Make bzcat, lzcat, xzcat, zcat, lzopcat, unlzop individually selectable
[oweals/busybox.git] / archival / lzop.c
1 /*
2    This file is part of the lzop file compressor.
3
4    Copyright (C) 1996..2003 Markus Franz Xaver Johannes Oberhumer
5    All Rights Reserved.
6
7    Markus F.X.J. Oberhumer <markus@oberhumer.com>
8    http://www.oberhumer.com/opensource/lzop/
9
10    lzop and the LZO library are free software; you can redistribute them
11    and/or modify them under the terms of the GNU General Public License as
12    published by the Free Software Foundation; either version 2 of
13    the License, or (at your option) any later version.
14
15    This program is distributed in the hope that it will be useful,
16    but WITHOUT ANY WARRANTY; without even the implied warranty of
17    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18    GNU General Public License for more details.
19
20    You should have received a copy of the GNU General Public License
21    along with this program; see the file COPYING.
22    If not, write to the Free Software Foundation, Inc.,
23    59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
24
25    "Minimalized" for busybox by Alain Knaff
26 */
27
28 //config:config LZOP
29 //config:       bool "lzop"
30 //config:       default y
31 //config:       help
32 //config:         Lzop compression/decompresion.
33 //config:
34 //config:config UNLZOP
35 //config:       bool "unlzop"
36 //config:       default y
37 //config:       help
38 //config:         Lzop decompresion.
39 //config:
40 //config:config LZOPCAT
41 //config:       bool "lzopcat"
42 //config:       default y
43 //config:       help
44 //config:         Alias to "unlzop -c".
45 //config:
46 //config:config LZOP_COMPR_HIGH
47 //config:       bool "lzop compression levels 7,8,9 (not very useful)"
48 //config:       default n
49 //config:       depends on LZOP || UNLZOP || LZOPCAT
50 //config:       help
51 //config:         High levels (7,8,9) of lzop compression. These levels
52 //config:         are actually slower than gzip at equivalent compression ratios
53 //config:         and take up 3.2K of code.
54
55 //applet:IF_LZOP(APPLET(lzop, BB_DIR_BIN, BB_SUID_DROP))
56 //applet:IF_UNLZOP(APPLET_ODDNAME(unlzop, lzop, BB_DIR_USR_BIN, BB_SUID_DROP, unlzop))
57 //applet:IF_LZOPCAT(APPLET_ODDNAME(lzopcat, lzop, BB_DIR_USR_BIN, BB_SUID_DROP, lzopcat))
58 //kbuild:lib-$(CONFIG_LZOP) += lzop.o
59 //kbuild:lib-$(CONFIG_UNLZOP) += lzop.o
60 //kbuild:lib-$(CONFIG_LZOPCAT) += lzop.o
61
62 //usage:#define lzop_trivial_usage
63 //usage:       "[-cfvd123456789CF] [FILE]..."
64 //usage:#define lzop_full_usage "\n\n"
65 //usage:       "        -1..9   Compression level"
66 //usage:     "\n        -d      Decompress"
67 //usage:     "\n        -c      Write to stdout"
68 //usage:     "\n        -f      Force"
69 //usage:     "\n        -v      Verbose"
70 //usage:     "\n        -F      Don't store or verify checksum"
71 //usage:     "\n        -C      Also write checksum of compressed block"
72 //usage:
73 //usage:#define lzopcat_trivial_usage
74 //usage:       "[-vF] [FILE]..."
75 //usage:#define lzopcat_full_usage "\n\n"
76 //usage:       "        -v      Verbose"
77 //usage:     "\n        -F      Don't verify checksum"
78 //usage:
79 //usage:#define unlzop_trivial_usage
80 //usage:       "[-cfvF] [FILE]..."
81 //usage:#define unlzop_full_usage "\n\n"
82 //usage:       "        -c      Write to stdout"
83 //usage:     "\n        -f      Force"
84 //usage:     "\n        -v      Verbose"
85 //usage:     "\n        -F      Don't verify checksum"
86
87 #include "libbb.h"
88 #include "common_bufsiz.h"
89 #include "bb_archive.h"
90 #include "liblzo_interface.h"
91
92 /* lzo-2.03/src/lzo_ptr.h */
93 #define pd(a,b)  ((unsigned)((a)-(b)))
94
95 #define lzo_version()                   LZO_VERSION
96 #define lzo_sizeof_dict_t               (sizeof(uint8_t*))
97
98 /* lzo-2.03/include/lzo/lzo1x.h */
99 #define LZO1X_1_MEM_COMPRESS    (16384 * lzo_sizeof_dict_t)
100 #define LZO1X_1_15_MEM_COMPRESS (32768 * lzo_sizeof_dict_t)
101 #define LZO1X_999_MEM_COMPRESS  (14 * 16384 * sizeof(short))
102
103 /* lzo-2.03/src/lzo1x_oo.c */
104 #define NO_LIT UINT_MAX
105
106 /**********************************************************************/
107 static void copy2(uint8_t* ip, const uint8_t* m_pos, unsigned off)
108 {
109         ip[0] = m_pos[0];
110         if (off == 1)
111                 ip[1] = m_pos[0];
112         else
113                 ip[1] = m_pos[1];
114 }
115
116 static void copy3(uint8_t* ip, const uint8_t* m_pos, unsigned off)
117 {
118         ip[0] = m_pos[0];
119         if (off == 1) {
120                 ip[2] = ip[1] = m_pos[0];
121         }
122         else if (off == 2) {
123                 ip[1] = m_pos[1];
124                 ip[2] = m_pos[0];
125         }
126         else {
127                 ip[1] = m_pos[1];
128                 ip[2] = m_pos[2];
129         }
130 }
131
132 /**********************************************************************/
133 // optimize a block of data.
