/src/qpdf/libqpdf/Pl_LZWDecoder.cc
Line | Count | Source (jump to first uncovered line) |
1 | | #include <qpdf/Pl_LZWDecoder.hh> |
2 | | |
3 | | #include <qpdf/QIntC.hh> |
4 | | #include <qpdf/QTC.hh> |
5 | | #include <qpdf/QUtil.hh> |
6 | | #include <cstring> |
7 | | #include <stdexcept> |
8 | | |
9 | | Pl_LZWDecoder::Pl_LZWDecoder(char const* identifier, Pipeline* next, bool early_code_change) : |
10 | 1.73k | Pipeline(identifier, next), |
11 | 1.73k | code_change_delta(early_code_change) |
12 | 1.73k | { |
13 | 1.73k | if (!next) { |
14 | 0 | throw std::logic_error("Attempt to create Pl_LZWDecoder with nullptr as next"); |
15 | 0 | } |
16 | 1.73k | } |
17 | | |
18 | | void |
19 | | Pl_LZWDecoder::write(unsigned char const* bytes, size_t len) |
20 | 4.88k | { |
21 | 1.37M | for (size_t i = 0; i < len; ++i) { |
22 | 1.36M | buf[next_char_++] = bytes[i]; |
23 | 1.36M | if (next_char_ == 3) { |
24 | 455k | next_char_ = 0; |
25 | 455k | } |
26 | 1.36M | this->bits_available += 8; |
27 | 1.36M | if (this->bits_available >= this->code_size) { |
28 | 1.18M | sendNextCode(); |
29 | 1.18M | } |
30 | 1.36M | } |
31 | 4.88k | } |
32 | | |
33 | | void |
34 | | Pl_LZWDecoder::finish() |
35 | 858 | { |
36 | 858 | next()->finish(); |
37 | 858 | } |
38 | | |
39 | | void |
40 | | Pl_LZWDecoder::sendNextCode() |
41 | 1.18M | { |
42 | 1.18M | unsigned int high = this->byte_pos; |
43 | 1.18M | unsigned int med = (this->byte_pos + 1) % 3; |
44 | 1.18M | unsigned int low = (this->byte_pos + 2) % 3; |
45 | | |
46 | 1.18M | unsigned int bits_from_high = 8 - this->bit_pos; |
47 | 1.18M | unsigned int bits_from_med = this->code_size - bits_from_high; |
48 | 1.18M | unsigned int bits_from_low = 0; |
49 | 1.18M | if (bits_from_med > 8) { |
50 | 28.5k | bits_from_low = bits_from_med - 8; |
51 | 28.5k | bits_from_med = 8; |
52 | 28.5k | } |
53 | 1.18M | unsigned int high_mask = (1U << bits_from_high) - 1U; |
54 | 1.18M | unsigned int med_mask = 0xff - ((1U << (8 - bits_from_med)) - 1U); |
55 | 1.18M | unsigned int low_mask = 0xff - ((1U << (8 - bits_from_low)) - 1U); |
56 | 1.18M | unsigned int code = 0; |
57 | 1.18M | code += (this->buf[high] & high_mask) << bits_from_med; |
58 | 1.18M | code += ((this->buf[med] & med_mask) >> (8 - bits_from_med)); |
59 | 1.18M | if (bits_from_low) { |
60 | 28.5k | code <<= bits_from_low; |
61 | 28.5k | code += ((this->buf[low] & low_mask) >> (8 - bits_from_low)); |
62 | 28.5k | this->byte_pos = low; |
63 | 28.5k | this->bit_pos = bits_from_low; |
64 | 1.16M | } else { |
65 | 1.16M | this->byte_pos = med; |
66 | 1.16M | this->bit_pos = bits_from_med; |
67 | 1.16M | } |
68 | 1.18M | if (this->bit_pos == 8) { |
69 | 149k | this->bit_pos = 0; |
70 | 149k | ++this->byte_pos; |
71 | 149k | this->byte_pos %= 3; |
72 | 149k | } |
73 | 1.18M | this->bits_available -= this->code_size; |
74 | | |
75 | 1.18M | handleCode(code); |
76 | 1.18M | } |
77 | | |
78 | | unsigned char |
79 | | Pl_LZWDecoder::getFirstChar(unsigned int code) |
80 | 7.85k | { |
81 | 7.85k | unsigned char result = '\0'; |
82 | 7.85k | if (code < 256) { |
83 | 149 | result = static_cast<unsigned char>(code); |
84 | 7.70k | } else if (code > 257) { |
85 | 7.70k | unsigned int idx = code - 258; |
86 | 7.70k | if (idx >= table.size()) { |
87 | 0 | throw std::runtime_error("Pl_LZWDecoder::getFirstChar: table overflow"); |
88 | 0 | } |
89 | 7.70k | Buffer& b = table.at(idx); |
90 | 7.70k | result = b.getBuffer()[0]; |
91 | 7.70k | } else { |
92 | 0 | throw std::runtime_error( |
93 | 0 | "Pl_LZWDecoder::getFirstChar called with invalid code (" + std::to_string(code) + ")"); |
94 | 0 | } |
95 | 7.85k | return result; |
96 | 7.85k | } |
97 | | |
98 | | void |
99 | | Pl_LZWDecoder::addToTable(unsigned char c) |
100 | 356k | { |
101 | 356k | unsigned int last_size = 0; |
