1 2 3 4 5 6 7 Network Working Group P. Deutsch 8 Request for Comments: 1950 Aladdin Enterprises 9 Category: Informational J-L. Gailly 10 Info-ZIP 11 May 1996 12 13 14 ZLIB Compressed Data Format Specification version 3.3 15 16 Status of This Memo 17 18 This memo provides information for the Internet community. This memo 19 does not specify an Internet standard of any kind. Distribution of 20 this memo is unlimited. 21 22 IESG Note: 23 24 The IESG takes no position on the validity of any Intellectual 25 Property Rights statements contained in this document. 26 27 Notices 28 29 Copyright (c) 1996 L. Peter Deutsch and Jean-Loup Gailly 30 31 Permission is granted to copy and distribute this document for any 32 purpose and without charge, including translations into other 33 languages and incorporation into compilations, provided that the 34 copyright notice and this notice are preserved, and that any 35 substantive changes or deletions from the original are clearly 36 marked. 37 38 A pointer to the latest version of this and related documentation in 39 HTML format can be found at the URL 40 <ftp://ftp.uu.net/graphics/png/documents/zlib/zdoc-index.html>. 41 42 Abstract 43 44 This specification defines a lossless compressed data format. The 45 data can be produced or consumed, even for an arbitrarily long 46 sequentially presented input data stream, using only an a priori 47 bounded amount of intermediate storage. The format presently uses 48 the DEFLATE compression method but can be easily extended to use 49 other compression methods. It can be implemented readily in a manner 50 not covered by patents. This specification also defines the ADLER-32 51 checksum (an extension and improvement of the Fletcher checksum), 52 used for detection of data corruption, and provides an algorithm for 53 computing it. 54 55 56 57 58 Deutsch & Gailly Informational [Page 1] 59 61 RFC 1950 ZLIB Compressed Data Format Specification May 1996 62 63 64 Table of Contents 65 66 1. Introduction ................................................... 2 67 1.1. Purpose ................................................... 2 68 1.2. Intended audience ......................................... 3 69 1.3. Scope ..................................................... 3 70 1.4. Compliance ................................................ 3 71 1.5. Definitions of terms and conventions used ................ 3 72 1.6. Changes from previous versions ............................ 3 73 2. Detailed specification ......................................... 3 74 2.1. Overall conventions ....................................... 3 75 2.2. Data format ............................................... 4 76 2.3. Compliance ................................................ 7 77 3. References ..................................................... 7 78 4. Source code .................................................... 8 79 5. Security Considerations ........................................ 8 80 6. Acknowledgements ............................................... 8 81 7. Authors' Addresses ............................................. 8 82 8. Appendix: Rationale ............................................ 9 83 9. Appendix: Sample code ..........................................10 84 85 1. Introduction 86 87 1.1. Purpose 88 89 The purpose of this specification is to define a lossless 90 compressed data format that: 91 92 * Is independent of CPU type, operating system, file system, 93 and character set, and hence can be used for interchange; 94 95 * Can be produced or consumed, even for an arbitrarily long 96 sequentially presented input data stream, using only an a 97 priori bounded amount of intermediate storage, and hence can 98 be used in data communications or similar structures such as 99 Unix filters; 100 101 * Can use a number of different compression methods; 102 103 * Can be implemented readily in a manner not covered by 104 patents, and hence can be practiced freely. 105 106 The data format defined by this specification does not attempt to 107 allow random access to compressed data. 108 109 110 111 112 113 114 115 Deutsch & Gailly Informational [Page 2] 116 118 RFC 1950 ZLIB Compressed Data Format Specification May 1996 119 120 121 1.2. Intended audience 122 123 This specification is intended for use by implementors of software 124 to compress data into zlib format and/or decompress data from zlib 125 format. 126 127 The text of the specification assumes a basic background in 128 programming at the level of bits and other primitive data 129 representations. 130 131 1.3. Scope 132 133 The specification specifies a compressed data format that can be 134 used for in-memory compression of a sequence of arbitrary bytes. 