Annotation of src/usr.bin/compress/zopen.c, Revision 1.11
1.11 ! deraadt 1: /* $OpenBSD: zopen.c,v 1.10 2003/06/03 02:56:07 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.
1.10 millert 20: * 3. Neither the name of the University nor the names of its contributors
1.1 deraadt 21: * may be used to endorse or promote products derived from this software
22: * without specific prior written permission.
23: *
24: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34: * SUCH DAMAGE.
1.9 mickey 35: *
36: * From: @(#)zopen.c 8.1 (Berkeley) 6/27/93
1.1 deraadt 37: */
38:
39: #if 0
40: static char sccsid[] = "@(#)zopen.c 8.1 (Berkeley) 6/27/93";
41: #else
1.9 mickey 42: const char z_rcsid[] =
1.11 ! deraadt 43: "$OpenBSD: zopen.c,v 1.10 2003/06/03 02:56:07 millert Exp $";
1.1 deraadt 44: #endif
45:
46: /*-
47: * fcompress.c - File compression ala IEEE Computer, June 1984.
48: *
49: * Compress authors:
50: * Spencer W. Thomas (decvax!utah-cs!thomas)
51: * Jim McKie (decvax!mcvax!jim)
52: * Steve Davies (decvax!vax135!petsd!peora!srd)
53: * Ken Turkowski (decvax!decwrl!turtlevax!ken)
54: * James A. Woods (decvax!ihnp4!ames!jaw)
55: * Joe Orost (decvax!vax135!petsd!joe)
56: *
57: * Cleaned up and converted to library returning I/O streams by
58: * Diomidis Spinellis <dds@doc.ic.ac.uk>.
59: *
60: * zopen(filename, mode, bits)
61: * Returns a FILE * that can be used for read or write. The modes
62: * supported are only "r" and "w". Seeking is not allowed. On
63: * reading the file is decompressed, on writing it is compressed.
64: * The output is compatible with compress(1) with 16 bit tables.
65: * Any file produced by compress(1) can be read.
66: */
67:
68: #include <sys/param.h>
69: #include <sys/stat.h>
70:
71: #include <ctype.h>
72: #include <errno.h>
73: #include <signal.h>
74: #include <stdio.h>
75: #include <stdlib.h>
76: #include <string.h>
77: #include <unistd.h>
1.5 mickey 78: #include <fcntl.h>
79: #include "compress.h"
1.1 deraadt 80:
81: #define BITS 16 /* Default bits. */
82: #define HSIZE 69001 /* 95% occupancy */
1.6 mickey 83: #define ZBUFSIZ 8192 /* I/O buffer size */
1.1 deraadt 84:
85: /* A code_int must be able to hold 2**BITS values of type int, and also -1. */
86: typedef long code_int;
87: typedef long count_int;
88:
1.9 mickey 89: static const u_char z_magic[] =
1.1 deraadt 90: {'\037', '\235'}; /* 1F 9D */
91:
92: #define BIT_MASK 0x1f /* Defines for third byte of header. */
93: #define BLOCK_MASK 0x80
94:
95: /*
96: * Masks 0x40 and 0x20 are free. I think 0x20 should mean that there is
97: * a fourth header byte (for expansion).
98: */
99: #define INIT_BITS 9 /* Initial number of bits/code. */
100:
101: #define MAXCODE(n_bits) ((1 << (n_bits)) - 1)
102:
103: struct s_zstate {
1.5 mickey 104: int zs_fd; /* File stream for I/O */
1.1 deraadt 105: char zs_mode; /* r or w */
106: enum {
107: S_START, S_MIDDLE, S_EOF
108: } zs_state; /* State of computation */
109: int zs_n_bits; /* Number of bits/code. */
110: int zs_maxbits; /* User settable max # bits/code. */
111: code_int zs_maxcode; /* Maximum code, given n_bits. */
112: code_int zs_maxmaxcode; /* Should NEVER generate this code. */
113: count_int zs_htab [HSIZE];
114: u_short zs_codetab [HSIZE];
115: code_int zs_hsize; /* For dynamic table sizing. */
116: code_int zs_free_ent; /* First unused entry. */
117: /*
118: * Block compression parameters -- after all codes are used up,
119: * and compression rate changes, start over.
