summaryrefslogtreecommitdiff
path: root/src/addons/GifMaker/godot-gdgifexporter/gdgifexporter/gif-lzw/lzw.gd
blob: 9140e7cb73a2667d297303857c8bfb44d537ee01 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
extends Reference

var lsbbitpacker = preload("./lsbbitpacker.gd")
var lsbbitunpacker = preload("./lsbbitunpacker.gd")


class CodeEntry:
	var sequence: PoolByteArray
	var raw_array: Array

	func _init(_sequence):
		raw_array = _sequence
		sequence = _sequence

	func add(other):
		return CodeEntry.new(self.raw_array + other.raw_array)

	func to_string():
		var result: String = ""
		for element in self.sequence:
			result += str(element) + ", "
		return result.substr(0, result.length() - 2)


class CodeTable:
	var entries: Dictionary = {}
	var counter: int = 0
	var lookup: Dictionary = {}

	func add(entry) -> int:
		self.entries[self.counter] = entry
		self.lookup[entry.raw_array] = self.counter
		counter += 1
		return counter

	func find(entry) -> int:
		return self.lookup.get(entry.raw_array, -1)

	func has(entry) -> bool:
		return self.find(entry) != -1

	func get(index) -> CodeEntry:
		return self.entries.get(index, null)

	func to_string() -> String:
		var result: String = "CodeTable:\n"
		for id in self.entries:
			result += str(id) + ": " + self.entries[id].to_string() + "\n"
		result += "Counter: " + str(self.counter) + "\n"
		return result


func log2(value: float) -> float:
	return log(value) / log(2.0)


func get_bits_number_for(value: int) -> int:
	if value == 0:
		return 1
	return int(ceil(log2(value + 1)))


func initialize_color_code_table(colors: PoolByteArray) -> CodeTable:
	var result_code_table: CodeTable = CodeTable.new()
	for color_id in colors:
		# warning-ignore:return_value_discarded
		result_code_table.add(CodeEntry.new([color_id]))
	# move counter to the first available compression code index
	var last_color_index: int = colors.size() - 1
	var clear_code_index: int = pow(2, get_bits_number_for(last_color_index))
	result_code_table.counter = clear_code_index + 2
	return result_code_table


# compression and decompression done with source:
# http://www.matthewflickinger.com/lab/whatsinagif/lzw_image_data.asp


func compress_lzw(image: PoolByteArray, colors: PoolByteArray) -> Array:
	# Initialize code table
	var code_table: CodeTable = initialize_color_code_table(colors)
	# Clear Code index is 2**<code size>
	# <code size> is the amount of bits needed to write down all colors
	# from color table. We use last color index because we can write
	# all colors (for example 16 colors) with indexes from 0 to 15.
	# Number 15 is in binary 0b1111, so we'll need 4 bits to write all
	# colors down.
	var last_color_index: int = colors.size() - 1
	var clear_code_index: int = pow(2, get_bits_number_for(last_color_index))
	var index_stream: PoolByteArray = image
	var current_code_size: int = get_bits_number_for(clear_code_index)
	var binary_code_stream = lsbbitpacker.LSBLZWBitPacker.new()

	# initialize with Clear Code
	binary_code_stream.write_bits(clear_code_index, current_code_size)

	# Read first index from index stream.
	var index_buffer: CodeEntry = CodeEntry.new([index_stream[0]])
	var data_index: int = 1
	# <LOOP POINT>
	while data_index < index_stream.size():
		# Get the next index from the index stream.
		var k: CodeEntry = CodeEntry.new([index_stream[data_index]])
		data_index += 1
		# Is index buffer + k in our code table?
		var new_index_buffer: CodeEntry = index_buffer.add(k)
		if code_table.has(new_index_buffer):  # if YES
			# Add k to the end of the index buffer
			index_buffer = new_index_buffer
		else:  # if NO
			# Add a row for index buffer + k into our code table
			binary_code_stream.write_bits(code_table.find(index_buffer), current_code_size)