134 /**********************************************************************/
135 #define TEST_IP         (ip < ip_end)
136 #define TEST_OP         (op <= op_end)
137
138 static NOINLINE int lzo1x_optimize(uint8_t *in, unsigned in_len,
139                 uint8_t *out, unsigned *out_len,
140                 void* wrkmem UNUSED_PARAM)
141 {
142         uint8_t* op;
143         uint8_t* ip;
144         unsigned t;
145         uint8_t* m_pos;
146         uint8_t* const ip_end = in + in_len;
147         uint8_t* const op_end = out + *out_len;
148         uint8_t* litp = NULL;
149         unsigned lit = 0;
150         unsigned next_lit = NO_LIT;
151         unsigned nl;
152         unsigned long o_m1_a = 0, o_m1_b = 0, o_m2 = 0, o_m3_a = 0, o_m3_b = 0;
153
154 //      LZO_UNUSED(wrkmem);
155
156         *out_len = 0;
157
158         op = out;
159         ip = in;
160
161         if (*ip > 17) {
162                 t = *ip++ - 17;
163                 if (t < 4)
164                         goto match_next;
165                 goto first_literal_run;
166         }
167
168         while (TEST_IP && TEST_OP) {
169                 t = *ip++;
170                 if (t >= 16)
171                         goto match;
172                 /* a literal run */
173                 litp = ip - 1;
174                 if (t == 0) {
175                         t = 15;
176                         while (*ip == 0)
177                                 t += 255, ip++;
178                         t += *ip++;
179                 }
180                 lit = t + 3;
181                 /* copy literals */
182  copy_literal_run:
183                 *op++ = *ip++;
184                 *op++ = *ip++;
185                 *op++ = *ip++;
186  first_literal_run:
187                 do *op++ = *ip++; while (--t > 0);
188
189                 t = *ip++;
190
191                 if (t >= 16)
192                         goto match;
193 #if defined(LZO1X)
194                 m_pos = op - 1 - 0x800;
195 #elif defined(LZO1Y)
196                 m_pos = op - 1 - 0x400;
197 #endif
198                 m_pos -= t >> 2;
199                 m_pos -= *ip++ << 2;
200                 *op++ = *m_pos++;
201                 *op++ = *m_pos++;
202                 *op++ = *m_pos++;
203                 lit = 0;
204                 goto match_done;
205
206
207                 /* handle matches */
208                 do {
209                         if (t < 16) { /* a M1 match */
210                                 m_pos = op - 1;
211                                 m_pos -= t >> 2;
212                                 m_pos -= *ip++ << 2;
213
214                                 if (litp == NULL)
215                                         goto copy_m1;
216
217                                 nl = ip[-2] & 3;
218                                 /* test if a match follows */
219                                 if (nl == 0 && lit == 1 && ip[0] >= 16) {
220                                         next_lit = nl;
221                                         /* adjust length of previous short run */
222                                         lit += 2;
223                                         *litp = (unsigned char)((*litp & ~3) | lit);
224                                         /* copy over the 2 literals that replace the match */
225                                         copy2(ip-2, m_pos, pd(op, m_pos));
226                                         o_m1_a++;
227                                 }
228                                 /* test if a literal run follows */
229                                 else
230                                 if (nl == 0
231                                  && ip[0] < 16
232                                  && ip[0] != 0
233                                  && (lit + 2 + ip[0] < 16)
234                                 ) {
235                                         t = *ip++;
236                                         /* remove short run */
237                                         *litp &= ~3;
238                                         /* copy over the 2 literals that replace the match */
239                                         copy2(ip-3+1, m_pos, pd(op, m_pos));
240                                         /* move literals 1 byte ahead */
241                                         litp += 2;
242                                         if (lit > 0)
243                                                 memmove(litp+1, litp, lit);
244                                         /* insert new length of long literal run */
245                                         lit += 2 + t + 3;
246                                         *litp = (unsigned char)(lit - 3);
247
248                                         o_m1_b++;
249                                         *op++ = *m_pos++;
250                                         *op++ = *m_pos++;
251                                         goto copy_literal_run;
252                                 }
253  copy_m1:
254                                 *op++ = *m_pos++;
255                                 *op++ = *m_pos++;
256                         } else {
257  match:
258                                 if (t >= 64) {                          /* a M2 match */
259                                         m_pos = op - 1;
260 #if defined(LZO1X)
261                                         m_pos -= (t >> 2) & 7;
262                                         m_pos -= *ip++ << 3;
263                                         t = (t >> 5) - 1;
264 #elif defined(LZO1Y)
265                                         m_pos -= (t >> 2) & 3;
266                                         m_pos -= *ip++ << 2;
267                                         t = (t >> 4) - 3;
268 #endif
269                                         if (litp == NULL)