120: */
121: int zs_block_compress;
122: int zs_clear_flg;
123: long zs_ratio;
124: count_int zs_checkpoint;
125: long zs_in_count; /* Length of input. */
126: long zs_bytes_out; /* Length of compressed output. */
1.5 mickey 127: long zs_out_count; /* # of codes output (for debugging).*/
1.6 mickey 128: u_char zs_buf[ZBUFSIZ]; /* I/O buffer */
129: u_char *zs_bp; /* Current I/O window in the zs_buf */
130: int zs_offset; /* Number of bits in the zs_buf */
1.1 deraadt 131: union {
132: struct {
133: long zs_fcode;
134: code_int zs_ent;
135: code_int zs_hsize_reg;
136: int zs_hshift;
137: } w; /* Write paramenters */
138: struct {
1.6 mickey 139: u_char *zs_stackp, *zs_ebp;
1.1 deraadt 140: int zs_finchar;
141: code_int zs_code, zs_oldcode, zs_incode;
1.6 mickey 142: int zs_size;
1.1 deraadt 143: } r; /* Read parameters */
144: } u;
145: };
146:
147: /* Definitions to retain old variable names */
1.5 mickey 148: #define zs_fcode u.w.zs_fcode
149: #define zs_ent u.w.zs_ent
150: #define zs_hsize_reg u.w.zs_hsize_reg
151: #define zs_hshift u.w.zs_hshift
152: #define zs_stackp u.r.zs_stackp
153: #define zs_finchar u.r.zs_finchar
154: #define zs_code u.r.zs_code
155: #define zs_oldcode u.r.zs_oldcode
156: #define zs_incode u.r.zs_incode
157: #define zs_size u.r.zs_size
1.6 mickey 158: #define zs_ebp u.r.zs_ebp
1.1 deraadt 159:
160: /*
161: * To save much memory, we overlay the table used by compress() with those
162: * used by decompress(). The tab_prefix table is the same size and type as
163: * the codetab. The tab_suffix table needs 2**BITS characters. We get this
164: * from the beginning of htab. The output stack uses the rest of htab, and
165: * contains characters. There is plenty of room for any possible stack
166: * (stack used to be 8000 characters).
167: */
168:
1.5 mickey 169: #define htabof(i) zs->zs_htab[i]
170: #define codetabof(i) zs->zs_codetab[i]
1.1 deraadt 171:
172: #define tab_prefixof(i) codetabof(i)
1.5 mickey 173: #define tab_suffixof(i) ((u_char *)(zs->zs_htab))[i]
174: #define de_stack ((u_char *)&tab_suffixof(1 << BITS))
1.1 deraadt 175:
176: #define CHECK_GAP 10000 /* Ratio check interval. */
177:
178: /*
179: * the next two codes should not be changed lightly, as they must not
180: * lie within the contiguous general code space.
181: */
182: #define FIRST 257 /* First free entry. */
183: #define CLEAR 256 /* Table clear output code. */
184:
1.8 millert 185: static int cl_block(struct s_zstate *);
1.9 mickey 186: static void cl_hash(struct s_zstate *, count_int);
1.8 millert 187: static code_int getcode(struct s_zstate *);
188: static int output(struct s_zstate *, code_int);
1.1 deraadt 189:
190: /*-
191: * Algorithm from "A Technique for High Performance Data Compression",
192: * Terry A. Welch, IEEE Computer Vol 17, No 6 (June 1984), pp 8-19.
193: *
194: * Algorithm:
1.9 mickey 195: * Modified Lempel-Ziv method (LZW). Basically finds common
1.1 deraadt 196: * substrings and replaces them with a variable size code. This is
197: * deterministic, and can be done on the fly. Thus, the decompression
198: * procedure needs no input table, but tracks the way the table was built.
