Annotation of src/usr.bin/compress/zopen.c, Revision 1.9
1.9 ! mickey 1: /* $OpenBSD: zopen.c,v 1.8 2002/02/16 21:27:45 millert Exp $ */
1.1 deraadt 2: /* $NetBSD: zopen.c,v 1.5 1995/03/26 09:44:53 glass Exp $ */
3:
4: /*-
5: * Copyright (c) 1985, 1986, 1992, 1993
6: * The Regents of the University of California. All rights reserved.
7: *
8: * This code is derived from software contributed to Berkeley by
9: * Diomidis Spinellis and James A. Woods, derived from original
10: * work by Spencer Thomas and Joseph Orost.
11: *
12: * Redistribution and use in source and binary forms, with or without
13: * modification, are permitted provided that the following conditions
14: * are met:
15: * 1. Redistributions of source code must retain the above copyright
16: * notice, this list of conditions and the following disclaimer.
17: * 2. Redistributions in binary form must reproduce the above copyright
18: * notice, this list of conditions and the following disclaimer in the
19: * documentation and/or other materials provided with the distribution.
20: * 3. All advertising materials mentioning features or use of this software
21: * must display the following acknowledgement:
22: * This product includes software developed by the University of
23: * California, Berkeley and its contributors.
24: * 4. Neither the name of the University nor the names of its contributors
25: * may be used to endorse or promote products derived from this software
26: * without specific prior written permission.
27: *
28: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38: * SUCH DAMAGE.
1.9 ! mickey 39: *
! 40: * From: @(#)zopen.c 8.1 (Berkeley) 6/27/93
1.1 deraadt 41: */
42:
43: #if 0
44: static char sccsid[] = "@(#)zopen.c 8.1 (Berkeley) 6/27/93";
45: #else
1.9 ! mickey 46: const char z_rcsid[] =
! 47: "$OpenBSD: zopen.c,v 1.8 2002/02/16 21:27:45 millert Exp $";
1.1 deraadt 48: #endif
49:
50: /*-
51: * fcompress.c - File compression ala IEEE Computer, June 1984.
52: *
53: * Compress authors:
54: * Spencer W. Thomas (decvax!utah-cs!thomas)
55: * Jim McKie (decvax!mcvax!jim)
56: * Steve Davies (decvax!vax135!petsd!peora!srd)
57: * Ken Turkowski (decvax!decwrl!turtlevax!ken)
58: * James A. Woods (decvax!ihnp4!ames!jaw)
59: * Joe Orost (decvax!vax135!petsd!joe)
60: *
61: * Cleaned up and converted to library returning I/O streams by
62: * Diomidis Spinellis <dds@doc.ic.ac.uk>.
63: *
64: * zopen(filename, mode, bits)
65: * Returns a FILE * that can be used for read or write. The modes
66: * supported are only "r" and "w". Seeking is not allowed. On
67: * reading the file is decompressed, on writing it is compressed.
68: * The output is compatible with compress(1) with 16 bit tables.
69: * Any file produced by compress(1) can be read.
70: */
71:
72: #include <sys/param.h>
73: #include <sys/stat.h>
74:
75: #include <ctype.h>
76: #include <errno.h>
77: #include <signal.h>
78: #include <stdio.h>
79: #include <stdlib.h>
80: #include <string.h>
81: #include <unistd.h>
1.5 mickey 82: #include <fcntl.h>
83: #include "compress.h"
1.1 deraadt 84:
85: #define BITS 16 /* Default bits. */
86: #define HSIZE 69001 /* 95% occupancy */
1.6 mickey 87: #define ZBUFSIZ 8192 /* I/O buffer size */
1.1 deraadt 88:
89: /* A code_int must be able to hold 2**BITS values of type int, and also -1. */
90: typedef long code_int;
91: typedef long count_int;
92:
1.9 ! mickey 93: static const u_char z_magic[] =
1.1 deraadt 94: {'\037', '\235'}; /* 1F 9D */
95:
96: #define BIT_MASK 0x1f /* Defines for third byte of header. */
97: #define BLOCK_MASK 0x80
98:
99: /*
100: * Masks 0x40 and 0x20 are free. I think 0x20 should mean that there is
101: * a fourth header byte (for expansion).
