Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 1 | // Copyright 2018 The Wuffs Authors. |
| 2 | // |
| 3 | // Licensed under the Apache License, Version 2.0 (the "License"); |
| 4 | // you may not use this file except in compliance with the License. |
| 5 | // You may obtain a copy of the License at |
| 6 | // |
| 7 | // https://www.apache.org/licenses/LICENSE-2.0 |
| 8 | // |
| 9 | // Unless required by applicable law or agreed to in writing, software |
| 10 | // distributed under the License is distributed on an "AS IS" BASIS, |
| 11 | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 12 | // See the License for the specific language governing permissions and |
| 13 | // limitations under the License. |
| 14 | |
Nigel Tao | 788479d | 2021-08-22 10:52:51 +1000 | [diff] [blame] | 15 | //go:build ignore |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 16 | // +build ignore |
| 17 | |
| 18 | package main |
| 19 | |
| 20 | // make-artificial.go makes test data in various formats. |
| 21 | // |
| 22 | // See test/data/artificial/*.make.txt for examples. |
| 23 | // |
| 24 | // Usage: go run make-artificial.go < foo.make.txt |
| 25 | |
| 26 | import ( |
| 27 | "bytes" |
| 28 | "compress/lzw" |
| 29 | "fmt" |
Nigel Tao | 226c476 | 2021-08-22 11:05:43 +1000 | [diff] [blame] | 30 | "io" |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 31 | "os" |
| 32 | "strconv" |
| 33 | "strings" |
| 34 | ) |
| 35 | |
| 36 | type stateFunc func(line string) (stateFunc, error) |
| 37 | |
| 38 | type repeat struct { |
| 39 | count uint32 |
| 40 | remaining string |
| 41 | } |
| 42 | |
| 43 | var ( |
| 44 | state stateFunc |
| 45 | repeats []repeat |
| 46 | out []byte |
| 47 | formats = map[string]stateFunc{} |
| 48 | ) |
| 49 | |
| 50 | func main() { |
| 51 | if err := main1(); err != nil { |
| 52 | os.Stderr.WriteString(err.Error() + "\n") |
| 53 | os.Exit(1) |
| 54 | } |
| 55 | } |
| 56 | |
| 57 | func main1() error { |
Nigel Tao | 226c476 | 2021-08-22 11:05:43 +1000 | [diff] [blame] | 58 | stdin, err := io.ReadAll(os.Stdin) |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 59 | if err != nil { |
| 60 | return err |
| 61 | } |
| 62 | s := string(stdin) |
| 63 | for remaining := ""; len(s) > 0; s, remaining = remaining, "" { |
| 64 | if i := strings.IndexByte(s, '\n'); i >= 0 { |
| 65 | s, remaining = s[:i], s[i+1:] |
| 66 | } |
| 67 | s = strings.TrimSpace(s) |
| 68 | if s == "" || s[0] == '#' { |
| 69 | continue |
| 70 | } |
| 71 | |
| 72 | if state == nil { |
| 73 | const m = "make " |
| 74 | if !strings.HasPrefix(s, m) { |
| 75 | return fmt.Errorf(`input must start with "make foo"`) |
| 76 | } |
| 77 | s = s[len(m):] |
| 78 | state = formats[s] |
| 79 | if state == nil { |
| 80 | return fmt.Errorf("unsupported format %q", s) |
| 81 | } |
| 82 | continue |
| 83 | } |
| 84 | |
| 85 | const rep = "repeat " |
| 86 | if strings.HasPrefix(s, rep) { |
| 87 | args := s[len(rep):] |
| 88 | count, args, ok := parseNum(args) |
| 89 | if !ok || count <= 0 || args != "[" { |
| 90 | return fmt.Errorf("bad repeat command: %q", s) |
| 91 | } |
| 92 | repeats = append(repeats, repeat{ |
| 93 | count: count, |
| 94 | remaining: remaining, |
| 95 | }) |
| 96 | continue |
| 97 | } |
| 98 | |
| 99 | if s == "]" { |
| 100 | if len(repeats) <= 0 { |
| 101 | return fmt.Errorf(`unbalanced close-repeat command: "]"`) |
| 102 | } |
| 103 | i := len(repeats) - 1 |
| 104 | repeats[i].count-- |
| 105 | if repeats[i].count == 0 { |
| 106 | repeats = repeats[:i] |
| 107 | } else { |
| 108 | remaining = repeats[i].remaining |
| 109 | } |
| 110 | continue |
| 111 | } |
| 112 | |
| 113 | state, err = state(s) |
| 114 | if err != nil { |
| 115 | return err |
| 116 | } |
| 117 | if state == nil { |
| 118 | return fmt.Errorf("bad state transition") |
| 119 | } |
| 120 | } |
| 121 | |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 122 | if state == nil { |
| 123 | return fmt.Errorf("no 'make' line") |
| 124 | } else { |
| 125 | state, err = state("") |
| 126 | if err != nil { |
| 127 | return err |
| 128 | } |
| 129 | } |
| 130 | |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 131 | _, err = os.Stdout.Write(out) |
| 132 | return err |
| 133 | } |
| 134 | |
| 135 | // ---- |
| 136 | |
| 137 | func appendU16LE(b []byte, u uint16) []byte { |
| 138 | return append(b, uint8(u), uint8(u>>8)) |
| 139 | } |
| 140 | |
| 141 | func log2(u uint32) (i int32) { |
| 142 | for i, pow := uint32(0), uint32(1); i < 32; i, pow = i+1, pow<<1 { |
| 143 | if u == pow { |
| 144 | return int32(i) |
| 145 | } |
| 146 | if u < pow { |
| 147 | break |
| 148 | } |
| 149 | } |
| 150 | return -1 |
| 151 | } |
| 152 | |
| 153 | func parseHex(s string) (num uint32, remaining string, ok bool) { |
| 154 | if i := strings.IndexByte(s, ' '); i >= 0 { |
| 155 | s, remaining = s[:i], s[i+1:] |
| 156 | for len(remaining) > 0 && remaining[0] == ' ' { |
| 157 | remaining = remaining[1:] |
| 158 | } |
| 159 | } |
| 160 | |
| 161 | if len(s) < 2 || s[0] != '0' || s[1] != 'x' { |
| 162 | return 0, "", false |
| 163 | } |
| 164 | s = s[2:] |
| 165 | |
| 166 | u, err := strconv.ParseUint(s, 16, 32) |
| 167 | if err != nil { |
| 168 | return 0, "", false |
| 169 | } |
| 170 | return uint32(u), remaining, true |
| 171 | } |
| 172 | |
| 173 | func parseNum(s string) (num uint32, remaining string, ok bool) { |
| 174 | if i := strings.IndexByte(s, ' '); i >= 0 { |
| 175 | s, remaining = s[:i], s[i+1:] |
| 176 | for len(remaining) > 0 && remaining[0] == ' ' { |
| 177 | remaining = remaining[1:] |
| 178 | } |
| 179 | } |
| 180 | |
| 181 | u, err := strconv.ParseUint(s, 10, 32) |
| 182 | if err != nil { |
| 183 | return 0, "", false |
| 184 | } |
| 185 | return uint32(u), remaining, true |
| 186 | } |
| 187 | |
| 188 | func reverse(x uint32, n uint32) uint32 { |
| 189 | x = uint32(reverse8[0xFF&(x>>0)])<<24 | |
| 190 | uint32(reverse8[0xFF&(x>>8)])<<16 | |
| 191 | uint32(reverse8[0xFF&(x>>16)])<<8 | |