199: */
200:
201: /*-
202: * compress write
203: *
204: * Algorithm: use open addressing double hashing (no chaining) on the
205: * prefix code / next character combination. We do a variant of Knuth's
206: * algorithm D (vol. 3, sec. 6.4) along with G. Knott's relatively-prime
207: * secondary probe. Here, the modular division first probe is gives way
208: * to a faster exclusive-or manipulation. Also do block compression with
209: * an adaptive reset, whereby the code table is cleared when the compression
210: * ratio decreases, but after the table fills. The variable-length output
211: * codes are re-sized at this point, and a special CLEAR code is generated
212: * for the decompressor. Late addition: construct the table according to
213: * file size for noticeable speed improvement on small files. Please direct
214: * questions about this implementation to ames!jaw.
215: */
1.5 mickey 216: int
1.11 ! deraadt 217: zwrite(void *cookie, const char *wbp, int num)
1.1 deraadt 218: {
1.7 mpech 219: code_int i;
220: int c, disp;
1.1 deraadt 221: struct s_zstate *zs;
222: const u_char *bp;
223: u_char tmp;
224: int count;
225:
226: zs = cookie;
227: count = num;
228: bp = (u_char *)wbp;
1.6 mickey 229: switch (zs->zs_state) {
230: case S_EOF:
231: return 0;
232: case S_START:
233: zs->zs_state = S_MIDDLE;
1.1 deraadt 234:
1.6 mickey 235: zs->zs_maxmaxcode = 1L << zs->zs_maxbits;
236: if (write(zs->zs_fd, z_magic, sizeof(z_magic)) !=
237: sizeof(z_magic))
238: return (-1);
239: tmp = (u_char)(zs->zs_maxbits | zs->zs_block_compress);
240: if (write(zs->zs_fd, &tmp, sizeof(tmp)) != sizeof(tmp))
241: return (-1);
1.1 deraadt 242:
1.6 mickey 243: zs->zs_bp = zs->zs_buf;
244: zs->zs_offset = 0;
245: zs->zs_bytes_out = 3; /* Includes 3-byte header mojo. */
246: zs->zs_out_count = 0;
247: zs->zs_clear_flg = 0;
248: zs->zs_ratio = 0;
249: zs->zs_in_count = 1;
250: zs->zs_checkpoint = CHECK_GAP;
251: zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
252: zs->zs_free_ent = ((zs->zs_block_compress) ? FIRST : 256);
253:
254: zs->zs_ent = *bp++;
255: --count;
256:
257: zs->zs_hshift = 0;
258: for (zs->zs_fcode = (long)zs->zs_hsize; zs->zs_fcode < 65536L;
1.11 ! deraadt 259: zs->zs_fcode *= 2L)
1.6 mickey 260: zs->zs_hshift++;
261: /* Set hash code range bound. */
262: zs->zs_hshift = 8 - zs->zs_hshift;
263:
264: zs->zs_hsize_reg = zs->zs_hsize;
265: /* Clear hash table. */
266: cl_hash(zs, (count_int)zs->zs_hsize_reg);
1.1 deraadt 267:
1.6 mickey 268: case S_MIDDLE:
269: for (i = 0; count-- > 0;) {
270: c = *bp++;
271: zs->zs_in_count++;
272: zs->zs_fcode = (long)(((long)c << zs->zs_maxbits) +
1.11 ! deraadt 273: zs->zs_ent);
1.6 mickey 274: /* Xor hashing. */
275: i = ((c << zs->zs_hshift) ^ zs->zs_ent);
1.9 mickey 276:
1.6 mickey 277: if (htabof(i) == zs->zs_fcode) {
278: zs->zs_ent = codetabof(i);
279: continue;
280: } else if ((long)htabof(i) < 0) /* Empty slot. */
281: goto nomatch;
282: /* Secondary hash (after G. Knott). */
