diff options
Diffstat (limited to 'lib/libz/deflate.c')
| -rw-r--r-- | lib/libz/deflate.c | 1270 |
1 files changed, 848 insertions, 422 deletions
diff --git a/lib/libz/deflate.c b/lib/libz/deflate.c index 5cc3f7a..55cc69e 100644 --- a/lib/libz/deflate.c +++ b/lib/libz/deflate.c @@ -1,6 +1,6 @@ -/* $OpenBSD: deflate.c,v 1.11 2009/10/27 23:59:31 deraadt Exp $ */ +/* $OpenBSD: deflate.c,v 1.12 2021/07/04 14:24:49 tb Exp $ */ /* deflate.c -- compress data using the deflation algorithm - * Copyright (C) 1995-2005 Jean-loup Gailly. + * Copyright (C) 1995-2017 Jean-loup Gailly and Mark Adler * For conditions of distribution and use, see copyright notice in zlib.h */ @@ -38,7 +38,7 @@ * REFERENCES * * Deutsch, L.P.,"DEFLATE Compressed Data Format Specification". - * Available in http://www.ietf.org/rfc/rfc1951.txt + * Available in http://tools.ietf.org/html/rfc1951 * * A description of the Rabin and Karp algorithm is given in the book * "Algorithms" by R. Sedgewick, Addison-Wesley, p252. @@ -48,7 +48,6 @@ * */ - #include "deflate.h" /* @@ -71,27 +70,29 @@ typedef enum { typedef block_state (*compress_func) OF((deflate_state *s, int flush)); /* Compression function. Returns the block state after the call. */ +local int deflateStateCheck OF((z_streamp strm)); +local void slide_hash OF((deflate_state *s)); local void fill_window OF((deflate_state *s)); local block_state deflate_stored OF((deflate_state *s, int flush)); local block_state deflate_fast OF((deflate_state *s, int flush)); #ifndef FASTEST local block_state deflate_slow OF((deflate_state *s, int flush)); #endif +local block_state deflate_rle OF((deflate_state *s, int flush)); +local block_state deflate_huff OF((deflate_state *s, int flush)); local void lm_init OF((deflate_state *s)); local void putShortMSB OF((deflate_state *s, uInt b)); local void flush_pending OF((z_streamp strm)); -local int read_buf OF((z_streamp strm, Bytef *buf, unsigned size)); -#ifndef FASTEST +local unsigned read_buf OF((z_streamp strm, Bytef *buf, unsigned size)); #ifdef ASMV +# pragma message("Assembler code may have bugs -- use at your own risk") void match_init OF((void)); /* asm code initialization */ uInt longest_match OF((deflate_state *s, IPos cur_match)); #else local uInt longest_match OF((deflate_state *s, IPos cur_match)); #endif -#endif -local uInt longest_match_fast OF((deflate_state *s, IPos cur_match)); -#ifdef DEBUG +#ifdef ZLIB_DEBUG local void check_match OF((deflate_state *s, IPos start, IPos match, int length)); #endif @@ -108,11 +109,6 @@ local void check_match OF((deflate_state *s, IPos start, IPos match, #endif /* Matches of length 3 are discarded if their distance exceeds TOO_FAR */ -#define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1) -/* Minimum amount of lookahead, except at the end of the input file. - * See deflate.c for comments about the MIN_MATCH+1. - */ - /* Values for max_lazy_match, good_match and max_chain_length, depending on * the desired pack level (0..9). The values given below have been tuned to * exclude worst case performance for pathological files. Better values may be @@ -152,18 +148,14 @@ local const config configuration_table[10] = { * meaning. */ -#define EQUAL 0 -/* result of memcmp for equal strings */ - -#ifndef NO_DUMMY_DECL -struct static_tree_desc_s {int dummy;}; /* for buggy compilers */ -#endif +/* rank Z_BLOCK between Z_NO_FLUSH and Z_PARTIAL_FLUSH */ +#define RANK(f) (((f) * 2) - ((f) > 4 ? 9 : 0)) /* =========================================================================== * Update a hash value with the given input byte - * IN assertion: all calls to UPDATE_HASH are made with consecutive - * input characters, so that a running hash key can be computed from the - * previous key instead of complete recalculation each time. + * IN assertion: all calls to UPDATE_HASH are made with consecutive input + * characters, so that a running hash key can be computed from the previous + * key instead of complete recalculation each time. */ #define UPDATE_HASH(s,h,c) (h = (((h)<<s->hash_shift) ^ (c)) & s->hash_mask) @@ -174,9 +166,9 @@ struct static_tree_desc_s {int dummy;}; /* for buggy compilers */ * the previous length of the hash chain. * If this file is compiled with -DFASTEST, the compression level is forced * to 1, and no hash chains are maintained. - * IN assertion: all calls to INSERT_STRING are made with consecutive - * input characters and the first MIN_MATCH bytes of str are valid - * (except for the last MIN_MATCH-1 bytes of the input file). + * IN assertion: all calls to INSERT_STRING are made with consecutive input + * characters and the first MIN_MATCH bytes of str are valid (except for + * the last MIN_MATCH-1 bytes of the input file). */ #ifdef FASTEST #define INSERT_STRING(s, str, match_head) \ @@ -198,6 +190,37 @@ struct static_tree_desc_s {int dummy;}; /* for buggy compilers */ s->head[s->hash_size-1] = NIL; \ zmemzero((Bytef *)s->head, (unsigned)(s->hash_size-1)*sizeof(*s->head)); +/* =========================================================================== + * Slide the hash table when sliding the window down (could be avoided with 32 + * bit values at the expense of memory usage). We slide even when level == 0 to + * keep the hash table consistent if we switch back to level > 0 later. + */ +local void slide_hash(s) + deflate_state *s; +{ + unsigned n, m; + Posf *p; + uInt wsize = s->w_size; + + n = s->hash_size; + p = &s->head[n]; + do { + m = *--p; + *p = (Pos)(m >= wsize ? m - wsize : NIL); + } while (--n); + n = wsize; +#ifndef FASTEST + p = &s->prev[n]; + do { + m = *--p; + *p = (Pos)(m >= wsize ? m - wsize : NIL); + /* If n is not on any hash chain, prev[n] is garbage but + * its value will never be used. + */ + } while (--n); +#endif +} + /* ========================================================================= */ int ZEXPORT deflateInit_(strm, level, version, stream_size) z_streamp strm; @@ -239,10 +262,19 @@ int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy, strm->msg = Z_NULL; if (strm->zalloc == (alloc_func)0) { +#ifdef Z_SOLO + return Z_STREAM_ERROR; +#else strm->zalloc = zcalloc; strm->opaque = (voidpf)0; +#endif } - if (strm->zfree == (free_func)0) strm->zfree = zcfree; + if (strm->zfree == (free_func)0) +#ifdef Z_SOLO + return Z_STREAM_ERROR; +#else + strm->zfree = zcfree; +#endif #ifdef FASTEST if (level != 0) level = 1; @@ -262,7 +294,7 @@ int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy, #endif if (memLevel < 1 || memLevel > MAX_MEM_LEVEL || method != Z_DEFLATED || windowBits < 8 || windowBits > 15 || level < 0 || level > 9 || - strategy < 0 || strategy > Z_FIXED) { + strategy < 0 || strategy > Z_FIXED || (windowBits == 8 && wrap != 1)) { return Z_STREAM_ERROR; } if (windowBits == 8) windowBits = 9; /* until 256-byte window bug fixed */ @@ -270,14 +302,15 @@ int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy, if (s == Z_NULL) return Z_MEM_ERROR; strm->state = (struct internal_state FAR *)s; s->strm = strm; + s->status = INIT_STATE; /* to pass state test in deflateReset() */ s->wrap = wrap; s->gzhead = Z_NULL; - s->w_bits = windowBits; + s->w_bits = (uInt)windowBits; s->w_size = 1 << s->w_bits; s->w_mask = s->w_size - 1; - s->hash_bits = memLevel + 7; + s->hash_bits = (uInt)memLevel + 7; s->hash_size = 1 << s->hash_bits; s->hash_mask = s->hash_size - 1; s->hash_shift = ((s->hash_bits+MIN_MATCH-1)/MIN_MATCH); @@ -286,6 +319,8 @@ int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy, s->prev = (Posf *) ZALLOC(strm, s->w_size, sizeof(Pos)); s->head = (Posf *) ZALLOC(strm, s->hash_size, sizeof(Pos)); + s->high_water = 0; /* nothing written to s->window yet */ + s->lit_bufsize = 1 << (memLevel + 6); /* 16K elements by default */ overlay = (ushf *) ZALLOC(strm, s->lit_bufsize, sizeof(ush)+2); @@ -295,7 +330,7 @@ int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy, if (s->window == Z_NULL || s->prev == Z_NULL || s->head == Z_NULL || s->pending_buf == Z_NULL) { s->status = FINISH_STATE; - strm->msg = (char*)ERR_MSG(Z_MEM_ERROR); + strm->msg = ERR_MSG(Z_MEM_ERROR); deflateEnd (strm); return Z_MEM_ERROR; } @@ -309,6 +344,31 @@ int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy, return deflateReset(strm); } +/* ========================================================================= + * Check for a valid deflate stream state. Return 0 if ok, 1 if not. + */ +local int deflateStateCheck (strm) + z_streamp strm; +{ + deflate_state *s; + if (strm == Z_NULL || + strm->zalloc == (alloc_func)0 || strm->zfree == (free_func)0) + return 1; + s = strm->state; + if (s == Z_NULL || s->strm != strm || (s->status != INIT_STATE && +#ifdef GZIP + s->status != GZIP_STATE && +#endif + s->status != EXTRA_STATE && + s->status != NAME_STATE && + s->status != COMMENT_STATE && + s->status != HCRC_STATE && + s->status != BUSY_STATE && + s->status != FINISH_STATE)) + return 1; + return 0; +} + /* ========================================================================= */ int ZEXPORT deflateSetDictionary (strm, dictionary, dictLength) z_streamp strm; @@ -316,49 +376,97 @@ int ZEXPORT deflateSetDictionary (strm, dictionary, dictLength) uInt dictLength; { deflate_state *s; - uInt length = dictLength; - uInt n; - IPos hash_head = 0; + uInt str, n; + int wrap; + unsigned avail; + z_const unsigned char *next; - if (strm == Z_NULL || strm->state == Z_NULL || dictionary == Z_NULL || - strm->state->wrap == 2 || - (strm->state->wrap == 1 && strm->state->status != INIT_STATE)) + if (deflateStateCheck(strm) || dictionary == Z_NULL) return Z_STREAM_ERROR; - s = strm->state; - if (s->wrap) - strm->adler = adler32(strm->adler, dictionary, dictLength); + wrap = s->wrap; + if (wrap == 2 || (wrap == 1 && s->status != INIT_STATE) || s->lookahead) + return Z_STREAM_ERROR; - if (length < MIN_MATCH) return Z_OK; - if (length > MAX_DIST(s)) { - length = MAX_DIST(s); - dictionary += dictLength - length; /* use the tail of the dictionary */ + /* when using zlib wrappers, compute Adler-32 for provided dictionary */ + if (wrap == 1) + strm->adler = adler32(strm->adler, dictionary, dictLength); + s->wrap = 0; /* avoid computing Adler-32 in read_buf */ + + /* if dictionary would fill window, just replace the history */ + if (dictLength >= s->w_size) { + if (wrap == 0) { /* already empty otherwise */ + CLEAR_HASH(s); + s->strstart = 0; + s->block_start = 0L; + s->insert = 0; + } + dictionary += dictLength - s->w_size; /* use the tail */ + dictLength = s->w_size; } - zmemcpy(s->window, dictionary, length); - s->strstart = length; - s->block_start = (long)length; - /* Insert all strings in the hash table (except for the last two bytes). - * s->lookahead stays null, so s->ins_h will be recomputed at the next - * call of fill_window. - */ - s->ins_h = s->window[0]; - UPDATE_HASH(s, s->ins_h, s->window[1]); - for (n = 0; n <= length - MIN_MATCH; n++) { - INSERT_STRING(s, n, hash_head); + /* insert dictionary into window and hash */ + avail = strm->avail_in; + next = strm->next_in; + strm->avail_in = dictLength; + strm->next_in = (z_const Bytef *)dictionary; + fill_window(s); + while (s->lookahead >= MIN_MATCH) { + str = s->strstart; + n = s->lookahead - (MIN_MATCH-1); + do { + UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]); +#ifndef FASTEST + s->prev[str & s->w_mask] = s->head[s->ins_h]; +#endif + s->head[s->ins_h] = (Pos)str; + str++; + } while (--n); + s->strstart = str; + s->lookahead = MIN_MATCH-1; + fill_window(s); } - if (hash_head) hash_head = 0; /* to make compiler happy */ + s->strstart += s->lookahead; + s->block_start = (long)s->strstart; + s->insert = s->lookahead; + s->lookahead = 0; + s->match_length = s->prev_length = MIN_MATCH-1; + s->match_available = 0; + strm->next_in = next; + strm->avail_in = avail; + s->wrap = wrap; return Z_OK; } /* ========================================================================= */ -int ZEXPORT deflateReset (strm) +int ZEXPORT deflateGetDictionary (strm, dictionary, dictLength) z_streamp strm; + Bytef *dictionary; + uInt *dictLength; { deflate_state *s; + uInt len; - if (strm == Z_NULL || strm->state == Z_NULL || - strm->zalloc == (alloc_func)0 || strm->zfree == (free_func)0) { + if (deflateStateCheck(strm)) + return Z_STREAM_ERROR; + s = strm->state; + len = s->strstart + s->lookahead; + if (len > s->w_size) + len = s->w_size; + if (dictionary != Z_NULL && len) + zmemcpy(dictionary, s->window + s->strstart + s->lookahead - len, len); + if (dictLength != Z_NULL) + *dictLength = len; + return Z_OK; +} + +/* ========================================================================= */ +int ZEXPORT deflateResetKeep (strm) + z_streamp strm; +{ + deflate_state *s; + + if (deflateStateCheck(strm)) { return Z_STREAM_ERROR; } @@ -373,7 +481,11 @@ int ZEXPORT deflateReset (strm) if (s->wrap < 0) { s->wrap = -s->wrap; /* was made negative by deflate(..., Z_FINISH); */ } - s->status = s->wrap ? INIT_STATE : BUSY_STATE; + s->status = +#ifdef GZIP + s->wrap == 2 ? GZIP_STATE : +#endif + s->wrap ? INIT_STATE : BUSY_STATE; strm->adler = #ifdef GZIP s->wrap == 2 ? crc32(0L, Z_NULL, 0) : @@ -382,31 +494,70 @@ int ZEXPORT deflateReset (strm) s->last_flush = Z_NO_FLUSH; _tr_init(s); - lm_init(s); return Z_OK; } /* ========================================================================= */ +int ZEXPORT deflateReset (strm) + z_streamp strm; +{ + int ret; + + ret = deflateResetKeep(strm); + if (ret == Z_OK) + lm_init(strm->state); + return ret; +} + +/* ========================================================================= */ int ZEXPORT deflateSetHeader (strm, head) z_streamp strm; gz_headerp head; { - if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; - if (strm->state->wrap != 2) return Z_STREAM_ERROR; + if (deflateStateCheck(strm) || strm->state->wrap != 2) + return Z_STREAM_ERROR; strm->state->gzhead = head; return Z_OK; } /* ========================================================================= */ +int ZEXPORT deflatePending (strm, pending, bits) + unsigned *pending; + int *bits; + z_streamp strm; +{ + if (deflateStateCheck(strm)) return Z_STREAM_ERROR; + if (pending != Z_NULL) + *pending = strm->state->pending; + if (bits != Z_NULL) + *bits = strm->state->bi_valid; + return Z_OK; +} + +/* ========================================================================= */ int ZEXPORT deflatePrime (strm, bits, value) z_streamp strm; int bits; int value; { - if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; - strm->state->bi_valid = bits; - strm->state->bi_buf = (ush)(value & ((1 << bits) - 1)); + deflate_state *s; + int put; + + if (deflateStateCheck(strm)) return Z_STREAM_ERROR; + s = strm->state; + if ((Bytef *)(s->d_buf) < s->pending_out + ((Buf_size + 7) >> 3)) + return Z_BUF_ERROR; + do { + put = Buf_size - s->bi_valid; + if (put > bits) + put = bits; + s->bi_buf |= (ush)((value & ((1 << put) - 1)) << s->bi_valid); + s->bi_valid += put; + _tr_flush_bits(s); + value >>= put; + bits -= put; + } while (bits); return Z_OK; } @@ -418,9 +569,8 @@ int ZEXPORT deflateParams(strm, level, strategy) { deflate_state *s; compress_func func; - int err = Z_OK; - if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; + if (deflateStateCheck(strm)) return Z_STREAM_ERROR; s = strm->state; #ifdef FASTEST @@ -433,11 +583,23 @@ int ZEXPORT deflateParams(strm, level, strategy) } func = configuration_table[s->level].func; - if (func != configuration_table[level].func && strm->total_in != 0) { + if ((strategy != s->strategy || func != configuration_table[level].func) && + s->high_water) { /* Flush the last buffer: */ - err = deflate(strm, Z_PARTIAL_FLUSH); + int err = deflate(strm, Z_BLOCK); + if (err == Z_STREAM_ERROR) + return err; + if (strm->avail_out == 0) + return Z_BUF_ERROR; } if (s->level != level) { + if (s->level == 0 && s->matches != 0) { + if (s->matches == 1) + slide_hash(s); + else + CLEAR_HASH(s); + s->matches = 0; + } s->level = level; s->max_lazy_match = configuration_table[level].max_lazy; s->good_match = configuration_table[level].good_length; @@ -445,7 +607,7 @@ int ZEXPORT deflateParams(strm, level, strategy) s->max_chain_length = configuration_table[level].max_chain; } s->strategy = strategy; - return err; + return Z_OK; } /* ========================================================================= */ @@ -458,12 +620,12 @@ int ZEXPORT deflateTune(strm, good_length, max_lazy, nice_length, max_chain) { deflate_state *s; - if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; + if (deflateStateCheck(strm)) return Z_STREAM_ERROR; s = strm->state; - s->good_match = good_length; - s->max_lazy_match = max_lazy; + s->good_match = (uInt)good_length; + s->max_lazy_match = (uInt)max_lazy; s->nice_match = nice_length; - s->max_chain_length = max_chain; + s->max_chain_length = (uInt)max_chain; return Z_OK; } @@ -479,33 +641,68 @@ int ZEXPORT deflateTune(strm, good_length, max_lazy, nice_length, max_chain) * resulting from using fixed blocks instead of stored blocks, which deflate * can emit on compressed data for some combinations of the parameters. * - * This function could be more sophisticated to provide closer upper bounds - * for every combination of windowBits and memLevel, as well as wrap. - * But even the conservative upper bound of about 14% expansion does not - * seem onerous for output buffer allocation. + * This function could be more sophisticated to provide closer upper bounds for + * every combination of windowBits and memLevel. But even the conservative + * upper bound of about 14% expansion does not seem onerous for output buffer + * allocation. */ uLong ZEXPORT deflateBound(strm, sourceLen) z_streamp strm; uLong sourceLen; { deflate_state *s; - uLong destLen; + uLong complen, wraplen; - /* conservative upper bound */ - destLen = sourceLen + - ((sourceLen + 7) >> 3) + ((sourceLen + 63) >> 6) + 11; + /* conservative upper bound for compressed data */ + complen = sourceLen + + ((sourceLen + 7) >> 3) + ((sourceLen + 63) >> 6) + 5; - /* if can't get parameters, return conservative bound */ - if (strm == Z_NULL || strm->state == Z_NULL) - return destLen; + /* if can't get parameters, return conservative bound plus zlib wrapper */ + if (deflateStateCheck(strm)) + return complen + 6; - /* if not default parameters, return conservative bound */ + /* compute wrapper length */ s = strm->state; + switch (s->wrap) { + case 0: /* raw deflate */ + wraplen = 0; + break; + case 1: /* zlib wrapper */ + wraplen = 6 + (s->strstart ? 4 : 0); + break; +#ifdef GZIP + case 2: /* gzip wrapper */ + wraplen = 18; + if (s->gzhead != Z_NULL) { /* user-supplied gzip header */ + Bytef *str; + if (s->gzhead->extra != Z_NULL) + wraplen += 2 + s->gzhead->extra_len; + str = s->gzhead->name; + if (str != Z_NULL) + do { + wraplen++; + } while (*str++); + str = s->gzhead->comment; + if (str != Z_NULL) + do { + wraplen++; + } while (*str++); + if (s->gzhead->hcrc) + wraplen += 2; + } + break; +#endif + default: /* for compiler happiness */ + wraplen = 6; + } + + /* if not default parameters, return conservative bound */ if (s->w_bits != 15 || s->hash_bits != 8 + 7) - return destLen; + return complen + wraplen; /* default settings: return tight bound for that case */ - return compressBound(sourceLen); + return sourceLen + (sourceLen >> 12) + (sourceLen >> 14) + + (sourceLen >> 25) + 13 - 6 + wraplen; } /* ========================================================================= @@ -522,30 +719,43 @@ local void putShortMSB (s, b) } /* ========================================================================= - * Flush as much pending output as possible. All deflate() output goes - * through this function so some applications may wish to modify it - * to avoid allocating a large strm->next_out buffer and copying into it. - * (See also read_buf()). + * Flush as much pending output as possible. All deflate() output, except for + * some deflate_stored() output, goes through this function so some + * applications may wish to modify it to avoid allocating a large + * strm->next_out buffer and copying into it. (See also read_buf()). */ local void flush_pending(strm) z_streamp strm; { - unsigned len = strm->state->pending; + unsigned len; + deflate_state *s = strm->state; + _tr_flush_bits(s); + len = s->pending; if (len > strm->avail_out) len = strm->avail_out; if (len == 0) return; - zmemcpy(strm->next_out, strm->state->pending_out, len); + zmemcpy(strm->next_out, s->pending_out, len); strm->next_out += len; - strm->state->pending_out += len; + s->pending_out += len; strm->total_out += len; - strm->avail_out -= len; - strm->state->pending -= len; - if (strm->state->pending == 0) { - strm->state->pending_out = strm->state->pending_buf; + strm->avail_out -= len; + s->pending -= len; + if (s->pending == 0) { + s->pending_out = s->pending_buf; } } +/* =========================================================================== + * Update the header CRC with the bytes s->pending_buf[beg..s->pending - 1]. + */ +#define HCRC_UPDATE(beg) \ + do { \ + if (s->gzhead->hcrc && s->pending > (beg)) \ + strm->adler = crc32(strm->adler, s->pending_buf + (beg), \ + s->pending - (beg)); \ + } while (0) + /* ========================================================================= */ int ZEXPORT deflate (strm, flush) z_streamp strm; @@ -554,230 +764,229 @@ int ZEXPORT deflate (strm, flush) int old_flush; /* value of flush param for previous deflate call */ deflate_state *s; - if (strm == Z_NULL || strm->state == Z_NULL || - flush > Z_FINISH || flush < 0) { + if (deflateStateCheck(strm) || flush > Z_BLOCK || flush < 0) { return Z_STREAM_ERROR; } s = strm->state; if (strm->next_out == Z_NULL || - (strm->next_in == Z_NULL && strm->avail_in != 0) || + (strm->avail_in != 0 && strm->next_in == Z_NULL) || (s->status == FINISH_STATE && flush != Z_FINISH)) { ERR_RETURN(strm, Z_STREAM_ERROR); } if (strm->avail_out == 0) ERR_RETURN(strm, Z_BUF_ERROR); - s->strm = strm; /* just in case */ old_flush = s->last_flush; s->last_flush = flush; + /* Flush as much pending output as possible */ + if (s->pending != 0) { + flush_pending(strm); + if (strm->avail_out == 0) { + /* Since avail_out is 0, deflate will be called again with + * more output space, but possibly with both pending and + * avail_in equal to zero. There won't be anything to do, + * but this is not an error situation so make sure we + * return OK instead of BUF_ERROR at next call of deflate: + */ + s->last_flush = -1; + return Z_OK; + } + + /* Make sure there is something to do and avoid duplicate consecutive + * flushes. For repeated and useless calls with Z_FINISH, we keep + * returning Z_STREAM_END instead of Z_BUF_ERROR. + */ + } else if (strm->avail_in == 0 && RANK(flush) <= RANK(old_flush) && + flush != Z_FINISH) { + ERR_RETURN(strm, Z_BUF_ERROR); + } + + /* User must not provide more input after the first FINISH: */ + if (s->status == FINISH_STATE && strm->avail_in != 0) { + ERR_RETURN(strm, Z_BUF_ERROR); + } + /* Write the header */ if (s->status == INIT_STATE) { -#ifdef GZIP - if (s->wrap == 2) { - strm->adler = crc32(0L, Z_NULL, 0); - put_byte(s, 31); - put_byte(s, 139); - put_byte(s, 8); - if (s->gzhead == NULL) { - put_byte(s, 0); - put_byte(s, 0); - put_byte(s, 0); - put_byte(s, 0); - put_byte(s, 0); - put_byte(s, s->level == 9 ? 