| 192 | uint32(reverse8[0xFF&(x>>24)])<<0 |
| 193 | return x >> (32 - n) |
| 194 | } |
| 195 | |
| 196 | var reverse8 = [256]byte{ |
| 197 | 0x00, 0x80, 0x40, 0xC0, 0x20, 0xA0, 0x60, 0xE0, // 0x00 - 0x07 |
| 198 | 0x10, 0x90, 0x50, 0xD0, 0x30, 0xB0, 0x70, 0xF0, // 0x08 - 0x0F |
| 199 | 0x08, 0x88, 0x48, 0xC8, 0x28, 0xA8, 0x68, 0xE8, // 0x10 - 0x17 |
| 200 | 0x18, 0x98, 0x58, 0xD8, 0x38, 0xB8, 0x78, 0xF8, // 0x18 - 0x1F |
| 201 | 0x04, 0x84, 0x44, 0xC4, 0x24, 0xA4, 0x64, 0xE4, // 0x20 - 0x27 |
| 202 | 0x14, 0x94, 0x54, 0xD4, 0x34, 0xB4, 0x74, 0xF4, // 0x28 - 0x2F |
| 203 | 0x0C, 0x8C, 0x4C, 0xCC, 0x2C, 0xAC, 0x6C, 0xEC, // 0x30 - 0x37 |
| 204 | 0x1C, 0x9C, 0x5C, 0xDC, 0x3C, 0xBC, 0x7C, 0xFC, // 0x38 - 0x3F |
| 205 | 0x02, 0x82, 0x42, 0xC2, 0x22, 0xA2, 0x62, 0xE2, // 0x40 - 0x47 |
| 206 | 0x12, 0x92, 0x52, 0xD2, 0x32, 0xB2, 0x72, 0xF2, // 0x48 - 0x4F |
| 207 | 0x0A, 0x8A, 0x4A, 0xCA, 0x2A, 0xAA, 0x6A, 0xEA, // 0x50 - 0x57 |
| 208 | 0x1A, 0x9A, 0x5A, 0xDA, 0x3A, 0xBA, 0x7A, 0xFA, // 0x58 - 0x5F |
| 209 | 0x06, 0x86, 0x46, 0xC6, 0x26, 0xA6, 0x66, 0xE6, // 0x60 - 0x67 |
| 210 | 0x16, 0x96, 0x56, 0xD6, 0x36, 0xB6, 0x76, 0xF6, // 0x68 - 0x6F |
| 211 | 0x0E, 0x8E, 0x4E, 0xCE, 0x2E, 0xAE, 0x6E, 0xEE, // 0x70 - 0x77 |
| 212 | 0x1E, 0x9E, 0x5E, 0xDE, 0x3E, 0xBE, 0x7E, 0xFE, // 0x78 - 0x7F |
| 213 | 0x01, 0x81, 0x41, 0xC1, 0x21, 0xA1, 0x61, 0xE1, // 0x80 - 0x87 |
| 214 | 0x11, 0x91, 0x51, 0xD1, 0x31, 0xB1, 0x71, 0xF1, // 0x88 - 0x8F |
| 215 | 0x09, 0x89, 0x49, 0xC9, 0x29, 0xA9, 0x69, 0xE9, // 0x90 - 0x97 |
| 216 | 0x19, 0x99, 0x59, 0xD9, 0x39, 0xB9, 0x79, 0xF9, // 0x98 - 0x9F |
| 217 | 0x05, 0x85, 0x45, 0xC5, 0x25, 0xA5, 0x65, 0xE5, // 0xA0 - 0xA7 |
| 218 | 0x15, 0x95, 0x55, 0xD5, 0x35, 0xB5, 0x75, 0xF5, // 0xA8 - 0xAF |
| 219 | 0x0D, 0x8D, 0x4D, 0xCD, 0x2D, 0xAD, 0x6D, 0xED, // 0xB0 - 0xB7 |
| 220 | 0x1D, 0x9D, 0x5D, 0xDD, 0x3D, 0xBD, 0x7D, 0xFD, // 0xB8 - 0xBF |
| 221 | 0x03, 0x83, 0x43, 0xC3, 0x23, 0xA3, 0x63, 0xE3, // 0xC0 - 0xC7 |
| 222 | 0x13, 0x93, 0x53, 0xD3, 0x33, 0xB3, 0x73, 0xF3, // 0xC8 - 0xCF |
| 223 | 0x0B, 0x8B, 0x4B, 0xCB, 0x2B, 0xAB, 0x6B, 0xEB, // 0xD0 - 0xD7 |
| 224 | 0x1B, 0x9B, 0x5B, 0xDB, 0x3B, 0xBB, 0x7B, 0xFB, // 0xD8 - 0xDF |
| 225 | 0x07, 0x87, 0x47, 0xC7, 0x27, 0xA7, 0x67, 0xE7, // 0xE0 - 0xE7 |
| 226 | 0x17, 0x97, 0x57, 0xD7, 0x37, 0xB7, 0x77, 0xF7, // 0xE8 - 0xEF |
| 227 | 0x0F, 0x8F, 0x4F, 0xCF, 0x2F, 0xAF, 0x6F, 0xEF, // 0xF0 - 0xF7 |
| 228 | 0x1F, 0x9F, 0x5F, 0xDF, 0x3F, 0xBF, 0x7F, 0xFF, // 0xF8 - 0xFF |
| 229 | } |
| 230 | |
| 231 | // ---- |
| 232 | |
| 233 | func init() { |
| 234 | formats["deflate"] = stateDeflate |
| 235 | } |
| 236 | |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 237 | var deflateCodeOrder = [19]uint32{ |
| 238 | 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15, |
| 239 | } |
| 240 | |
| 241 | type deflateHuffmanTable map[uint32]string |
| 242 | |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 243 | var deflateGlobals struct { |
| 244 | bncData []byte |
| 245 | stream deflateBitStream |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 246 | |
| 247 | // Dynamic Huffman state. |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 248 | whichHuffman uint32 |
Nigel Tao | dc1218f | 2019-10-14 09:28:28 +1100 | [diff] [blame] | 249 | // 0=Unused, 1=CodeLength, 2=Literal/Length, 3=Distance. |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 250 | huffmans [4]deflateHuffmanTable |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 251 | |
| 252 | // DHH (Dynamic Huffman, inside a Huffman table) state. |
| 253 | prevLine string |
| 254 | etcetera bool |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 255 | } |
| 256 | |
| 257 | func deflateGlobalsClearDynamicHuffmanState() { |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 258 | deflateGlobals.whichHuffman = 0 |
| 259 | deflateGlobals.huffmans = [4]deflateHuffmanTable{} |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 260 | deflateGlobals.prevLine = "" |
| 261 | deflateGlobals.etcetera = false |
| 262 | } |
| 263 | |
| 264 | func deflateGlobalsCountCodes() (numLCodes uint32, numDCodes uint32, numCLCodeLengths uint32, retErr error) { |
| 265 | for k := range deflateGlobals.huffmans[2] { |
| 266 | if numLCodes < (k + 1) { |
| 267 | numLCodes = (k + 1) |
| 268 | } |
| 269 | } |
| 270 | |
| 271 | for k := range deflateGlobals.huffmans[3] { |
| 272 | if numDCodes < (k + 1) { |
| 273 | numDCodes = (k + 1) |
| 274 | } |
| 275 | } |
| 276 | |
| 277 | for k := range deflateGlobals.huffmans[1] { |
| 278 | if (k < 0) || (18 < k) { |
| 279 | return 0, 0, 0, fmt.Errorf("bad CodeLength: %d", k) |
| 280 | } |
| 281 | } |
| 282 | for i := len(deflateCodeOrder) - 1; i >= 0; i-- { |
| 283 | cl := deflateCodeOrder[i] |
| 284 | if _, ok := deflateGlobals.huffmans[1][cl]; ok { |
| 285 | numCLCodeLengths = uint32(i + 1) |
| 286 | break |
| 287 | } |
| 288 | } |
| 289 | if numCLCodeLengths < 4 { |
| 290 | numCLCodeLengths = 4 |
| 291 | } |
| 292 | |
| 293 | return numLCodes, numDCodes, numCLCodeLengths, nil |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 294 | } |
| 295 | |
| 296 | func deflateGlobalsWriteDynamicHuffmanTables() error { |
| 297 | g := &deflateGlobals |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 298 | numLCodes, numDCodes, numCLCodeLengths, err := deflateGlobalsCountCodes() |
| 299 | if err != nil { |
| 300 | return err |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 301 | } |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 302 | if (numLCodes < 257) || (257+31 < numLCodes) { |
| 303 | return fmt.Errorf("bad numLCodes: %d", numLCodes) |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 304 | } |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 305 | g.stream.writeBits(numLCodes-257, 5) |
| 306 | if (numDCodes < 1) || (1+31 < numDCodes) { |
