Annotation of src/usr.bin/compress/zopen.c, Revision 1.5
1.5 ! mickey 1: /* $OpenBSD: zopen.c,v 1.4 1997/01/19 17:25:14 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.
39: */
40:
41: #if defined(LIBC_SCCS) && !defined(lint)
42: #if 0
43: static char sccsid[] = "@(#)zopen.c 8.1 (Berkeley) 6/27/93";
44: #else
1.5 ! mickey 45: static char rcsid[] = "$OpenBSD: zopen.c,v 1.4 1997/01/19 17:25:14 millert Exp $";
1.1 deraadt 46: #endif
47: #endif /* LIBC_SCCS and not lint */
48:
49: /*-
50: * fcompress.c - File compression ala IEEE Computer, June 1984.
51: *
52: * Compress authors:
53: * Spencer W. Thomas (decvax!utah-cs!thomas)
54: * Jim McKie (decvax!mcvax!jim)
55: * Steve Davies (decvax!vax135!petsd!peora!srd)
56: * Ken Turkowski (decvax!decwrl!turtlevax!ken)
57: * James A. Woods (decvax!ihnp4!ames!jaw)
58: * Joe Orost (decvax!vax135!petsd!joe)
59: *
60: * Cleaned up and converted to library returning I/O streams by
61: * Diomidis Spinellis <dds@doc.ic.ac.uk>.
62: *
63: * zopen(filename, mode, bits)
64: * Returns a FILE * that can be used for read or write. The modes
65: * supported are only "r" and "w". Seeking is not allowed. On
66: * reading the file is decompressed, on writing it is compressed.
67: * The output is compatible with compress(1) with 16 bit tables.
68: * Any file produced by compress(1) can be read.
69: */
70:
71: #include <sys/param.h>
72: #include <sys/stat.h>
73:
74: #include <ctype.h>
75: #include <errno.h>
76: #include <signal.h>
77: #include <stdio.h>
78: #include <stdlib.h>
79: #include <string.h>
80: #include <unistd.h>
1.5 ! mickey 81: #include <fcntl.h>
! 82: #include "compress.h"
1.1 deraadt 83:
84: #define BITS 16 /* Default bits. */
85: #define HSIZE 69001 /* 95% occupancy */
86:
87: /* A code_int must be able to hold 2**BITS values of type int, and also -1. */
88: typedef long code_int;
89: typedef long count_int;
90:
1.5 ! mickey 91: static u_char z_magic[] =
1.1 deraadt 92: {'\037', '\235'}; /* 1F 9D */
93:
94: #define BIT_MASK 0x1f /* Defines for third byte of header. */
95: #define BLOCK_MASK 0x80
96:
97: /*
98: * Masks 0x40 and 0x20 are free. I think 0x20 should mean that there is
99: * a fourth header byte (for expansion).
100: */
101: #define INIT_BITS 9 /* Initial number of bits/code. */
102:
103: #define MAXCODE(n_bits) ((1 << (n_bits)) - 1)
104:
105: struct s_zstate {
1.5 ! mickey 106: int zs_fd; /* File stream for I/O */
1.1 deraadt 107: char zs_mode; /* r or w */
108: enum {
109: S_START, S_MIDDLE, S_EOF
110: } zs_state; /* State of computation */
111: int zs_n_bits; /* Number of bits/code. */
112: int zs_maxbits; /* User settable max # bits/code. */
113: code_int zs_maxcode; /* Maximum code, given n_bits. */
114: code_int zs_maxmaxcode; /* Should NEVER generate this code. */
115: count_int zs_htab [HSIZE];
116: u_short zs_codetab [HSIZE];
117: code_int zs_hsize; /* For dynamic table sizing. */
118: code_int zs_free_ent; /* First unused entry. */
119: /*
120: * Block compression parameters -- after all codes are used up,
121: * and compression rate changes, start over.