283: disp = zs->zs_hsize_reg - i;
284: if (i == 0)
1.1 deraadt 285: disp = 1;
1.6 mickey 286: probe: if ((i -= disp) < 0)
287: i += zs->zs_hsize_reg;
1.1 deraadt 288:
1.6 mickey 289: if (htabof(i) == zs->zs_fcode) {
290: zs->zs_ent = codetabof(i);
291: continue;
292: }
293: if ((long)htabof(i) >= 0)
294: goto probe;
295: nomatch: if (output(zs, (code_int) zs->zs_ent) == -1)
1.1 deraadt 296: return (-1);
1.6 mickey 297: zs->zs_out_count++;
298: zs->zs_ent = c;
299: if (zs->zs_free_ent < zs->zs_maxmaxcode) {
300: /* code -> hashtable */
301: codetabof(i) = zs->zs_free_ent++;
302: htabof(i) = zs->zs_fcode;
303: } else if ((count_int)zs->zs_in_count >=
304: zs->zs_checkpoint && zs->zs_block_compress) {
305: if (cl_block(zs) == -1)
306: return (-1);
307: }
1.1 deraadt 308: }
309: }
310: return (num);
311: }
312:
1.5 mickey 313: int
1.11 ! deraadt 314: zclose(void *cookie)
1.1 deraadt 315: {
316: struct s_zstate *zs;
317: int rval;
318:
319: zs = cookie;
1.5 mickey 320: if (zs->zs_mode == 'w') { /* Put out the final code. */
321: if (output(zs, (code_int) zs->zs_ent) == -1) {
322: (void)close(zs->zs_fd);
1.1 deraadt 323: free(zs);
324: return (-1);
325: }
1.5 mickey 326: zs->zs_out_count++;
1.1 deraadt 327: if (output(zs, (code_int) - 1) == -1) {
1.5 mickey 328: (void)close(zs->zs_fd);
1.1 deraadt 329: free(zs);
330: return (-1);
331: }
332: }
1.5 mickey 333: rval = close(zs->zs_fd);
1.1 deraadt 334: free(zs);
335: return (rval);
336: }
337:
338: /*-
339: * Output the given code.
340: * Inputs:
1.9 mickey 341: * code: A n_bits-bit integer. If == -1, then EOF. This assumes
1.1 deraadt 342: * that n_bits =< (long)wordsize - 1.
343: * Outputs:
1.9 mickey 344: * Outputs code to the file.
1.1 deraadt 345: * Assumptions:
346: * Chars are 8 bits long.
347: * Algorithm:
1.9 mickey 348: * Maintain a BITS character long buffer (so that 8 codes will
1.1 deraadt 349: * fit in it exactly). Use the VAX insv instruction to insert each
350: * code in turn. When the buffer fills up empty it and start over.
351: */
352:
1.6 mickey 353: static const u_char lmask[9] =
1.1 deraadt 354: {0xff, 0xfe, 0xfc, 0xf8, 0xf0, 0xe0, 0xc0, 0x80, 0x00};
1.6 mickey 355: static const u_char rmask[9] =
1.1 deraadt 356: {0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff};
357:
358: static int
1.11 ! deraadt 359: output(struct s_zstate *zs, code_int ocode)
1.1 deraadt 360: {
1.7 mpech 361: int bits;
1.1 deraadt 362:
363: if (ocode >= 0) {
1.7 mpech 364: int r_off;
365: u_char *bp;
1.6 mickey 366:
1.1 deraadt 367: /* Get to the first byte. */
1.6 mickey 368: bp = zs->zs_bp + (zs->zs_offset >> 3);
369: r_off = zs->zs_offset & 7;
370: bits = zs->zs_n_bits;
371:
1.1 deraadt 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)) {
1.6 mickey 391: zs->zs_bp += zs->zs_n_bits;
1.5 mickey 392: zs->zs_offset = 0;
1.1 deraadt 393: }
394: /*
395: * If the next entry is going to be too big for the ocode size,
396: * then increase it, if possible.
