1 /*
  2  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  3  *
  4  * This code is free software; you can redistribute it and/or modify it
  5  * under the terms of the GNU General Public License version 2 only, as
  6  * published by the Free Software Foundation.  Oracle designates this
  7  * particular file as subject to the "Classpath" exception as provided
  8  * by Oracle in the LICENSE file that accompanied this code.
  9  *
 10  * This code is distributed in the hope that it will be useful, but WITHOUT
 11  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 12  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
 13  * version 2 for more details (a copy is included in the LICENSE file that
 14  * accompanied this code).
 15  *
 16  * You should have received a copy of the GNU General Public License version
 17  * 2 along with this work; if not, write to the Free Software Foundation,
 18  * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
 19  *
 20  * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
 21  * or visit www.oracle.com if you need additional information or have any
 22  * questions.
 23  */
 24 
 25 /* deflate.h -- internal compression state
 26  * Copyright (C) 1995-2018 Jean-loup Gailly
 27  * For conditions of distribution and use, see copyright notice in zlib.h
 28  */
 29 
 30 /* WARNING: this file should *not* be used by applications. It is
 31    part of the implementation of the compression library and is
 32    subject to change. Applications should only use zlib.h.
 33  */
 34 
 35 /* @(#) $Id$ */
 36 
 37 #ifndef DEFLATE_H
 38 #define DEFLATE_H
 39 
 40 #include "zutil.h"
 41 
 42 /* define NO_GZIP when compiling if you want to disable gzip header and
 43    trailer creation by deflate().  NO_GZIP would be used to avoid linking in
 44    the crc code when it is not needed.  For shared libraries, gzip encoding
 45    should be left enabled. */
 46 #ifndef NO_GZIP
 47 #  define GZIP
 48 #endif
 49 
 50 /* ===========================================================================
 51  * Internal compression state.
 52  */
 53 
 54 #define LENGTH_CODES 29
 55 /* number of length codes, not counting the special END_BLOCK code */
 56 
 57 #define LITERALS  256
 58 /* number of literal bytes 0..255 */
 59 
 60 #define L_CODES (LITERALS+1+LENGTH_CODES)
 61 /* number of Literal or Length codes, including the END_BLOCK code */
 62 
 63 #define D_CODES   30
 64 /* number of distance codes */
 65 
 66 #define BL_CODES  19
 67 /* number of codes used to transfer the bit lengths */
 68 
 69 #define HEAP_SIZE (2*L_CODES+1)
 70 /* maximum heap size */
 71 
 72 #define MAX_BITS 15
 73 /* All codes must not exceed MAX_BITS bits */
 74 
 75 #define Buf_size 16
 76 /* size of bit buffer in bi_buf */
 77 
 78 #define INIT_STATE    42    /* zlib header -> BUSY_STATE */
 79 #ifdef GZIP
 80 #  define GZIP_STATE  57    /* gzip header -> BUSY_STATE | EXTRA_STATE */
 81 #endif
 82 #define EXTRA_STATE   69    /* gzip extra block -> NAME_STATE */
 83 #define NAME_STATE    73    /* gzip file name -> COMMENT_STATE */
 84 #define COMMENT_STATE 91    /* gzip comment -> HCRC_STATE */
 85 #define HCRC_STATE   103    /* gzip header CRC -> BUSY_STATE */
 86 #define BUSY_STATE   113    /* deflate -> FINISH_STATE */
 87 #define FINISH_STATE 666    /* stream complete */
 88 /* Stream status */
 89 
 90 
 91 /* Data structure describing a single value and its code string. */
 92 typedef struct ct_data_s {
 93     union {
 94         ush  freq;       /* frequency count */
 95         ush  code;       /* bit string */
 96     } fc;
 97     union {
 98         ush  dad;        /* father node in Huffman tree */
 99         ush  len;        /* length of bit string */
100     } dl;
101 } FAR ct_data;
102 
103 #define Freq fc.freq
104 #define Code fc.code
105 #define Dad  dl.dad
106 #define Len  dl.len
107 
108 typedef struct static_tree_desc_s  static_tree_desc;
109 
110 typedef struct tree_desc_s {
111     ct_data *dyn_tree;           /* the dynamic tree */
112     int     max_code;            /* largest code with non zero frequency */
113     const static_tree_desc *stat_desc;  /* the corresponding static tree */
114 } FAR tree_desc;
115 
116 typedef ush Pos;
117 typedef Pos FAR Posf;
118 typedef unsigned IPos;
119 
120 /* A Pos is an index in the character window. We use short instead of int to
121  * save space in the various tables. IPos is used only for parameter passing.
