (0, len_actors_file):
        actor = actor_array[actor_num];
        #
        # set matrix
        #
        if myfile.find(actor) == -1:
            matrix[actor_num][show_num] = 0;
        else:
            matrix[actor_num][show_num] = 1;

            tot = tot + 1;

            actor_count_array[actor_num] = actor_count_array[actor_num] + 1;
            show_count_array[show_num] = show_count_array[show_num] + 1;

            if actor_count_array[actor_num] > actor_count_max:
                actor_count_max = actor_count_array[actor_num];

            # don't count the first column
            if show_num != 0 and show_count_array[show_num] > show_count_max:
                show_count_max = show_count_array[show_num];
shows_file.close()
#
# loop through shows and print
#
print '<OL>'
for show_num in range (0, len_shows_file):
    print '<LI> {0}'.format(show_array[show_num]);
print '</OL>'
print '<P>'
print '<TABLE BORDER=2>'
#
# loop through matrix and print
#
# loop through actors array
#
for actor_num in range (0, len_actors_file):
    if (actor_num/20.0).is_integer():
        #
        # print ruler
        #
        print_ruler(len_shows_file);
    print '<TR> <TD><B><FONT SIZE=-1>{0}</FONT></B></TD>'.format(actor_array[actor_num])
    #
    # loop through shows
    #
    for show_num in range (0, len_shows_file):
        if matrix[actor_num][show_num] == 0:
            print '    <TD BGCOLOR=WHITE> </TD>'
        else:
            print '    <TD BGCOLOR=BLACK> </TD>'
    if actor_count_array[actor_num] == actor_count_max:
        print '    <TD BGCOLOR=PINK><FONT SIZE=-2>{0}</FONT></TD>'.format(actor_count_array[actor_num]);
    elif actor_count_array[actor_num] == actor_count_min:
        print '    <TD BGCOLOR=#ffff80><FONT SIZE=-2>{0}</FONT></TD>'.format(actor_count_array[actor_num]);
    else:
        print '    <TD><FONT SIZE=-2>{0}</FONT></TD></TR>'.format(actor_count_array[actor_num]);
#
# print show totals
#
print '<TR> <TD>totals</TD>'
for show_num in range (0, len_shows_file):
    if show_count_array[show_num] == show_count_max:
        print '    <TD BGCOLOR=PINK><FONT SIZE=-2>{0}</TD>'.format(show_count_array[show_num]);
    elif show_count_array[show_num] == show_count_min:
        print '    <TD BGCOLOR=#ffff80><FONT SIZE=-2>{0}</TD>'.format(show_count_array[show_num]);
    else:
        print '    <TD><FONT SIZE=-2>{0}</TD>'.format(show_count_array[show_num]);
print '    <TD><FONT SIZE=-2>{0}</TD></TR>'.format(tot);
#
# end of table
#
print '</TABLE>'