1 /***************************************************************************
2 * lzx.c - LZX decompression routines *
3 * ------------------- *
5 * maintainer: Jed Wing <jedwin@ugcs.caltech.edu> *
6 * source: modified lzx.c from cabextract v0.5 *
7 * notes: This file was taken from cabextract v0.5, which was, *
8 * itself, a modified version of the lzx decompression code *
11 * platforms: In its current incarnation, this file has been tested on *
12 * two different Linux platforms (one, redhat-based, with a *
13 * 2.1.2 glibc and gcc 2.95.x, and the other, Debian, with *
14 * 2.2.4 glibc and both gcc 2.95.4 and gcc 3.0.2). Both were *
15 * Intel x86 compatible machines. *
16 ***************************************************************************/
18 /***************************************************************************
20 * Copyright(C) Stuart Caie *
22 * This library is free software; you can redistribute it and/or modify *
23 * it under the terms of the GNU Lesser General Public License as *
24 * published by the Free Software Foundation; either version 2.1 of the *
25 * License, or (at your option) any later version. *
27 ***************************************************************************/
35 typedef unsigned char UBYTE; /* 8 bits exactly */
36 typedef unsigned short UWORD; /* 16 bits (or more) */
37 typedef unsigned int ULONG; /* 32 bits (or more) */
38 typedef signed int LONG; /* 32 bits (or more) */
40 /* some constants defined by the LZX specification */
41 #define LZX_MIN_MATCH (2)
42 #define LZX_MAX_MATCH (257)
43 #define LZX_NUM_CHARS (256)
44 #define LZX_BLOCKTYPE_INVALID (0) /* also blocktypes 4-7 invalid */
45 #define LZX_BLOCKTYPE_VERBATIM (1)
46 #define LZX_BLOCKTYPE_ALIGNED (2)
47 #define LZX_BLOCKTYPE_UNCOMPRESSED (3)
48 #define LZX_PRETREE_NUM_ELEMENTS (20)
49 #define LZX_ALIGNED_NUM_ELEMENTS (8) /* aligned offset tree #elements */
50 #define LZX_NUM_PRIMARY_LENGTHS (7) /* this one missing from spec! */
51 #define LZX_NUM_SECONDARY_LENGTHS (249) /* length tree #elements */
53 /* LZX huffman defines: tweak tablebits as desired */
54 #define LZX_PRETREE_MAXSYMBOLS (LZX_PRETREE_NUM_ELEMENTS)
55 #define LZX_PRETREE_TABLEBITS (6)
56 #define LZX_MAINTREE_MAXSYMBOLS (LZX_NUM_CHARS + 50*8)
57 #define LZX_MAINTREE_TABLEBITS (12)
58 #define LZX_LENGTH_MAXSYMBOLS (LZX_NUM_SECONDARY_LENGTHS+1)
59 #define LZX_LENGTH_TABLEBITS (12)
60 #define LZX_ALIGNED_MAXSYMBOLS (LZX_ALIGNED_NUM_ELEMENTS)
61 #define LZX_ALIGNED_TABLEBITS (7)
63 #define LZX_LENTABLE_SAFETY (64) /* we allow length table decoding overruns */
65 #define LZX_DECLARE_TABLE(tbl) \
66 UWORD tbl##_table[(1<<LZX_##tbl##_TABLEBITS) + (LZX_##tbl##_MAXSYMBOLS<<1)];\
67 UBYTE tbl##_len [LZX_##tbl##_MAXSYMBOLS + LZX_LENTABLE_SAFETY]
71 UBYTE *window; /* the actual decoding window */
72 ULONG window_size; /* window size (32Kb through 2Mb) */
73 ULONG actual_size; /* window size when it was first allocated */
74 ULONG window_posn; /* current offset within the window */
75 ULONG R0, R1, R2; /* for the LRU offset system */
76 UWORD main_elements; /* number of main tree elements */
77 int header_read; /* have we started decoding at all yet? */
78 UWORD block_type; /* type of this block */
79 ULONG block_length; /* uncompressed length of this block */
80 ULONG block_remaining; /* uncompressed bytes still left to decode */
81 ULONG frames_read; /* the number of CFDATA blocks processed */
82 LONG intel_filesize; /* magic header value used for transform */
83 LONG intel_curpos; /* current offset in transform space */
84 int intel_started; /* have we seen any translatable data yet? */
86 LZX_DECLARE_TABLE(PRETREE);
87 LZX_DECLARE_TABLE(MAINTREE);
88 LZX_DECLARE_TABLE(LENGTH);
89 LZX_DECLARE_TABLE(ALIGNED);
94 /* Microsoft's LZX document and their implementation of the
95 * com.ms.util.cab Java package do not concur.