102: */
103: #define INIT_BITS 9 /* Initial number of bits/code. */
104:
105: #define MAXCODE(n_bits) ((1 << (n_bits)) - 1)
106:
107: struct s_zstate {
1.5 mickey 108: int zs_fd; /* File stream for I/O */
1.1 deraadt 109: char zs_mode; /* r or w */
110: enum {
111: S_START, S_MIDDLE, S_EOF
112: } zs_state; /* State of computation */
113: int zs_n_bits; /* Number of bits/code. */
114: int zs_maxbits; /* User settable max # bits/code. */
115: code_int zs_maxcode; /* Maximum code, given n_bits. */
116: code_int zs_maxmaxcode; /* Should NEVER generate this code. */
117: count_int zs_htab [HSIZE];
118: u_short zs_codetab [HSIZE];
119: code_int zs_hsize; /* For dynamic table sizing. */
120: code_int zs_free_ent; /* First unused entry. */
121: /*
122: * Block compression parameters -- after all codes are used up,
123: * and compression rate changes, start over.
124: */
125: int zs_block_compress;
126: int zs_clear_flg;
127: long zs_ratio;
128: count_int zs_checkpoint;
129: long zs_in_count; /* Length of input. */
130: long zs_bytes_out; /* Length of compressed output. */
1.5 mickey 131: long zs_out_count; /* # of codes output (for debugging).*/
1.6 mickey 132: u_char zs_buf[ZBUFSIZ]; /* I/O buffer */
133: u_char *zs_bp; /* Current I/O window in the zs_buf */
134: int zs_offset; /* Number of bits in the zs_buf */
1.1 deraadt 135: union {
136: struct {
137: long zs_fcode;
138: code_int zs_ent;
139: code_int zs_hsize_reg;
140: int zs_hshift;
141: } w; /* Write paramenters */
142: struct {
1.6 mickey 143: u_char *zs_stackp, *zs_ebp;
1.1 deraadt 144: int zs_finchar;
145: code_int zs_code, zs_oldcode, zs_incode;
1.6 mickey 146: int zs_size;
1.1 deraadt 147: } r; /* Read parameters */
148: } u;
149: };
150:
151: /* Definitions to retain old variable names */
1.5 mickey 152: #define zs_fcode u.w.zs_fcode
153: #define zs_ent u.w.zs_ent
154: #define zs_hsize_reg u.w.zs_hsize_reg
155: #define zs_hshift u.w.zs_hshift
156: #define zs_stackp u.r.zs_stackp
157: #define zs_finchar u.r.zs_finchar
158: #define zs_code u.r.zs_code
159: #define zs_oldcode u.r.zs_oldcode
160: #define zs_incode u.r.zs_incode
161: #define zs_size u.r.zs_size
1.6 mickey 162: #define zs_ebp u.r.zs_ebp
1.1 deraadt 163:
164: /*
165: * To save much memory, we overlay the table used by compress() with those
166: * used by decompress(). The tab_prefix table is the same size and type as
167: * the codetab. The tab_suffix table needs 2**BITS characters. We get this
168: * from the beginning of htab. The output stack uses the rest of htab, and
169: * contains characters. There is plenty of room for any possible stack
170: * (stack used to be 8000 characters).
171: */
172:
1.5 mickey 173: #define htabof(i) zs->zs_htab[i]
174: #define codetabof(i) zs->zs_codetab[i]
1.1 deraadt 175:
176: #define tab_prefixof(i) codetabof(i)
1.5 mickey 177: #define tab_suffixof(i) ((u_char *)(zs->zs_htab))[i]
178: #define de_stack ((u_char *)&tab_suffixof(1 << BITS))
1.1 deraadt 179:
180: #define CHECK_GAP 10000 /* Ratio check interval. */
181:
182: /*
183: * the next two codes should not be changed lightly, as they must not
184: * lie within the contiguous general code space.
185: */
186: #define FIRST 257 /* First free entry. */
187: #define CLEAR 256 /* Table clear output code. */
188:
1.8 millert 189: static int cl_block(struct s_zstate *);
1.9 ! mickey 190: static void cl_hash(struct s_zstate *, count_int);
1.8 millert 191: static code_int getcode(struct s_zstate *);
192: static int output(struct s_zstate *, code_int);
1.1 deraadt 193:
194: /*-
195: * Algorithm from "A Technique for High Performance Data Compression",
196: * Terry A. Welch, IEEE Computer Vol 17, No 6 (June 1984), pp 8-19.