270                                                 goto copy_m;
271
272                                         nl = ip[-2] & 3;
273                                         /* test if in beetween two long literal runs */
274                                         if (t == 1 && lit > 3 && nl == 0
275                                          && ip[0] < 16 && ip[0] != 0 && (lit + 3 + ip[0] < 16)
276                                         ) {
277                                                 t = *ip++;
278                                                 /* copy over the 3 literals that replace the match */
279                                                 copy3(ip-1-2, m_pos, pd(op, m_pos));
280                                                 /* set new length of previous literal run */
281                                                 lit += 3 + t + 3;
282                                                 *litp = (unsigned char)(lit - 3);
283                                                 o_m2++;
284                                                 *op++ = *m_pos++;
285                                                 *op++ = *m_pos++;
286                                                 *op++ = *m_pos++;
287                                                 goto copy_literal_run;
288                                         }
289                                 } else {
290                                         if (t >= 32) {                  /* a M3 match */
291                                                 t &= 31;
292                                                 if (t == 0) {
293                                                         t = 31;
294                                                         while (*ip == 0)
295                                                                 t += 255, ip++;
296                                                         t += *ip++;
297                                                 }
298                                                 m_pos = op - 1;
299                                                 m_pos -= *ip++ >> 2;
300                                                 m_pos -= *ip++ << 6;
301                                         } else {                                        /* a M4 match */
302                                                 m_pos = op;
303                                                 m_pos -= (t & 8) << 11;
304                                                 t &= 7;
305                                                 if (t == 0) {
306                                                         t = 7;
307                                                         while (*ip == 0)
308                                                                 t += 255, ip++;
309                                                         t += *ip++;
310                                                 }
311                                                 m_pos -= *ip++ >> 2;
312                                                 m_pos -= *ip++ << 6;
313                                                 if (m_pos == op)
314                                                         goto eof_found;
315                                                 m_pos -= 0x4000;
316                                         }
317                                         if (litp == NULL)
318                                                 goto copy_m;
319
320                                         nl = ip[-2] & 3;
321                                         /* test if in beetween two matches */
322                                         if (t == 1 && lit == 0 && nl == 0 && ip[0] >= 16) {
323                                                 next_lit = nl;
324                                                 /* make a previous short run */
325                                                 lit += 3;
326                                                 *litp = (unsigned char)((*litp & ~3) | lit);
327                                                 /* copy over the 3 literals that replace the match */
328                                                 copy3(ip-3, m_pos, pd(op, m_pos));
329                                                 o_m3_a++;
330                                         }
331                                         /* test if a literal run follows */
332                                         else if (t == 1 && lit <= 3 && nl == 0
333                                          && ip[0] < 16 && ip[0] != 0 && (lit + 3 + ip[0] < 16)
334                                         ) {
335                                                 t = *ip++;
336                                                 /* remove short run */
337                                                 *litp &= ~3;
338                                                 /* copy over the 3 literals that replace the match */
339                                                 copy3(ip-4+1, m_pos, pd(op, m_pos));
340                                                 /* move literals 1 byte ahead */
341                                                 litp += 2;
342                                                 if (lit > 0)
343                                                         memmove(litp+1,litp,lit);
344                                                 /* insert new length of long literal run */
345                                                 lit += 3 + t + 3;
346                                                 *litp = (unsigned char)(lit - 3);
347
348                                                 o_m3_b++;
349                                                 *op++ = *m_pos++;
350                                                 *op++ = *m_pos++;
351                                                 *op++ = *m_pos++;
