lz78/lz_78_2x2.py

   1 import sys
   2 import struct
   3
   4 ########################################################################################
   5 # Apenas comprimir
   6 #
   7 # python lz_78_2x2.py <caminho_para_o_ficheiro_original>
   8 #
   9 #
  10 # Os ficheiros gerados ficarão na mesma pasta que o script
  11 ########################################################################################
  12
  13 def lz78_compression(image):
  14     dictionary = {0: ""}
  15     output = []
  16     symbol = ""
  17
  18     for i in image:
  19         if (symbol + i) in dictionary.values():
  20             symbol += i
  21         else:
  22             if symbol == "":
  23                 output.append([0, i])
  24                 dictionary[len(dictionary)] = i
  25             else:
  26                 idx = list(dictionary.keys())[list(dictionary.values()).index(symbol)]
  27                 output.append([idx, i])
  28                 dictionary[len(dictionary)] = symbol + i
  29             symbol = ""
  30
  31     if symbol != "":
  32         idx = list(dictionary.keys())[list(dictionary.values()).index(symbol)]
  33         output.append([idx, ""])
  34
  35     return output, dictionary
  36
  37 def clean_image_data(image_data):
  38     clean_text = ""
  39     for line in image_data:
  40         line = line.strip().replace(" ", "")
  41         if not line or line.startswith('#'):
  42             continue
  43         clean_text += line
  44     return clean_text
  45
  46 def chunkstring(string, length):
  47     return [string[i : length+i] for i in range(0, len(string), length)]
  48
  49 def get_blocks(bitstream, largura, altura):
  50     matriz = [bitstream[i*largura:(i+1)*largura] for i in range(altura)]
  51     # Padding
  52     if largura % 2 != 0:
  53         matriz = [linha + '0' for linha in matriz]
  54         largura += 1
  55     if altura % 2 != 0:
  56         matriz.append('0' * largura)
  57         altura += 1
  58
  59     blocos = []
  60     for r in range(0, altura, 2):
  61         for c in range(0, largura, 2):
  62             bloco = matriz[r][c] + matriz[r][c+1] + matriz[r+1][c] + matriz[r+1][c+1]
  63             blocos.append(bloco)
  64     return blocos, largura, altura
  65
  66
  67 with open(sys.argv[1], "r") as image:
  68     lines = image.readlines()
  69     image_size = lines[1].split()
  70
  71     if lines[0].strip() != 'P1':
  72         raise ValueError("This is not a pbm file")
  73
  74     largura_orig = int(image_size[0])
  75     altura_orig = int(image_size[1])
  76
  77     clean_text = clean_image_data(lines[2:])
  78
  79     image_blocks, largura_pad, altura_pad = get_blocks(clean_text, largura_orig, altura_orig)
  80
  81     output, dic = lz78_compression(image_blocks)
  82
  83     with open(sys.argv[1] + "_compressed", "w") as save_file:
  84         save_file.write(f"{output}\n{largura_orig} {altura_orig} {largura_pad} {altura_pad}")
  85
  86     with open(sys.argv[1] + ".bin", "wb") as save_file:
  87         save_file.write(struct.pack('IIII', largura_orig, altura_orig, largura_pad, altura_pad))
  88         for indice, simbolo in output:
  89             simbolo_val = int(simbolo, 2) if simbolo != "" else 16
  90             save_file.write(struct.pack('HB', indice, simbolo_val))