135 136 1.4. Compliance 137 138 Unless otherwise indicated below, a compliant decompressor must be 139 able to accept and decompress any data set that conforms to all 140 the specifications presented here; a compliant compressor must 141 produce data sets that conform to all the specifications presented 142 here. 143 144 1.5. Definitions of terms and conventions used 145 146 byte: 8 bits stored or transmitted as a unit (same as an octet). 147 (For this specification, a byte is exactly 8 bits, even on 148 machines which store a character on a number of bits different 149 from 8.) See below, for the numbering of bits within a byte. 150 151 1.6. Changes from previous versions 152 153 Version 3.1 was the first public release of this specification. 154 In version 3.2, some terminology was changed and the Adler-32 155 sample code was rewritten for clarity. In version 3.3, the 156 support for a preset dictionary was introduced, and the 157 specification was converted to RFC style. 158 159 2. Detailed specification 160 161 2.1. Overall conventions 162 163 In the diagrams below, a box like this: 164 165 +---+ 166 | | <-- the vertical bars might be missing 167 +---+ 168 169 170 171 172 Deutsch & Gailly Informational [Page 3] 173 175 RFC 1950 ZLIB Compressed Data Format Specification May 1996 176 177 178 represents one byte; a box like this: 179 180 +==============+ 181 | | 182 +==============+ 183 184 represents a variable number of bytes. 185 186 Bytes stored within a computer do not have a "bit order", since 187 they are always treated as a unit. However, a byte considered as 188 an integer between 0 and 255 does have a most- and least- 189 significant bit, and since we write numbers with the most- 190 significant digit on the left, we also write bytes with the most- 191 significant bit on the left. In the diagrams below, we number the 192 bits of a byte so that bit 0 is the least-significant bit, i.e., 193 the bits are numbered: 194 195 +--------+ 196 |76543210| 197 +--------+ 198 199 Within a computer, a number may occupy multiple bytes. All 200 multi-byte numbers in the format described here are stored with 201 the MOST-significant byte first (at the lower memory address). 202 For example, the decimal number 520 is stored as: 203 204 0 1 205 +--------+--------+ 206 |00000010|00001000| 207 +--------+--------+ 208 ^ ^ 209 | | 210 | + less significant byte = 8 211 + more significant byte = 2 x 256 212 213 2.2. Data format 214 215 A zlib stream has the following structure: 216 217 0 1 218 +---+---+ 219 |CMF|FLG| (more-->) 220 +---+---+ 221 222 223 224 225 226 227 228 229 Deutsch & Gailly Informational [Page 4] 230 232 RFC 1950 ZLIB Compressed Data Format Specification May 1996 233 234 235 (if FLG.FDICT set) 236 237 0 1 2 3 238 +---+---+---+---+ 239 | DICTID | (more-->) 240 +---+---+---+---+ 241 242 +=====================+---+---+---+---+ 243 |...compressed data...| ADLER32 | 244 +=====================+---+---+---+---+ 245 246 Any data which may appear after ADLER32 are not part of the zlib 247 stream. 248 249 CMF (Compression Method and flags) 250 This byte is divided into a 4-bit compression method and a 4- 251 bit information field depending on the compression method. 252 253 bits 0 to 3 CM Compression method 254 bits 4 to 7 CINFO Compression info 255 256 CM (Compression method) 257 This identifies the compression method used in the file. CM = 8 258 denotes the "deflate" compression method with a window size up 259 to 32K. This is the method used by gzip and PNG (see 260 references [1] and [2] in Chapter 3, below, for the reference 261 documents). CM = 15 is reserved. It might be used in a future 262 version of this specification to indicate the presence of an 263 extra field before the compressed data. 264 265 CINFO (Compression info) 266 For CM = 8, CINFO is the base-2 logarithm of the LZ77 window 267 size, minus eight (CINFO=7 indicates a 32K window size). Values 268 of CINFO above 7 are not allowed in this version of the 269 specification. CINFO is not defined in this specification for 270 CM not equal to 8. 271 272 FLG (FLaGs) 273 This flag byte is divided as follows: 274 275 bits 0 to 4 FCHECK (check bits for CMF and FLG) 276 bit 5 FDICT (preset dictionary) 277 bits 6 to 7 FLEVEL (compression level) 278 279 The FCHECK value must be such that CMF and FLG, when viewed as 280 a 16-bit unsigned integer stored in MSB order (CMF*256 + FLG), 281 is a multiple of 31. 282 283 284 285 286 Deutsch & Gailly Informational [Page 5] 287 289 RFC 1950 ZLIB Compressed Data Format Specification May 1996 290 291 292 FDICT (Preset dictionary) 293 If FDICT is set, a DICT dictionary identifier is present 294 immediately after the FLG byte. The dictionary is a sequence of 295 bytes which are initially fed to the compressor without 296 producing any compressed output. DICT is the Adler-32 checksum 297 of this sequence of bytes (see the definition of ADLER32 298 below). The decompressor can use this identifier to determine 299 which dictionary has been used by the compressor. 