352                                                 goto copy_literal_run;
353                                         }
354                                 }
355  copy_m:
356                                 *op++ = *m_pos++;
357                                 *op++ = *m_pos++;
358                                 do *op++ = *m_pos++; while (--t > 0);
359                         }
360
361  match_done:
362                         if (next_lit == NO_LIT) {
363                                 t = ip[-2] & 3;
364                                 lit = t;
365                                 litp = ip - 2;
366                         }
367                         else
368                                 t = next_lit;
369                         next_lit = NO_LIT;
370                         if (t == 0)
371                                 break;
372                         /* copy literals */
373  match_next:
374                         do *op++ = *ip++; while (--t > 0);
375                         t = *ip++;
376                 } while (TEST_IP && TEST_OP);
377         }
378
379         /* no EOF code was found */
380         *out_len = pd(op, out);
381         return LZO_E_EOF_NOT_FOUND;
382
383  eof_found:
384 //      LZO_UNUSED(o_m1_a); LZO_UNUSED(o_m1_b); LZO_UNUSED(o_m2);
385 //      LZO_UNUSED(o_m3_a); LZO_UNUSED(o_m3_b);
386         *out_len = pd(op, out);
387         return (ip == ip_end ? LZO_E_OK :
388                 (ip < ip_end ? LZO_E_INPUT_NOT_CONSUMED : LZO_E_INPUT_OVERRUN));
389 }
390
391 /**********************************************************************/
392 #define F_OS F_OS_UNIX
393 #define F_CS F_CS_NATIVE
394
395 /**********************************************************************/
396 #define ADLER32_INIT_VALUE 1
397 #define CRC32_INIT_VALUE   0
398
399 /**********************************************************************/
400 enum {
401         M_LZO1X_1    = 1,
402         M_LZO1X_1_15 = 2,
403         M_LZO1X_999  = 3,
404 };
405
406 /**********************************************************************/
407 /* header flags */
408 #define F_ADLER32_D     0x00000001L
409 #define F_ADLER32_C     0x00000002L
410 #define F_H_EXTRA_FIELD 0x00000040L
411 #define F_H_GMTDIFF     0x00000080L
412 #define F_CRC32_D       0x00000100L
413 #define F_CRC32_C       0x00000200L
414 #define F_H_FILTER      0x00000800L
415 #define F_H_CRC32       0x00001000L
416 #define F_MASK          0x00003FFFL
417
418 /* operating system & file system that created the file [mostly unused] */
419 #define F_OS_UNIX       0x03000000L
420 #define F_OS_SHIFT      24
421 #define F_OS_MASK       0xff000000L
422
423 /* character set for file name encoding [mostly unused] */
424 #define F_CS_NATIVE     0x00000000L
425 #define F_CS_SHIFT      20
426 #define F_CS_MASK       0x00f00000L
427
428 /* these bits must be zero */
429 #define F_RESERVED      ((F_MASK | F_OS_MASK | F_CS_MASK) ^ 0xffffffffL)
430
431 typedef struct chksum_t {
432         uint32_t f_adler32;
433         uint32_t f_crc32;
434 } chksum_t;
435
436 typedef struct header_t {
437         unsigned version;
438         unsigned lib_version;
439         unsigned version_needed_to_extract;
440         uint32_t flags;
441         uint32_t mode;
442         uint32_t mtime;
443         uint32_t gmtdiff;
444         uint32_t header_checksum;
445
446         uint32_t extra_field_len;
447         uint32_t extra_field_checksum;
448
449         unsigned char method;
450         unsigned char level;
451
452         /* info */
453         char name[255+1];
454 } header_t;
455
456 struct globals {
457         /*const uint32_t *lzo_crc32_table;*/
458         chksum_t chksum_in;
459         chksum_t chksum_out;
460 } FIX_ALIASING;
461 #define G (*(struct globals*)bb_common_bufsiz1)
462 #define INIT_G() do { setup_common_bufsiz(); } while (0)
463 //#define G (*ptr_to_globals)
464 //#define INIT_G() do {
465 //      SET_PTR_TO_GLOBALS(xzalloc(sizeof(G)));
466 //} while (0)
467
468
469 /**********************************************************************/
470 #define LZOP_VERSION            0x1010
471 //#define LZOP_VERSION_STRING     "1.01"
472 //#define LZOP_VERSION_DATE       "Apr 27th 2003"
473
474 #define OPTION_STRING "cfvqdt123456789CF"
475
476 /* Note: must be kept in sync with archival/bbunzip.c */
477 enum {
478         OPT_STDOUT      = (1 << 0),
479         OPT_FORCE       = (1 << 1),
480         OPT_VERBOSE     = (1 << 2),
481         OPT_QUIET       = (1 << 3),
482         OPT_DECOMPRESS  = (1 << 4),
483         OPT_TEST        = (1 << 5),
484         OPT_1           = (1 << 6),
485         OPT_2           = (1 << 7),
486         OPT_3           = (1 << 8),
487         OPT_4           = (1 << 9),
488         OPT_5           = (1 << 10),
489         OPT_6           = (1 << 11),
490         OPT_789         = (7 << 12),
491         OPT_7           = (1 << 13),
492         OPT_8           = (1 << 14),
493         OPT_C           = (1 << 15),
494         OPT_F           = (1 << 16),
495 };
496
497 /**********************************************************************/
498 // adler32 checksum
499 // adapted from free code by Mark Adler <madler@alumni.caltech.edu>
500 // see http://www.zlib.org/
501 /**********************************************************************/
502 static FAST_FUNC uint32_t
503 lzo_adler32(uint32_t adler, const uint8_t* buf, unsigned len)
504 {
505         enum {
506                 LZO_BASE = 65521, /* largest prime smaller than 65536 */
507                 /* NMAX is the largest n such that
508                  * 255n(n+1)/2 + (n+1)(BASE-1) <= 2^32-1 */
509                 LZO_NMAX = 5552,
510         };
511         uint32_t s1 = adler & 0xffff;
512         uint32_t s2 = (adler >> 16) & 0xffff;
513         unsigned k;