2 : - (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? - 4 : 0)); - put_byte(s, OS_CODE); - s->status = BUSY_STATE; - } - else { - put_byte(s, (s->gzhead->text ? 1 : 0) + - (s->gzhead->hcrc ? 2 : 0) + - (s->gzhead->extra == Z_NULL ? 0 : 4) + - (s->gzhead->name == Z_NULL ? 0 : 8) + - (s->gzhead->comment == Z_NULL ? 0 : 16) - ); - put_byte(s, (Byte)(s->gzhead->time & 0xff)); - put_byte(s, (Byte)((s->gzhead->time >> 8) & 0xff)); - put_byte(s, (Byte)((s->gzhead->time >> 16) & 0xff)); - put_byte(s, (Byte)((s->gzhead->time >> 24) & 0xff)); - put_byte(s, s->level == 9 ? 2 : - (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? - 4 : 0)); - put_byte(s, s->gzhead->os & 0xff); - if (s->gzhead->extra != NULL) { - put_byte(s, s->gzhead->extra_len & 0xff); - put_byte(s, (s->gzhead->extra_len >> 8) & 0xff); - } - if (s->gzhead->hcrc) - strm->adler = crc32(strm->adler, s->pending_buf, - s->pending); - s->gzindex = 0; - s->status = EXTRA_STATE; - } - } + /* zlib header */ + uInt header = (Z_DEFLATED + ((s->w_bits-8)<<4)) << 8; + uInt level_flags; + + if (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2) + level_flags = 0; + else if (s->level < 6) + level_flags = 1; + else if (s->level == 6) + level_flags = 2; else -#endif - { - uInt header = (Z_DEFLATED + ((s->w_bits-8)<<4)) << 8; - uInt level_flags; - - if (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2) - level_flags = 0; - else if (s->level < 6) - level_flags = 1; - else if (s->level == 6) - level_flags = 2; - else - level_flags = 3; - header |= (level_flags << 6); - if (s->strstart != 0) header |= PRESET_DICT; - header += 31 - (header % 31); + level_flags = 3; + header |= (level_flags << 6); + if (s->strstart != 0) header |= PRESET_DICT; + header += 31 - (header % 31); + + putShortMSB(s, header); + /* Save the adler32 of the preset dictionary: */ + if (s->strstart != 0) { + putShortMSB(s, (uInt)(strm->adler >> 16)); + putShortMSB(s, (uInt)(strm->adler & 0xffff)); + } + strm->adler = adler32(0L, Z_NULL, 0); + s->status = BUSY_STATE; + + /* Compression must start with an empty pending buffer */ + flush_pending(strm); + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; + } + } +#ifdef GZIP + if (s->status == GZIP_STATE) { + /* gzip header */ + strm->adler = crc32(0L, Z_NULL, 0); + put_byte(s, 31); + put_byte(s, 139); + put_byte(s, 8); + if (s->gzhead == Z_NULL) { + put_byte(s, 0); + put_byte(s, 0); + put_byte(s, 0); + put_byte(s, 0); + put_byte(s, 0); + put_byte(s, s->level == 9 ? 2 : + (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? + 4 : 0)); + put_byte(s, OS_CODE); s->status = BUSY_STATE; - putShortMSB(s, header); - /* Save the adler32 of the preset dictionary: */ - if (s->strstart != 0) { - putShortMSB(s, (uInt)(strm->adler >> 16)); - putShortMSB(s, (uInt)(strm->adler & 0xffff)); + /* Compression must start with an empty pending buffer */ + flush_pending(strm); + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; + } + } + else { + put_byte(s, (s->gzhead->text ? 1 : 0) + + (s->gzhead->hcrc ? 2 : 0) + + (s->gzhead->extra == Z_NULL ? 0 : 4) + + (s->gzhead->name == Z_NULL ? 0 : 8) + + (s->gzhead->comment == Z_NULL ? 0 : 16) + ); + put_byte(s, (Byte)(s->gzhead->time & 0xff)); + put_byte(s, (Byte)((s->gzhead->time >> 8) & 0xff)); + put_byte(s, (Byte)((s->gzhead->time >> 16) & 0xff)); + put_byte(s, (Byte)((s->gzhead->time >> 24) & 0xff)); + put_byte(s, s->level == 9 ? 2 : + (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? + 4 : 0)); + put_byte(s, s->gzhead->os & 0xff); + if (s->gzhead->extra != Z_NULL) { + put_byte(s, s->gzhead->extra_len & 0xff); + put_byte(s, (s->gzhead->extra_len >> 8) & 0xff); } - strm->adler = adler32(0L, Z_NULL, 0); + if (s->gzhead->hcrc) + strm->adler = crc32(strm->adler, s->pending_buf, + s->pending); + s->gzindex = 0; + s->status = EXTRA_STATE; } } -#ifdef GZIP if (s->status == EXTRA_STATE) { - if (s->gzhead->extra != NULL) { - uInt beg = s->pending; /* start of bytes to update crc */ - - while (s->gzindex < (s->gzhead->extra_len & 0xffff)) { - if (s->pending == s->pending_buf_size) { - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); - flush_pending(strm); - beg = s->pending; - if (s->pending == s->pending_buf_size) - break; + if (s->gzhead->extra != Z_NULL) { + ulg beg = s->pending; /* start of bytes to update crc */ + uInt left = (s->gzhead->extra_len & 0xffff) - s->gzindex; + while (s->pending + left > s->pending_buf_size) { + uInt copy = s->pending_buf_size - s->pending; + zmemcpy(s->pending_buf + s->pending, + s->gzhead->extra + s->gzindex, copy); + s->pending = s->pending_buf_size; + HCRC_UPDATE(beg); + s->gzindex += copy; + flush_pending(strm); + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; } - put_byte(s, s->gzhead->extra[s->gzindex]); - s->gzindex++; - } - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); - if (s->gzindex == s->gzhead->extra_len) { - s->gzindex = 0; - s->status = NAME_STATE; + beg = 0; + left -= copy; } + zmemcpy(s->pending_buf + s->pending, + s->gzhead->extra + s->gzindex, left); + s->pending += left; + HCRC_UPDATE(beg); + s->gzindex = 0; } - else - s->status = NAME_STATE; + s->status = NAME_STATE; } if (s->status == NAME_STATE) { - if (s->gzhead->name != NULL) { - uInt beg = s->pending; /* start of bytes to update crc */ + if (s->gzhead->name != Z_NULL) { + ulg beg = s->pending; /* start of bytes to update crc */ int val; - do { if (s->pending == s->pending_buf_size) { - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); + HCRC_UPDATE(beg); flush_pending(strm); - beg = s->pending; - if (s->pending == s->pending_buf_size) { - val = 1; - break; + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; } + beg = 0; } val = s->gzhead->name[s->gzindex++]; put_byte(s, val); } while (val != 0); - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); - if (val == 0) { - s->gzindex = 0; - s->status = COMMENT_STATE; - } + HCRC_UPDATE(beg); + s->gzindex = 0; } - else - s->status = COMMENT_STATE; + s->status = COMMENT_STATE; } if (s->status == COMMENT_STATE) { - if (s->gzhead->comment != NULL) { - uInt beg = s->pending; /* start of bytes to update crc */ + if (s->gzhead->comment != Z_NULL) { + ulg beg = s->pending; /* start of bytes to update crc */ int val; - do { if (s->pending == s->pending_buf_size) { - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); + HCRC_UPDATE(beg); flush_pending(strm); - beg = s->pending; - if (s->pending == s->pending_buf_size) { - val = 1; - break; + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; } + beg = 0; } val = s->gzhead->comment[s->gzindex++]; put_byte(s, val); } while (val != 0); - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); - if (val == 0) - s->status = HCRC_STATE; + HCRC_UPDATE(beg); } - else - s->status = HCRC_STATE; + s->status = HCRC_STATE; } if (s->status == HCRC_STATE) { if (s->gzhead->hcrc) { - if (s->pending + 2 > s->pending_buf_size) + if (s->pending + 2 > s->pending_buf_size) { flush_pending(strm); - if (s->pending + 2 <= s->pending_buf_size) { - put_byte(s, (Byte)(strm->adler & 0xff)); - put_byte(s, (Byte)((strm->adler >> 8) & 0xff)); - strm->adler = crc32(0L, Z_NULL, 0); - s->status = BUSY_STATE; + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; + } } + put_byte(s, (Byte)(strm->adler & 0xff)); + put_byte(s, (Byte)((strm->adler >> 8) & 0xff)); + strm->adler = crc32(0L, Z_NULL, 0); } - else - s->status = BUSY_STATE; - } -#endif + s->status = BUSY_STATE; - /* Flush as much pending output as possible */ - if (s->pending != 0) { + /* Compression must start with an empty pending buffer */ flush_pending(strm); - if (strm->avail_out == 0) { - /* Since avail_out is 0, deflate will be called again with - * more output space, but possibly with both pending and - * avail_in equal to zero. There won't be anything to do, - * but this is not an error situation so make sure we - * return OK instead of BUF_ERROR at next call of deflate: - */ + if (s->pending != 0) { s->last_flush = -1; return Z_OK; } - - /* Make sure there is something to do and avoid duplicate consecutive - * flushes. For repeated and useless calls with Z_FINISH, we keep - * returning Z_STREAM_END instead of Z_BUF_ERROR. - */ - } else if (strm->avail_in == 0 && flush <= old_flush && - flush != Z_FINISH) { - ERR_RETURN(strm, Z_BUF_ERROR); - } - - /* User must not provide more input after the first FINISH: */ - if (s->status == FINISH_STATE && strm->avail_in != 0) { - ERR_RETURN(strm, Z_BUF_ERROR); } +#endif /* Start a new block or continue the current one. */ @@ -785,7 +994,10 @@ int ZEXPORT deflate (strm, flush) (flush != Z_NO_FLUSH && s->status != FINISH_STATE)) { block_state bstate; - bstate = (*(configuration_table[s->level].func))(s, flush); + bstate = s->level == 0 ? deflate_stored(s, flush) : + s->strategy == Z_HUFFMAN_ONLY ? deflate_huff(s, flush) : + s->strategy == Z_RLE ? deflate_rle(s, flush) : + (*(configuration_table[s->level].func))(s, flush); if (bstate == finish_started || bstate == finish_done) { s->status = FINISH_STATE; @@ -806,13 +1018,18 @@ int ZEXPORT deflate (strm, flush) if (bstate == block_done) { if (flush == Z_PARTIAL_FLUSH) { _tr_align(s); - } else { /* FULL_FLUSH or SYNC_FLUSH */ + } else if (flush != Z_BLOCK) { /* FULL_FLUSH or SYNC_FLUSH */ _tr_stored_block(s, (char*)0, 0L, 0); /* For a full flush, this empty block will be recognized * as a special marker by inflate_sync(). */ if (flush == Z_FULL_FLUSH) { CLEAR_HASH(s); /* forget history */ + if (s->lookahead == 0) { + s->strstart = 0; + s->block_start = 0L; + s->insert = 0; + } } } flush_pending(strm); @@ -822,7 +1039,6 @@ int ZEXPORT deflate (strm, flush) } } } - Assert(strm->avail_out > 0, "bug2"); if (flush != Z_FINISH) return Z_OK; if (s->wrap <= 0) return Z_STREAM_END; @@ -859,18 +1075,9 @@ int ZEXPORT deflateEnd (strm) { int status; - if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; + if (deflateStateCheck(strm)) return Z_STREAM_ERROR; status = strm->state->status; - if (status != INIT_STATE && - status != EXTRA_STATE && - status != NAME_STATE && - status != COMMENT_STATE && - status != HCRC_STATE && - status != BUSY_STATE && - status != FINISH_STATE) { - return Z_STREAM_ERROR; - } /* Deallocate in reverse order of allocations: */ TRY_FREE(strm, strm->state->pending_buf); @@ -901,18 +1108,18 @@ int ZEXPORT deflateCopy (dest, source) ushf *overlay; - if (source == Z_NULL || dest == Z_NULL || source->state == Z_NULL) { + if (deflateStateCheck(source) || dest == Z_NULL) { return Z_STREAM_ERROR; } ss = source->state; - zmemcpy(dest, source, sizeof(z_stream)); + zmemcpy((voidpf)dest, (voidpf)source, sizeof(z_stream)); ds = (deflate_state *) ZALLOC(dest, 1, sizeof(deflate_state)); if (ds == Z_NULL) return Z_MEM_ERROR; dest->state = (struct internal_state FAR *) ds; - zmemcpy(ds, ss, sizeof(deflate_state)); + zmemcpy((voidpf)ds, (voidpf)ss, sizeof(deflate_state)); ds->strm = dest; ds->window = (Bytef *) ZALLOC(dest, ds->w_size, 2*sizeof(Byte)); @@ -928,8 +1135,8 @@ int ZEXPORT deflateCopy (dest, source) } /* following zmemcpy do not work for 16-bit MSDOS */ zmemcpy(ds->window, ss->window, ds->w_size * 2 * sizeof(Byte)); - zmemcpy(ds->prev, ss->prev, ds->w_size * sizeof(Pos)); - zmemcpy(ds->head, ss->head, ds->hash_size * sizeof(Pos)); + zmemcpy((voidpf)ds->prev, (voidpf)ss->prev, ds->w_size * sizeof(Pos)); + zmemcpy((voidpf)ds->head, (voidpf)ss->head, ds->hash_size * sizeof(Pos)); zmemcpy(ds->pending_buf, ss->pending_buf, (uInt)ds->pending_buf_size); ds->pending_out = ds->pending_buf + (ss->pending_out - ss->pending_buf); @@ -951,7 +1158,7 @@ int ZEXPORT deflateCopy (dest, source) * allocating a large strm->next_in buffer and copying from it. * (See also flush_pending()). */ -local int read_buf(strm, buf, size) +local unsigned read_buf(strm, buf, size) z_streamp strm; Bytef *buf; unsigned size; @@ -963,19 +1170,19 @@ local int read_buf(strm, buf, size) strm->avail_in -= len; + zmemcpy(buf, strm->next_in, len); if (strm->state->wrap == 1) { - strm->adler = adler32(strm->adler, strm->next_in, len); + strm->adler = adler32(strm->adler, buf, len); } #ifdef GZIP else if (strm->state->wrap == 2) { - strm->adler = crc32(strm->adler, strm->next_in, len); + strm->adler = crc32(strm->adler, buf, len); } #endif - zmemcpy(buf, strm->next_in, len); strm->next_in += len; strm->total_in += len; - return (int)len; + return len; } /* =========================================================================== @@ -998,6 +1205,7 @@ local void lm_init (s) s->strstart = 0; s->block_start = 0L; s->lookahead = 0; + s->insert = 0; s->match_length = s->prev_length = MIN_MATCH-1; s->match_available = 0; s->ins_h = 0; @@ -1028,9 +1236,9 @@ local uInt longest_match(s, cur_match) { unsigned chain_length = s->max_chain_length;/* max hash chain length */ register Bytef *scan = s->window + s->strstart; /* current string */ - register Bytef *match; /* matched string */ + register Bytef *match; /* matched string */ register int len; /* length of current match */ - int best_len = s->prev_length; /* best match length so far */ + int best_len = (int)s->prev_length; /* best match length so far */ int nice_match = s->nice_match; /* stop if match long enough */ IPos limit = s->strstart > (IPos)MAX_DIST(s) ? s->strstart - (IPos)MAX_DIST(s) : NIL; @@ -1065,7 +1273,7 @@ local uInt longest_match(s, cur_match) /* Do not look for matches beyond the end of the input. This is necessary * to make deflate deterministic. */ - if ((uInt)nice_match > s->lookahead) nice_match = s->lookahead; + if ((uInt)nice_match > s->lookahead) nice_match = (int)s->lookahead; Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead"); @@ -1165,12 +1373,13 @@ local uInt longest_match(s, cur_match) return s->lookahead; } #endif /* ASMV */ -#endif /* FASTEST */ + +#else /* FASTEST */ /* --------------------------------------------------------------------------- - * Optimized version for level == 1 or strategy == Z_RLE only + * Optimized version for FASTEST only */ -local uInt longest_match_fast(s, cur_match) +local uInt longest_match(s, cur_match) deflate_state *s; IPos