122: */
123: int zs_block_compress;
124: int zs_clear_flg;
125: long zs_ratio;
126: count_int zs_checkpoint;
127: int zs_offset;
128: long zs_in_count; /* Length of input. */
129: long zs_bytes_out; /* Length of compressed output. */
1.5 ! mickey 130: long zs_out_count; /* # of codes output (for debugging).*/
! 131: u_char zs_buf[BITS];
1.1 deraadt 132: union {
133: struct {
134: long zs_fcode;
135: code_int zs_ent;
136: code_int zs_hsize_reg;
137: int zs_hshift;
138: } w; /* Write paramenters */
139: struct {
1.5 ! mickey 140: u_char *zs_stackp;
1.1 deraadt 141: int zs_finchar;
142: code_int zs_code, zs_oldcode, zs_incode;
143: int zs_roffset, zs_size;
144: } r; /* Read parameters */
145: } u;
146: };
147:
148: /* Definitions to retain old variable names */
1.5 ! mickey 149: #define zs_fcode u.w.zs_fcode
! 150: #define zs_ent u.w.zs_ent
! 151: #define zs_hsize_reg u.w.zs_hsize_reg
! 152: #define zs_hshift u.w.zs_hshift
! 153: #define zs_stackp u.r.zs_stackp
! 154: #define zs_finchar u.r.zs_finchar
! 155: #define zs_code u.r.zs_code
! 156: #define zs_oldcode u.r.zs_oldcode
! 157: #define zs_incode u.r.zs_incode
! 158: #define zs_roffset u.r.zs_roffset
! 159: #define zs_size u.r.zs_size
1.1 deraadt 160:
161: /*
162: * To save much memory, we overlay the table used by compress() with those
163: * used by decompress(). The tab_prefix table is the same size and type as
164: * the codetab. The tab_suffix table needs 2**BITS characters. We get this
165: * from the beginning of htab. The output stack uses the rest of htab, and
166: * contains characters. There is plenty of room for any possible stack
167: * (stack used to be 8000 characters).
168: */
169:
1.5 ! mickey 170: #define htabof(i) zs->zs_htab[i]
! 171: #define codetabof(i) zs->zs_codetab[i]
1.1 deraadt 172:
173: #define tab_prefixof(i) codetabof(i)
1.5 ! mickey 174: #define tab_suffixof(i) ((u_char *)(zs->zs_htab))[i]
! 175: #define de_stack ((u_char *)&tab_suffixof(1 << BITS))
1.1 deraadt 176:
177: #define CHECK_GAP 10000 /* Ratio check interval. */
178:
179: /*
180: * the next two codes should not be changed lightly, as they must not
181: * lie within the contiguous general code space.
182: */
183: #define FIRST 257 /* First free entry. */
184: #define CLEAR 256 /* Table clear output code. */
185:
1.5 ! mickey 186: static int cl_block __P((register struct s_zstate *));
! 187: static void cl_hash __P((register struct s_zstate *, register count_int));
! 188: static code_int getcode __P((register struct s_zstate *));
! 189: static int output __P((register struct s_zstate *, code_int));
1.1 deraadt 190:
191: /*-
192: * Algorithm from "A Technique for High Performance Data Compression",
193: * Terry A. Welch, IEEE Computer Vol 17, No 6 (June 1984), pp 8-19.
194: *
195: * Algorithm:
196: * Modified Lempel-Ziv method (LZW). Basically finds common
197: * substrings and replaces them with a variable size code. This is
198: * deterministic, and can be done on the fly. Thus, the decompression
199: * procedure needs no input table, but tracks the way the table was built.
200: */
201:
202: /*-
203: * compress write
204: *
205: * Algorithm: use open addressing double hashing (no chaining) on the
206: * prefix code / next character combination. We do a variant of Knuth's
207: * algorithm D (vol. 3, sec. 6.4) along with G. Knott's relatively-prime
208: * secondary probe. Here, the modular division first probe is gives way
209: * to a faster exclusive-or manipulation. Also do block compression with
210: * an adaptive reset, whereby the code table is cleared when the compression
211: * ratio decreases, but after the table fills. The variable-length output
212: * codes are re-sized at this point, and a special CLEAR code is generated
213: * for the decompressor. Late addition: construct the table according to
214: * file size for noticeable speed improvement on small files. Please direct
215: * questions about this implementation to ames!jaw.