197: *
198: * Algorithm:
1.9 ! mickey 199: * Modified Lempel-Ziv method (LZW). Basically finds common
1.1 deraadt 200: * substrings and replaces them with a variable size code. This is
201: * deterministic, and can be done on the fly. Thus, the decompression
202: * procedure needs no input table, but tracks the way the table was built.
203: */
204:
205: /*-
206: * compress write
207: *
208: * Algorithm: use open addressing double hashing (no chaining) on the
209: * prefix code / next character combination. We do a variant of Knuth's
210: * algorithm D (vol. 3, sec. 6.4) along with G. Knott's relatively-prime
211: * secondary probe. Here, the modular division first probe is gives way
212: * to a faster exclusive-or manipulation. Also do block compression with
213: * an adaptive reset, whereby the code table is cleared when the compression
214: * ratio decreases, but after the table fills. The variable-length output
215: * codes are re-sized at this point, and a special CLEAR code is generated
216: * for the decompressor. Late addition: construct the table according to
217: * file size for noticeable speed improvement on small files. Please direct
218: * questions about this implementation to ames!jaw.
219: */
1.5 mickey 220: int
1.1 deraadt 221: zwrite(cookie, wbp, num)
222: void *cookie;
223: const char *wbp;
224: int num;
225: {
1.7 mpech 226: code_int i;
227: int c, disp;
1.1 deraadt 228: struct s_zstate *zs;
229: const u_char *bp;
230: u_char tmp;
231: int count;
232:
233: zs = cookie;
234: count = num;
235: bp = (u_char *)wbp;
1.6 mickey 236: switch (zs->zs_state) {
237: case S_EOF:
238: return 0;
239: case S_START:
240: zs->zs_state = S_MIDDLE;
1.1 deraadt 241:
1.6 mickey 242: zs->zs_maxmaxcode = 1L << zs->zs_maxbits;
243: if (write(zs->zs_fd, z_magic, sizeof(z_magic)) !=
244: sizeof(z_magic))
245: return (-1);
246: tmp = (u_char)(zs->zs_maxbits | zs->zs_block_compress);
247: if (write(zs->zs_fd, &tmp, sizeof(tmp)) != sizeof(tmp))
248: return (-1);
1.1 deraadt 249:
1.6 mickey 250: zs->zs_bp = zs->zs_buf;
251: zs->zs_offset = 0;
252: zs->zs_bytes_out = 3; /* Includes 3-byte header mojo. */
253: zs->zs_out_count = 0;
254: zs->zs_clear_flg = 0;
255: zs->zs_ratio = 0;
256: zs->zs_in_count = 1;
257: zs->zs_checkpoint = CHECK_GAP;
258: zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
259: zs->zs_free_ent = ((zs->zs_block_compress) ? FIRST : 256);
260:
261: zs->zs_ent = *bp++;
262: --count;
263:
264: zs->zs_hshift = 0;
265: for (zs->zs_fcode = (long)zs->zs_hsize; zs->zs_fcode < 65536L;
266: zs->zs_fcode *= 2L)
267: zs->zs_hshift++;
268: /* Set hash code range bound. */
269: zs->zs_hshift = 8 - zs->zs_hshift;
270:
271: zs->zs_hsize_reg = zs->zs_hsize;
272: /* Clear hash table. */
273: cl_hash(zs, (count_int)zs->zs_hsize_reg);
1.1 deraadt 274:
1.6 mickey 275: case S_MIDDLE:
276: for (i = 0; count-- > 0;) {
277: c = *bp++;
278: zs->zs_in_count++;
279: zs->zs_fcode = (long)(((long)c << zs->zs_maxbits) +
280: zs->zs_ent);
281: /* Xor hashing. */
282: i = ((c << zs->zs_hshift) ^ zs->zs_ent);
1.9 ! mickey 283:
1.6 mickey 284: if (htabof(i) == zs->zs_fcode) {
285: zs->zs_ent = codetabof(i);
286: continue;
287: } else if ((long)htabof(i) < 0) /* Empty slot. */
288: goto nomatch;
289: /* Secondary hash (after G. Knott). */
290: disp = zs->zs_hsize_reg - i;
291: if (i == 0)
1.1 deraadt 292: disp = 1;
1.6 mickey 293: probe: if ((i -= disp) < 0)