397: */
1.5 mickey 398: if (zs->zs_free_ent > zs->zs_maxcode ||
399: (zs->zs_clear_flg > 0)) {
1.11 ! deraadt 400: /*
! 401: * Write the whole buffer, because the input side won't
! 402: * discover the size increase until after it has read it
! 403: */
1.5 mickey 404: if (zs->zs_offset > 0) {
1.6 mickey 405: zs->zs_bp += zs->zs_n_bits;
406: zs->zs_offset = 0;
1.1 deraadt 407: }
408:
1.5 mickey 409: if (zs->zs_clear_flg) {
410: zs->zs_maxcode =
411: MAXCODE(zs->zs_n_bits = INIT_BITS);
412: zs->zs_clear_flg = 0;
1.1 deraadt 413: } else {
1.5 mickey 414: zs->zs_n_bits++;
415: if (zs->zs_n_bits == zs->zs_maxbits)
416: zs->zs_maxcode = zs->zs_maxmaxcode;
1.1 deraadt 417: else
1.5 mickey 418: zs->zs_maxcode =
419: MAXCODE(zs->zs_n_bits);
1.1 deraadt 420: }
421: }
1.6 mickey 422:
423: if (zs->zs_bp + zs->zs_n_bits > &zs->zs_buf[ZBUFSIZ]) {
424: bits = zs->zs_bp - zs->zs_buf;
425: if (write(zs->zs_fd, zs->zs_buf, bits) != bits)
426: return (-1);
427: zs->zs_bytes_out += bits;
428: if (zs->zs_offset > 0)
429: fprintf (stderr, "zs_offset != 0\n");
430: zs->zs_bp = zs->zs_buf;
431: }
1.1 deraadt 432: } else {
433: /* At EOF, write the rest of the buffer. */
1.6 mickey 434: if (zs->zs_offset > 0)
435: zs->zs_bp += (zs->zs_offset + 7) / 8;
436: if (zs->zs_bp > zs->zs_buf) {
437: bits = zs->zs_bp - zs->zs_buf;
438: if (write(zs->zs_fd, zs->zs_buf, bits) != bits)
1.1 deraadt 439: return (-1);
1.6 mickey 440: zs->zs_bytes_out += bits;
1.1 deraadt 441: }
1.5 mickey 442: zs->zs_offset = 0;
1.6 mickey 443: zs->zs_bp = zs->zs_buf;
1.1 deraadt 444: }
445: return (0);
446: }
447:
448: /*
449: * Decompress read. This routine adapts to the codes in the file building
450: * the "string" table on-the-fly; requiring no table to be stored in the
451: * compressed file. The tables used herein are shared with those of the
452: * compress() routine. See the definitions above.
453: */
1.5 mickey 454: int
1.11 ! deraadt 455: zread(void *cookie, char *rbp, int num)
1.1 deraadt 456: {
1.7 mpech 457: u_int count;
1.1 deraadt 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;
1.6 mickey 470: zs->zs_bp = zs->zs_buf;
1.1 deraadt 471: break;
472: case S_MIDDLE:
473: goto middle;
474: case S_EOF:
475: goto eof;
476: }
477:
478: /* Check the magic number */
1.5 mickey 479: if (read(zs->zs_fd, header, sizeof(header)) != sizeof(header) ||
480: memcmp(header, z_magic, sizeof(z_magic)) != 0) {
1.1 deraadt 481: errno = EFTYPE;
482: return (-1);
483: }
1.5 mickey 484: zs->zs_maxbits = header[2]; /* Set -b from file. */
485: zs->zs_block_compress = zs->zs_maxbits & BLOCK_MASK;
486: zs->zs_maxbits &= BIT_MASK;
487: zs->zs_maxmaxcode = 1L << zs->zs_maxbits;
488: if (zs->zs_maxbits > BITS) {
1.1 deraadt 489: errno = EFTYPE;
490: return (-1);
491: }
492: /* As above, initialize the first 256 entries in the table. */
1.5 mickey 493: zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
494: for (zs->zs_code = 255; zs->zs_code >= 0; zs->zs_code--) {