122  */
123 
124 typedef struct internal_state {
125     z_streamp strm;      /* pointer back to this zlib stream */
126     int   status;        /* as the name implies */
127     Bytef *pending_buf;  /* output still pending */
128     ulg   pending_buf_size; /* size of pending_buf */
129     Bytef *pending_out;  /* next pending byte to output to the stream */
130     ulg   pending;       /* nb of bytes in the pending buffer */
131     int   wrap;          /* bit 0 true for zlib, bit 1 true for gzip */
132     gz_headerp  gzhead;  /* gzip header information to write */
133     ulg   gzindex;       /* where in extra, name, or comment */
134     Byte  method;        /* can only be DEFLATED */
135     int   last_flush;    /* value of flush param for previous deflate call */
136 
137                 /* used by deflate.c: */
138 
139     uInt  w_size;        /* LZ77 window size (32K by default) */
140     uInt  w_bits;        /* log2(w_size)  (8..16) */
141     uInt  w_mask;        /* w_size - 1 */
142 
143     Bytef *window;
144     /* Sliding window. Input bytes are read into the second half of the window,
145      * and move to the first half later to keep a dictionary of at least wSize
146      * bytes. With this organization, matches are limited to a distance of
147      * wSize-MAX_MATCH bytes, but this ensures that IO is always
148      * performed with a length multiple of the block size. Also, it limits
149      * the window size to 64K, which is quite useful on MSDOS.
150      * To do: use the user input buffer as sliding window.
151      */
152 
153     ulg window_size;
154     /* Actual size of window: 2*wSize, except when the user input buffer
155      * is directly used as sliding window.
156      */
157 
158     Posf *prev;
159     /* Link to older string with same hash index. To limit the size of this
160      * array to 64K, this link is maintained only for the last 32K strings.
161      * An index in this array is thus a window index modulo 32K.
162      */
163 
164     Posf *head; /* Heads of the hash chains or NIL. */
165 
166     uInt  ins_h;          /* hash index of string to be inserted */
167     uInt  hash_size;      /* number of elements in hash table */
168     uInt  hash_bits;      /* log2(hash_size) */
169     uInt  hash_mask;      /* hash_size-1 */
170 
171     uInt  hash_shift;
172     /* Number of bits by which ins_h must be shifted at each input
173      * step. It must be such that after MIN_MATCH steps, the oldest
174      * byte no longer takes part in the hash key, that is:
175      *   hash_shift * MIN_MATCH >= hash_bits
176      */
177 
178     long block_start;
179     /* Window position at the beginning of the current output block. Gets
180      * negative when the window is moved backwards.
181      */
182 
183     uInt match_length;           /* length of best match */
184     IPos prev_match;             /* previous match */
185     int match_available;         /* set if previous match exists */
186     uInt strstart;               /* start of string to insert */
187     uInt match_start;            /* start of matching string */
188     uInt lookahead;              /* number of valid bytes ahead in window */
189 
190     uInt prev_length;
191     /* Length of the best match at previous step. Matches not greater than this
192      * are discarded. This is used in the lazy match evaluation.
193      */
194 
195     uInt max_chain_length;
196     /* To speed up deflation, hash chains are never searched beyond this
197      * length.  A higher limit improves compression ratio but degrades the
198      * speed.
