|
| 1 | +def expand(data, uncompressed_size=None): |
| 2 | + compress = EmbCompress() |
| 3 | + return compress.decompress(data, uncompressed_size) |
| 4 | + |
| 5 | + |
| 6 | +class Huffman: |
| 7 | + def __init__(self, lengths=None, value=0): |
| 8 | + self.default_value = value |
| 9 | + self.lengths = lengths |
| 10 | + self.table = None |
| 11 | + self.table_width = 0 |
| 12 | + |
| 13 | + def build_table(self): |
| 14 | + """Build an index huffman table based on the lengths. lowest index value wins in a tie.""" |
| 15 | + self.table_width = max(self.lengths) |
| 16 | + self.table = [] |
| 17 | + size = (1 << self.table_width) |
| 18 | + for bit_length in range(1, self.table_width + 1): |
| 19 | + size /= 2 |
| 20 | + for len_index in range(0, len(self.lengths)): |
| 21 | + length = self.lengths[len_index] |
| 22 | + if length == bit_length: |
| 23 | + self.table += [len_index] * size |
| 24 | + |
| 25 | + def lookup(self, byte_lookup): |
| 26 | + """lookup into the index, returns value and length |
| 27 | + must be requested with 2 bytes.""" |
| 28 | + if self.table is None: |
| 29 | + return self.default_value, 0 |
| 30 | + v = self.table[byte_lookup >> (16 - self.table_width)] |
| 31 | + return v, self.lengths[v] |
| 32 | + |
| 33 | + |
| 34 | +class EmbCompress: |
| 35 | + def __init__(self): |
| 36 | + self.bit_position = 0 |
| 37 | + self.input_data = None |
| 38 | + self.block_elements = None |
| 39 | + self.character_huffman = None |
| 40 | + self.distance_huffman = None |
| 41 | + |
| 42 | + def get_bits(self, start_pos_in_bits, length): |
| 43 | + end_pos_in_bits = start_pos_in_bits + length - 1 |
| 44 | + start_pos_in_bytes = int(start_pos_in_bits / 8) |
| 45 | + end_pos_in_bytes = int(end_pos_in_bits / 8) |
| 46 | + value = 0 |
| 47 | + for i in range(start_pos_in_bytes, end_pos_in_bytes + 1): |
| 48 | + value <<= 8 |
| 49 | + try: |
| 50 | + value |= self.input_data[i] & 0xFF |
| 51 | + except IndexError: |
| 52 | + pass |
| 53 | + unused_bits_right_of_sample = (8 - (end_pos_in_bits + 1) % 8) % 8 |
| 54 | + mask_sample_bits = (1 << length) - 1 |
| 55 | + original = (value >> unused_bits_right_of_sample) & mask_sample_bits |
| 56 | + return original |
| 57 | + |
| 58 | + def pop(self, bit_count): |
| 59 | + value = self.peek(bit_count) |
| 60 | + self.slide(bit_count) |
| 61 | + return value |
| 62 | + |
| 63 | + def peek(self, bit_count): |
| 64 | + return self.get_bits(self.bit_position, bit_count) |
| 65 | + |
| 66 | + def slide(self, bit_count): |
| 67 | + self.bit_position += bit_count |
| 68 | + |
| 69 | + def read_variable_length(self): |
| 70 | + m = self.pop(3) |
| 71 | + if m != 7: |
| 72 | + return m |
| 73 | + for q in range(0, 13): # max read is 16 bit, 3 bits already used. It can't exceed 16-3 |
| 74 | + s = self.pop(1) |
| 75 | + if s == 1: |
| 76 | + m += 1 |
| 77 | + else: |
| 78 | + break |
| 79 | + return m |
| 80 | + |
| 81 | + def load_character_length_huffman(self): |
| 82 | + count = self.pop(5) |
| 83 | + if count == 0: |
| 84 | + v = self.pop(5) |
| 85 | + huffman = Huffman(value=v) |
| 86 | + else: |
| 87 | + huffman_code_lengths = [0] * count |
| 88 | + index = 0 |
| 89 | + while index < count: |
| 90 | + if index == 3: # Special index 3, skip up to 3 elements. |
| 91 | + index += self.pop(2) |
| 92 | + huffman_code_lengths[index] = self.read_variable_length() |
| 93 | + index += 1 |
| 94 | + huffman = Huffman(huffman_code_lengths, 8) |