294: i += zs->zs_hsize_reg;
1.1 deraadt 295:
1.6 mickey 296: if (htabof(i) == zs->zs_fcode) {
297: zs->zs_ent = codetabof(i);
298: continue;
299: }
300: if ((long)htabof(i) >= 0)
301: goto probe;
302: nomatch: if (output(zs, (code_int) zs->zs_ent) == -1)
1.1 deraadt 303: return (-1);
1.6 mickey 304: zs->zs_out_count++;
305: zs->zs_ent = c;
306: if (zs->zs_free_ent < zs->zs_maxmaxcode) {
307: /* code -> hashtable */
308: codetabof(i) = zs->zs_free_ent++;
309: htabof(i) = zs->zs_fcode;
310: } else if ((count_int)zs->zs_in_count >=
311: zs->zs_checkpoint && zs->zs_block_compress) {
312: if (cl_block(zs) == -1)
313: return (-1);
314: }
1.1 deraadt 315: }
316: }
317: return (num);
318: }
319:
1.5 mickey 320: int
1.1 deraadt 321: zclose(cookie)
322: void *cookie;
323: {
324: struct s_zstate *zs;
325: int rval;
326:
327: zs = cookie;
1.5 mickey 328: if (zs->zs_mode == 'w') { /* Put out the final code. */
329: if (output(zs, (code_int) zs->zs_ent) == -1) {
330: (void)close(zs->zs_fd);
1.1 deraadt 331: free(zs);
332: return (-1);
333: }
1.5 mickey 334: zs->zs_out_count++;
1.1 deraadt 335: if (output(zs, (code_int) - 1) == -1) {
1.5 mickey 336: (void)close(zs->zs_fd);
1.1 deraadt 337: free(zs);
338: return (-1);
339: }
340: }
1.5 mickey 341: rval = close(zs->zs_fd);
1.1 deraadt 342: free(zs);
343: return (rval);
344: }
345:
346: /*-
347: * Output the given code.
348: * Inputs:
1.9 ! mickey 349: * code: A n_bits-bit integer. If == -1, then EOF. This assumes
1.1 deraadt 350: * that n_bits =< (long)wordsize - 1.
351: * Outputs:
1.9 ! mickey 352: * Outputs code to the file.
1.1 deraadt 353: * Assumptions:
354: * Chars are 8 bits long.
355: * Algorithm:
1.9 ! mickey 356: * Maintain a BITS character long buffer (so that 8 codes will
1.1 deraadt 357: * fit in it exactly). Use the VAX insv instruction to insert each
358: * code in turn. When the buffer fills up empty it and start over.
359: */
360:
1.6 mickey 361: static const u_char lmask[9] =
1.1 deraadt 362: {0xff, 0xfe, 0xfc, 0xf8, 0xf0, 0xe0, 0xc0, 0x80, 0x00};
1.6 mickey 363: static const u_char rmask[9] =
1.1 deraadt 364: {0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff};
365:
366: static int
367: output(zs, ocode)
1.7 mpech 368: struct s_zstate *zs;
1.1 deraadt 369: code_int ocode;
370: {
1.7 mpech 371: int bits;
1.1 deraadt 372:
373: if (ocode >= 0) {
1.7 mpech 374: int r_off;
375: u_char *bp;
1.6 mickey 376:
1.1 deraadt 377: /* Get to the first byte. */
1.6 mickey 378: bp = zs->zs_bp + (zs->zs_offset >> 3);
379: r_off = zs->zs_offset & 7;
380: bits = zs->zs_n_bits;
381:
1.1 deraadt 382: /*
383: * Since ocode is always >= 8 bits, only need to mask the first
384: * hunk on the left.
385: */
1.4 millert 386: *bp = (*bp & rmask[r_off]) | ((ocode << r_off) & lmask[r_off]);
1.1 deraadt 387: bp++;
388: bits -= (8 - r_off);
389: ocode >>= 8 - r_off;
1.5 mickey 390: /* Get any 8 bit parts in the middle (<=1 for up to 16 bits) */
1.1 deraadt 391: if (bits >= 8) {
392: *bp++ = ocode;
393: ocode >>= 8;
394: bits -= 8;
395: }
396: /* Last bits. */
397: if (bits)
398: *bp = ocode;
1.5 mickey 399: zs->zs_offset += zs->zs_n_bits;
400: if (zs->zs_offset == (zs->zs_n_bits << 3)) {
1.6 mickey 401: zs->zs_bp += zs->zs_n_bits;
1.5 mickey 402: zs->zs_offset = 0;
1.1 deraadt 403: }
404: /*
405: * If the next entry is going to be too big for the ocode size,
406: * then increase it, if possible.