199      */
200 
201     uInt max_lazy_match;
202     /* Attempt to find a better match only when the current match is strictly
203      * smaller than this value. This mechanism is used only for compression
204      * levels >= 4.
205      */
206 #   define max_insert_length  max_lazy_match
207     /* Insert new strings in the hash table only if the match length is not
208      * greater than this length. This saves time but degrades compression.
209      * max_insert_length is used only for compression levels <= 3.
210      */
211 
212     int level;    /* compression level (1..9) */
213     int strategy; /* favor or force Huffman coding*/
214 
215     uInt good_match;
216     /* Use a faster search when the previous match is longer than this */
217 
218     int nice_match; /* Stop searching when current match exceeds this */
219 
220                 /* used by trees.c: */
221     /* Didn't use ct_data typedef below to suppress compiler warning */
222     struct ct_data_s dyn_ltree[HEAP_SIZE];   /* literal and length tree */
223     struct ct_data_s dyn_dtree[2*D_CODES+1]; /* distance tree */
224     struct ct_data_s bl_tree[2*BL_CODES+1];  /* Huffman tree for bit lengths */
225 
226     struct tree_desc_s l_desc;               /* desc. for literal tree */
227     struct tree_desc_s d_desc;               /* desc. for distance tree */
228     struct tree_desc_s bl_desc;              /* desc. for bit length tree */
229 
230     ush bl_count[MAX_BITS+1];
231     /* number of codes at each bit length for an optimal tree */
232 
233     int heap[2*L_CODES+1];      /* heap used to build the Huffman trees */
234     int heap_len;               /* number of elements in the heap */
235     int heap_max;               /* element of largest frequency */
236     /* The sons of heap[n] are heap[2*n] and heap[2*n+1]. heap[0] is not used.
237      * The same heap array is used to build all trees.
238      */
239 
240     uch depth[2*L_CODES+1];
241     /* Depth of each subtree used as tie breaker for trees of equal frequency
242      */
243 
244     uchf *sym_buf;        /* buffer for distances and literals/lengths */
245 
246     uInt  lit_bufsize;
247     /* Size of match buffer for literals/lengths.  There are 4 reasons for
248      * limiting lit_bufsize to 64K:
249      *   - frequencies can be kept in 16 bit counters
250      *   - if compression is not successful for the first block, all input
251      *     data is still in the window so we can still emit a stored block even
252      *     when input comes from standard input.  (This can also be done for
253      *     all blocks if lit_bufsize is not greater than 32K.)
254      *   - if compression is not successful for a file smaller than 64K, we can
255      *     even emit a stored file instead of a stored block (saving 5 bytes).
256      *     This is applicable only for zip (not gzip or zlib).
257      *   - creating new Huffman trees less frequently may not provide fast
258      *     adaptation to changes in the input data statistics. (Take for
259      *     example a binary file with poorly compressible code followed by
260      *     a highly compressible string table.) Smaller buffer sizes give
261      *     fast adaptation but have of course the overhead of transmitting
262      *     trees more frequently.
263      *   - I can't count above 4
264      */
265 
266     uInt sym_next;      /* running index in sym_buf */
267     uInt sym_end;       /* symbol table full when sym_next reaches this */
268 
269     ulg opt_len;        /* bit length of current block with optimal trees */
270     ulg static_len;     /* bit length of current block with static trees */
271     uInt matches;       /* number of string matches in current block */
272     uInt insert;        /* bytes at end of window left to insert */
273 
274 #ifdef ZLIB_DEBUG
275     ulg compressed_len; /* total bit length of compressed file mod 2^32 */
276     ulg bits_sent;      /* bit length of compressed data sent mod 2^32 */
277 #endif
278 
279     ush bi_buf;
280     /* Output buffer. bits are inserted starting at the bottom (least
281      * significant bits).
282      */
283     int bi_valid;
284     /* Number of valid bits in bi_buf.  All bits above the last valid bit
285      * are always zero.