514
515         if (buf == NULL)
516                 return 1;
517
518         while (len > 0) {
519                 k = len < LZO_NMAX ? (unsigned) len : LZO_NMAX;
520                 len -= k;
521                 if (k != 0) do {
522                         s1 += *buf++;
523                         s2 += s1;
524                 } while (--k > 0);
525                 s1 %= LZO_BASE;
526                 s2 %= LZO_BASE;
527         }
528         return (s2 << 16) | s1;
529 }
530
531 static FAST_FUNC uint32_t
532 lzo_crc32(uint32_t c, const uint8_t* buf, unsigned len)
533 {
534         //if (buf == NULL) - impossible
535         //      return 0;
536
537         return ~crc32_block_endian0(~c, buf, len, global_crc32_table);
538 }
539
540 /**********************************************************************/
541 static void init_chksum(chksum_t *ct)
542 {
543         ct->f_adler32 = ADLER32_INIT_VALUE;
544         ct->f_crc32 = CRC32_INIT_VALUE;
545 }
546
547 static void add_bytes_to_chksum(chksum_t *ct, const void* buf, int cnt)
548 {
549         /* We need to handle the two checksums at once, because at the
550          * beginning of the header, we don't know yet which one we'll
551          * eventually need */
552         ct->f_adler32 = lzo_adler32(ct->f_adler32, (const uint8_t*)buf, cnt);
553         ct->f_crc32 = lzo_crc32(ct->f_crc32, (const uint8_t*)buf, cnt);
554 }
555
556 static uint32_t chksum_getresult(chksum_t *ct, const header_t *h)
557 {
558         return (h->flags & F_H_CRC32) ? ct->f_crc32 : ct->f_adler32;
559 }
560
561 /**********************************************************************/
562 static uint32_t read32(void)
563 {
564         uint32_t v;
565         xread(0, &v, 4);
566         return ntohl(v);
567 }
568
569 static void write32(uint32_t v)
570 {
571         v = htonl(v);
572         xwrite(1, &v, 4);
573 }
574
575 static void f_write(const void* buf, int cnt)
576 {
577         xwrite(1, buf, cnt);
578         add_bytes_to_chksum(&G.chksum_out, buf, cnt);
579 }
580
581 static void f_read(void* buf, int cnt)
582 {
583         xread(0, buf, cnt);
584         add_bytes_to_chksum(&G.chksum_in, buf, cnt);
585 }
586
587 static int f_read8(void)
588 {
589         uint8_t v;
590         f_read(&v, 1);
591         return v;
592 }
593
594 static void f_write8(uint8_t v)
595 {
596         f_write(&v, 1);
597 }
598
599 static unsigned f_read16(void)
600 {
601         uint16_t v;
602         f_read(&v, 2);
603         return ntohs(v);
604 }
605
606 static void f_write16(uint16_t v)
607 {
608         v = htons(v);
609         f_write(&v, 2);
610 }
611
612 static uint32_t f_read32(void)
613 {
614         uint32_t v;
615         f_read(&v, 4);
616         return ntohl(v);
617 }
618
619 static void f_write32(uint32_t v)
620 {
621         v = htonl(v);
622         f_write(&v, 4);
623 }
624
625 /**********************************************************************/
626 static int lzo_get_method(header_t *h)
627 {
628         /* check method */
629         if (h->method == M_LZO1X_1) {
630                 if (h->level == 0)
631                         h->level = 3;
632         } else if (h->method == M_LZO1X_1_15) {
633                 if (h->level == 0)
634                         h->level = 1;
635         } else if (h->method == M_LZO1X_999) {
636                 if (h->level == 0)
637                         h->level = 9;
638         } else
639                 return -1;              /* not a LZO method */
640
641         /* check compression level */
642         if (h->level < 1 || h->level > 9)
643                 return 15;
644
645         return 0;
646 }
647
648 /**********************************************************************/
649 #define LZO_BLOCK_SIZE  (256 * 1024l)
650 #define MAX_BLOCK_SIZE  (64 * 1024l * 1024l)    /* DO NOT CHANGE */
651
652 /* LZO may expand uncompressible data by a small amount */
653 #define MAX_COMPRESSED_SIZE(x)  ((x) + (x) / 16 + 64 + 3)
654
655 /**********************************************************************/
656 // compress a file
657 /**********************************************************************/
658 static NOINLINE int lzo_compress(const header_t *h)
659 {
660         unsigned block_size = LZO_BLOCK_SIZE;
661         int r = 0; /* LZO_E_OK */
662         uint8_t *const b1 = xzalloc(block_size);
663         uint8_t *const b2 = xzalloc(MAX_COMPRESSED_SIZE(block_size));
664         unsigned src_len = 0, dst_len = 0;
665         uint32_t d_adler32 = ADLER32_INIT_VALUE;
666         uint32_t d_crc32 = CRC32_INIT_VALUE;
667         int l;
668         uint8_t *wrk_mem = NULL;
669
670         if (h->method == M_LZO1X_1)
671                 wrk_mem = xzalloc(LZO1X_1_MEM_COMPRESS);
672         else if (h->method == M_LZO1X_1_15)
673                 wrk_mem = xzalloc(LZO1X_1_15_MEM_COMPRESS);
674         else if (h->method == M_LZO1X_999)
675                 wrk_mem = xzalloc(LZO1X_999_MEM_COMPRESS);
676
677         for (;;) {
678                 /* read a block */
679                 l = full_read(0, b1, block_size);
680                 src_len = (l > 0 ? l : 0);
681
682                 /* write uncompressed block size */
683                 write32(src_len);
684
685                 /* exit if last block */
686                 if (src_len == 0)
687                         break;
688
689                 /* compute checksum of uncompressed block */
690                 if (h->flags & F_ADLER32_D)
691                         d_adler32 = lzo_adler32(ADLER32_INIT_VALUE, b1, src_len);
692                 if (h->flags & F_CRC32_D)
693                         d_crc32 = lzo_crc32(CRC32_INIT_VALUE, b1, src_len);
694
695                 /* compress */
696                 if (h->method == M_LZO1X_1)
697                         r = lzo1x_1_compress(b1, src_len, b2, &dst_len, wrk_mem);