			# We don't want to add new code to code table if we've exceeded 4095
			# index.
			var last_entry_index: int = code_table.counter - 1
			if last_entry_index != 4095:
				# Output the code for just the index buffer to our code stream
				# warning-ignore:return_value_discarded
				code_table.add(new_index_buffer)
			else:
				# if we exceeded 4095 index (code table is full), we should
				# output Clear Code and reset everything.
				binary_code_stream.write_bits(clear_code_index, current_code_size)
				code_table = initialize_color_code_table(colors)
				# get_bits_number_for(clear_code_index) is the same as
				# LZW code size + 1
				current_code_size = get_bits_number_for(clear_code_index)

			# Detect when you have to save new codes in bigger bits boxes
			# change current code size when it happens because we want to save
			# flexible code sized codes
			var new_code_size_candidate: int = get_bits_number_for(code_table.counter - 1)
			if new_code_size_candidate > current_code_size:
				current_code_size = new_code_size_candidate

			# Index buffer is set to k
			index_buffer = k
	# Output code for contents of index buffer
	binary_code_stream.write_bits(code_table.find(index_buffer), current_code_size)

	# output end with End Of Information Code
	binary_code_stream.write_bits(clear_code_index + 1, current_code_size)

	var min_code_size: int = get_bits_number_for(clear_code_index) - 1

	return [binary_code_stream.pack(), min_code_size]


# gdlint: ignore=max-line-length
func decompress_lzw(code_stream_data: PoolByteArray, min_code_size: int, colors: PoolByteArray) -> PoolByteArray:
	var code_table: CodeTable = initialize_color_code_table(colors)
	var index_stream: PoolByteArray = PoolByteArray([])
	var binary_code_stream = lsbbitunpacker.LSBLZWBitUnpacker.new(code_stream_data)
	var current_code_size: int = min_code_size + 1
	var clear_code_index: int = pow(2, min_code_size)

	# CODE is an index of code table, {CODE} is sequence inside
	# code table with index CODE. The same goes for PREVCODE.

	# Remove first Clear Code from stream. We don't need it.
	binary_code_stream.remove_bits(current_code_size)

	# let CODE be the first code in the code stream
	var code: int = binary_code_stream.read_bits(current_code_size)
	# output {CODE} to index stream
	index_stream.append_array(code_table.get(code).sequence)
	# set PREVCODE = CODE
	var prevcode: int = code
	# <LOOP POINT>
	while true:
		# let CODE be the next code in the code stream
		code = binary_code_stream.read_bits(current_code_size)
		# Detect Clear Code. When detected reset everything and get next code.
		if code == clear_code_index:
			code_table = initialize_color_code_table(colors)
			current_code_size = min_code_size + 1
			code = binary_code_stream.read_bits(current_code_size)
		elif code == clear_code_index + 1:  # Stop when detected EOI Code.
			break
		# is CODE in the code table?
		var code_entry: CodeEntry = code_table.get(code)
		if code_entry != null:  # if YES
			# output {CODE} to index stream
			index_stream.append_array(code_entry.sequence)
			# let k be the first index in {CODE}
			var k: CodeEntry = CodeEntry.new([code_entry.sequence[0]])
			# warning-ignore:return_value_discarded
			# add {PREVCODE} + k to the code table
			code_table.add(code_table.get(prevcode).add(k))
			# set PREVCODE = CODE
			prevcode = code
		else:  # if NO
			# let k be the first index of {PREVCODE}
			var prevcode_entry: CodeEntry = code_table.get(prevcode)
			var k: CodeEntry = CodeEntry.new([prevcode_entry.sequence[0]])
			# output {PREVCODE} + k to index stream
			index_stream.append_array(prevcode_entry.add(k).sequence)
			# add {PREVCODE} + k to code table
			# warning-ignore:return_value_discarded
			code_table.add(prevcode_entry.add(k))
			# set PREVCODE = CODE
			prevcode = code

		# Detect when we should increase current code size and increase it.
		var new_code_size_candidate: int = get_bits_number_for(code_table.counter)
		if new_code_size_candidate > current_code_size:
			current_code_size = new_code_size_candidate

	return index_stream