216: */
1.5 ! mickey 217: int
1.1 deraadt 218: zwrite(cookie, wbp, num)
219: void *cookie;
220: const char *wbp;
221: int num;
222: {
223: register code_int i;
224: register int c, disp;
225: struct s_zstate *zs;
226: const u_char *bp;
227: u_char tmp;
228: int count;
229:
230: if (num == 0)
231: return (0);
232:
233: zs = cookie;
234: count = num;
235: bp = (u_char *)wbp;
1.5 ! mickey 236: if (zs->zs_state == S_MIDDLE)
1.1 deraadt 237: goto middle;
1.5 ! mickey 238: zs->zs_state = S_MIDDLE;
1.1 deraadt 239:
1.5 ! mickey 240: zs->zs_maxmaxcode = 1L << zs->zs_maxbits;
! 241: if (write(zs->zs_fd, z_magic, sizeof(z_magic)) != sizeof(z_magic))
1.1 deraadt 242: return (-1);
1.5 ! mickey 243: tmp = (u_char)(zs->zs_maxbits | zs->zs_block_compress);
! 244: if (write(zs->zs_fd, &tmp, sizeof(tmp)) != sizeof(tmp))
1.1 deraadt 245: return (-1);
246:
1.5 ! mickey 247: zs->zs_offset = 0;
! 248: zs->zs_bytes_out = 3; /* Includes 3-byte header mojo. */
! 249: zs->zs_out_count = 0;
! 250: zs->zs_clear_flg = 0;
! 251: zs->zs_ratio = 0;
! 252: zs->zs_in_count = 1;
! 253: zs->zs_checkpoint = CHECK_GAP;
! 254: zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
! 255: zs->zs_free_ent = ((zs->zs_block_compress) ? FIRST : 256);
1.1 deraadt 256:
1.5 ! mickey 257: zs->zs_ent = *bp++;
1.1 deraadt 258: --count;
259:
1.5 ! mickey 260: zs->zs_hshift = 0;
! 261: for (zs->zs_fcode = (long)zs->zs_hsize; zs->zs_fcode < 65536L;
! 262: zs->zs_fcode *= 2L)
! 263: zs->zs_hshift++;
! 264: zs->zs_hshift = 8 - zs->zs_hshift; /* Set hash code range bound. */
1.1 deraadt 265:
1.5 ! mickey 266: zs->zs_hsize_reg = zs->zs_hsize;
! 267: cl_hash(zs, (count_int)zs->zs_hsize_reg); /* Clear hash table. */
1.1 deraadt 268:
269: middle: for (i = 0; count--;) {
270: c = *bp++;
1.5 ! mickey 271: zs->zs_in_count++;
! 272: zs->zs_fcode = (long)(((long)c << zs->zs_maxbits) +
! 273: zs->zs_ent);
! 274: i = ((c << zs->zs_hshift) ^ zs->zs_ent); /* Xor hashing. */
1.1 deraadt 275:
1.5 ! mickey 276: if (htabof(i) == zs->zs_fcode) {
! 277: zs->zs_ent = codetabof(i);
1.1 deraadt 278: continue;
279: } else if ((long)htabof(i) < 0) /* Empty slot. */
280: goto nomatch;
1.5 ! mickey 281: /* Secondary hash (after G. Knott). */
! 282: disp = zs->zs_hsize_reg - i;
1.1 deraadt 283: if (i == 0)
284: disp = 1;
285: probe: if ((i -= disp) < 0)
1.5 ! mickey 286: i += zs->zs_hsize_reg;
1.1 deraadt 287:
1.5 ! mickey 288: if (htabof(i) == zs->zs_fcode) {
! 289: zs->zs_ent = codetabof(i);
1.1 deraadt 290: continue;
291: }
292: if ((long)htabof(i) >= 0)
293: goto probe;
1.5 ! mickey 294: nomatch: if (output(zs, (code_int) zs->zs_ent) == -1)
1.1 deraadt 295: return (-1);
1.5 ! mickey 296: zs->zs_out_count++;
! 297: zs->zs_ent = c;
! 298: if (zs->zs_free_ent < zs->zs_maxmaxcode) {
! 299: /* code -> hashtable */
! 300: codetabof(i) = zs->zs_free_ent++;
! 301: htabof(i) = zs->zs_fcode;
! 302: } else if ((count_int)zs->zs_in_count >=