698                 else if (h->method == M_LZO1X_1_15)
699                         r = lzo1x_1_15_compress(b1, src_len, b2, &dst_len, wrk_mem);
700 #if ENABLE_LZOP_COMPR_HIGH
701                 else if (h->method == M_LZO1X_999)
702                         r = lzo1x_999_compress_level(b1, src_len, b2, &dst_len,
703                                                 wrk_mem, h->level);
704 #endif
705                 else
706                         bb_error_msg_and_die("internal error");
707
708                 if (r != 0) /* not LZO_E_OK */
709                         bb_error_msg_and_die("internal error - compression failed");
710
711                 /* write compressed block size */
712                 if (dst_len < src_len) {
713                         /* optimize */
714                         if (h->method == M_LZO1X_999) {
715                                 unsigned new_len = src_len;
716                                 r = lzo1x_optimize(b2, dst_len, b1, &new_len, NULL);
717                                 if (r != 0 /*LZO_E_OK*/ || new_len != src_len)
718                                         bb_error_msg_and_die("internal error - optimization failed");
719                         }
720                         write32(dst_len);
721                 } else {
722                         /* data actually expanded => store data uncompressed */
723                         write32(src_len);
724                 }
725
726                 /* write checksum of uncompressed block */
727                 if (h->flags & F_ADLER32_D)
728                         write32(d_adler32);
729                 if (h->flags & F_CRC32_D)
730                         write32(d_crc32);
731
732                 if (dst_len < src_len) {
733                         /* write checksum of compressed block */
734                         if (h->flags & F_ADLER32_C)
735                                 write32(lzo_adler32(ADLER32_INIT_VALUE, b2, dst_len));
736                         if (h->flags & F_CRC32_C)
737                                 write32(lzo_crc32(CRC32_INIT_VALUE, b2, dst_len));
738                         /* write compressed block data */
739                         xwrite(1, b2, dst_len);
740                 } else {
741                         /* write uncompressed block data */
742                         xwrite(1, b1, src_len);
743                 }
744         }
745
746         free(wrk_mem);
747         free(b1);
748         free(b2);
749         return 1;
750 }
751
752 static FAST_FUNC void lzo_check(
753                 uint32_t init,
754                 uint8_t* buf, unsigned len,
755                 uint32_t FAST_FUNC (*fn)(uint32_t, const uint8_t*, unsigned),
756                 uint32_t ref)
757 {
758         /* This function, by having the same order of parameters
759          * as fn, and by being marked FAST_FUNC (same as fn),
760          * saves a dozen bytes of code.
761          */
762         uint32_t c = fn(init, buf, len);
763         if (c != ref)
764                 bb_error_msg_and_die("checksum error");
765 }
766
767 /**********************************************************************/
768 // decompress a file
769 /**********************************************************************/
770 static NOINLINE int lzo_decompress(const header_t *h)
771 {
772         unsigned block_size = LZO_BLOCK_SIZE;
773         int r;
774         uint32_t src_len, dst_len;
775         uint32_t c_adler32 = ADLER32_INIT_VALUE;
776         uint32_t d_adler32 = ADLER32_INIT_VALUE;
777         uint32_t c_crc32 = CRC32_INIT_VALUE, d_crc32 = CRC32_INIT_VALUE;
778         uint8_t *b1;
779         uint32_t mcs_block_size = MAX_COMPRESSED_SIZE(block_size);
780         uint8_t *b2 = NULL;
781
782         for (;;) {
783                 uint8_t *dst;
784
785                 /* read uncompressed block size */
786                 dst_len = read32();
787
788                 /* exit if last block */
789                 if (dst_len == 0)
790                         break;
791
792                 /* error if split file */
793                 if (dst_len == 0xffffffffL)
794                         /* should not happen - not yet implemented */
795                         bb_error_msg_and_die("this file is a split lzop file");
796
797                 if (dst_len > MAX_BLOCK_SIZE)
798                         bb_error_msg_and_die("corrupted data");
799
800                 /* read compressed block size */
801                 src_len = read32();
802                 if (src_len <= 0 || src_len > dst_len)
803                         bb_error_msg_and_die("corrupted data");
804
805                 if (dst_len > block_size) {
806                         if (b2) {
807                                 free(b2);
808                                 b2 = NULL;
809                         }
810                         block_size = dst_len;
811                         mcs_block_size = MAX_COMPRESSED_SIZE(block_size);
812                 }
813
814                 /* read checksum of uncompressed block */
815                 if (h->flags & F_ADLER32_D)
816                         d_adler32 = read32();
817                 if (h->flags & F_CRC32_D)
818                         d_crc32 = read32();
819
820                 /* read checksum of compressed block */
821                 if (src_len < dst_len) {
822                         if (h->flags & F_ADLER32_C)
823                                 c_adler32 = read32();
824                         if (h->flags & F_CRC32_C)
825                                 c_crc32 = read32();
826                 }
827
828                 if (b2 == NULL)
829                         b2 = xzalloc(mcs_block_size);
830                 /* read the block into the end of our buffer */