495: tab_prefixof(zs->zs_code) = 0;
496: tab_suffixof(zs->zs_code) = (u_char) zs->zs_code;
1.1 deraadt 497: }
1.5 mickey 498: zs->zs_free_ent = zs->zs_block_compress ? FIRST : 256;
1.1 deraadt 499:
1.5 mickey 500: zs->zs_finchar = zs->zs_oldcode = getcode(zs);
501: if (zs->zs_oldcode == -1) /* EOF already? */
1.1 deraadt 502: return (0); /* Get out of here */
503:
504: /* First code must be 8 bits = char. */
1.5 mickey 505: *bp++ = (u_char)zs->zs_finchar;
1.1 deraadt 506: count--;
1.5 mickey 507: zs->zs_stackp = de_stack;
1.1 deraadt 508:
1.5 mickey 509: while ((zs->zs_code = getcode(zs)) > -1) {
1.1 deraadt 510:
1.5 mickey 511: if ((zs->zs_code == CLEAR) && zs->zs_block_compress) {
512: for (zs->zs_code = 255; zs->zs_code >= 0;
1.11 ! deraadt 513: zs->zs_code--)
1.5 mickey 514: tab_prefixof(zs->zs_code) = 0;
515: zs->zs_clear_flg = 1;
516: zs->zs_free_ent = FIRST - 1;
517: if ((zs->zs_code = getcode(zs)) == -1) /* O, untimely death! */
1.1 deraadt 518: break;
519: }
1.5 mickey 520: zs->zs_incode = zs->zs_code;
1.1 deraadt 521:
522: /* Special case for KwKwK string. */
1.5 mickey 523: if (zs->zs_code >= zs->zs_free_ent) {
524: *zs->zs_stackp++ = zs->zs_finchar;
525: zs->zs_code = zs->zs_oldcode;
1.1 deraadt 526: }
527:
528: /* Generate output characters in reverse order. */
1.5 mickey 529: while (zs->zs_code >= 256) {
530: *zs->zs_stackp++ = tab_suffixof(zs->zs_code);
531: zs->zs_code = tab_prefixof(zs->zs_code);
1.1 deraadt 532: }
1.5 mickey 533: *zs->zs_stackp++ = zs->zs_finchar = tab_suffixof(zs->zs_code);
1.1 deraadt 534:
535: /* And put them out in forward order. */
536: middle: do {
537: if (count-- == 0)
538: return (num);
1.5 mickey 539: *bp++ = *--zs->zs_stackp;
540: } while (zs->zs_stackp > de_stack);
1.1 deraadt 541:
542: /* Generate the new entry. */
1.5 mickey 543: if ((zs->zs_code = zs->zs_free_ent) < zs->zs_maxmaxcode) {
544: tab_prefixof(zs->zs_code) = (u_short) zs->zs_oldcode;
545: tab_suffixof(zs->zs_code) = zs->zs_finchar;
546: zs->zs_free_ent = zs->zs_code + 1;
1.1 deraadt 547: }
548:
549: /* Remember previous code. */
1.5 mickey 550: zs->zs_oldcode = zs->zs_incode;
1.1 deraadt 551: }
1.5 mickey 552: zs->zs_state = S_EOF;
1.1 deraadt 553: eof: return (num - count);
554: }
555:
556: /*-
557: * Read one code from the standard input. If EOF, return -1.
558: * Inputs:
1.9 mickey 559: * stdin
1.1 deraadt 560: * Outputs:
1.9 mickey 561: * code or -1 is returned.
1.1 deraadt 562: */
563: static code_int
1.11 ! deraadt 564: getcode(struct s_zstate *zs)
1.1 deraadt 565: {
1.7 mpech 566: code_int gcode;
567: int r_off, bits;
568: u_char *bp;
1.1 deraadt 569:
1.6 mickey 570: if (zs->zs_clear_flg > 0 || zs->zs_offset >= zs->zs_size ||
1.5 mickey 571: zs->zs_free_ent > zs->zs_maxcode) {
1.6 mickey 572:
573: zs->zs_bp += zs->zs_n_bits;
1.1 deraadt 574: /*
575: * If the next entry will be too big for the current gcode
576: * size, then we must increase the size. This implies reading
577: * a new buffer full, too.