! 303: zs->zs_checkpoint && zs->zs_block_compress) {
1.1 deraadt 304: if (cl_block(zs) == -1)
305: return (-1);
306: }
307: }
308: return (num);
309: }
310:
1.5 ! mickey 311: int
1.1 deraadt 312: zclose(cookie)
313: void *cookie;
314: {
315: struct s_zstate *zs;
316: int rval;
317:
318: zs = cookie;
1.5 ! mickey 319: if (zs->zs_mode == 'w') { /* Put out the final code. */
! 320: if (output(zs, (code_int) zs->zs_ent) == -1) {
! 321: (void)close(zs->zs_fd);
1.1 deraadt 322: free(zs);
323: return (-1);
324: }
1.5 ! mickey 325: zs->zs_out_count++;
1.1 deraadt 326: if (output(zs, (code_int) - 1) == -1) {
1.5 ! mickey 327: (void)close(zs->zs_fd);
1.1 deraadt 328: free(zs);
329: return (-1);
330: }
331: }
1.5 ! mickey 332: rval = close(zs->zs_fd);
1.1 deraadt 333: free(zs);
334: return (rval);
335: }
336:
337: /*-
338: * Output the given code.
339: * Inputs:
340: * code: A n_bits-bit integer. If == -1, then EOF. This assumes
341: * that n_bits =< (long)wordsize - 1.
342: * Outputs:
343: * Outputs code to the file.
344: * Assumptions:
345: * Chars are 8 bits long.
346: * Algorithm:
347: * Maintain a BITS character long buffer (so that 8 codes will
348: * fit in it exactly). Use the VAX insv instruction to insert each
349: * code in turn. When the buffer fills up empty it and start over.
350: */
351:
1.5 ! mickey 352: static u_char lmask[9] =
1.1 deraadt 353: {0xff, 0xfe, 0xfc, 0xf8, 0xf0, 0xe0, 0xc0, 0x80, 0x00};
1.5 ! mickey 354: static u_char rmask[9] =
1.1 deraadt 355: {0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff};
356:
357: static int
358: output(zs, ocode)
1.5 ! mickey 359: register struct s_zstate *zs;
1.1 deraadt 360: code_int ocode;
361: {
362: register int bits, r_off;
1.5 ! mickey 363: register u_char *bp;
1.1 deraadt 364:
1.5 ! mickey 365: r_off = zs->zs_offset;
! 366: bits = zs->zs_n_bits;
! 367: bp = zs->zs_buf;
1.1 deraadt 368: if (ocode >= 0) {
369: /* Get to the first byte. */
370: bp += (r_off >> 3);
371: r_off &= 7;
372: /*
373: * Since ocode is always >= 8 bits, only need to mask the first
374: * hunk on the left.
375: */
1.4 millert 376: *bp = (*bp & rmask[r_off]) | ((ocode << r_off) & lmask[r_off]);
1.1 deraadt 377: bp++;
378: bits -= (8 - r_off);
379: ocode >>= 8 - r_off;
1.5 ! mickey 380: /* Get any 8 bit parts in the middle (<=1 for up to 16 bits) */
1.1 deraadt 381: if (bits >= 8) {
382: *bp++ = ocode;
383: ocode >>= 8;
384: bits -= 8;
385: }
386: /* Last bits. */
387: if (bits)
388: *bp = ocode;
1.5 ! mickey 389: zs->zs_offset += zs->zs_n_bits;
! 390: if (zs->zs_offset == (zs->zs_n_bits << 3)) {
! 391: bp = zs->zs_buf;
! 392: bits = zs->zs_n_bits;
! 393: zs->zs_bytes_out += bits;
! 394: if (write(zs->zs_fd, bp, bits) != bits)
1.1 deraadt 395: return (-1);
396: bp += bits;
397: bits = 0;
1.5 ! mickey 398: zs->zs_offset = 0;
1.1 deraadt 399: }
400: /*
401: * If the next entry is going to be too big for the ocode size,
402: * then increase it, if possible.
