def make_customer_file(filename, d, max_id=17770): print "Leyendo archivos movie" m = Printer(max_id) new_filename = d['output_name'] + ".csv" fd = open(new_filename, "w") fd.write("customerid," + d['column_name'] + "\n") customer_array = MyArrayList() for i in range(1, max_id + 1): m.update() with open(filename + str(i) + ".txt") as infile: for line in infile: if ':' in line: continue else: customer_array.insert(line[:-1].split(",")) m.close() print "Creando archivo " + new_filename m.reset(customer_array.get_length()) compare = d['compare'] customer_array.sort(compare) for i in range(0, customer_array.get_length()): m.update() fd.write(customer_array.toLine(i)) m.close() fd.close() return customer_array.get_length() + 1
def reduce_movie_files(filename, d, max_id=17770): m = Printer(max_id) fd = open(d['output_name'] + ".csv", "w") fd.write("movieid," + d['column_name'] + "\n") for i in range(1, max_id + 1): m.update() movie_id = 0 lines_array = [] with open(filename + str(i) + ".txt") as infile: for line in infile: if movie_id == 0: movie_id = line.split(":", 1)[0] else: lines_array.append(line[:-1].split(",")) fun = d['fun'] n = None if fun is not None: n = fun(lines_array, d['valor_inicial']) fd.write(str(movie_id) + "," + str(n) + "\n") m.close() fd.close()
def fix_movie_titles(filename, d): print "Leyendo archivo " + filename n = numero_lineas(filename) new_filename = d['output_name'] + ".csv" fd = open(new_filename, "w") fd.write(d['column_name'] + "\n") print "Creando archivo " + new_filename m = Printer(n) with open(filename) as infile: for line in infile: m.update() fd.write(line) m.close() fd.close() return n + 1
def create_super_file(filename, d, max_id=17770): m = Printer(max_id) fd = open(d['output_name'] + ".csv", "w") fd.write("movieid," + d['column_name'] + "\n") for i in range(1, max_id + 1): m.update() movie_id = 0 lines_array = [] with open(filename + str(i) + ".txt") as infile: for line in infile: if movie_id == 0: movie_id = line.split(":", 1)[0] else: fd.write(str(movie_id) + "," + line) m.close() fd.close()