| 307 | return fmt.Errorf("bad numDCodes: %d", numDCodes) |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 308 | } |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 309 | g.stream.writeBits(numDCodes-1, 5) |
| 310 | if (numCLCodeLengths < 4) || (4+15 < numCLCodeLengths) { |
| 311 | return fmt.Errorf("bad numCLCodeLengths: %d", numCLCodeLengths) |
| 312 | } |
| 313 | g.stream.writeBits(numCLCodeLengths-4, 4) |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 314 | |
| 315 | // Write the Huffman table for CodeLength. |
| 316 | { |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 317 | for i := uint32(0); i < numCLCodeLengths; i++ { |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 318 | n := len(g.huffmans[1][deflateCodeOrder[i]]) |
| 319 | g.stream.writeBits(uint32(n), 3) |
| 320 | } |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 321 | for i := numCLCodeLengths; i < uint32(len(deflateCodeOrder)); i++ { |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 322 | n := len(g.huffmans[1][deflateCodeOrder[i]]) |
| 323 | if n > 0 { |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 324 | return fmt.Errorf("short numCLCodeLengths: %d", numCLCodeLengths) |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 325 | } |
| 326 | } |
| 327 | } |
| 328 | |
Nigel Tao | dc1218f | 2019-10-14 09:28:28 +1100 | [diff] [blame] | 329 | // Write the Huffman tables for Literal/Length and Distance. |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 330 | { |
| 331 | numZeroes := uint32(0) |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 332 | for i := uint32(0); i < numLCodes+numDCodes; i++ { |
| 333 | codeLen := uint32(0) |
| 334 | if i < numLCodes { |
| 335 | codeLen = uint32(len(g.huffmans[2][i])) |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 336 | } else { |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 337 | codeLen = uint32(len(g.huffmans[3][i-numLCodes])) |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 338 | } |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 339 | if codeLen == 0 { |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 340 | numZeroes++ |
| 341 | continue |
| 342 | } |
| 343 | |
| 344 | if err := deflateGlobalsWriteDynamicHuffmanZeroes(numZeroes); err != nil { |
| 345 | return err |
| 346 | } |
| 347 | numZeroes = 0 |
| 348 | |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 349 | codeLenCode := g.huffmans[1][codeLen] |
| 350 | if codeLenCode == "" { |
| 351 | return fmt.Errorf("no code for code-length %d", codeLen) |
| 352 | } |
| 353 | deflateGlobalsWriteDynamicHuffmanBits(g.huffmans[1][codeLen]) |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 354 | } |
| 355 | if err := deflateGlobalsWriteDynamicHuffmanZeroes(numZeroes); err != nil { |
| 356 | return err |
| 357 | } |
| 358 | } |
| 359 | |
| 360 | return nil |
| 361 | } |
| 362 | |
| 363 | func deflateGlobalsWriteDynamicHuffmanZeroes(numZeroes uint32) error { |
| 364 | g := &deflateGlobals |
| 365 | if numZeroes == 0 { |
| 366 | return nil |
| 367 | } |
| 368 | |
| 369 | if s := g.huffmans[1][18]; s != "" { |
| 370 | for numZeroes >= 11 { |
| 371 | extra := numZeroes - 11 |
| 372 | if extra > 127 { |
| 373 | extra = 127 |
| 374 | } |
| 375 | deflateGlobalsWriteDynamicHuffmanBits(s) |
| 376 | g.stream.writeBits(extra, 7) |
| 377 | numZeroes -= 11 + extra |
| 378 | } |
| 379 | } |
| 380 | |
| 381 | if s := g.huffmans[1][17]; s != "" { |
| 382 | for numZeroes >= 3 { |
| 383 | extra := numZeroes - 3 |
| 384 | if extra > 7 { |
| 385 | extra = 7 |
| 386 | } |
| 387 | deflateGlobalsWriteDynamicHuffmanBits(s) |
| 388 | g.stream.writeBits(extra, 3) |
| 389 | numZeroes -= 3 + extra |
| 390 | } |
| 391 | } |
| 392 | |
| 393 | if s := g.huffmans[1][0]; s != "" { |
| 394 | for ; numZeroes > 0; numZeroes-- { |
| 395 | deflateGlobalsWriteDynamicHuffmanBits(s) |
| 396 | } |
| 397 | } |
| 398 | |
| 399 | if numZeroes > 0 { |
| 400 | return fmt.Errorf("could not write a run of zero-valued code lengths") |
| 401 | } |
| 402 | return nil |
| 403 | } |
| 404 | |
| 405 | func deflateGlobalsWriteDynamicHuffmanBits(s string) { |
| 406 | g := &deflateGlobals |
| 407 | for i := 0; i < len(s); i++ { |
| 408 | g.stream.writeBits(uint32(s[i]&1), 1) |
| 409 | } |
| 410 | } |
| 411 | |
| 412 | func parseDeflateWhichHuffman(s string) (num uint32, remaining string, ok bool) { |
| 413 | if i := strings.IndexByte(s, ' '); i >= 0 { |
| 414 | s, remaining = s[:i], s[i+1:] |
| 415 | for len(remaining) > 0 && remaining[0] == ' ' { |
| 416 | remaining = remaining[1:] |
| 417 | } |
| 418 | } |
| 419 | |
| 420 | switch s { |
| 421 | case "CodeLength": |
| 422 | return 1, remaining, true |
Nigel Tao | dc1218f | 2019-10-14 09:28:28 +1100 | [diff] [blame] | 423 | case "Literal/Length": |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 424 | return 2, remaining, true |
| 425 | case "Distance": |
| 426 | return 3, remaining, true |
| 427 | } |
| 428 | return 0, "", false |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 429 | } |
| 430 | |
| 431 | func stateDeflate(line string) (stateFunc, error) { |
| 432 | g := &deflateGlobals |
| 433 | const ( |
Nigel Tao | 0cc1798 | 2019-05-19 23:43:20 +1000 | [diff] [blame] | 434 | cmdB = "bytes " |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 435 | cmdBDH = "blockDynamicHuffman " |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 436 | cmdBFH = "blockFixedHuffman " |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 437 | cmdBNC = "blockNoCompression " |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 438 | ) |
| 439 | bits := uint32(0) |
| 440 | s := "" |
| 441 | |
| 442 | retState := stateFunc(nil) |
| 443 | switch { |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 444 | case line == "": |
| 445 | g.stream.flush() |
| 446 | return stateDeflate, nil |
| 447 | |
Nigel Tao | 0cc1798 | 2019-05-19 23:43:20 +1000 | [diff] [blame] | 448 | case strings.HasPrefix(line, cmdB): |
| 449 | s := line[len(cmdB):] |
| 450 | for s != "" { |