| 95 | + huffman.build_table() |
| 96 | + return huffman |
| 97 | + |
| 98 | + def load_character_huffman(self, length_huffman): |
| 99 | + count = self.pop(9) |
| 100 | + if count == 0: |
| 101 | + v = self.pop(9) |
| 102 | + huffman = Huffman(value=v) |
| 103 | + else: |
| 104 | + huffman_code_lengths = [0] * count |
| 105 | + index = 0 |
| 106 | + while index < count: |
| 107 | + h = length_huffman.lookup(self.peek(16)) |
| 108 | + c = h[0] |
| 109 | + self.slide(h[1]) |
| 110 | + if c == 0: # C == 0, skip 1. |
| 111 | + c = 1 |
| 112 | + index += c |
| 113 | + elif c == 1: # C == 1, skip 3 + read(4) |
| 114 | + c = 3 + self.pop(4) |
| 115 | + index += c |
| 116 | + elif c == 2: # C == 2, skip 20 + read(9) |
| 117 | + c = 20 + self.pop(9) |
| 118 | + index += c |
| 119 | + else: |
| 120 | + c -= 2 |
| 121 | + huffman_code_lengths[index] = c |
| 122 | + index += 1 |
| 123 | + huffman = Huffman(huffman_code_lengths) |
| 124 | + huffman.build_table() |
| 125 | + return huffman |
| 126 | + |
| 127 | + def load_distance_huffman(self): |
| 128 | + count = self.pop(5) |
| 129 | + if count == 0: |
| 130 | + v = self.pop(5) |
| 131 | + huffman = Huffman(value=v) |
| 132 | + else: |
| 133 | + index = 0 |
| 134 | + lengths = [0] * count |
| 135 | + for i in range(0, count): |
| 136 | + lengths[index] = self.read_variable_length() |
| 137 | + index += 1 |
| 138 | + huffman = Huffman(lengths) |
| 139 | + huffman.build_table() |
| 140 | + return huffman |
| 141 | + |
| 142 | + def load_block(self): |
| 143 | + self.block_elements = self.pop(16) |
| 144 | + character_length_huffman = self.load_character_length_huffman() |
| 145 | + self.character_huffman = self.load_character_huffman(character_length_huffman) |
| 146 | + self.distance_huffman = self.load_distance_huffman() |
| 147 | + |
| 148 | + def get_token(self): |
| 149 | + if self.block_elements <= 0: |
| 150 | + self.load_block() |
| 151 | + self.block_elements -= 1 |
| 152 | + h = self.character_huffman.lookup(self.peek(16)) |
| 153 | + self.slide(h[1]) |
| 154 | + return h[0] |
| 155 | + |
| 156 | + def get_position(self): |
| 157 | + h = self.distance_huffman.lookup(self.peek(16)) |
| 158 | + self.slide(h[1]) |
| 159 | + if h[0] == 0: |
| 160 | + return 0 |
| 161 | + v = h[0] - 1 |
| 162 | + v = (1 << v) + self.pop(v) |
| 163 | + return v |
| 164 | + |
| 165 | + def decompress(self, input_data, uncompressed_size=None): |
| 166 | + self.input_data = input_data |
| 167 | + output_data = [] |
| 168 | + self.block_elements = -1 |
| 169 | + bits_total = (len(input_data) * 8) |
| 170 | + while bits_total > self.bit_position and (uncompressed_size is None or len(output_data) <= uncompressed_size): |
| 171 | + character = self.get_token() |
| 172 | + if character <= 255: # literal. |
| 173 | + output_data.append(character) |
| 174 | + elif character == 510: |
| 175 | + break # END |
| 176 | + else: |
| 177 | + length = character - 253 # Min length is 3. 256-253=3. |
| 178 | + back = self.get_position() + 1 |
| 179 | + position = len(output_data) - back |
| 180 | + if back > length: |
| 181 | + # Entire lookback is already within output data. |
| 182 | + output_data += output_data[position:position + length] |
| 183 | + else: |
| 184 | + # Will read & write the same data at some point. |
| 185 | + for i in range(position, position + length): |
| 186 | + output_data.append(output_data[i]) |
| 187 | + return output_data |
0 commit comments