300 301 FLEVEL (Compression level) 302 These flags are available for use by specific compression 303 methods. The "deflate" method (CM = 8) sets these flags as 304 follows: 305 306 0 - compressor used fastest algorithm 307 1 - compressor used fast algorithm 308 2 - compressor used default algorithm 309 3 - compressor used maximum compression, slowest algorithm 310 311 The information in FLEVEL is not needed for decompression; it 312 is there to indicate if recompression might be worthwhile. 313 314 compressed data 315 For compression method 8, the compressed data is stored in the 316 deflate compressed data format as described in the document 317 "DEFLATE Compressed Data Format Specification" by L. Peter 318 Deutsch. (See reference [3] in Chapter 3, below) 319 320 Other compressed data formats are not specified in this version 321 of the zlib specification. 322 323 ADLER32 (Adler-32 checksum) 324 This contains a checksum value of the uncompressed data 325 (excluding any dictionary data) computed according to Adler-32 326 algorithm. This algorithm is a 32-bit extension and improvement 327 of the Fletcher algorithm, used in the ITU-T X.224 / ISO 8073 328 standard. See references [4] and [5] in Chapter 3, below) 329 330 Adler-32 is composed of two sums accumulated per byte: s1 is 331 the sum of all bytes, s2 is the sum of all s1 values. Both sums 332 are done modulo 65521. s1 is initialized to 1, s2 to zero. The 333 Adler-32 checksum is stored as s2*65536 + s1 in most- 334 significant-byte first (network) order. 335 336 337 338 339 340 341 342 343 Deutsch & Gailly Informational [Page 6] 344 346 RFC 1950 ZLIB Compressed Data Format Specification May 1996 347 348 349 2.3. Compliance 350 351 A compliant compressor must produce streams with correct CMF, FLG 352 and ADLER32, but need not support preset dictionaries. When the 353 zlib data format is used as part of another standard data format, 354 the compressor may use only preset dictionaries that are specified 355 by this other data format. If this other format does not use the 356 preset dictionary feature, the compressor must not set the FDICT 357 flag. 358 359 A compliant decompressor must check CMF, FLG, and ADLER32, and 360 provide an error indication if any of these have incorrect values. 361 A compliant decompressor must give an error indication if CM is 362 not one of the values defined in this specification (only the 363 value 8 is permitted in this version), since another value could 364 indicate the presence of new features that would cause subsequent 365 data to be interpreted incorrectly. A compliant decompressor must 366 give an error indication if FDICT is set and DICTID is not the 367 identifier of a known preset dictionary. A decompressor may 368 ignore FLEVEL and still be compliant. When the zlib data format 369 is being used as a part of another standard format, a compliant 370 decompressor must support all the preset dictionaries specified by 371 the other format. When the other format does not use the preset 372 dictionary feature, a compliant decompressor must reject any 373 stream in which the FDICT flag is set. 374 375 3. References 376 377 [1] Deutsch, L.P.,"GZIP Compressed Data Format Specification", 378 available in ftp://ftp.uu.net/pub/archiving/zip/doc/ 379 380 [2] Thomas Boutell, "PNG (Portable Network Graphics) specification", 381 available in ftp://ftp.uu.net/graphics/png/documents/ 382 383 [3] Deutsch, L.P.,"DEFLATE Compressed Data Format Specification", 384 available in ftp://ftp.uu.net/pub/archiving/zip/doc/ 385 386 [4] Fletcher, J. G., "An Arithmetic Checksum for Serial 387 Transmissions," IEEE Transactions on Communications, Vol. COM-30, 388 No. 1, January 1982, pp. 247-252. 389 390 [5] ITU-T Recommendation X.224, Annex D, "Checksum Algorithms," 391 November, 1993, pp. 144, 145. (Available from 392 gopher://info.itu.ch). ITU-T X.244 is also the same as ISO 8073. 393 394 395 396 397 398 399 400 Deutsch & Gailly Informational [Page 7] 401 403 RFC 1950 ZLIB Compressed Data Format Specification May 1996 404 405 406 4. Source code 407 408 Source code for a C language implementation of a "zlib" compliant 409 library is available at ftp://ftp.uu.net/pub/archiving/zip/zlib/. 410 411 5. Security Considerations 412 413 A decoder that fails to check the ADLER32 checksum value may be 414 subject to undetected data corruption. 415 416 6. Acknowledgements 417 418 Trademarks cited in this document are the property of their 419 respective owners. 420 421 Jean-Loup Gailly and Mark Adler designed the zlib format and wrote 422 the related software described in this specification. Glenn 423 Randers-Pehrson converted this document to RFC and HTML format. 