| 451 | x, ok := uint32(0), false |
| 452 | x, s, ok = parseHex(s) |
| 453 | if !ok { |
| 454 | return nil, fmt.Errorf("bad stateDeflate command: %q", line) |
| 455 | } |
| 456 | out = append(out, uint8(x)) |
| 457 | } |
| 458 | return stateDeflate, nil |
| 459 | |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 460 | case strings.HasPrefix(line, cmdBNC): |
| 461 | s = line[len(cmdBNC):] |
| 462 | retState = stateDeflateNoCompression |
| 463 | bits |= 0 << 1 |
| 464 | case strings.HasPrefix(line, cmdBFH): |
| 465 | s = line[len(cmdBFH):] |
| 466 | retState = stateDeflateFixedHuffman |
| 467 | bits |= 1 << 1 |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 468 | case strings.HasPrefix(line, cmdBDH): |
| 469 | s = line[len(cmdBDH):] |
| 470 | retState = stateDeflateDynamicHuffman |
| 471 | bits |= 2 << 1 |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 472 | default: |
| 473 | return nil, fmt.Errorf("bad stateDeflate command: %q", line) |
| 474 | } |
| 475 | |
| 476 | switch s { |
| 477 | case "(final) {": |
| 478 | bits |= 1 |
| 479 | case "(nonFinal) {": |
| 480 | // No-op. |
| 481 | default: |
| 482 | return nil, fmt.Errorf("bad stateDeflate command: %q", line) |
| 483 | } |
| 484 | |
| 485 | g.stream.writeBits(bits, 3) |
| 486 | return retState, nil |
| 487 | } |
| 488 | |
| 489 | func stateDeflateNoCompression(line string) (stateFunc, error) { |
| 490 | g := &deflateGlobals |
| 491 | if line == "}" { |
| 492 | if len(g.bncData) > 0xFFFF { |
| 493 | return nil, fmt.Errorf("bncData is too long") |
| 494 | } |
| 495 | n := uint32(len(g.bncData)) |
| 496 | g.stream.flush() |
| 497 | g.stream.writeBits(n, 16) |
| 498 | g.stream.writeBits(0xFFFF-n, 16) |
| 499 | g.stream.writeBytes(g.bncData) |
| 500 | g.bncData = g.bncData[:0] |
| 501 | return stateDeflate, nil |
| 502 | } |
| 503 | |
| 504 | if lit, ok := deflateParseLiteral(line); ok { |
| 505 | g.bncData = append(g.bncData, lit...) |
| 506 | return stateDeflateNoCompression, nil |
| 507 | } |
| 508 | |
| 509 | return nil, fmt.Errorf("bad blockNoCompression command: %q", line) |
| 510 | } |
| 511 | |
| 512 | func stateDeflateFixedHuffman(line string) (stateFunc, error) { |
| 513 | g := &deflateGlobals |
| 514 | if line == "}" { |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 515 | return stateDeflate, nil |
| 516 | } |
| 517 | |
| 518 | if line == "endOfBlock" { |
| 519 | g.stream.writeBits(0, 7) |
| 520 | return stateDeflateFixedHuffman, nil |
| 521 | } |
| 522 | |
| 523 | if lit, ok := deflateParseLiteral(line); ok { |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 524 | for i := 0; i < len(lit); i++ { |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 525 | g.stream.writeFixedHuffmanLCode(uint32(lit[i])) |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 526 | } |
| 527 | return stateDeflateFixedHuffman, nil |
| 528 | } |
| 529 | |
Nigel Tao | 11a6819 | 2019-02-02 13:04:09 +1100 | [diff] [blame] | 530 | if line == "len 3 distCode 31" { |
| 531 | lCode, lExtra, lNExtra := deflateEncodeLength(3) |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 532 | g.stream.writeFixedHuffmanLCode(lCode) |
Nigel Tao | 11a6819 | 2019-02-02 13:04:09 +1100 | [diff] [blame] | 533 | g.stream.writeBits(lExtra, lNExtra) |
| 534 | dCode, dExtra, dNExtra := uint32(31), uint32(0), uint32(0) |
| 535 | g.stream.writeBits(reverse(dCode, 5), 5) |
| 536 | g.stream.writeBits(dExtra, dNExtra) |
| 537 | return stateDeflateFixedHuffman, nil |
| 538 | } |
| 539 | |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 540 | if l, d, ok := deflateParseLenDist(line); ok { |
| 541 | lCode, lExtra, lNExtra := deflateEncodeLength(l) |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 542 | g.stream.writeFixedHuffmanLCode(lCode) |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 543 | g.stream.writeBits(lExtra, lNExtra) |
| 544 | dCode, dExtra, dNExtra := deflateEncodeDistance(d) |
| 545 | g.stream.writeBits(reverse(dCode, 5), 5) |
| 546 | g.stream.writeBits(dExtra, dNExtra) |
| 547 | return stateDeflateFixedHuffman, nil |
| 548 | } |
| 549 | |
| 550 | return nil, fmt.Errorf("bad stateDeflateFixedHuffman command: %q", line) |
| 551 | } |
| 552 | |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 553 | func stateDeflateDynamicHuffman(line string) (stateFunc, error) { |
| 554 | g := &deflateGlobals |
| 555 | const ( |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 556 | cmdH = "huffman " |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 557 | ) |
| 558 | switch { |
| 559 | case line == "}": |
| 560 | deflateGlobalsClearDynamicHuffmanState() |
| 561 | return stateDeflate, nil |
| 562 | |
| 563 | case strings.HasPrefix(line, cmdH): |
| 564 | s := line[len(cmdH):] |
| 565 | n, s, ok := parseDeflateWhichHuffman(s) |
| 566 | if !ok { |
| 567 | break |
| 568 | } |
| 569 | g.whichHuffman = n |
| 570 | return stateDeflateDynamicHuffmanHuffman, nil |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 571 | } |
| 572 | |
| 573 | if lit, ok := deflateParseLiteral(line); ok { |
| 574 | for i := 0; i < len(lit); i++ { |
| 575 | s := g.huffmans[2][uint32(lit[i])] |
| 576 | if s == "" { |
| 577 | return nil, fmt.Errorf("no code for literal %q (%d)", lit[i:i+1], lit[i]) |
| 578 | } |
| 579 | deflateGlobalsWriteDynamicHuffmanBits(s) |
| 580 | } |
| 581 | return stateDeflateDynamicHuffman, nil |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 582 | |
| 583 | } else if l, d, ok := deflateParseLenDist(line); ok { |
| 584 | if (l != 3) || (d != 2) { |
| 585 | return nil, fmt.Errorf("TODO: support len/dist pairs for dynamic Huffman blocks") |
| 586 | } |
| 587 | // len 3 is code 257, with no extra bits. |
| 588 | if s := g.huffmans[2][257]; s != "" { |
| 589 | deflateGlobalsWriteDynamicHuffmanBits(s) |
| 590 | } else { |
| 591 | return nil, fmt.Errorf("no code for literal/length symbol 257") |
| 592 | } |
| 593 | // dist 2 is code 1, with no extra bits. |