97 * In the LZX document, there is a table showing the correlation between
98 * window size and the number of position slots. It states that the 1MB
99 * window = 40 slots and the 2MB window = 42 slots. In the implementation,
100 * 1MB = 42 slots, 2MB = 50 slots. The actual calculation is 'find the
101 * first slot whose position base is equal to or more than the required
102 * window size'. This would explain why other tables in the document refer
103 * to 50 slots rather than 42.
105 * The constant NUM_PRIMARY_LENGTHS used in the decompression pseudocode
106 * is not defined in the specification.
108 * The LZX document does not state the uncompressed block has an
109 * uncompressed length field. Where does this length field come from, so
110 * we can know how large the block is? The implementation has it as the 24
111 * bits following after the 3 blocktype bits, before the alignment
114 * The LZX document states that aligned offset blocks have their aligned
115 * offset huffman tree AFTER the main and length trees. The implementation
116 * suggests that the aligned offset tree is BEFORE the main and length
119 * The LZX document decoding algorithm states that, in an aligned offset
120 * block, if an extra_bits value is 1, 2 or 3, then that number of bits
121 * should be read and the result added to the match offset. This is
122 * correct for 1 and 2, but not 3, where just a huffman symbol (using the
123 * aligned tree) should be read.
125 * Regarding the E8 preprocessing, the LZX document states 'No translation
126 * may be performed on the last 6 bytes of the input block'. This is
127 * correct. However, the pseudocode provided checks for the *E8 leader*
128 * up to the last 6 bytes. If the leader appears between -10 and -7 bytes
129 * from the end, this would cause the next four bytes to be modified, at
130 * least one of which would be in the last 6 bytes, which is not allowed
131 * according to the spec.
133 * The specification states that the huffman trees must always contain at
134 * least one element. However, many CAB files contain blocks where the
135 * length tree is completely empty (because there are no matches), and
136 * this is expected to succeed.
140 /* LZX uses what it calls 'position slots' to represent match offsets.
141 * What this means is that a small 'position slot' number and a small
142 * offset from that slot are encoded instead of one large offset for
144 * - position_base is an index to the position slot bases
145 * - extra_bits states how many bits of offset-from-base data is needed.
147 static const UBYTE extra_bits[51] = {
148 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
149 7, 7, 8, 8, 9, 9, 10, 10, 11, 11, 12, 12, 13, 13, 14, 14,
150 15, 15, 16, 16, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17,
154 static const ULONG position_base[51] = {
155 0, 1, 2, 3, 4, 6, 8, 12, 16, 24, 32, 48, 64, 96, 128, 192,
156 256, 384, 512, 768, 1024, 1536, 2048, 3072, 4096, 6144, 8192, 12288, 16384, 24576, 32768, 49152,
157 65536, 98304, 131072, 196608, 262144, 393216, 524288, 655360, 786432, 917504, 1048576, 1179648, 1310720, 1441792, 1572864, 1703936,
158 1835008, 1966080, 2097152
161 struct LZXstate *LZXinit(int window)
163 struct LZXstate *pState=NULL;
164 ULONG wndsize = 1 << window;
167 /* LZX supports window sizes of 2^15 (32Kb) through 2^21 (2Mb) */
168 /* if a previously allocated window is big enough, keep it */
169 if (window < 15 || window > 21) return NULL;