403: */
1.5 ! mickey 404: if (zs->zs_free_ent > zs->zs_maxcode ||
! 405: (zs->zs_clear_flg > 0)) {
1.1 deraadt 406: /*
407: * Write the whole buffer, because the input side won't
1.5 ! mickey 408: * discover the size increase until after it has read it
1.1 deraadt 409: */
1.5 ! mickey 410: if (zs->zs_offset > 0) {
! 411: if (write(zs->zs_fd, zs->zs_buf, zs->zs_n_bits)
! 412: != zs->zs_n_bits)
1.1 deraadt 413: return (-1);
1.5 ! mickey 414: zs->zs_bytes_out += zs->zs_n_bits;
1.1 deraadt 415: }
1.5 ! mickey 416: zs->zs_offset = 0;
1.1 deraadt 417:
1.5 ! mickey 418: if (zs->zs_clear_flg) {
! 419: zs->zs_maxcode =
! 420: MAXCODE(zs->zs_n_bits = INIT_BITS);
! 421: zs->zs_clear_flg = 0;
1.1 deraadt 422: } else {
1.5 ! mickey 423: zs->zs_n_bits++;
! 424: if (zs->zs_n_bits == zs->zs_maxbits)
! 425: zs->zs_maxcode = zs->zs_maxmaxcode;
1.1 deraadt 426: else
1.5 ! mickey 427: zs->zs_maxcode =
! 428: MAXCODE(zs->zs_n_bits);
1.1 deraadt 429: }
430: }
431: } else {
432: /* At EOF, write the rest of the buffer. */
1.5 ! mickey 433: if (zs->zs_offset > 0) {
! 434: zs->zs_offset = (zs->zs_offset + 7) / 8;
! 435: if (write(zs->zs_fd, zs->zs_buf, zs->zs_offset)
! 436: != zs->zs_offset)
1.1 deraadt 437: return (-1);
1.5 ! mickey 438: zs->zs_bytes_out += zs->zs_offset;
1.1 deraadt 439: }
1.5 ! mickey 440: zs->zs_offset = 0;
1.1 deraadt 441: }
442: return (0);
443: }
444:
445: /*
446: * Decompress read. This routine adapts to the codes in the file building
447: * the "string" table on-the-fly; requiring no table to be stored in the
448: * compressed file. The tables used herein are shared with those of the
449: * compress() routine. See the definitions above.
450: */
1.5 ! mickey 451: int
1.1 deraadt 452: zread(cookie, rbp, num)
453: void *cookie;
454: char *rbp;
455: int num;
456: {
457: register u_int count;
458: struct s_zstate *zs;
459: u_char *bp, header[3];
460:
461: if (num == 0)
462: return (0);
463:
464: zs = cookie;
465: count = num;
466: bp = (u_char *)rbp;
1.5 ! mickey 467: switch (zs->zs_state) {
1.1 deraadt 468: case S_START:
1.5 ! mickey 469: zs->zs_state = S_MIDDLE;
1.1 deraadt 470: break;
471: case S_MIDDLE:
472: goto middle;
473: case S_EOF:
474: goto eof;
475: }
476:
477: /* Check the magic number */
1.5 ! mickey 478: if (read(zs->zs_fd, header, sizeof(header)) != sizeof(header) ||
! 479: memcmp(header, z_magic, sizeof(z_magic)) != 0) {
1.1 deraadt 480: errno = EFTYPE;
481: return (-1);
482: }
1.5 ! mickey 483: zs->zs_maxbits = header[2]; /* Set -b from file. */
! 484: zs->zs_block_compress = zs->zs_maxbits & BLOCK_MASK;
! 485: zs->zs_maxbits &= BIT_MASK;
! 486: zs->zs_maxmaxcode = 1L << zs->zs_maxbits;
! 487: if (zs->zs_maxbits > BITS) {
1.1 deraadt 488: errno = EFTYPE;
489: return (-1);
490: }
491: /* As above, initialize the first 256 entries in the table. */
1.5 ! mickey 492: zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
! 493: for (zs->zs_code = 255; zs->zs_code >= 0; zs->zs_code--) {