| 594 | if s := g.huffmans[3][1]; s != "" { |
| 595 | deflateGlobalsWriteDynamicHuffmanBits(s) |
| 596 | } else { |
| 597 | return nil, fmt.Errorf("no code for distance symbol 2") |
| 598 | } |
| 599 | return stateDeflateDynamicHuffman, nil |
| 600 | |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 601 | } else if line == "endOfBlock" { |
| 602 | s := g.huffmans[2][256] |
| 603 | if s == "" { |
| 604 | return nil, fmt.Errorf("no code for end-of-block (256)") |
| 605 | } |
| 606 | deflateGlobalsWriteDynamicHuffmanBits(s) |
| 607 | return stateDeflateDynamicHuffman, nil |
| 608 | } |
| 609 | |
| 610 | return nil, fmt.Errorf("bad stateDeflateDynamicHuffman command: %q", line) |
| 611 | } |
| 612 | |
| 613 | func stateDeflateDynamicHuffmanHuffman(line string) (stateFunc, error) { |
| 614 | g := &deflateGlobals |
| 615 | outer: |
| 616 | switch { |
| 617 | case line == "}": |
| 618 | g.whichHuffman = 0 |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 619 | g.prevLine = "" |
| 620 | g.etcetera = false |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 621 | |
| 622 | // If we have all three Huffman tables, write them. |
| 623 | for i := 1; ; i++ { |
| 624 | if i == 4 { |
| 625 | if err := deflateGlobalsWriteDynamicHuffmanTables(); err != nil { |
| 626 | return nil, err |
| 627 | } |
| 628 | break |
| 629 | } |
| 630 | if g.huffmans[i] == nil { |
| 631 | break |
| 632 | } |
| 633 | } |
| 634 | |
| 635 | return stateDeflateDynamicHuffman, nil |
| 636 | |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 637 | case line == "etcetera": |
| 638 | g.etcetera = true |
| 639 | return stateDeflateDynamicHuffmanHuffman, nil |
| 640 | |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 641 | default: |
| 642 | s := line |
| 643 | n, s, ok := parseNum(s) |
| 644 | if !ok || s == "" { |
| 645 | break |
| 646 | } |
| 647 | for i := 0; i < len(s); i++ { |
| 648 | if c := s[i]; c != '0' && c != '1' { |
| 649 | break outer |
| 650 | } |
| 651 | } |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 652 | if (len(s) < 1) || (15 < len(s)) { |
| 653 | return nil, fmt.Errorf("%q code length, %d, is out of range", s, len(s)) |
| 654 | } |
| 655 | |
| 656 | if g.etcetera { |
| 657 | g.etcetera = false |
| 658 | n0, s0, ok := parseNum(g.prevLine) |
| 659 | if !ok { |
| 660 | return nil, fmt.Errorf("bad etcetera command") |
| 661 | } |
| 662 | if err := stateDeflateDHHEtcetera(n0, s0, n, s); err != nil { |
| 663 | return nil, err |
| 664 | } |
| 665 | } |
| 666 | |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 667 | if g.huffmans[g.whichHuffman] == nil { |
| 668 | g.huffmans[g.whichHuffman] = deflateHuffmanTable{} |
| 669 | } |
| 670 | g.huffmans[g.whichHuffman][n] = s |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 671 | g.prevLine = line |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 672 | return stateDeflateDynamicHuffmanHuffman, nil |
| 673 | } |
| 674 | |
| 675 | return nil, fmt.Errorf("bad stateDeflateDynamicHuffmanHuffman command: %q", line) |
| 676 | } |
| 677 | |
Nigel Tao | ff7bffc | 2019-10-13 23:23:48 +1100 | [diff] [blame] | 678 | // stateDeflateDHHEtcetera expands the "etcetera" line in: |
| 679 | // |
| 680 | // 0 0000 |
| 681 | // 1 0001 |
| 682 | // etcetera |
| 683 | // 7 0111 |
| 684 | // |
| 685 | // to produce the implicit lines: |
| 686 | // |
| 687 | // 0 0000 |
| 688 | // 1 0001 |
| 689 | // 2 0010 |
| 690 | // 3 0011 |
| 691 | // 4 0100 |
| 692 | // 5 0101 |
| 693 | // 6 0110 |
| 694 | // 7 0111 |
| 695 | func stateDeflateDHHEtcetera(n0 uint32, s0 string, n1 uint32, s1 string) error { |
| 696 | b := []byte(s0) |
| 697 | if !incrementBitstring(b) { |
| 698 | return fmt.Errorf("etcetera: could not increment bitstring") |
| 699 | } |
| 700 | for n := n0 + 1; n < n1; n++ { |
| 701 | line := fmt.Sprintf("%d %s", n, b) |
| 702 | if _, err := stateDeflateDynamicHuffmanHuffman(line); err != nil { |
| 703 | return err |
| 704 | } |
| 705 | if !incrementBitstring(b) { |
| 706 | return fmt.Errorf("etcetera: could not increment bitstring") |
| 707 | } |
| 708 | } |
| 709 | if string(b) != s1 { |
| 710 | return fmt.Errorf("etcetera: final bitstring: got %q, want %q", b, s1) |
| 711 | } |
| 712 | return nil |
| 713 | } |
| 714 | |
| 715 | func incrementBitstring(b []byte) (ok bool) { |
| 716 | for i := len(b) - 1; i >= 0; i-- { |
| 717 | switch b[i] { |
| 718 | case '0': |
| 719 | b[i] = '1' |
| 720 | return true |
| 721 | case '1': |
| 722 | b[i] = '0' |
| 723 | default: |
| 724 | return false |
| 725 | } |
| 726 | } |
| 727 | return false |
| 728 | } |
| 729 | |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 730 | type deflateBitStream struct { |
| 731 | bits uint32 |
| 732 | nBits uint32 // Always within [0, 7]. |
| 733 | } |
| 734 | |
| 735 | // writeBits writes the low n bits of b to z. |
| 736 | func (z *deflateBitStream) writeBits(b uint32, n uint32) { |
| 737 | if n > 24 { |
| 738 | panic("writeBits: n is too large") |
| 739 | } |
| 740 | z.bits |= b << z.nBits |
| 741 | z.nBits += n |
| 742 | for z.nBits >= 8 { |
| 743 | out = append(out, uint8(z.bits)) |
| 744 | z.bits >>= 8 |
| 745 | z.nBits -= 8 |
| 746 | } |
| 747 | } |
| 748 | |
| 749 | func (z *deflateBitStream) writeBytes(b []byte) { |
| 750 | z.flush() |
| 751 | out = append(out, b...) |
| 752 | } |
| 753 | |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 754 | func (z *deflateBitStream) writeFixedHuffmanLCode(lCode uint32) { |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 755 | switch { |
| 756 | case lCode < 144: // 0b._0011_0000 through 0b._1011_1111 |
| 757 | lCode += 0x030 |
| 758 | z.writeBits(reverse(lCode, 8), 8) |
| 759 | case lCode < 256: // 0b1_1001_0000 through 0b1_1111_1111 |
| 760 | lCode += 0x190 - 144 |
| 761 | z.writeBits(reverse(lCode, 9), 9) |
| 762 | case lCode < 280: // 0b._.000_0000 through 0b._.001_0111 |
| 763 | lCode -= 256 - 0x000 |
| 764 | z.writeBits(reverse(lCode, 7), 7) |