171 /* allocate state and associated window */
172 pState = (struct LZXstate *)malloc(sizeof(struct LZXstate));
173 if (!(pState->window = (UBYTE *)malloc(wndsize)))
178 pState->actual_size = wndsize;
179 pState->window_size = wndsize;
181 /* calculate required position slots */
182 if (window == 20) posn_slots = 42;
183 else if (window == 21) posn_slots = 50;
184 else posn_slots = window << 1;
186 /** alternatively **/
187 /* posn_slots=i=0; while (i < wndsize) i += 1 << extra_bits[posn_slots++]; */
189 /* initialize other state */
190 pState->R0 = pState->R1 = pState->R2 = 1;
191 pState->main_elements = LZX_NUM_CHARS + (posn_slots << 3);
192 pState->header_read = 0;
193 pState->frames_read = 0;
194 pState->block_remaining = 0;
195 pState->block_type = LZX_BLOCKTYPE_INVALID;
196 pState->intel_curpos = 0;
197 pState->intel_started = 0;
198 pState->window_posn = 0;
200 /* initialise tables to 0 (because deltas will be applied to them) */
201 for (i = 0; i < LZX_MAINTREE_MAXSYMBOLS; i++) pState->MAINTREE_len[i] = 0;
202 for (i = 0; i < LZX_LENGTH_MAXSYMBOLS; i++) pState->LENGTH_len[i] = 0;
207 void LZXteardown(struct LZXstate *pState)
212 free(pState->window);
217 int LZXreset(struct LZXstate *pState)
221 pState->R0 = pState->R1 = pState->R2 = 1;
222 pState->header_read = 0;
223 pState->frames_read = 0;
224 pState->block_remaining = 0;
225 pState->block_type = LZX_BLOCKTYPE_INVALID;
226 pState->intel_curpos = 0;
227 pState->intel_started = 0;
228 pState->window_posn = 0;
230 for (i = 0; i < LZX_MAINTREE_MAXSYMBOLS + LZX_LENTABLE_SAFETY; i++) pState->MAINTREE_len[i] = 0;
231 for (i = 0; i < LZX_LENGTH_MAXSYMBOLS + LZX_LENTABLE_SAFETY; i++) pState->LENGTH_len[i] = 0;
237 /* Bitstream reading macros:
239 * INIT_BITSTREAM should be used first to set up the system
240 * READ_BITS(var,n) takes N bits from the buffer and puts them in var
242 * ENSURE_BITS(n) ensures there are at least N bits in the bit buffer
243 * PEEK_BITS(n) extracts (without removing) N bits from the bit buffer
244 * REMOVE_BITS(n) removes N bits from the bit buffer
246 * These bit access routines work by using the area beyond the MSB and the
247 * LSB as a free source of zeroes. This avoids having to mask any bits.
248 * So we have to know the bit width of the bitbuffer variable. This is
249 * sizeof(ULONG) * 8, also defined as ULONG_BITS
252 /* number of bits in ULONG. Note: This must be at multiple of 16, and at
253 * least 32 for the bitbuffer code to work (ie, it must be able to ensure
254 * up to 17 bits - that's adding 16 bits when there's one bit left, or
255 * adding 32 bits when there are no bits left. The code should work fine
256 * for machines where ULONG >= 32 bits.
258 #define ULONG_BITS (sizeof(ULONG)<<3)
260 #define INIT_BITSTREAM do { bitsleft = 0; bitbuf = 0; } while (0)
262 #define ENSURE_BITS(n) \
263 while (bitsleft < (n)) { \
264 bitbuf |= ((inpos[1]<<8)|inpos[0]) << (ULONG_BITS-16 - bitsleft); \
265 bitsleft += 16; inpos+=2; \
268 #define PEEK_BITS(n) (bitbuf >> (ULONG_BITS - (n)))
269 #define REMOVE_BITS(n) ((bitbuf <<= (n)), (bitsleft -= (n)))
271 #define READ_BITS(v,n) do { \
273 (v) = PEEK_BITS(n); \
280 #define TABLEBITS(tbl) (LZX_##tbl##_TABLEBITS)
281 #define MAXSYMBOLS(tbl) (LZX_##tbl##_MAXSYMBOLS)
282 #define SYMTABLE(tbl) (pState->tbl##_table)
283 #define LENTABLE(tbl) (pState->tbl##_len)
285 /* BUILD_TABLE(tablename) builds a huffman lookup table from code lengths.
