sort_vods

import os
import sys
import string
import codecs
import unicodedata
from time import gmtime, strftime

if (len(sys.argv) != 2):
    print("Usage: sort_vods.py <twitch_vods_info_file>")
    sys.exit(0)

vods_info_file_path = sys.argv[1];
if not (os.path.isfile(vods_info_file_path)):
    print("Error: file", vods_info_file_path, "not found")
    sys.exit(0)

all_unicode = ''.join(chr(i) for i in range(65536))
unicode_letters = ''.join(c for c in all_unicode
                          if unicodedata.category(c)=='Lu' or unicodedata.category(c)=='Ll')

valid_chars = ";-_.() %s%s" % (unicode_letters, string.digits)
max_file_name_length = 260

log_file = open('sort_vods.log', 'w')
log_file_time_format = "[%Y-%m-%d %H:%M:%S]"

vods_info_file = codecs.open(vods_info_file_path, 'r', encoding='utf8')

folders_created = 0
files_moved = 0
first_line_text = "Video Id";
current_line = 1

# read each line from vods info file
for line in vods_info_file:
    vod_info_array = line.split(';')
    vod_info_array_size = len(vod_info_array)

    # skip header
    if (vod_info_array[0] == first_line_text):
        continue

    # check if vod title\description contains semicolons
    vod_title = ''
    expected_vod_info_array_size = 12
    if (vod_info_array_size > expected_vod_info_array_size):
        redundant_semicolons = vod_info_array_size - expected_vod_info_array_size
        for i in range(2, 3 + redundant_semicolons):
            vod_title = vod_title + vod_info_array[i]
            if (i != (2 + redundant_semicolons)):
                vod_title = vod_title + ';'
    else:
        vod_title = vod_info_array[2]

    vod_date = vod_info_array[vod_info_array_size - 8][0:10]
    # get video files links
    vod_flv_files = vod_info_array[vod_info_array_size - 2].split(',')

    for vod_flv_file in vod_flv_files:

        vod_flv_file_splitted = vod_flv_file.split('/')

        vod_flv_file_name = ''
        try:
            if (len(vod_flv_file_splitted) == 6): #old link format?
                vod_flv_file_name = vod_flv_file_splitted[5]
            else:
                vod_flv_file_name = vod_flv_file_splitted[6]
        except:
            vod_file_name_read_error = "line " + str(current_line)  + " " \
                                       ". ERROR: cannot get vod flv file name from " + vod_flv_file
            print(vod_file_name_read_error)
            log_file.write(strftime(log_file_time_format, gmtime()) + " " +
                               vod_file_name_read_error + '\n')
            continue

        if (os.path.isfile(vod_flv_file_name)):
            vod_folder = vod_date + " - " + vod_title
            if (len(vod_folder) > max_file_name_length):
                vod_folder = vod_folder[0:max_file_name_length - 1]
            vod_folder = ''.join(c for c in vod_folder if c in valid_chars)
            vod_folder = vod_folder.rstrip()

            # create new directory for each stream
            if (os.path.isdir(vod_folder)):
                folder_exists_message = "line " + str(current_line) + " " \
                                        "Folder " + vod_date + " - ... already exists"
                print(folder_exists_message)
                log_file.write(strftime(log_file_time_format, gmtime()) + " " +
                               folder_exists_message + '\n')
            else:
                try:
                    os.mkdir(vod_folder)
                    if (os.path.isdir(vod_folder)):
                        folder_created_message = "line " + str(current_line) + " " \
                                                 "Folder " + vod_date + " - ... created"
                        print(folder_created_message)
                        log_file.write(strftime(log_file_time_format, gmtime()) + " " +
                                       folder_created_message + '\n')
                        folders_created = folders_created + 1
                except:
                    folder_create_exception = "line " + str(current_line) + " " \
                                              "ERROR: cannot create folder " + vod_folder
                    print(folder_create_exception)
                    log_file.write(strftime(log_file_time_format, gmtime()) + " " +
                                   folder_create_exception + '\n')

            # move video file to corresponding stream directory
            new_vod_flv_file_location = vod_folder + "/" + vod_flv_file_name
            try:
                os.rename(vod_flv_file_name, new_vod_flv_file_location)
                file_rename_message = "line " + str(current_line) + " " \
                                      "File " + vod_flv_file_name + " moved to the new folder"
                if (os.path.isfile(new_vod_flv_file_location)):
                    print(file_rename_message)
                    log_file.write(strftime(log_file_time_format, gmtime()) + " " +
                                   file_rename_message + '\n')
                    files_moved = files_moved + 1
            except:
                file_rename_exception = "line " + str(current_line) + " " \
                                        "ERROR: cannot move file " + vod_flv_file_name + " " \
                                        "to " + new_vod_flv_file_location
                print(file_rename_exception)
                log_file.write(strftime(log_file_time_format, gmtime()) + " " +
                               file_rename_exception + '\n')

    current_line = current_line + 1

print('')
log_file.write('\n')

vods_info_file.close()

total_folders_created_mesage = "Total folders created: " + str(folders_created)
print(total_folders_created_mesage)
log_file.write(strftime(log_file_time_format, gmtime()) + " " + total_folders_created_mesage + '\n')

total_files_moved_mesage = "Total files moved: " + str(files_moved)
print(total_files_moved_mesage)
log_file.write(strftime(log_file_time_format, gmtime()) + " " + total_files_moved_mesage + '\n')

log_file.close()