| 765 | default: // 0b._1100_0000 through 0b._1100_0111 |
| 766 | lCode -= 280 - 0x0C0 |
| 767 | z.writeBits(reverse(lCode, 8), 8) |
| 768 | } |
| 769 | } |
| 770 | |
| 771 | func (z *deflateBitStream) flush() { |
| 772 | if z.nBits > 0 { |
| 773 | out = append(out, uint8(z.bits)) |
| 774 | z.bits = 0 |
| 775 | z.nBits = 0 |
| 776 | } |
| 777 | } |
| 778 | |
| 779 | func deflateEncodeLength(l uint32) (code uint32, extra uint32, nExtra uint32) { |
| 780 | switch { |
| 781 | case l < 3: |
| 782 | // No-op. |
| 783 | case l < 11: |
| 784 | l -= 3 |
| 785 | return (l >> 0) + 257, l & 0x0000, 0 |
| 786 | case l < 19: |
| 787 | l -= 11 |
| 788 | return (l >> 1) + 265, l & 0x0001, 1 |
| 789 | case l < 35: |
| 790 | l -= 19 |
| 791 | return (l >> 2) + 269, l & 0x0003, 2 |
| 792 | case l < 67: |
| 793 | l -= 35 |
| 794 | return (l >> 3) + 273, l & 0x0007, 3 |
| 795 | case l < 131: |
| 796 | l -= 67 |
| 797 | return (l >> 4) + 277, l & 0x000F, 4 |
| 798 | case l < 258: |
| 799 | l -= 131 |
| 800 | return (l >> 5) + 281, l & 0x001F, 5 |
| 801 | case l == 258: |
| 802 | return 285, 0, 0 |
| 803 | } |
| 804 | panic(fmt.Sprintf("deflateEncodeLength: l=%d", l)) |
| 805 | } |
| 806 | |
| 807 | func deflateEncodeDistance(d uint32) (code uint32, extra uint32, nExtra uint32) { |
| 808 | switch { |
| 809 | case d < 1: |
| 810 | // No-op. |
| 811 | case d < 5: |
| 812 | d -= 1 |
| 813 | return (d >> 0) + 0, d & 0x0000, 0 |
| 814 | case d < 9: |
| 815 | d -= 5 |
| 816 | return (d >> 1) + 4, d & 0x0001, 1 |
| 817 | case d < 17: |
| 818 | d -= 9 |
| 819 | return (d >> 2) + 6, d & 0x0003, 2 |
| 820 | case d < 33: |
| 821 | d -= 17 |
| 822 | return (d >> 3) + 8, d & 0x0007, 3 |
| 823 | case d < 65: |
| 824 | d -= 33 |
| 825 | return (d >> 4) + 10, d & 0x000F, 4 |
| 826 | case d < 129: |
| 827 | d -= 65 |
| 828 | return (d >> 5) + 12, d & 0x001F, 5 |
| 829 | case d < 257: |
| 830 | d -= 129 |
| 831 | return (d >> 6) + 14, d & 0x003F, 6 |
| 832 | case d < 513: |
| 833 | d -= 257 |
| 834 | return (d >> 7) + 16, d & 0x007F, 7 |
| 835 | case d < 1025: |
| 836 | d -= 513 |
| 837 | return (d >> 8) + 18, d & 0x00FF, 8 |
| 838 | case d < 2049: |
| 839 | d -= 1025 |
| 840 | return (d >> 9) + 20, d & 0x01FF, 9 |
| 841 | case d < 4097: |
| 842 | d -= 2049 |
| 843 | return (d >> 10) + 22, d & 0x03FF, 10 |
| 844 | case d < 8193: |
| 845 | d -= 4097 |
| 846 | return (d >> 11) + 24, d & 0x07FF, 11 |
| 847 | case d < 16385: |
| 848 | d -= 8193 |
| 849 | return (d >> 12) + 26, d & 0x0FFF, 12 |
| 850 | case d < 32769: |
| 851 | d -= 16385 |
| 852 | return (d >> 13) + 28, d & 0x1FFF, 13 |
| 853 | } |
| 854 | panic(fmt.Sprintf("deflateEncodeDistance: d=%d", d)) |
| 855 | } |
| 856 | |
| 857 | func deflateParseLiteral(s string) (lit string, ok bool) { |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 858 | // TODO: support "\xAB" escape codes in the script? |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 859 | const ( |
| 860 | prefix = `literal "` |
| 861 | suffix = `"` |
| 862 | ) |
| 863 | if strings.HasPrefix(s, prefix) { |
| 864 | s = s[len(prefix):] |
| 865 | if strings.HasSuffix(s, suffix) { |
| 866 | return s[:len(s)-len(suffix)], true |
| 867 | } |
| 868 | } |
| 869 | return "", false |
| 870 | } |
| 871 | |
| 872 | func deflateParseLenDist(line string) (l uint32, d uint32, ok bool) { |
| 873 | const ( |
| 874 | lStr = "len " |
| 875 | dStr = "dist " |
| 876 | ) |
| 877 | s := line |
| 878 | if strings.HasPrefix(s, lStr) { |
| 879 | s = s[len(lStr):] |
| 880 | if l, s, ok := parseNum(s); ok && 3 <= l && l <= 258 { |
| 881 | if strings.HasPrefix(s, dStr) { |
| 882 | s = s[len(dStr):] |
| 883 | if d, s, ok := parseNum(s); ok && 1 <= d && d <= 32768 && s == "" { |
| 884 | return l, d, true |
| 885 | } |
| 886 | } |
| 887 | } |
| 888 | } |
| 889 | return 0, 0, false |
| 890 | } |
| 891 | |
| 892 | // ---- |
| 893 | |
| 894 | func init() { |
| 895 | formats["gif"] = stateGif |
| 896 | } |
| 897 | |
| 898 | var gifGlobals struct { |
Nigel Tao | 31a2bc9 | 2019-05-18 17:32:24 +1000 | [diff] [blame] | 899 | imageWidth uint32 |
| 900 | imageHeight uint32 |
| 901 | imageBackgroundColorIndex uint32 |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 902 | |
Nigel Tao | efb1803 | 2019-10-05 11:40:24 +1000 | [diff] [blame] | 903 | frameInterlaced bool |
| 904 | frameLeft uint32 |
| 905 | frameTop uint32 |
| 906 | frameWidth uint32 |
| 907 | frameHeight uint32 |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 908 | |
| 909 | globalPalette [][4]uint8 |
Nigel Tao | 746681e | 2019-05-04 13:49:22 +1000 | [diff] [blame] | 910 | localPalette [][4]uint8 |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 911 | } |
| 912 | |
| 913 | func stateGif(line string) (stateFunc, error) { |
| 914 | const ( |
Nigel Tao | 6700c72 | 2019-05-26 09:41:55 +1000 | [diff] [blame] | 915 | cmdB = "bytes " |
| 916 | cmdGC = "graphicControl " |
| 917 | cmdL = "lzw " |
| 918 | cmdLC = "loopCount " |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 919 | ) |
| 920 | outer: |
| 921 | switch { |
Nigel Tao | 1234760 | 2019-06-16 23:15:05 +1000 | [diff] [blame] | 922 | case line == "": |
| 923 | return stateGif, nil |
| 924 | |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 925 | case line == "frame {": |
Nigel Tao | efb1803 | 2019-10-05 11:40:24 +1000 | [diff] [blame] | 926 | gifGlobals.frameInterlaced = false |
| 927 | gifGlobals.frameLeft = 0 |
| 928 | gifGlobals.frameTop = 0 |
| 929 | gifGlobals.frameWidth = 0 |
| 930 | gifGlobals.frameHeight = 0 |
Nigel Tao | 746681e | 2019-05-04 13:49:22 +1000 | [diff] [blame] | 931 | gifGlobals.localPalette = nil |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 932 | out = append(out, 0x2C) |
| 933 | return stateGifFrame, nil |
| 934 | |
| 935 | case line == "header": |
| 936 | out = append(out, "GIF89a"...) |
| 937 | return stateGif, nil |
| 938 | |
| 939 | case line == "image {": |
| 940 | return stateGifImage, nil |