424 425 7. Authors' Addresses 426 427 L. Peter Deutsch 428 Aladdin Enterprises 429 203 Santa Margarita Ave. 430 Menlo Park, CA 94025 431 432 Phone: (415) 322-0103 (AM only) 433 FAX: (415) 322-1734 434 EMail: <ghost (a] aladdin.com> 435 436 437 Jean-Loup Gailly 438 439 EMail: <gzip (a] prep.ai.mit.edu> 440 441 Questions about the technical content of this specification can be 442 sent by email to 443 444 Jean-Loup Gailly <gzip (a] prep.ai.mit.edu> and 445 Mark Adler <madler (a] alumni.caltech.edu> 446 447 Editorial comments on this specification can be sent by email to 448 449 L. Peter Deutsch <ghost (a] aladdin.com> and 450 Glenn Randers-Pehrson <randeg (a] alumni.rpi.edu> 451 452 453 454 455 456 457 Deutsch & Gailly Informational [Page 8] 458 460 RFC 1950 ZLIB Compressed Data Format Specification May 1996 461 462 463 8. Appendix: Rationale 464 465 8.1. Preset dictionaries 466 467 A preset dictionary is specially useful to compress short input 468 sequences. The compressor can take advantage of the dictionary 469 context to encode the input in a more compact manner. The 470 decompressor can be initialized with the appropriate context by 471 virtually decompressing a compressed version of the dictionary 472 without producing any output. However for certain compression 473 algorithms such as the deflate algorithm this operation can be 474 achieved without actually performing any decompression. 475 476 The compressor and the decompressor must use exactly the same 477 dictionary. The dictionary may be fixed or may be chosen among a 478 certain number of predefined dictionaries, according to the kind 479 of input data. The decompressor can determine which dictionary has 480 been chosen by the compressor by checking the dictionary 481 identifier. This document does not specify the contents of 482 predefined dictionaries, since the optimal dictionaries are 483 application specific. Standard data formats using this feature of 484 the zlib specification must precisely define the allowed 485 dictionaries. 486 487 8.2. The Adler-32 algorithm 488 489 The Adler-32 algorithm is much faster than the CRC32 algorithm yet 490 still provides an extremely low probability of undetected errors. 491 492 The modulo on unsigned long accumulators can be delayed for 5552 493 bytes, so the modulo operation time is negligible. If the bytes 494 are a, b, c, the second sum is 3a + 2b + c + 3, and so is position 495 and order sensitive, unlike the first sum, which is just a 496 checksum. That 65521 is prime is important to avoid a possible 497 large class of two-byte errors that leave the check unchanged. 498 (The Fletcher checksum uses 255, which is not prime and which also 499 makes the Fletcher check insensitive to single byte changes 0 <-> 500 255.) 501 502 The sum s1 is initialized to 1 instead of zero to make the length 503 of the sequence part of s2, so that the length does not have to be 504 checked separately. (Any sequence of zeroes has a Fletcher 505 checksum of zero.) 506 507 508 509 510 511 512 513 514 Deutsch & Gailly Informational [Page 9] 515 517 RFC 1950 ZLIB Compressed Data Format Specification May 1996 518 519 520 9. Appendix: Sample code 521 522 The following C code computes the Adler-32 checksum of a data buffer. 523 It is written for clarity, not for speed. The sample code is in the 524 ANSI C programming language. Non C users may find it easier to read 525 with these hints: 526 527 & Bitwise AND operator. 528 >> Bitwise right shift operator. When applied to an 529 unsigned quantity, as here, right shift inserts zero bit(s) 530 at the left. 531 << Bitwise left shift operator. Left shift inserts zero 532 bit(s) at the right. 533 ++ "n++" increments the variable n. 534 % modulo operator: a % b is the remainder of a divided by b. 535 536 #define BASE 65521 /* largest prime smaller than 65536 */ 537 538 /* 539 Update a running Adler-32 checksum with the bytes buf[0..len-1] 540 and return the updated checksum. The Adler-32 checksum should be 541 initialized to 1. 542 543 Usage example: 544 545 unsigned long adler = 1L; 546 547 while (read_buffer(buffer, length) != EOF) { 548 adler = update_adler32(adler, buffer, length); 549 } 550 if (adler != original_adler) error(); 551 */ 552 unsigned long update_adler32(unsigned long adler, 553 unsigned char *buf, int len) 554 { 555 unsigned long s1 = adler & 0xffff; 556 unsigned long s2 = (adler >> 16) & 0xffff; 557 int n; 558 559 for (n = 0; n < len; n++) { 560 s1 = (s1 + buf[n]) % BASE; 561 s2 = (s2 + s1) % BASE; 562 } 563 return (s2 << 16) + s1; 564 } 565 566 /* Return the adler32 of the bytes buf[0..len-1] */ 567 568 569 570 571 Deutsch & Gailly Informational [Page 10] 572 574 RFC 1950 ZLIB Compressed Data Format Specification May 1996 575 576 577 unsigned long adler32(unsigned char *buf, int len) 578 { 579 return update_adler32(1L, buf, len); 580 } 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 Deutsch & Gailly Informational [Page 11] 629 631