407: */
1.5 mickey 408: if (zs->zs_free_ent > zs->zs_maxcode ||
409: (zs->zs_clear_flg > 0)) {
1.1 deraadt 410: /*
411: * Write the whole buffer, because the input side won't
1.5 mickey 412: * discover the size increase until after it has read it
1.1 deraadt 413: */
1.5 mickey 414: if (zs->zs_offset > 0) {
1.6 mickey 415: zs->zs_bp += zs->zs_n_bits;
416: zs->zs_offset = 0;
1.1 deraadt 417: }
418:
1.5 mickey 419: if (zs->zs_clear_flg) {
420: zs->zs_maxcode =
421: MAXCODE(zs->zs_n_bits = INIT_BITS);
422: zs->zs_clear_flg = 0;
1.1 deraadt 423: } else {
1.5 mickey 424: zs->zs_n_bits++;
425: if (zs->zs_n_bits == zs->zs_maxbits)
426: zs->zs_maxcode = zs->zs_maxmaxcode;
1.1 deraadt 427: else
1.5 mickey 428: zs->zs_maxcode =
429: MAXCODE(zs->zs_n_bits);
1.1 deraadt 430: }
431: }
1.6 mickey 432:
433: if (zs->zs_bp + zs->zs_n_bits > &zs->zs_buf[ZBUFSIZ]) {
434: bits = zs->zs_bp - zs->zs_buf;
435: if (write(zs->zs_fd, zs->zs_buf, bits) != bits)
436: return (-1);
437: zs->zs_bytes_out += bits;
438: if (zs->zs_offset > 0)
439: fprintf (stderr, "zs_offset != 0\n");
440: zs->zs_bp = zs->zs_buf;
441: }
1.1 deraadt 442: } else {
443: /* At EOF, write the rest of the buffer. */
1.6 mickey 444: if (zs->zs_offset > 0)
445: zs->zs_bp += (zs->zs_offset + 7) / 8;
446: if (zs->zs_bp > zs->zs_buf) {
447: bits = zs->zs_bp - zs->zs_buf;
448: if (write(zs->zs_fd, zs->zs_buf, bits) != bits)
1.1 deraadt 449: return (-1);
1.6 mickey 450: zs->zs_bytes_out += bits;
1.1 deraadt 451: }
1.5 mickey 452: zs->zs_offset = 0;
1.6 mickey 453: zs->zs_bp = zs->zs_buf;
1.1 deraadt 454: }
455: return (0);
456: }
457:
458: /*
459: * Decompress read. This routine adapts to the codes in the file building
460: * the "string" table on-the-fly; requiring no table to be stored in the
461: * compressed file. The tables used herein are shared with those of the
462: * compress() routine. See the definitions above.
463: */
1.5 mickey 464: int
1.1 deraadt 465: zread(cookie, rbp, num)
466: void *cookie;
467: char *rbp;
468: int num;
469: {
1.7 mpech 470: u_int count;
1.1 deraadt 471: struct s_zstate *zs;
472: u_char *bp, header[3];
473:
474: if (num == 0)
475: return (0);
476:
477: zs = cookie;
478: count = num;
479: bp = (u_char *)rbp;
1.5 mickey 480: switch (zs->zs_state) {
1.1 deraadt 481: case S_START:
1.5 mickey 482: zs->zs_state = S_MIDDLE;
1.6 mickey 483: zs->zs_bp = zs->zs_buf;
1.1 deraadt 484: break;
485: case S_MIDDLE:
486: goto middle;
487: case S_EOF:
488: goto eof;
489: }
490:
491: /* Check the magic number */
1.5 mickey 492: if (read(zs->zs_fd, header, sizeof(header)) != sizeof(header) ||
493: memcmp(header, z_magic, sizeof(z_magic)) != 0) {
1.1 deraadt 494: errno = EFTYPE;
495: return (-1);
496: }
1.5 mickey 497: zs->zs_maxbits = header[2]; /* Set -b from file. */
498: zs->zs_block_compress = zs->zs_maxbits & BLOCK_MASK;
499: zs->zs_maxbits &= BIT_MASK;
500: zs->zs_maxmaxcode = 1L << zs->zs_maxbits;
501: if (zs->zs_maxbits > BITS) {
1.1 deraadt 502: errno = EFTYPE;
503: return (-1);
504: }
505: /* As above, initialize the first 256 entries in the table. */