831                 b1 = b2 + mcs_block_size - src_len;
832                 xread(0, b1, src_len);
833
834                 if (src_len < dst_len) {
835                         unsigned d = dst_len;
836
837                         if (!(option_mask32 & OPT_F)) {
838                                 /* verify checksum of compressed block */
839                                 if (h->flags & F_ADLER32_C)
840                                         lzo_check(ADLER32_INIT_VALUE,
841                                                         b1, src_len,
842                                                         lzo_adler32, c_adler32);
843                                 if (h->flags & F_CRC32_C)
844                                         lzo_check(CRC32_INIT_VALUE,
845                                                         b1, src_len,
846                                                         lzo_crc32, c_crc32);
847                         }
848
849                         /* decompress */
850 //                      if (option_mask32 & OPT_F)
851 //                              r = lzo1x_decompress(b1, src_len, b2, &d, NULL);
852 //                      else
853                                 r = lzo1x_decompress_safe(b1, src_len, b2, &d, NULL);
854
855                         if (r != 0 /*LZO_E_OK*/ || dst_len != d) {
856                                 bb_error_msg_and_die("corrupted data");
857                         }
858                         dst = b2;
859                 } else {
860                         /* "stored" block => no decompression */
861                         dst = b1;
862                 }
863
864                 if (!(option_mask32 & OPT_F)) {
865                         /* verify checksum of uncompressed block */
866                         if (h->flags & F_ADLER32_D)
867                                 lzo_check(ADLER32_INIT_VALUE,
868                                         dst, dst_len,
869                                         lzo_adler32, d_adler32);
870                         if (h->flags & F_CRC32_D)
871                                 lzo_check(CRC32_INIT_VALUE,
872                                         dst, dst_len,
873                                         lzo_crc32, d_crc32);
874                 }
875
876                 /* write uncompressed block data */
877                 xwrite(1, dst, dst_len);
878         }
879
880         free(b2);
881         return 1;
882 }
883
884 /**********************************************************************/
885 // lzop file signature (shamelessly borrowed from PNG)
886 /**********************************************************************/
887 /*
888  * The first nine bytes of a lzop file always contain the following values:
889  *
890  *                                 0   1   2   3   4   5   6   7   8
891  *                               --- --- --- --- --- --- --- --- ---
892  * (hex)                          89  4c  5a  4f  00  0d  0a  1a  0a
893  * (decimal)                     137  76  90  79   0  13  10  26  10
894  * (C notation - ASCII)         \211   L   Z   O  \0  \r  \n \032 \n
895  */
896
897 /* (vda) comparison with lzop v1.02rc1 ("lzop -1 <FILE" cmd):
898  * Only slight differences in header:
899  * -00000000  89 4c 5a 4f 00 0d 0a 1a 0a 10 20 20 20 09 40 02
900  * +00000000  89 4c 5a 4f 00 0d 0a 1a 0a 10 10 20 30 09 40 02
901  *                                       ^^^^^ ^^^^^
902  *                                     version lib_version
903  * -00000010  01 03 00 00 0d 00 00 81 a4 49 f7 a6 3f 00 00 00
904  * +00000010  01 03 00 00 01 00 00 00 00 00 00 00 00 00 00 00
905  *               ^^^^^^^^^^^ ^^^^^^^^^^^ ^^^^^^^^^^^
906  *               flags       mode        mtime
907  * -00000020  00 00 2d 67 04 17 00 04 00 00 00 03 ed ec 9d 6d
908  * +00000020  00 00 10 5f 00 c1 00 04 00 00 00 03 ed ec 9d 6d
909  *                  ^^^^^^^^^^^
910  *                  chksum_out
911  * The rest is identical.
912 */
913 static const unsigned char lzop_magic[9] ALIGN1 = {
914         0x89, 0x4c, 0x5a, 0x4f, 0x00, 0x0d, 0x0a, 0x1a, 0x0a
915 };
916
917 /* This coding is derived from Alexander Lehmann's pngcheck code. */
918 static void check_magic(void)
919 {
920         unsigned char magic[sizeof(lzop_magic)];
921         xread(0, magic, sizeof(magic));
922         if (memcmp(magic, lzop_magic, sizeof(lzop_magic)) != 0)
923                 bb_error_msg_and_die("bad magic number");
924 }
925
926 /**********************************************************************/
927 // lzop file header
928 /**********************************************************************/
929 static void write_header(const header_t *h)
930 {
931         int l;
932
933         xwrite(1, lzop_magic, sizeof(lzop_magic));
934
935         init_chksum(&G.chksum_out);
936
937         f_write16(h->version);
938         f_write16(h->lib_version);
939         f_write16(h->version_needed_to_extract);
940         f_write8(h->method);
941         f_write8(h->level);
942         f_write32(h->flags);
943         f_write32(h->mode);
944         f_write32(h->mtime);
945         f_write32(h->gmtdiff);
946
947         l = (int) strlen(h->name);
948         f_write8(l);
949         if (l)
950                 f_write(h->name, l);
951
952         f_write32(chksum_getresult(&G.chksum_out, h));
953 }
954
955 static int read_header(header_t *h)
956 {
957         int r;
958         int l;
959         uint32_t checksum;
960
961         memset(h, 0, sizeof(*h));
962         h->version_needed_to_extract = 0x0900;  /* first lzop version */
963         h->level = 0;
964
965         init_chksum(&G.chksum_in);
966
967         h->version = f_read16();
968         if (h->version < 0x0900)
969                 return 3;
970         h->lib_version = f_read16();
971         if (h->version >= 0x0940) {