286      */
287 
288     ulg high_water;
289     /* High water mark offset in window for initialized bytes -- bytes above
290      * this are set to zero in order to avoid memory check warnings when
291      * longest match routines access bytes past the input.  This is then
292      * updated to the new high water mark.
293      */
294 
295 } FAR deflate_state;
296 
297 /* Output a byte on the stream.
298  * IN assertion: there is enough room in pending_buf.
299  */
300 #define put_byte(s, c) {s->pending_buf[s->pending++] = (Bytef)(c);}
301 
302 
303 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
304 /* Minimum amount of lookahead, except at the end of the input file.
305  * See deflate.c for comments about the MIN_MATCH+1.
306  */
307 
308 #define MAX_DIST(s)  ((s)->w_size-MIN_LOOKAHEAD)
309 /* In order to simplify the code, particularly on 16 bit machines, match
310  * distances are limited to MAX_DIST instead of WSIZE.
311  */
312 
313 #define WIN_INIT MAX_MATCH
314 /* Number of bytes after end of data in window to initialize in order to avoid
315    memory checker errors from longest match routines */
316 
317         /* in trees.c */
318 void ZLIB_INTERNAL _tr_init OF((deflate_state *s));
319 int ZLIB_INTERNAL _tr_tally OF((deflate_state *s, unsigned dist, unsigned lc));
320 void ZLIB_INTERNAL _tr_flush_block OF((deflate_state *s, charf *buf,
321                         ulg stored_len, int last));
322 void ZLIB_INTERNAL _tr_flush_bits OF((deflate_state *s));
323 void ZLIB_INTERNAL _tr_align OF((deflate_state *s));
324 void ZLIB_INTERNAL _tr_stored_block OF((deflate_state *s, charf *buf,
325                         ulg stored_len, int last));
326 
327 #define d_code(dist) \
328    ((dist) < 256 ? _dist_code[dist] : _dist_code[256+((dist)>>7)])
329 /* Mapping from a distance to a distance code. dist is the distance - 1 and
330  * must not have side effects. _dist_code[256] and _dist_code[257] are never
331  * used.
332  */
333 
334 #ifndef ZLIB_DEBUG
335 /* Inline versions of _tr_tally for speed: */
336 
337 #if defined(GEN_TREES_H) || !defined(STDC)
338   extern uch ZLIB_INTERNAL _length_code[];
339   extern uch ZLIB_INTERNAL _dist_code[];
340 #else
341   extern const uch ZLIB_INTERNAL _length_code[];
342   extern const uch ZLIB_INTERNAL _dist_code[];
343 #endif
344 
345 # define _tr_tally_lit(s, c, flush) \
346   { uch cc = (c); \
347     s->sym_buf[s->sym_next++] = 0; \
348     s->sym_buf[s->sym_next++] = 0; \
349     s->sym_buf[s->sym_next++] = cc; \
350     s->dyn_ltree[cc].Freq++; \
351     flush = (s->sym_next == s->sym_end); \
352    }
353 # define _tr_tally_dist(s, distance, length, flush) \
354   { uch len = (uch)(length); \
355     ush dist = (ush)(distance); \
356     s->sym_buf[s->sym_next++] = (uch)dist; \
357     s->sym_buf[s->sym_next++] = (uch)(dist >> 8); \
358     s->sym_buf[s->sym_next++] = len; \
359     dist--; \
360     s->dyn_ltree[_length_code[len]+LITERALS+1].Freq++; \
361     s->dyn_dtree[d_code(dist)].Freq++; \
362     flush = (s->sym_next == s->sym_end); \
363   }
364 #else
365 # define _tr_tally_lit(s, c, flush) flush = _tr_tally(s, 0, c)
366 # define _tr_tally_dist(s, distance, length, flush) \
367               flush = _tr_tally(s, distance, length)
368 #endif
369 
370 #endif /* DEFLATE_H */