! 494: tab_prefixof(zs->zs_code) = 0;
! 495: tab_suffixof(zs->zs_code) = (u_char) zs->zs_code;
1.1 deraadt 496: }
1.5 ! mickey 497: zs->zs_free_ent = zs->zs_block_compress ? FIRST : 256;
1.1 deraadt 498:
1.5 ! mickey 499: zs->zs_finchar = zs->zs_oldcode = getcode(zs);
! 500: if (zs->zs_oldcode == -1) /* EOF already? */
1.1 deraadt 501: return (0); /* Get out of here */
502:
503: /* First code must be 8 bits = char. */
1.5 ! mickey 504: *bp++ = (u_char)zs->zs_finchar;
1.1 deraadt 505: count--;
1.5 ! mickey 506: zs->zs_stackp = de_stack;
1.1 deraadt 507:
1.5 ! mickey 508: while ((zs->zs_code = getcode(zs)) > -1) {
1.1 deraadt 509:
1.5 ! mickey 510: if ((zs->zs_code == CLEAR) && zs->zs_block_compress) {
! 511: for (zs->zs_code = 255; zs->zs_code >= 0;
! 512: zs->zs_code--)
! 513: tab_prefixof(zs->zs_code) = 0;
! 514: zs->zs_clear_flg = 1;
! 515: zs->zs_free_ent = FIRST - 1;
! 516: if ((zs->zs_code = getcode(zs)) == -1) /* O, untimely death! */
1.1 deraadt 517: break;
518: }
1.5 ! mickey 519: zs->zs_incode = zs->zs_code;
1.1 deraadt 520:
521: /* Special case for KwKwK string. */
1.5 ! mickey 522: if (zs->zs_code >= zs->zs_free_ent) {
! 523: *zs->zs_stackp++ = zs->zs_finchar;
! 524: zs->zs_code = zs->zs_oldcode;
1.1 deraadt 525: }
526:
527: /* Generate output characters in reverse order. */
1.5 ! mickey 528: while (zs->zs_code >= 256) {
! 529: *zs->zs_stackp++ = tab_suffixof(zs->zs_code);
! 530: zs->zs_code = tab_prefixof(zs->zs_code);
1.1 deraadt 531: }
1.5 ! mickey 532: *zs->zs_stackp++ = zs->zs_finchar = tab_suffixof(zs->zs_code);
1.1 deraadt 533:
534: /* And put them out in forward order. */
535: middle: do {
536: if (count-- == 0)
537: return (num);
1.5 ! mickey 538: *bp++ = *--zs->zs_stackp;
! 539: } while (zs->zs_stackp > de_stack);
1.1 deraadt 540:
541: /* Generate the new entry. */
1.5 ! mickey 542: if ((zs->zs_code = zs->zs_free_ent) < zs->zs_maxmaxcode) {
! 543: tab_prefixof(zs->zs_code) = (u_short) zs->zs_oldcode;
! 544: tab_suffixof(zs->zs_code) = zs->zs_finchar;
! 545: zs->zs_free_ent = zs->zs_code + 1;
1.1 deraadt 546: }
547:
548: /* Remember previous code. */
1.5 ! mickey 549: zs->zs_oldcode = zs->zs_incode;
1.1 deraadt 550: }
1.5 ! mickey 551: zs->zs_state = S_EOF;
1.1 deraadt 552: eof: return (num - count);
553: }
554:
555: /*-
556: * Read one code from the standard input. If EOF, return -1.
557: * Inputs:
558: * stdin
559: * Outputs:
560: * code or -1 is returned.
561: */
562: static code_int
563: getcode(zs)
1.5 ! mickey 564: register struct s_zstate *zs;
1.1 deraadt 565: {
566: register code_int gcode;
567: register int r_off, bits;
1.5 ! mickey 568: register u_char *bp;
1.1 deraadt 569:
1.5 ! mickey 570: bp = zs->zs_buf;
! 571: if (zs->zs_clear_flg > 0 || zs->zs_roffset >= zs->zs_size ||
! 572: zs->zs_free_ent > zs->zs_maxcode) {
1.1 deraadt 573: /*
574: * If the next entry will be too big for the current gcode
575: * size, then we must increase the size. This implies reading
576: * a new buffer full, too.