1.5 mickey 506: zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
507: for (zs->zs_code = 255; zs->zs_code >= 0; zs->zs_code--) {
508: tab_prefixof(zs->zs_code) = 0;
509: tab_suffixof(zs->zs_code) = (u_char) zs->zs_code;
1.1 deraadt 510: }
1.5 mickey 511: zs->zs_free_ent = zs->zs_block_compress ? FIRST : 256;
1.1 deraadt 512:
1.5 mickey 513: zs->zs_finchar = zs->zs_oldcode = getcode(zs);
514: if (zs->zs_oldcode == -1) /* EOF already? */
1.1 deraadt 515: return (0); /* Get out of here */
516:
517: /* First code must be 8 bits = char. */
1.5 mickey 518: *bp++ = (u_char)zs->zs_finchar;
1.1 deraadt 519: count--;
1.5 mickey 520: zs->zs_stackp = de_stack;
1.1 deraadt 521:
1.5 mickey 522: while ((zs->zs_code = getcode(zs)) > -1) {
1.1 deraadt 523:
1.5 mickey 524: if ((zs->zs_code == CLEAR) && zs->zs_block_compress) {
525: for (zs->zs_code = 255; zs->zs_code >= 0;
526: zs->zs_code--)
527: tab_prefixof(zs->zs_code) = 0;
528: zs->zs_clear_flg = 1;
529: zs->zs_free_ent = FIRST - 1;
530: if ((zs->zs_code = getcode(zs)) == -1) /* O, untimely death! */
1.1 deraadt 531: break;
532: }
1.5 mickey 533: zs->zs_incode = zs->zs_code;
1.1 deraadt 534:
535: /* Special case for KwKwK string. */
1.5 mickey 536: if (zs->zs_code >= zs->zs_free_ent) {
537: *zs->zs_stackp++ = zs->zs_finchar;
538: zs->zs_code = zs->zs_oldcode;
1.1 deraadt 539: }
540:
541: /* Generate output characters in reverse order. */
1.5 mickey 542: while (zs->zs_code >= 256) {
543: *zs->zs_stackp++ = tab_suffixof(zs->zs_code);
544: zs->zs_code = tab_prefixof(zs->zs_code);
1.1 deraadt 545: }
1.5 mickey 546: *zs->zs_stackp++ = zs->zs_finchar = tab_suffixof(zs->zs_code);
1.1 deraadt 547:
548: /* And put them out in forward order. */
549: middle: do {
550: if (count-- == 0)
551: return (num);
1.5 mickey 552: *bp++ = *--zs->zs_stackp;
553: } while (zs->zs_stackp > de_stack);
1.1 deraadt 554:
555: /* Generate the new entry. */
1.5 mickey 556: if ((zs->zs_code = zs->zs_free_ent) < zs->zs_maxmaxcode) {
557: tab_prefixof(zs->zs_code) = (u_short) zs->zs_oldcode;
558: tab_suffixof(zs->zs_code) = zs->zs_finchar;
559: zs->zs_free_ent = zs->zs_code + 1;
1.1 deraadt 560: }
561:
562: /* Remember previous code. */
1.5 mickey 563: zs->zs_oldcode = zs->zs_incode;
1.1 deraadt 564: }
1.5 mickey 565: zs->zs_state = S_EOF;
1.1 deraadt 566: eof: return (num - count);
567: }
568:
569: /*-
570: * Read one code from the standard input. If EOF, return -1.
571: * Inputs:
1.9 ! mickey 572: * stdin
1.1 deraadt 573: * Outputs:
1.9 ! mickey 574: * code or -1 is returned.
1.1 deraadt 575: */
576: static code_int
577: getcode(zs)
1.7 mpech 578: struct s_zstate *zs;
1.1 deraadt 579: {
1.7 mpech 580: code_int gcode;
581: int r_off, bits;
582: u_char *bp;
1.1 deraadt 583:
1.6 mickey 584: if (zs->zs_clear_flg > 0 || zs->zs_offset >= zs->zs_size ||
1.5 mickey 585: zs->zs_free_ent > zs->zs_maxcode) {
1.6 mickey 586:
587: zs->zs_bp += zs->zs_n_bits;
1.1 deraadt 588: /*
589: * If the next entry will be too big for the current gcode
590: * size, then we must increase the size. This implies reading
591: * a new buffer full, too.