cur_match; /* current match */ { @@ -1223,7 +1432,13 @@ local uInt longest_match_fast(s, cur_match) return (uInt)len <= s->lookahead ? (uInt)len : s->lookahead; } -#ifdef DEBUG +#endif /* FASTEST */ + +#ifdef ZLIB_DEBUG + +#define EQUAL 0 +/* result of memcmp for equal strings */ + /* =========================================================================== * Check that the match at match_start is indeed a match. */ @@ -1249,7 +1464,7 @@ local void check_match(s, start, match, length) } #else # define check_match(s, start, match, length) -#endif /* DEBUG */ +#endif /* ZLIB_DEBUG */ /* =========================================================================== * Fill the window when the lookahead becomes insufficient. @@ -1264,11 +1479,12 @@ local void check_match(s, start, match, length) local void fill_window(s) deflate_state *s; { - register unsigned n, m; - register Posf *p; + unsigned n; unsigned more; /* Amount of free space at the end of the window. */ uInt wsize = s->w_size; + Assert(s->lookahead < MIN_LOOKAHEAD, "already enough lookahead"); + do { more = (unsigned)(s->window_size -(ulg)s->lookahead -(ulg)s->strstart); @@ -1290,39 +1506,14 @@ local void fill_window(s) */ if (s->strstart >= wsize+MAX_DIST(s)) { - zmemcpy(s->window, s->window+wsize, (unsigned)wsize); + zmemcpy(s->window, s->window+wsize, (unsigned)wsize - more); s->match_start -= wsize; s->strstart -= wsize; /* we now have strstart >= MAX_DIST */ s->block_start -= (long) wsize; - - /* Slide the hash table (could be avoided with 32 bit values - at the expense of memory usage). We slide even when level == 0 - to keep the hash table consistent if we switch back to level > 0 - later. (Using level 0 permanently is not an optimal usage of - zlib, so we don't care about this pathological case.) - */ - /* %%% avoid this when Z_RLE */ - n = s->hash_size; - p = &s->head[n]; - do { - m = *--p; - *p = (Pos)(m >= wsize ? m-wsize : NIL); - } while (--n); - - n = wsize; -#ifndef FASTEST - p = &s->prev[n]; - do { - m = *--p; - *p = (Pos)(m >= wsize ? m-wsize : NIL); - /* If n is not on any hash chain, prev[n] is garbage but - * its value will never be used. - */ - } while (--n); -#endif + slide_hash(s); more += wsize; } - if (s->strm->avail_in == 0) return; + if (s->strm->avail_in == 0) break; /* If there was no sliding: * strstart <= WSIZE+MAX_DIST-1 && lookahead <= MIN_LOOKAHEAD - 1 && @@ -1341,99 +1532,283 @@ local void fill_window(s) s->lookahead += n; /* Initialize the hash value now that we have some input: */ - if (s->lookahead >= MIN_MATCH) { - s->ins_h = s->window[s->strstart]; - UPDATE_HASH(s, s->ins_h, s->window[s->strstart+1]); + if (s->lookahead + s->insert >= MIN_MATCH) { + uInt str = s->strstart - s->insert; + s->ins_h = s->window[str]; + UPDATE_HASH(s, s->ins_h, s->window[str + 1]); #if MIN_MATCH != 3 Call UPDATE_HASH() MIN_MATCH-3 more times #endif + while (s->insert) { + UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]); +#ifndef FASTEST + s->prev[str & s->w_mask] = s->head[s->ins_h]; +#endif + s->head[s->ins_h] = (Pos)str; + str++; + s->insert--; + if (s->lookahead + s->insert < MIN_MATCH) + break; + } } /* If the whole input has less than MIN_MATCH bytes, ins_h is garbage, * but this is not important since only literal bytes will be emitted. */ } while (s->lookahead < MIN_LOOKAHEAD && s->strm->avail_in != 0); + + /* If the WIN_INIT bytes after the end of the current data have never been + * written, then zero those bytes in order to avoid memory check reports of + * the use of uninitialized (or uninitialised as Julian writes) bytes by + * the longest match routines. Update the high water mark for the next + * time through here. WIN_INIT is set to MAX_MATCH since the longest match + * routines allow scanning to strstart + MAX_MATCH, ignoring lookahead. + */ + if (s->high_water < s->window_size) { + ulg curr = s->strstart + (ulg)(s->lookahead); + ulg init; + + if (s->high_water < curr) { + /* Previous high water mark below current data -- zero WIN_INIT + * bytes or up to end of window, whichever is less. + */ + init = s->window_size - curr; + if (init > WIN_INIT) + init = WIN_INIT; + zmemzero(s->window + curr, (unsigned)init); + s->high_water = curr + init; + } + else if (s->high_water < (ulg)curr + WIN_INIT) { + /* High water mark at or above current data, but below current data + * plus WIN_INIT -- zero out to current data plus WIN_INIT, or up + * to end of window, whichever is less. + */ + init = (ulg)curr + WIN_INIT - s->high_water; + if (init > s->window_size - s->high_water) + init = s->window_size - s->high_water; + zmemzero(s->window + s->high_water, (unsigned)init); + s->high_water += init; + } + } + + Assert((ulg)s->strstart <= s->window_size - MIN_LOOKAHEAD, + "not enough room for search"); } /* =========================================================================== * Flush the current block, with given end-of-file flag. * IN assertion: strstart is set to the end of the current match. */ -#define FLUSH_BLOCK_ONLY(s, eof) { \ +#define FLUSH_BLOCK_ONLY(s, last) { \ _tr_flush_block(s, (s->block_start >= 0L ? \ (charf *)&s->window[(unsigned)s->block_start] : \ (charf *)Z_NULL), \ (ulg)((long)s->strstart - s->block_start), \ - (eof)); \ + (last)); \ s->block_start = s->strstart; \ flush_pending(s->strm); \ Tracev((stderr,"[FLUSH]")); \ } /* Same but force premature exit if necessary. */ -#define FLUSH_BLOCK(s, eof) { \ - FLUSH_BLOCK_ONLY(s, eof); \ - if (s->strm->avail_out == 0) return (eof) ? finish_started : need_more; \ +#define FLUSH_BLOCK(s, last) { \ + FLUSH_BLOCK_ONLY(s, last); \ + if (s->strm->avail_out == 0) return (last) ? finish_started : need_more; \ } +/* Maximum stored block length in deflate format (not including header). */ +#define MAX_STORED 65535 + +/* Minimum of a and b. */ +#define MIN(a, b) ((a) > (b) ? (b) : (a)) + /* =========================================================================== * Copy without compression as much as possible from the input stream, return * the current block state. - * This function does not insert new strings in the dictionary since - * uncompressible data is probably not useful. This function is used - * only for the level=0 compression option. - * NOTE: this function should be optimized to avoid extra copying from - * window to pending_buf. + * + * In case deflateParams() is used to later switch to a non-zero compression + * level, s->matches (otherwise unused when storing) keeps track of the number + * of hash table slides to perform. If s->matches is 1, then one hash table + * slide will be done when switching. If s->matches is 2, the maximum value + * allowed here, then the hash table will be cleared, since two or more slides + * is the same as a clear. + * + * deflate_stored() is written to minimize the number of times an input byte is + * copied. It is most efficient with large input and output buffers, which + * maximizes the opportunites to have a single copy from next_in to next_out. */ local block_state deflate_stored(s, flush) deflate_state *s; int flush; { - /* Stored blocks are limited to 0xffff bytes, pending_buf is limited - * to pending_buf_size, and each stored block has a 5 byte header: + /* Smallest worthy block size when not flushing or finishing. By default + * this is 32K. This can be as small as 507 bytes for memLevel == 1. For + * large input and output buffers, the stored block size will be larger. */ - ulg max_block_size = 0xffff; - ulg