| 941 | |
| 942 | case line == "trailer": |
| 943 | out = append(out, 0x3B) |
| 944 | return stateGif, nil |
| 945 | |
Nigel Tao | 21f2890 | 2019-04-20 16:33:11 +1000 | [diff] [blame] | 946 | case strings.HasPrefix(line, cmdB): |
| 947 | s := line[len(cmdB):] |
| 948 | for s != "" { |
| 949 | x, ok := uint32(0), false |
| 950 | x, s, ok = parseHex(s) |
| 951 | if !ok { |
| 952 | break outer |
| 953 | } |
| 954 | out = append(out, uint8(x)) |
| 955 | } |
| 956 | return stateGif, nil |
| 957 | |
Nigel Tao | 6700c72 | 2019-05-26 09:41:55 +1000 | [diff] [blame] | 958 | case strings.HasPrefix(line, cmdGC): |
| 959 | s := line[len(cmdGC):] |
| 960 | |
| 961 | flags := uint8(0) |
Nigel Tao | 344123b | 2020-09-25 23:00:01 +1000 | [diff] [blame] | 962 | duration := uint32(0) |
Nigel Tao | 6c2fb9a | 2020-09-27 00:09:25 +1000 | [diff] [blame] | 963 | transparentIndex := uint8(0) |
Nigel Tao | 344123b | 2020-09-25 23:00:01 +1000 | [diff] [blame] | 964 | for s != "" { |
| 965 | term := "" |
| 966 | if i := strings.IndexByte(s, ' '); i >= 0 { |
| 967 | term, s = s[:i], s[i+1:] |
| 968 | } else { |
| 969 | term, s = s, "" |
| 970 | } |
| 971 | |
| 972 | const ( |
| 973 | ms = "ms" |
| 974 | trans = "transparentIndex=" |
| 975 | ) |
| 976 | switch { |
| 977 | case term == "animationDisposalNone": |
Nigel Tao | 6700c72 | 2019-05-26 09:41:55 +1000 | [diff] [blame] | 978 | flags |= 0x00 |
Nigel Tao | 344123b | 2020-09-25 23:00:01 +1000 | [diff] [blame] | 979 | case term == "animationDisposalRestoreBackground": |
Nigel Tao | 6700c72 | 2019-05-26 09:41:55 +1000 | [diff] [blame] | 980 | flags |= 0x08 |
Nigel Tao | 344123b | 2020-09-25 23:00:01 +1000 | [diff] [blame] | 981 | case term == "animationDisposalRestorePrevious": |
Nigel Tao | 6700c72 | 2019-05-26 09:41:55 +1000 | [diff] [blame] | 982 | flags |= 0x0C |
Nigel Tao | 344123b | 2020-09-25 23:00:01 +1000 | [diff] [blame] | 983 | case strings.HasPrefix(term, trans): |
Nigel Tao | 6c2fb9a | 2020-09-27 00:09:25 +1000 | [diff] [blame] | 984 | num, err := strconv.ParseUint(term[len(trans):], 0, 8) |
| 985 | if err != nil { |
Nigel Tao | 344123b | 2020-09-25 23:00:01 +1000 | [diff] [blame] | 986 | break outer |
| 987 | } |
| 988 | flags |= 0x01 |
Nigel Tao | 6c2fb9a | 2020-09-27 00:09:25 +1000 | [diff] [blame] | 989 | transparentIndex = uint8(num) |
Nigel Tao | 344123b | 2020-09-25 23:00:01 +1000 | [diff] [blame] | 990 | case strings.HasSuffix(term, ms): |
| 991 | num, remaining, ok := parseNum(term[:len(term)-len(ms)]) |
| 992 | if !ok || remaining != "" { |
| 993 | break outer |
| 994 | } |
| 995 | duration = num / 10 // GIF's unit of time is 10ms. |
Nigel Tao | 6700c72 | 2019-05-26 09:41:55 +1000 | [diff] [blame] | 996 | default: |
| 997 | break outer |
| 998 | } |
Nigel Tao | 6700c72 | 2019-05-26 09:41:55 +1000 | [diff] [blame] | 999 | } |
| 1000 | |
Nigel Tao | b1ff27f | 2019-05-12 22:00:44 +1000 | [diff] [blame] | 1001 | out = append(out, |
Nigel Tao | 6700c72 | 2019-05-26 09:41:55 +1000 | [diff] [blame] | 1002 | 0x21, 0xF9, 0x04, |
| 1003 | flags, |
Nigel Tao | b1ff27f | 2019-05-12 22:00:44 +1000 | [diff] [blame] | 1004 | uint8(duration>>0), |
| 1005 | uint8(duration>>8), |
Nigel Tao | 6c2fb9a | 2020-09-27 00:09:25 +1000 | [diff] [blame] | 1006 | transparentIndex, |
Nigel Tao | 6700c72 | 2019-05-26 09:41:55 +1000 | [diff] [blame] | 1007 | 0x00, |
Nigel Tao | b1ff27f | 2019-05-12 22:00:44 +1000 | [diff] [blame] | 1008 | ) |
| 1009 | return stateGif, nil |
| 1010 | |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 1011 | case strings.HasPrefix(line, cmdL): |
| 1012 | s := line[len(cmdL):] |
| 1013 | litWidth, s, ok := parseNum(s) |
| 1014 | if !ok || litWidth < 2 || 8 < litWidth { |
| 1015 | break |
| 1016 | } |
| 1017 | out = append(out, uint8(litWidth)) |
| 1018 | |
| 1019 | uncompressed := []byte(nil) |
| 1020 | for s != "" { |
| 1021 | x := uint32(0) |
| 1022 | x, s, ok = parseHex(s) |
| 1023 | if !ok { |
| 1024 | break outer |
| 1025 | } |
| 1026 | uncompressed = append(uncompressed, uint8(x)) |
| 1027 | } |
| 1028 | |
| 1029 | buf := bytes.NewBuffer(nil) |
| 1030 | w := lzw.NewWriter(buf, lzw.LSB, int(litWidth)) |
| 1031 | if _, err := w.Write(uncompressed); err != nil { |
| 1032 | return nil, err |
| 1033 | } |
| 1034 | if err := w.Close(); err != nil { |
| 1035 | return nil, err |
| 1036 | } |
| 1037 | compressed := buf.Bytes() |
| 1038 | |
| 1039 | for len(compressed) > 0 { |
| 1040 | if len(compressed) <= 0xFF { |
| 1041 | out = append(out, uint8(len(compressed))) |
| 1042 | out = append(out, compressed...) |
| 1043 | compressed = nil |
| 1044 | } else { |
| 1045 | out = append(out, 0xFF) |
| 1046 | out = append(out, compressed[:0xFF]...) |
| 1047 | compressed = compressed[0xFF:] |
| 1048 | } |
| 1049 | } |
| 1050 | out = append(out, 0x00) |
| 1051 | return stateGif, nil |
Nigel Tao | 7be9c39 | 2018-10-13 17:00:45 +1100 | [diff] [blame] | 1052 | |
| 1053 | case strings.HasPrefix(line, cmdLC): |
| 1054 | s := line[len(cmdLC):] |
| 1055 | loopCount, _, ok := parseNum(s) |
| 1056 | if !ok || 0xFFFF < loopCount { |
| 1057 | break |
| 1058 | } |
| 1059 | out = append(out, |
| 1060 | 0x21, // Extension Introducer. |
| 1061 | 0xFF, // Application Extension Label. |
| 1062 | 0x0B, // Block Size. |
| 1063 | ) |
| 1064 | out = append(out, "NETSCAPE2.0"...) |
| 1065 | out = append(out, |
| 1066 | 0x03, // Block Size. |
| 1067 | 0x01, // Magic Number. |
| 1068 | byte(loopCount), |
| 1069 | byte(loopCount>>8), |
| 1070 | 0x00, // Block Terminator. |
| 1071 | ) |
| 1072 | return stateGif, nil |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 1073 | } |
| 1074 | |
| 1075 | return nil, fmt.Errorf("bad stateGif command: %q", line) |
| 1076 | } |
| 1077 | |
| 1078 | func stateGifImage(line string) (stateFunc, error) { |
| 1079 | g := &gifGlobals |
| 1080 | if line == "}" { |
| 1081 | out = appendU16LE(out, uint16(g.imageWidth)) |
| 1082 | out = appendU16LE(out, uint16(g.imageHeight)) |