592: */
1.5 mickey 593: if (zs->zs_free_ent > zs->zs_maxcode) {
594: zs->zs_n_bits++;
595: if (zs->zs_n_bits == zs->zs_maxbits) /* Won't get any bigger now. */
596: zs->zs_maxcode = zs->zs_maxmaxcode;
1.1 deraadt 597: else
1.5 mickey 598: zs->zs_maxcode = MAXCODE(zs->zs_n_bits);
1.1 deraadt 599: }
1.5 mickey 600: if (zs->zs_clear_flg > 0) {
601: zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
602: zs->zs_clear_flg = 0;
1.1 deraadt 603: }
1.6 mickey 604:
605: /* fill the buffer up to the neck */
606: if (zs->zs_bp + zs->zs_n_bits > zs->zs_ebp) {
607: for (bp = zs->zs_buf; zs->zs_bp < zs->zs_ebp;
608: *bp++ = *zs->zs_bp++);
609: if ((bits = read(zs->zs_fd, bp, ZBUFSIZ -
610: (bp - zs->zs_buf))) < 0)
611: return -1;
612: zs->zs_bp = zs->zs_buf;
613: zs->zs_ebp = bp + bits;
614: }
615: zs->zs_offset = 0;
616: zs->zs_size = MIN(zs->zs_n_bits, zs->zs_ebp - zs->zs_bp);
617: if (zs->zs_size == 0)
618: return -1;
1.1 deraadt 619: /* Round size down to integral number of codes. */
1.5 mickey 620: zs->zs_size = (zs->zs_size << 3) - (zs->zs_n_bits - 1);
1.1 deraadt 621: }
1.6 mickey 622:
623: bp = zs->zs_bp;
624: r_off = zs->zs_offset;
1.5 mickey 625: bits = zs->zs_n_bits;
1.1 deraadt 626:
627: /* Get to the first byte. */
628: bp += (r_off >> 3);
629: r_off &= 7;
630:
631: /* Get first part (low order bits). */
632: gcode = (*bp++ >> r_off);
633: bits -= (8 - r_off);
634: r_off = 8 - r_off; /* Now, roffset into gcode word. */
635:
636: /* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
637: if (bits >= 8) {
638: gcode |= *bp++ << r_off;
639: r_off += 8;
640: bits -= 8;
641: }
642:
643: /* High order bits. */
644: gcode |= (*bp & rmask[bits]) << r_off;
1.6 mickey 645: zs->zs_offset += zs->zs_n_bits;
1.1 deraadt 646:
647: return (gcode);
648: }
649:
650: static int
651: cl_block(zs) /* Table clear for block compress. */
1.7 mpech 652: struct s_zstate *zs;
1.1 deraadt 653: {
1.7 mpech 654: long rat;
1.1 deraadt 655:
1.5 mickey 656: zs->zs_checkpoint = zs->zs_in_count + CHECK_GAP;
1.1 deraadt 657:
1.5 mickey 658: if (zs->zs_in_count > 0x007fffff) { /* Shift will overflow. */
659: rat = zs->zs_bytes_out >> 8;
1.1 deraadt 660: if (rat == 0) /* Don't divide by zero. */
661: rat = 0x7fffffff;
662: else
1.5 mickey 663: rat = zs->zs_in_count / rat;
1.1 deraadt 664: } else
1.5 mickey 665: rat = (zs->zs_in_count << 8) / zs->zs_bytes_out; /* 8 fractional bits. */
666: if (rat > zs->zs_ratio)
667: zs->zs_ratio = rat;
1.1 deraadt 668: else {
1.5 mickey 669: zs->zs_ratio = 0;
670: cl_hash(zs, (count_int) zs->zs_hsize);
671: zs->zs_free_ent = FIRST;
672: zs->zs_clear_flg = 1;
1.1 deraadt 673: if (output(zs, (code_int) CLEAR) == -1)
674: return (-1);
675: }
676: return (0);
677: }
678:
679: static void
680: cl_hash(zs, cl_hsize) /* Reset code table. */
1.7 mpech 681: struct s_zstate *zs;
682: count_int cl_hsize;
1.1 deraadt 683: {
1.7 mpech 684: count_int *htab_p;