max_start; + unsigned min_block = MIN(s->pending_buf_size - 5, s->w_size); - if (max_block_size > s->pending_buf_size - 5) { - max_block_size = s->pending_buf_size - 5; - } - - /* Copy as much as possible from input to output: */ - for (;;) { - /* Fill the window as much as possible: */ - if (s->lookahead <= 1) { - - Assert(s->strstart < s->w_size+MAX_DIST(s) || - s->block_start >= (long)s->w_size, "slide too late"); + /* Copy as many min_block or larger stored blocks directly to next_out as + * possible. If flushing, copy the remaining available input to next_out as + * stored blocks, if there is enough space. + */ + unsigned len, left, have, last = 0; + unsigned used = s->strm->avail_in; + do { + /* Set len to the maximum size block that we can copy directly with the + * available input data and output space. Set left to how much of that + * would be copied from what's left in the window. + */ + len = MAX_STORED; /* maximum deflate stored block length */ + have = (s->bi_valid + 42) >> 3; /* number of header bytes */ + if (s->strm->avail_out < have) /* need room for header */ + break; + /* maximum stored block length that will fit in avail_out: */ + have = s->strm->avail_out - have; + left = s->strstart - s->block_start; /* bytes left in window */ + if (len > (ulg)left + s->strm->avail_in) + len = left + s->strm->avail_in; /* limit len to the input */ + if (len > have) + len = have; /* limit len to the output */ + + /* If the stored block would be less than min_block in length, or if + * unable to copy all of the available input when flushing, then try + * copying to the window and the pending buffer instead. Also don't + * write an empty block when flushing -- deflate() does that. + */ + if (len < min_block && ((len == 0 && flush != Z_FINISH) || + flush == Z_NO_FLUSH || + len != left + s->strm->avail_in)) + break; - fill_window(s); - if (s->lookahead == 0 && flush == Z_NO_FLUSH) return need_more; + /* Make a dummy stored block in pending to get the header bytes, + * including any pending bits. This also updates the debugging counts. + */ + last = flush == Z_FINISH && len == left + s->strm->avail_in ? 1 : 0; + _tr_stored_block(s, (char *)0, 0L, last); + + /* Replace the lengths in the dummy stored block with len. */ + s->pending_buf[s->pending - 4] = len; + s->pending_buf[s->pending - 3] = len >> 8; + s->pending_buf[s->pending - 2] = ~len; + s->pending_buf[s->pending - 1] = ~len >> 8; + + /* Write the stored block header bytes. */ + flush_pending(s->strm); + +#ifdef ZLIB_DEBUG + /* Update debugging counts for the data about to be copied. */ + s->compressed_len += len << 3; + s->bits_sent += len << 3; +#endif - if (s->lookahead == 0) break; /* flush the current block */ + /* Copy uncompressed bytes from the window to next_out. */ + if (left) { + if (left > len) + left = len; + zmemcpy(s->strm->next_out, s->window + s->block_start, left); + s->strm->next_out += left; + s->strm->avail_out -= left; + s->strm->total_out += left; + s->block_start += left; + len -= left; } - Assert(s->block_start >= 0L, "block gone"); - - s->strstart += s->lookahead; - s->lookahead = 0; - - /* Emit a stored block if pending_buf will be full: */ - max_start = s->block_start + max_block_size; - if (s->strstart == 0 || (ulg)s->strstart >= max_start) { - /* strstart == 0 is possible when wraparound on 16-bit machine */ - s->lookahead = (uInt)(s->strstart - max_start); - s->strstart = (uInt)max_start; - FLUSH_BLOCK(s, 0); + + /* Copy uncompressed bytes directly from next_in to next_out, updating + * the check value. + */ + if (len) { + read_buf(s->strm, s->strm->next_out, len); + s->strm->next_out += len; + s->strm->avail_out -= len; + s->strm->total_out += len; } - /* Flush if we may have to slide, otherwise block_start may become - * negative and the data will be gone: + } while (last == 0); + + /* Update the sliding window with the last s->w_size bytes of the copied + * data, or append all of the copied data to the existing window if less + * than s->w_size bytes were copied. Also update the number of bytes to + * insert in the hash tables, in the event that deflateParams() switches to + * a non-zero compression level. + */ + used -= s->strm->avail_in; /* number of input bytes directly copied */ + if (used) { + /* If any input was used, then no unused input remains in the window, + * therefore s->block_start == s->strstart. */ - if (s->strstart - (uInt)s->block_start >= MAX_DIST(s)) { - FLUSH_BLOCK(s, 0); + if (used >= s->w_size) { /* supplant the previous history */ + s->matches = 2; /* clear hash */ + zmemcpy(s->window, s->strm->next_in - s->w_size, s->w_size); + s->strstart = s->w_size; } + else { + if (s->window_size - s->strstart <= used) { + /* Slide the window down. */ + s->strstart -= s->w_size; + zmemcpy(s->window, s->window + s->w_size, s->strstart); + if (s->matches < 2) + s->matches++; /* add a pending slide_hash() */ + } + zmemcpy(s->window + s->strstart, s->strm->next_in - used, used); + s->strstart += used; + } + s->block_start = s->strstart; + s->insert += MIN(used, s->w_size - s->insert); + } + if (s->high_water < s->strstart) + s->high_water = s->strstart; + + /* If the last block was written to next_out, then done. */ + if (last) + return finish_done; + + /* If flushing and all input has been consumed, then done. */ + if (flush != Z_NO_FLUSH && flush != Z_FINISH && + s->strm->avail_in == 0 && (long)s->strstart == s->block_start) + return block_done; + + /* Fill the window with any remaining input. */ + have = s->window_size - s->strstart - 1; + if (s->strm->avail_in > have && s->block_start >= (long)s->w_size) { + /* Slide the window down. */ + s->block_start -= s->w_size; + s->strstart -= s->w_size; + zmemcpy(s->window, s->window + s->w_size, s->strstart); + if (s->matches < 2) + s->matches++; /* add a pending slide_hash() */ + have += s->w_size; /* more space now */ + } + if (have > s->strm->avail_in) + have = s->strm->avail_in; + if (have) { + read_buf(s->strm, s->window + s->strstart, have); + s->strstart += have; + } + if (s->high_water < s->strstart) + s->high_water = s->strstart; + + /* There was not enough avail_out to write a complete worthy or flushed + * stored block to next_out. Write a stored block to pending instead, if we + * have enough input for a worthy block, or if flushing and there is enough + * room for the remaining input as a stored block in the pending buffer. + */ + have = (s->bi_valid + 42) >> 3; /* number of header bytes */ + /* maximum stored block length that will fit in pending: */ + have = MIN(s->pending_buf_size - have, MAX_STORED); + min_block = MIN(have, s->w_size); + left = s->strstart - s->block_start; + if (left >= min_block || + ((left || flush == Z_FINISH) && flush != Z_NO_FLUSH && + s->strm->avail_in == 0 && left <= have)) { + len = MIN(left, have); + last = flush == Z_FINISH && s->strm->avail_in == 0 && + len == left ? 