578: */
1.5 mickey 579: if (zs->zs_free_ent > zs->zs_maxcode) {
580: zs->zs_n_bits++;
1.11 ! deraadt 581: if (zs->zs_n_bits == zs->zs_maxbits) {
! 582: /* Won't get any bigger now. */
1.5 mickey 583: zs->zs_maxcode = zs->zs_maxmaxcode;
1.11 ! deraadt 584: } else
1.5 mickey 585: zs->zs_maxcode = MAXCODE(zs->zs_n_bits);
1.1 deraadt 586: }
1.5 mickey 587: if (zs->zs_clear_flg > 0) {
588: zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
589: zs->zs_clear_flg = 0;
1.1 deraadt 590: }
1.6 mickey 591:
592: /* fill the buffer up to the neck */
593: if (zs->zs_bp + zs->zs_n_bits > zs->zs_ebp) {
594: for (bp = zs->zs_buf; zs->zs_bp < zs->zs_ebp;
595: *bp++ = *zs->zs_bp++);
596: if ((bits = read(zs->zs_fd, bp, ZBUFSIZ -
597: (bp - zs->zs_buf))) < 0)
598: return -1;
599: zs->zs_bp = zs->zs_buf;
600: zs->zs_ebp = bp + bits;
601: }
602: zs->zs_offset = 0;
603: zs->zs_size = MIN(zs->zs_n_bits, zs->zs_ebp - zs->zs_bp);
604: if (zs->zs_size == 0)
605: return -1;
1.1 deraadt 606: /* Round size down to integral number of codes. */
1.5 mickey 607: zs->zs_size = (zs->zs_size << 3) - (zs->zs_n_bits - 1);
1.1 deraadt 608: }
1.6 mickey 609:
610: bp = zs->zs_bp;
611: r_off = zs->zs_offset;
1.5 mickey 612: bits = zs->zs_n_bits;
1.1 deraadt 613:
614: /* Get to the first byte. */
615: bp += (r_off >> 3);
616: r_off &= 7;
617:
618: /* Get first part (low order bits). */
619: gcode = (*bp++ >> r_off);
620: bits -= (8 - r_off);
621: r_off = 8 - r_off; /* Now, roffset into gcode word. */
622:
623: /* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
624: if (bits >= 8) {
625: gcode |= *bp++ << r_off;
626: r_off += 8;
627: bits -= 8;
628: }
629:
630: /* High order bits. */
631: gcode |= (*bp & rmask[bits]) << r_off;
1.6 mickey 632: zs->zs_offset += zs->zs_n_bits;
1.1 deraadt 633:
634: return (gcode);
635: }
636:
1.11 ! deraadt 637: /* Table clear for block compress. */
1.1 deraadt 638: static int
1.11 ! deraadt 639: cl_block(struct s_zstate *zs)
1.1 deraadt 640: {
1.7 mpech 641: long rat;
1.1 deraadt 642:
1.5 mickey 643: zs->zs_checkpoint = zs->zs_in_count + CHECK_GAP;
1.1 deraadt 644:
1.5 mickey 645: if (zs->zs_in_count > 0x007fffff) { /* Shift will overflow. */
646: rat = zs->zs_bytes_out >> 8;
1.1 deraadt 647: if (rat == 0) /* Don't divide by zero. */
648: rat = 0x7fffffff;
649: else
1.5 mickey 650: rat = zs->zs_in_count / rat;
1.11 ! deraadt 651: } else {
! 652: /* 8 fractional bits. */
! 653: rat = (zs->zs_in_count << 8) / zs->zs_bytes_out;
! 654: }
1.5 mickey 655: if (rat > zs->zs_ratio)
656: zs->zs_ratio = rat;
1.1 deraadt 657: else {
1.5 mickey 658: zs->zs_ratio = 0;
659: cl_hash(zs, (count_int) zs->zs_hsize);
660: zs->zs_free_ent = FIRST;
661: zs->zs_clear_flg = 1;
1.1 deraadt 662: if (output(zs, (code_int) CLEAR) == -1)
663: return (-1);
664: }
665: return (0);
666: }
667:
1.11 ! deraadt 668: /* Reset code table. */
1.1 deraadt 669: static void
1.11 ! deraadt 670: cl_hash(struct s_zstate *zs, count_int cl_hsize)
1.1 deraadt 671: {