102 | 356k | unsigned char const* last_data = nullptr; |
103 | 356k | unsigned char tmp[1]; |
104 | | |
105 | 356k | if (this->last_code < 256) { |
106 | 348k | tmp[0] = static_cast<unsigned char>(this->last_code); |
107 | 348k | last_data = tmp; |
108 | 348k | last_size = 1; |
109 | 348k | } else if (this->last_code > 257) { |
110 | 7.73k | unsigned int idx = this->last_code - 258; |
111 | 7.73k | if (idx >= table.size()) { |
112 | 0 | throw std::runtime_error("Pl_LZWDecoder::addToTable: table overflow"); |
113 | 0 | } |
114 | 7.73k | Buffer& b = table.at(idx); |
115 | 7.73k | last_data = b.getBuffer(); |
116 | 7.73k | last_size = QIntC::to_uint(b.getSize()); |
117 | 7.73k | } else { |
118 | 0 | throw std::runtime_error( |
119 | 0 | "Pl_LZWDecoder::addToTable called with invalid code (" + |
120 | 0 | std::to_string(this->last_code) + ")"); |
121 | 0 | } |
122 | | |
123 | 356k | Buffer entry(1 + last_size); |
124 | 356k | unsigned char* new_data = entry.getBuffer(); |
125 | 356k | memcpy(new_data, last_data, last_size); |
126 | 356k | new_data[last_size] = c; |
127 | 356k | this->table.push_back(std::move(entry)); |
128 | 356k | } |
129 | | |
130 | | void |
131 | | Pl_LZWDecoder::handleCode(unsigned int code) |
132 | 1.18M | { |
133 | 1.18M | if (this->eod) { |
134 | 829k | return; |
135 | 829k | } |
136 | | |
137 | 359k | if (code == 256) { |
138 | 1.25k | if (!this->table.empty()) { |
139 | 1.17k | QTC::TC("libtests", "Pl_LZWDecoder intermediate reset"); |
140 | 1.17k | } |
141 | 1.25k | this->table.clear(); |
142 | 1.25k | this->code_size = 9; |
143 | 358k | } else if (code == 257) { |
144 | 88 | this->eod = true; |
145 | 358k | } else { |
146 | 358k | if (this->last_code != 256) { |
147 | | // Add to the table from last time. New table entry would be what we read last plus the |
148 | | // first character of what we're reading now. |
149 | 356k | unsigned char next_c = '\0'; |
150 | 356k | unsigned int table_size = QIntC::to_uint(table.size()); |
151 | 356k | if (code < 256) { |
152 | | // just read < 256; last time's next_c was code |
153 | 348k | next_c = static_cast<unsigned char>(code); |
154 | 348k | } else if (code > 257) { |
155 | 8.14k | size_t idx = code - 258; |
156 | 8.14k | if (idx > table_size) { |
157 | 290 | throw std::runtime_error("LZWDecoder: bad code received"); |
158 | 7.85k | } else if (idx == table_size) { |
159 | | // The encoder would have just created this entry, so the first character of |
160 | | // this entry would have been the same as the first character of the last entry. |
161 | 235 | QTC::TC("libtests", "Pl_LZWDecoder last was table size"); |
162 | 235 | next_c = getFirstChar(this->last_code); |
163 | 7.61k | } else { |
164 | 7.61k | next_c = getFirstChar(code); |
165 | 7.61k | } |
166 | 8.14k | } |
167 | 356k | unsigned int new_idx = 258 + table_size; |
168 | 356k | if (new_idx == 4096) { |
169 | 7 | throw std::runtime_error("LZWDecoder: table full"); |
170 | 7 | } |
171 | 356k | addToTable(next_c); |
172 | 356k | unsigned int change_idx = new_idx + code_change_delta; |
173 | 356k | if ((change_idx == 511) || (change_idx == 1023) || (change_idx == 2047)) { |
174 | 686 | ++this->code_size; |
175 | 686 | } |
176 | 356k | } |
177 | | |
178 | 358k | if (code < 256) { |
179 | 350k | auto ch = static_cast<unsigned char>(code); |
180 | 350k | next()->write(&ch, 1); |
181 | 350k | } else { |
182 | 7.88k | unsigned int idx = code - 258; |
183 | 7.88k | if (idx >= table.size()) { |
184 | 35 | throw std::runtime_error("Pl_LZWDecoder::handleCode: table overflow"); |
185 | 35 | } |
186 | 7.84k | Buffer& b = table.at(idx); |
187 | 7.84k | next()->write(b.getBuffer(), b.getSize()); |
188 | 7.84k | } |
189 | 358k | } |
190 | | |
191 | 359k | this->last_code = code; |
192 | 359k | } |