685: long i, m1;
1.1 deraadt 686:
687: m1 = -1;
1.5 mickey 688: htab_p = zs->zs_htab + cl_hsize;
1.1 deraadt 689: i = cl_hsize - 16;
690: do { /* Might use Sys V memset(3) here. */
691: *(htab_p - 16) = m1;
692: *(htab_p - 15) = m1;
693: *(htab_p - 14) = m1;
694: *(htab_p - 13) = m1;
695: *(htab_p - 12) = m1;
696: *(htab_p - 11) = m1;
697: *(htab_p - 10) = m1;
698: *(htab_p - 9) = m1;
699: *(htab_p - 8) = m1;
700: *(htab_p - 7) = m1;
701: *(htab_p - 6) = m1;
702: *(htab_p - 5) = m1;
703: *(htab_p - 4) = m1;
704: *(htab_p - 3) = m1;
705: *(htab_p - 2) = m1;
706: *(htab_p - 1) = m1;
707: htab_p -= 16;
708: } while ((i -= 16) >= 0);
709: for (i += 16; i > 0; i--)
710: *--htab_p = m1;
1.3 tholo 711: }
712:
713: FILE *
1.5 mickey 714: zopen(name, mode, bits)
715: const char *name;
716: const char *mode;
717: int bits;
718: {
719: int fd;
720: void *cookie;
721: if ((fd = open(name, (*mode=='r'? O_RDONLY:O_WRONLY|O_CREAT),
722: S_IRUSR|S_IWUSR|S_IRGRP|S_IROTH)) == -1)
723: return NULL;
724: if ((cookie = z_open(fd, mode, bits)) == NULL) {
725: close(fd);
726: return NULL;
727: }
728: return funopen(cookie, (*mode == 'r'?zread:NULL),
729: (*mode == 'w'?zwrite:NULL), NULL, zclose);
730: }
731:
732: void *
733: z_open(fd, mode, bits)
1.3 tholo 734: int fd;
735: const char *mode;
736: int bits;
737: {
1.7 mpech 738: struct s_zstate *zs;
1.3 tholo 739:
1.4 millert 740: if ((mode[0] != 'r' && mode[0] != 'w') || mode[1] != '\0' ||
1.3 tholo 741: bits < 0 || bits > BITS) {
742: errno = EINVAL;
743: return (NULL);
744: }
745:
746: if ((zs = calloc(1, sizeof(struct s_zstate))) == NULL)
747: return (NULL);
748:
1.5 mickey 749: /* User settable max # bits/code. */
750: zs->zs_maxbits = bits ? bits : BITS;
751: /* Should NEVER generate this code. */
752: zs->zs_maxmaxcode = 1 << zs->zs_maxbits;
753: zs->zs_hsize = HSIZE; /* For dynamic table sizing. */
754: zs->zs_free_ent = 0; /* First unused entry. */
755: zs->zs_block_compress = BLOCK_MASK;
756: zs->zs_clear_flg = 0;
757: zs->zs_ratio = 0;
758: zs->zs_checkpoint = CHECK_GAP;
759: zs->zs_in_count = 1; /* Length of input. */
760: zs->zs_out_count = 0; /* # of codes output (for debugging).*/
761: zs->zs_state = S_START;
1.6 mickey 762: zs->zs_offset = 0;
1.5 mickey 763: zs->zs_size = 0;
764: zs->zs_mode = mode[0];
1.6 mickey 765: zs->zs_bp = zs->zs_ebp = zs->zs_buf;
1.3 tholo 766:
1.5 mickey 767: zs->zs_fd = fd;
768: return zs;
1.1 deraadt 769: }
770:
1.5 mickey 771: int
772: z_check_header(fd, sb, ofn)
773: int fd;
774: struct stat *sb;
775: const char *ofn;
1.1 deraadt 776: {
1.5 mickey 777: int f;
778: u_char buf[sizeof(z_magic)];
779: off_t off = lseek(fd, 0, SEEK_CUR);
1.1 deraadt 780:
1.5 mickey 781: f = (read(fd, buf, sizeof(buf)) == sizeof(buf) &&
782: !memcmp(buf, z_magic, sizeof(buf)));
1.1 deraadt 783:
1.5 mickey 784: lseek (fd, off, SEEK_SET);
1.1 deraadt 785:
1.5 mickey 786: return f;
1.1 deraadt 787: }