577: */
1.5 ! mickey 578: if (zs->zs_free_ent > zs->zs_maxcode) {
! 579: zs->zs_n_bits++;
! 580: if (zs->zs_n_bits == zs->zs_maxbits) /* Won't get any bigger now. */
! 581: zs->zs_maxcode = zs->zs_maxmaxcode;
1.1 deraadt 582: else
1.5 ! mickey 583: zs->zs_maxcode = MAXCODE(zs->zs_n_bits);
1.1 deraadt 584: }
1.5 ! mickey 585: if (zs->zs_clear_flg > 0) {
! 586: zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
! 587: zs->zs_clear_flg = 0;
1.1 deraadt 588: }
1.5 ! mickey 589: zs->zs_size = read(zs->zs_fd, zs->zs_buf, zs->zs_n_bits);
! 590: if (zs->zs_size <= 0) /* End of file. */
1.1 deraadt 591: return (-1);
1.5 ! mickey 592: zs->zs_roffset = 0;
1.1 deraadt 593: /* Round size down to integral number of codes. */
1.5 ! mickey 594: zs->zs_size = (zs->zs_size << 3) - (zs->zs_n_bits - 1);
1.1 deraadt 595: }
1.5 ! mickey 596: r_off = zs->zs_roffset;
! 597: bits = zs->zs_n_bits;
1.1 deraadt 598:
599: /* Get to the first byte. */
600: bp += (r_off >> 3);
601: r_off &= 7;
602:
603: /* Get first part (low order bits). */
604: gcode = (*bp++ >> r_off);
605: bits -= (8 - r_off);
606: r_off = 8 - r_off; /* Now, roffset into gcode word. */
607:
608: /* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
609: if (bits >= 8) {
610: gcode |= *bp++ << r_off;
611: r_off += 8;
612: bits -= 8;
613: }
614:
615: /* High order bits. */
616: gcode |= (*bp & rmask[bits]) << r_off;
1.5 ! mickey 617: zs->zs_roffset += zs->zs_n_bits;
1.1 deraadt 618:
619: return (gcode);
620: }
621:
622: static int
623: cl_block(zs) /* Table clear for block compress. */
1.5 ! mickey 624: register struct s_zstate *zs;
1.1 deraadt 625: {
626: register long rat;
627:
1.5 ! mickey 628: zs->zs_checkpoint = zs->zs_in_count + CHECK_GAP;
1.1 deraadt 629:
1.5 ! mickey 630: if (zs->zs_in_count > 0x007fffff) { /* Shift will overflow. */
! 631: rat = zs->zs_bytes_out >> 8;
1.1 deraadt 632: if (rat == 0) /* Don't divide by zero. */
633: rat = 0x7fffffff;
634: else
1.5 ! mickey 635: rat = zs->zs_in_count / rat;
1.1 deraadt 636: } else
1.5 ! mickey 637: rat = (zs->zs_in_count << 8) / zs->zs_bytes_out; /* 8 fractional bits. */
! 638: if (rat > zs->zs_ratio)
! 639: zs->zs_ratio = rat;
1.1 deraadt 640: else {
1.5 ! mickey 641: zs->zs_ratio = 0;
! 642: cl_hash(zs, (count_int) zs->zs_hsize);
! 643: zs->zs_free_ent = FIRST;
! 644: zs->zs_clear_flg = 1;
1.1 deraadt 645: if (output(zs, (code_int) CLEAR) == -1)
646: return (-1);
647: }
648: return (0);
649: }
650:
651: static void
652: cl_hash(zs, cl_hsize) /* Reset code table. */
1.5 ! mickey 653: register struct s_zstate *zs;
1.1 deraadt 654: register count_int cl_hsize;
655: {
656: register count_int *htab_p;
657: register long i, m1;
658:
659: m1 = -1;
1.5 ! mickey 660: htab_p = zs->zs_htab + cl_hsize;
1.1 deraadt 661: i = cl_hsize - 16;