1.7 mpech 672: count_int *htab_p;
673: long i, m1;
1.1 deraadt 674:
675: m1 = -1;
1.5 mickey 676: htab_p = zs->zs_htab + cl_hsize;
1.1 deraadt 677: i = cl_hsize - 16;
678: do { /* Might use Sys V memset(3) here. */
679: *(htab_p - 16) = m1;
680: *(htab_p - 15) = m1;
681: *(htab_p - 14) = m1;
682: *(htab_p - 13) = m1;
683: *(htab_p - 12) = m1;
684: *(htab_p - 11) = m1;
685: *(htab_p - 10) = m1;
686: *(htab_p - 9) = m1;
687: *(htab_p - 8) = m1;
688: *(htab_p - 7) = m1;
689: *(htab_p - 6) = m1;
690: *(htab_p - 5) = m1;
691: *(htab_p - 4) = m1;
692: *(htab_p - 3) = m1;
693: *(htab_p - 2) = m1;
694: *(htab_p - 1) = m1;
695: htab_p -= 16;
696: } while ((i -= 16) >= 0);
697: for (i += 16; i > 0; i--)
698: *--htab_p = m1;
1.3 tholo 699: }
700:
701: FILE *
1.11 ! deraadt 702: zopen(const char *name, const char *mode, int bits)
1.5 mickey 703: {
704: int fd;
705: void *cookie;
706: if ((fd = open(name, (*mode=='r'? O_RDONLY:O_WRONLY|O_CREAT),
1.11 ! deraadt 707: S_IRUSR|S_IWUSR|S_IRGRP|S_IROTH)) == -1)
1.5 mickey 708: return NULL;
709: if ((cookie = z_open(fd, mode, bits)) == NULL) {
710: close(fd);
711: return NULL;
712: }
713: return funopen(cookie, (*mode == 'r'?zread:NULL),
1.11 ! deraadt 714: (*mode == 'w'?zwrite:NULL), NULL, zclose);
1.5 mickey 715: }
716:
717: void *
1.11 ! deraadt 718: z_open(int fd, const char *mode, int bits)
1.3 tholo 719: {
1.7 mpech 720: struct s_zstate *zs;
1.3 tholo 721:
1.4 millert 722: if ((mode[0] != 'r' && mode[0] != 'w') || mode[1] != '\0' ||
1.3 tholo 723: bits < 0 || bits > BITS) {
724: errno = EINVAL;
725: return (NULL);
726: }
727:
728: if ((zs = calloc(1, sizeof(struct s_zstate))) == NULL)
729: return (NULL);
730:
1.5 mickey 731: /* User settable max # bits/code. */
732: zs->zs_maxbits = bits ? bits : BITS;
733: /* Should NEVER generate this code. */
734: zs->zs_maxmaxcode = 1 << zs->zs_maxbits;
735: zs->zs_hsize = HSIZE; /* For dynamic table sizing. */
736: zs->zs_free_ent = 0; /* First unused entry. */
737: zs->zs_block_compress = BLOCK_MASK;
738: zs->zs_clear_flg = 0;
739: zs->zs_ratio = 0;
740: zs->zs_checkpoint = CHECK_GAP;
741: zs->zs_in_count = 1; /* Length of input. */
742: zs->zs_out_count = 0; /* # of codes output (for debugging).*/
743: zs->zs_state = S_START;
1.6 mickey 744: zs->zs_offset = 0;
1.5 mickey 745: zs->zs_size = 0;
746: zs->zs_mode = mode[0];
1.6 mickey 747: zs->zs_bp = zs->zs_ebp = zs->zs_buf;
1.3 tholo 748:
1.5 mickey 749: zs->zs_fd = fd;
750: return zs;
1.1 deraadt 751: }
752:
1.5 mickey 753: int
1.11 ! deraadt 754: z_check_header(int fd, struct stat *sb, const char *ofn)
1.1 deraadt 755: {
1.5 mickey 756: int f;
757: u_char buf[sizeof(z_magic)];
758: off_t off = lseek(fd, 0, SEEK_CUR);
1.1 deraadt 759:
1.5 mickey 760: f = (read(fd, buf, sizeof(buf)) == sizeof(buf) &&
1.11 ! deraadt 761: !memcmp(buf, z_magic, sizeof(buf)));
1.1 deraadt 762:
1.5 mickey 763: lseek (fd, off, SEEK_SET);
1.1 deraadt 764:
1.5 mickey 765: return f;
1.1 deraadt 766: }