| 1083 | if g.globalPalette == nil { |
| 1084 | out = append(out, 0x00) |
| 1085 | } else if n := log2(uint32(len(g.globalPalette))); n < 2 || 8 < n { |
| 1086 | return nil, fmt.Errorf("bad len(g.globalPalette): %d", len(g.globalPalette)) |
| 1087 | } else { |
| 1088 | out = append(out, 0x80|uint8(n-1)) |
| 1089 | } |
Nigel Tao | 31a2bc9 | 2019-05-18 17:32:24 +1000 | [diff] [blame] | 1090 | out = append(out, uint8(g.imageBackgroundColorIndex)) |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 1091 | out = append(out, 0x00) |
| 1092 | for _, x := range g.globalPalette { |
| 1093 | out = append(out, x[0], x[1], x[2]) |
| 1094 | } |
| 1095 | return stateGif, nil |
| 1096 | } |
| 1097 | |
| 1098 | const ( |
Nigel Tao | 31a2bc9 | 2019-05-18 17:32:24 +1000 | [diff] [blame] | 1099 | cmdBCI = "backgroundColorIndex " |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 1100 | cmdIWH = "imageWidthHeight " |
| 1101 | cmdP = "palette {" |
| 1102 | ) |
| 1103 | switch { |
Nigel Tao | 31a2bc9 | 2019-05-18 17:32:24 +1000 | [diff] [blame] | 1104 | case strings.HasPrefix(line, cmdBCI): |
| 1105 | s := line[len(cmdBCI):] |
| 1106 | if i, _, ok := parseNum(s); ok { |
| 1107 | g.imageBackgroundColorIndex = i |
| 1108 | } |
| 1109 | return stateGifImage, nil |
| 1110 | |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 1111 | case strings.HasPrefix(line, cmdIWH): |
| 1112 | s := line[len(cmdIWH):] |
| 1113 | if w, s, ok := parseNum(s); ok { |
| 1114 | if h, _, ok := parseNum(s); ok { |
| 1115 | g.imageWidth = w |
| 1116 | g.imageHeight = h |
| 1117 | return stateGifImage, nil |
| 1118 | } |
| 1119 | } |
| 1120 | |
| 1121 | case strings.HasPrefix(line, cmdP): |
| 1122 | return stateGifImagePalette, nil |
| 1123 | } |
| 1124 | |
| 1125 | return nil, fmt.Errorf("bad stateGifImage command: %q", line) |
| 1126 | } |
| 1127 | |
| 1128 | func stateGifFrame(line string) (stateFunc, error) { |
| 1129 | g := &gifGlobals |
| 1130 | if line == "}" { |
| 1131 | out = appendU16LE(out, uint16(g.frameLeft)) |
| 1132 | out = appendU16LE(out, uint16(g.frameTop)) |
| 1133 | out = appendU16LE(out, uint16(g.frameWidth)) |
| 1134 | out = appendU16LE(out, uint16(g.frameHeight)) |
Nigel Tao | efb1803 | 2019-10-05 11:40:24 +1000 | [diff] [blame] | 1135 | flags := byte(0x00) |
| 1136 | if g.frameInterlaced { |
| 1137 | flags |= 0x40 |
| 1138 | } |
Nigel Tao | 746681e | 2019-05-04 13:49:22 +1000 | [diff] [blame] | 1139 | if g.localPalette == nil { |
Nigel Tao | efb1803 | 2019-10-05 11:40:24 +1000 | [diff] [blame] | 1140 | out = append(out, flags) |
Nigel Tao | 746681e | 2019-05-04 13:49:22 +1000 | [diff] [blame] | 1141 | } else if n := log2(uint32(len(g.localPalette))); n < 2 || 8 < n { |
| 1142 | return nil, fmt.Errorf("bad len(g.localPalette): %d", len(g.localPalette)) |
| 1143 | } else { |
Nigel Tao | efb1803 | 2019-10-05 11:40:24 +1000 | [diff] [blame] | 1144 | out = append(out, flags|0x80|uint8(n-1)) |
Nigel Tao | 746681e | 2019-05-04 13:49:22 +1000 | [diff] [blame] | 1145 | } |
| 1146 | for _, x := range g.localPalette { |
| 1147 | out = append(out, x[0], x[1], x[2]) |
| 1148 | } |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 1149 | return stateGif, nil |
| 1150 | } |
| 1151 | |
| 1152 | const ( |
| 1153 | cmdFLTWH = "frameLeftTopWidthHeight " |
Nigel Tao | 746681e | 2019-05-04 13:49:22 +1000 | [diff] [blame] | 1154 | cmdP = "palette {" |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 1155 | ) |
| 1156 | switch { |
Nigel Tao | efb1803 | 2019-10-05 11:40:24 +1000 | [diff] [blame] | 1157 | case line == "interlaced": |
| 1158 | g.frameInterlaced = true |
| 1159 | return stateGifFrame, nil |
| 1160 | |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 1161 | case strings.HasPrefix(line, cmdFLTWH): |
| 1162 | s := line[len(cmdFLTWH):] |
| 1163 | if l, s, ok := parseNum(s); ok { |
| 1164 | if t, s, ok := parseNum(s); ok { |
| 1165 | if w, s, ok := parseNum(s); ok { |
| 1166 | if h, _, ok := parseNum(s); ok { |
| 1167 | g.frameLeft = l |
| 1168 | g.frameTop = t |
| 1169 | g.frameWidth = w |
| 1170 | g.frameHeight = h |
| 1171 | return stateGifFrame, nil |
| 1172 | } |
| 1173 | } |
| 1174 | } |
| 1175 | } |
Nigel Tao | 746681e | 2019-05-04 13:49:22 +1000 | [diff] [blame] | 1176 | |
| 1177 | case strings.HasPrefix(line, cmdP): |
| 1178 | return stateGifFramePalette, nil |
Nigel Tao | 981bf19 | 2018-05-23 12:17:55 +1000 | [diff] [blame] | 1179 | } |
| 1180 | |
| 1181 | return nil, fmt.Errorf("bad stateGifFrame command: %q", line) |
| 1182 | } |
| 1183 | |
| 1184 | func stateGifImagePalette(line string) (stateFunc, error) { |
| 1185 | g := &gifGlobals |
| 1186 | if line == "}" { |
| 1187 | return stateGifImage, nil |
| 1188 | } |
| 1189 | |
| 1190 | s := line |
| 1191 | if rgb0, s, ok := parseHex(s); ok { |
| 1192 | if rgb1, s, ok := parseHex(s); ok { |
| 1193 | if rgb2, _, ok := parseHex(s); ok { |
| 1194 | g.globalPalette = append(g.globalPalette, |
| 1195 | [4]uint8{uint8(rgb0), uint8(rgb1), uint8(rgb2), 0xFF}) |
| 1196 | return stateGifImagePalette, nil |
| 1197 | } |
| 1198 | } |
| 1199 | } |
| 1200 | |
| 1201 | return nil, fmt.Errorf("bad stateGifImagePalette command: %q", line) |
| 1202 | } |
Nigel Tao | 746681e | 2019-05-04 13:49:22 +1000 | [diff] [blame] | 1203 | |
| 1204 | func stateGifFramePalette(line string) (stateFunc, error) { |
| 1205 | g := &gifGlobals |
| 1206 | if line == "}" { |
| 1207 | return stateGifFrame, nil |
| 1208 | } |
| 1209 | |
| 1210 | s := line |
| 1211 | if rgb0, s, ok := parseHex(s); ok { |
| 1212 | if rgb1, s, ok := parseHex(s); ok { |
| 1213 | if rgb2, _, ok := parseHex(s); ok { |
| 1214 | g.localPalette = append(g.localPalette, |
| 1215 | [4]uint8{uint8(rgb0), uint8(rgb1), uint8(rgb2), 0xFF}) |
| 1216 | return stateGifFramePalette, nil |
| 1217 | } |
| 1218 | } |
| 1219 | } |
| 1220 | |
| 1221 | return nil, fmt.Errorf("bad stateGifFramePalette command: %q", line) |
| 1222 | } |