972                 h->version_needed_to_extract = f_read16();
973                 if (h->version_needed_to_extract > LZOP_VERSION)
974                         return 16;
975                 if (h->version_needed_to_extract < 0x0900)
976                         return 3;
977         }
978         h->method = f_read8();
979         if (h->version >= 0x0940)
980                 h->level = f_read8();
981         h->flags = f_read32();
982         if (h->flags & F_H_FILTER)
983                 return 16; /* filter not supported */
984         h->mode = f_read32();
985         h->mtime = f_read32();
986         if (h->version >= 0x0940)
987                 h->gmtdiff = f_read32();
988
989         l = f_read8();
990         if (l > 0)
991                 f_read(h->name, l);
992         h->name[l] = 0;
993
994         checksum = chksum_getresult(&G.chksum_in, h);
995         h->header_checksum = f_read32();
996         if (h->header_checksum != checksum)
997                 return 2;
998
999         if (h->method <= 0)
1000                 return 14;
1001         r = lzo_get_method(h);
1002         if (r != 0)
1003                 return r;
1004
1005         /* check reserved flags */
1006         if (h->flags & F_RESERVED)
1007                 return -13;
1008
1009         /* skip extra field [not used yet] */
1010         if (h->flags & F_H_EXTRA_FIELD) {
1011                 uint32_t k;
1012
1013                 /* note: the checksum also covers the length */
1014                 init_chksum(&G.chksum_in);
1015                 h->extra_field_len = f_read32();
1016                 for (k = 0; k < h->extra_field_len; k++)
1017                         f_read8();
1018                 checksum = chksum_getresult(&G.chksum_in, h);
1019                 h->extra_field_checksum = f_read32();
1020                 if (h->extra_field_checksum != checksum)
1021                         return 3;
1022         }
1023
1024         return 0;
1025 }
1026
1027 static void p_header(header_t *h)
1028 {
1029         int r;
1030
1031         r = read_header(h);
1032         if (r == 0)
1033                 return;
1034         bb_error_msg_and_die("header_error %d", r);
1035 }
1036
1037 /**********************************************************************/
1038 // compress
1039 /**********************************************************************/
1040 static void lzo_set_method(header_t *h)
1041 {
1042         int level = 1;
1043
1044         if (option_mask32 & OPT_1) {
1045                 h->method = M_LZO1X_1_15;
1046         } else if (option_mask32 & OPT_789) {
1047 #if ENABLE_LZOP_COMPR_HIGH
1048                 h->method = M_LZO1X_999;
1049                 if (option_mask32 & OPT_7)
1050                         level = 7;
1051                 else if (option_mask32 & OPT_8)
1052                         level = 8;
1053                 else
1054                         level = 9;
1055 #else
1056                 bb_error_msg_and_die("high compression not compiled in");
1057 #endif
1058         } else { /* levels 2..6 or none (defaults to level 3) */
1059                 h->method = M_LZO1X_1;
1060                 level = 5; /* levels 2-6 are actually the same */
1061         }
1062
1063         h->level = level;
1064 }
1065
1066 static int do_lzo_compress(void)
1067 {
1068         header_t header;
1069
1070 #define h (&header)
1071         memset(h, 0, sizeof(*h));
1072
1073         lzo_set_method(h);
1074
1075         h->version = (LZOP_VERSION & 0xffff);
1076         h->version_needed_to_extract = 0x0940;
1077         h->lib_version = lzo_version() & 0xffff;
1078
1079         h->flags = (F_OS & F_OS_MASK) | (F_CS & F_CS_MASK);
1080
1081         if (!(option_mask32 & OPT_F) || h->method == M_LZO1X_999) {
1082                 h->flags |= F_ADLER32_D;
1083                 if (option_mask32 & OPT_C)
1084                         h->flags |= F_ADLER32_C;
1085         }
1086         write_header(h);
1087         return lzo_compress(h);
1088 #undef h
1089 }
1090
1091 /**********************************************************************/
1092 // decompress
1093 /**********************************************************************/
1094 static int do_lzo_decompress(void)
1095 {
1096         header_t header;
1097
1098         check_magic();
1099         p_header(&header);
1100         return lzo_decompress(&header);
1101 }
1102
1103 static char* FAST_FUNC make_new_name_lzop(char *filename, const char *expected_ext UNUSED_PARAM)
1104 {
1105         if (option_mask32 & OPT_DECOMPRESS) {
1106                 char *extension = strrchr(filename, '.');
1107                 if (!extension || strcmp(extension + 1, "lzo") != 0)
1108                         return xasprintf("%s.out", filename);
1109                 *extension = '\0';
1110                 return filename;
1111         }
1112         return xasprintf("%s.lzo", filename);
1113 }
1114
1115 static IF_DESKTOP(long long) int FAST_FUNC pack_lzop(transformer_state_t *xstate UNUSED_PARAM)
1116 {
1117         if (option_mask32 & OPT_DECOMPRESS)
1118                 return do_lzo_decompress();
1119         return do_lzo_compress();
1120 }
1121
1122 int lzop_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
1123 int lzop_main(int argc UNUSED_PARAM, char **argv)
1124 {
1125         getopt32(argv, OPTION_STRING);
1126         argv += optind;
1127         /* lzopcat? */
1128         if (ENABLE_LZOPCAT && applet_name[4] == 'c')
1129                 option_mask32 |= (OPT_STDOUT | OPT_DECOMPRESS);
1130         /* unlzop? */
1131         if (ENABLE_UNLZOP && applet_name[4] == 'o')
1132                 option_mask32 |= OPT_DECOMPRESS;
1133
1134         global_crc32_table = crc32_filltable(NULL, 0);
1135         return bbunpack(argv, pack_lzop, make_new_name_lzop, /*unused:*/ NULL);
1136 }