286 * In reality, it just calls make_decode_table() with the appropriate
287 * values - they're all fixed by some #defines anyway, so there's no point
288 * writing each call out in full by hand.
290 #define BUILD_TABLE(tbl) \
291 if (make_decode_table( \
292 MAXSYMBOLS(tbl), TABLEBITS(tbl), LENTABLE(tbl), SYMTABLE(tbl) \
293 )) { return DECR_ILLEGALDATA; }
296 /* READ_HUFFSYM(tablename, var) decodes one huffman symbol from the
297 * bitstream using the stated table and puts it in var.
299 #define READ_HUFFSYM(tbl,var) do { \
301 hufftbl = SYMTABLE(tbl); \
302 if ((i = hufftbl[PEEK_BITS(TABLEBITS(tbl))]) >= MAXSYMBOLS(tbl)) { \
303 j = 1 << (ULONG_BITS - TABLEBITS(tbl)); \
305 j >>= 1; i <<= 1; i |= (bitbuf & j) ? 1 : 0; \
306 if (!j) { return DECR_ILLEGALDATA; } \
307 } while ((i = hufftbl[i]) >= MAXSYMBOLS(tbl)); \
309 j = LENTABLE(tbl)[(var) = i]; \
314 /* READ_LENGTHS(tablename, first, last) reads in code lengths for symbols
315 * first to last in the given table. The code lengths are stored in their
316 * own special LZX way.
318 #define READ_LENGTHS(tbl,first,last) do { \
319 lb.bb = bitbuf; lb.bl = bitsleft; lb.ip = inpos; \
320 if (lzx_read_lens(pState, LENTABLE(tbl),(first),(last),&lb)) { \
321 return DECR_ILLEGALDATA; \
323 bitbuf = lb.bb; bitsleft = lb.bl; inpos = lb.ip; \
327 /* make_decode_table(nsyms, nbits, length[], table[])
329 * This function was coded by David Tritscher. It builds a fast huffman
330 * decoding table out of just a canonical huffman code lengths table.
332 * nsyms = total number of symbols in this huffman tree.
333 * nbits = any symbols with a code length of nbits or less can be decoded
334 * in one lookup of the table.
335 * length = A table to get code lengths from [0 to syms-1]
336 * table = The table to fill up with decoded symbols and pointers.
338 * Returns 0 for OK or 1 for error
341 static int make_decode_table(ULONG nsyms, ULONG nbits, UBYTE *length, UWORD *table) {
344 register UBYTE bit_num = 1;
346 ULONG pos = 0; /* the current position in the decode table */
347 ULONG table_mask = 1 << nbits;
348 ULONG bit_mask = table_mask >> 1; /* don't do 0 length codes */
349 ULONG next_symbol = bit_mask; /* base of allocation for long codes */
351 /* fill entries for codes short enough for a direct mapping */
352 while (bit_num <= nbits) {
353 for (sym = 0; sym < nsyms; sym++) {
354 if (length[sym] == bit_num) {
357 if((pos += bit_mask) > table_mask) return 1; /* table overrun */
359 /* fill all possible lookups of this symbol with the symbol itself */
361 while (fill-- > 0) table[leaf++] = sym;
368 /* if there are any codes longer than nbits */
369 if (pos != table_mask) {
370 /* clear the remainder of the table */
371 for (sym = pos; sym < table_mask; sym++) table[sym] = 0;
373 /* give ourselves room for codes to grow by up to 16 more bits */
378 while (bit_num <= 16) {
379 for (sym = 0; sym < nsyms; sym++) {
380 if (length[sym] == bit_num) {
382 for (fill = 0; fill < bit_num - nbits; fill++) {
383 /* if this path hasn't been taken yet, 'allocate' two entries */
384 if (table[leaf] == 0) {
385 table[(next_symbol << 1)] = 0;
386 table[(next_symbol << 1) + 1] = 0;
387 table[leaf] = next_symbol++;
389 /* follow the path and select either left or right for next bit */
390 leaf = table[leaf] << 1;
391 if ((pos >> (15-fill)) & 1) leaf++;