662: do { /* Might use Sys V memset(3) here. */
663: *(htab_p - 16) = m1;
664: *(htab_p - 15) = m1;
665: *(htab_p - 14) = m1;
666: *(htab_p - 13) = m1;
667: *(htab_p - 12) = m1;
668: *(htab_p - 11) = m1;
669: *(htab_p - 10) = m1;
670: *(htab_p - 9) = m1;
671: *(htab_p - 8) = m1;
672: *(htab_p - 7) = m1;
673: *(htab_p - 6) = m1;
674: *(htab_p - 5) = m1;
675: *(htab_p - 4) = m1;
676: *(htab_p - 3) = m1;
677: *(htab_p - 2) = m1;
678: *(htab_p - 1) = m1;
679: htab_p -= 16;
680: } while ((i -= 16) >= 0);
681: for (i += 16; i > 0; i--)
682: *--htab_p = m1;
1.3 tholo 683: }
684:
685: FILE *
1.5 ! mickey 686: zopen(name, mode, bits)
! 687: const char *name;
! 688: const char *mode;
! 689: int bits;
! 690: {
! 691: int fd;
! 692: void *cookie;
! 693: if ((fd = open(name, (*mode=='r'? O_RDONLY:O_WRONLY|O_CREAT),
! 694: S_IRUSR|S_IWUSR|S_IRGRP|S_IROTH)) == -1)
! 695: return NULL;
! 696: if ((cookie = z_open(fd, mode, bits)) == NULL) {
! 697: close(fd);
! 698: return NULL;
! 699: }
! 700: return funopen(cookie, (*mode == 'r'?zread:NULL),
! 701: (*mode == 'w'?zwrite:NULL), NULL, zclose);
! 702: }
! 703:
! 704: void *
! 705: z_open(fd, mode, bits)
1.3 tholo 706: int fd;
707: const char *mode;
708: int bits;
709: {
1.5 ! mickey 710: register struct s_zstate *zs;
1.3 tholo 711:
1.4 millert 712: if ((mode[0] != 'r' && mode[0] != 'w') || mode[1] != '\0' ||
1.3 tholo 713: bits < 0 || bits > BITS) {
714: errno = EINVAL;
715: return (NULL);
716: }
717:
718: if ((zs = calloc(1, sizeof(struct s_zstate))) == NULL)
719: return (NULL);
720:
1.5 ! mickey 721: /* User settable max # bits/code. */
! 722: zs->zs_maxbits = bits ? bits : BITS;
! 723: /* Should NEVER generate this code. */
! 724: zs->zs_maxmaxcode = 1 << zs->zs_maxbits;
! 725: zs->zs_hsize = HSIZE; /* For dynamic table sizing. */
! 726: zs->zs_free_ent = 0; /* First unused entry. */
! 727: zs->zs_block_compress = BLOCK_MASK;
! 728: zs->zs_clear_flg = 0;
! 729: zs->zs_ratio = 0;
! 730: zs->zs_checkpoint = CHECK_GAP;
! 731: zs->zs_in_count = 1; /* Length of input. */
! 732: zs->zs_out_count = 0; /* # of codes output (for debugging).*/
! 733: zs->zs_state = S_START;
! 734: zs->zs_roffset = 0;
! 735: zs->zs_size = 0;
! 736: zs->zs_mode = mode[0];
1.3 tholo 737:
1.5 ! mickey 738: zs->zs_fd = fd;
! 739: return zs;
1.1 deraadt 740: }
741:
1.5 ! mickey 742: int
! 743: z_check_header(fd, sb, ofn)
! 744: int fd;
! 745: struct stat *sb;
! 746: const char *ofn;
1.1 deraadt 747: {
1.5 ! mickey 748: int f;
! 749: u_char buf[sizeof(z_magic)];
! 750: off_t off = lseek(fd, 0, SEEK_CUR);
1.1 deraadt 751:
1.5 ! mickey 752: f = (read(fd, buf, sizeof(buf)) == sizeof(buf) &&
! 753: !memcmp(buf, z_magic, sizeof(buf)));
1.1 deraadt 754:
1.5 ! mickey 755: lseek (fd, off, SEEK_SET);
1.1 deraadt 756:
1.5 ! mickey 757: return f;
1.1 deraadt 758: }