1 : 0; + _tr_stored_block(s, (charf *)s->window + s->block_start, len, last); + s->block_start += len; + flush_pending(s->strm); } - FLUSH_BLOCK(s, flush == Z_FINISH); - return flush == Z_FINISH ? finish_done : block_done; + + /* We've done all we can with the available input and output. */ + return last ? finish_started : need_more; } /* =========================================================================== @@ -1447,7 +1822,7 @@ local block_state deflate_fast(s, flush) deflate_state *s; int flush; { - IPos hash_head = NIL; /* head of the hash chain */ + IPos hash_head; /* head of the hash chain */ int bflush; /* set if current block must be flushed */ for (;;) { @@ -1467,6 +1842,7 @@ local block_state deflate_fast(s, flush) /* Insert the string window[strstart .. strstart+2] in the * dictionary, and set hash_head to the head of the hash chain: */ + hash_head = NIL; if (s->lookahead >= MIN_MATCH) { INSERT_STRING(s, s->strstart, hash_head); } @@ -1479,19 +1855,8 @@ local block_state deflate_fast(s, flush) * of window index 0 (in particular we have to avoid a match * of the string with itself at the start of the input file). */ -#ifdef FASTEST - if ((s->strategy != Z_HUFFMAN_ONLY && s->strategy != Z_RLE) || - (s->strategy == Z_RLE && s->strstart - hash_head == 1)) { - s->match_length = longest_match_fast (s, hash_head); - } -#else - if (s->strategy != Z_HUFFMAN_ONLY && s->strategy != Z_RLE) { - s->match_length = longest_match (s, hash_head); - } else if (s->strategy == Z_RLE && s->strstart - hash_head == 1) { - s->match_length = longest_match_fast (s, hash_head); - } -#endif - /* longest_match() or longest_match_fast() sets match_start */ + s->match_length = longest_match (s, hash_head); + /* longest_match() sets match_start */ } if (s->match_length >= MIN_MATCH) { check_match(s, s->strstart, s->match_start, s->match_length); @@ -1539,8 +1904,14 @@ local block_state deflate_fast(s, flush) } if (bflush) FLUSH_BLOCK(s, 0); } - FLUSH_BLOCK(s, flush == Z_FINISH); - return flush == Z_FINISH ? finish_done : block_done; + s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1; + if (flush == Z_FINISH) { + FLUSH_BLOCK(s, 1); + return finish_done; + } + if (s->last_lit) + FLUSH_BLOCK(s, 0); + return block_done; } #ifndef FASTEST @@ -1553,7 +1924,7 @@ local block_state deflate_slow(s, flush) deflate_state *s; int flush; { - IPos hash_head = NIL; /* head of hash chain */ + IPos hash_head; /* head of hash chain */ int bflush; /* set if current block must be flushed */ /* Process the input block. */ @@ -1574,6 +1945,7 @@ local block_state deflate_slow(s, flush) /* Insert the string window[strstart .. strstart+2] in the * dictionary, and set hash_head to the head of the hash chain: */ + hash_head = NIL; if (s->lookahead >= MIN_MATCH) { INSERT_STRING(s, s->strstart, hash_head); } @@ -1589,12 +1961,8 @@ local block_state deflate_slow(s, flush) * of window index 0 (in particular we have to avoid a match * of the string with itself at the start of the input file). */ - if (s->strategy != Z_HUFFMAN_ONLY && s->strategy != Z_RLE) { - s->match_length = longest_match (s, hash_head); - } else if (s->strategy == Z_RLE && s->strstart - hash_head == 1) { - s->match_length = longest_match_fast (s, hash_head); - } - /* longest_match() or longest_match_fast() sets match_start */ + s->match_length = longest_match (s, hash_head); + /* longest_match() sets match_start */ if (s->match_length <= 5 && (s->strategy == Z_FILTERED #if TOO_FAR <= 32767 @@ -1667,12 +2035,17 @@ local block_state deflate_slow(s, flush) _tr_tally_lit(s, s->window[s->strstart-1], bflush); s->match_available = 0; } - FLUSH_BLOCK(s, flush == Z_FINISH); - return flush == Z_FINISH ? finish_done : block_done; + s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1; + if (flush == Z_FINISH) { + FLUSH_BLOCK(s, 1); + return finish_done; + } + if (s->last_lit) + FLUSH_BLOCK(s, 0); + return block_done; } #endif /* FASTEST */ -#if 0 /* =========================================================================== * For Z_RLE, simply look for runs of bytes, generate matches only of distance * one. Do not maintain a hash table. (It will be regenerated if this run of @@ -1682,43 +2055,52 @@ local block_state deflate_rle(s, flush) deflate_state *s; int flush; { - int bflush; /* set if current block must be flushed */ - uInt run; /* length of run */ - uInt max; /* maximum length of run */ - uInt prev; /* byte at distance one to match */ - Bytef *scan; /* scan for end of run */ + int bflush; /* set if current block must be flushed */ + uInt prev; /* byte at distance one to match */ + Bytef *scan, *strend; /* scan goes up to strend for length of run */ for (;;) { /* Make sure that we always have enough lookahead, except * at the end of the input file. We need MAX_MATCH bytes - * for the longest encodable run. + * for the longest run, plus one for the unrolled loop. */ - if (s->lookahead < MAX_MATCH) { + if (s->lookahead <= MAX_MATCH) { fill_window(s); - if (s->lookahead < MAX_MATCH && flush == Z_NO_FLUSH) { + if (s->lookahead <= MAX_MATCH && flush == Z_NO_FLUSH) { return need_more; } if (s->lookahead == 0) break; /* flush the current block */ } /* See how many times the previous byte repeats */ - run = 0; - if (s->strstart > 0) { /* if there is a previous byte, that is */ - max = s->lookahead < MAX_MATCH ? s->lookahead : MAX_MATCH; + s->match_length = 0; + if (s->lookahead >= MIN_MATCH && s->strstart > 0) { scan = s->window + s->strstart - 1; - prev = *scan++; - do { - if (*scan++ != prev) - break; - } while (++run < max); + prev = *scan; + if (prev == *++scan && prev == *++scan && prev == *++scan) { + strend = s->window + s->strstart + MAX_MATCH; + do { + } while (prev == *++scan && prev == *++scan && + prev == *++scan && prev == *++scan && + prev == *++scan && prev == *++scan && + prev == *++scan && prev == *++scan && + scan < strend); + s->match_length = MAX_MATCH - (uInt)(strend - scan); + if (s->match_length > s->lookahead) + s->match_length = s->lookahead; + } + Assert(scan <= s->window+(uInt)(s->window_size-1), "wild scan"); } /* Emit match if have run of MIN_MATCH or longer, else emit literal */ - if (run >= MIN_MATCH) { - check_match(s, s->strstart, s->strstart - 1, run); - _tr_tally_dist(s, 1, run - MIN_MATCH, bflush); - s->lookahead -= run; - s->strstart += run; + if (s->match_length >= MIN_MATCH) { + check_match(s, s->strstart, s->strstart - 1, s->match_length); + + _tr_tally_dist(s, 1, s->match_length - MIN_MATCH, bflush); + + s->lookahead -= s->match_length; + s->strstart += s->match_length; + s->match_length = 0; } else { /* No match, output a literal byte */ Tracevv((stderr,"%c", s->window[s->strstart])); @@ -1728,7 +2110,51 @@ local block_state deflate_rle(s, flush) } if (bflush) FLUSH_BLOCK(s, 0); } - FLUSH_BLOCK(s, flush == Z_FINISH); - return flush == Z_FINISH ? finish_done : block_done; + s->insert = 0; + if (flush == Z_FINISH) { + FLUSH_BLOCK(s, 1); + return finish_done; + } + if (s->last_lit) + FLUSH_BLOCK(s, 0); + return block_done; +} + +/* =========================================================================== + * For Z_HUFFMAN_ONLY, do not look for matches. Do not maintain a hash table. + * (It will be regenerated if this run of deflate switches away from Huffman.) + */ +local block_state deflate_huff(s, flush) + deflate_state *s; + int flush; +{ + int bflush; /* set if current block must be flushed */ + + for (;;) { + /* Make sure that we have a literal to write. */ + if (s->lookahead == 0) { + fill_window(s); + if (s->lookahead == 0) { + if (flush == Z_NO_FLUSH) + return need_more; + break; /* flush the current block */ + } + } + + /* Output a literal byte */ + s->match_length = 0; + Tracevv((stderr,"%c", s->window[s->strstart])); + _tr_tally_lit (s, s->window[s->strstart], bflush); + s->lookahead--; + s->strstart++; + if (bflush) FLUSH_BLOCK(s, 0); + } + s->insert = 0; + if (flush == Z_FINISH) { + FLUSH_BLOCK(s, 1); + return finish_done; + } + if (s->last_lit) + FLUSH_BLOCK(s, 0); + return block_done; } -#endif |