395 if ((pos += bit_mask) > table_mask) return 1; /* table overflow */
404 if (pos == table_mask) return 0;
406 /* either erroneous table, or all elements are 0 - let's find out. */
407 for (sym = 0; sym < nsyms; sym++) if (length[sym]) return 1;
417 static int lzx_read_lens(struct LZXstate *pState, UBYTE *lens, ULONG first, ULONG last, struct lzx_bits *lb) {
421 register ULONG bitbuf = lb->bb;
422 register int bitsleft = lb->bl;
423 UBYTE *inpos = lb->ip;
426 for (x = 0; x < 20; x++) {
428 LENTABLE(PRETREE)[x] = y;
430 BUILD_TABLE(PRETREE);
432 for (x = first; x < last; ) {
433 READ_HUFFSYM(PRETREE, z);
435 READ_BITS(y, 4); y += 4;
436 while (y--) lens[x++] = 0;
439 READ_BITS(y, 5); y += 20;
440 while (y--) lens[x++] = 0;
443 READ_BITS(y, 1); y += 4;
444 READ_HUFFSYM(PRETREE, z);
445 z = lens[x] - z; if (z < 0) z += 17;
446 while (y--) lens[x++] = z;
449 z = lens[x] - z; if (z < 0) z += 17;
460 int LZXdecompress(struct LZXstate *pState, unsigned char *inpos, unsigned char *outpos, int inlen, int outlen) {
461 UBYTE *endinp = inpos + inlen;
462 UBYTE *window = pState->window;
463 UBYTE *runsrc, *rundest;
464 UWORD *hufftbl; /* used in READ_HUFFSYM macro as chosen decoding table */
466 ULONG window_posn = pState->window_posn;
467 ULONG window_size = pState->window_size;
468 ULONG R0 = pState->R0;
469 ULONG R1 = pState->R1;
470 ULONG R2 = pState->R2;
472 register ULONG bitbuf;
473 register int bitsleft;
474 ULONG match_offset, i,j,k; /* ijk used in READ_HUFFSYM macro */
475 struct lzx_bits lb; /* used in READ_LENGTHS macro */
477 int togo = outlen, this_run, main_element, aligned_bits;
478 int match_length, length_footer, extra, verbatim_bits;
483 /* read header if necessary */
484 if (!pState->header_read) {
486 READ_BITS(k, 1); if (k) { READ_BITS(i,16); READ_BITS(j,16); }
487 pState->intel_filesize = (i << 16) | j; /* or 0 if not encoded */
488 pState->header_read = 1;
491 /* main decoding loop */
493 /* last block finished, new block expected */
494 if (pState->block_remaining == 0) {
495 if (pState->block_type == LZX_BLOCKTYPE_UNCOMPRESSED) {
496 if (pState->block_length & 1) inpos++; /* realign bitstream to word */
500 READ_BITS(pState->block_type, 3);
503 pState->block_remaining = pState->block_length = (i << 8) | j;
505 switch (pState->block_type) {
506 case LZX_BLOCKTYPE_ALIGNED:
507 for (i = 0; i < 8; i++) { READ_BITS(j, 3); LENTABLE(ALIGNED)[i] = j; }
508 BUILD_TABLE(ALIGNED);
509 /* rest of aligned header is same as verbatim */
511 case LZX_BLOCKTYPE_VERBATIM:
512 READ_LENGTHS(MAINTREE, 0, 256);
513 READ_LENGTHS(MAINTREE, 256, pState->main_elements);
514 BUILD_TABLE(MAINTREE);
515 if (LENTABLE(MAINTREE)[0xE8] != 0) pState->intel_started = 1;
517 READ_LENGTHS(LENGTH, 0, LZX_NUM_SECONDARY_LENGTHS);
521 case LZX_BLOCKTYPE_UNCOMPRESSED:
522 pState->intel_started = 1; /* because we can't assume otherwise */
523 ENSURE_BITS(16); /* get up to 16 pad bits into the buffer */
524 if (bitsleft > 16) inpos -= 2; /* and align the bitstream! */
525 R0 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
526 R1 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
527 R2 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
531 return DECR_ILLEGALDATA;
535 /* buffer exhaustion check */
536 if (inpos > endinp) {
537 /* it's possible to have a file where the next run is less than
538 * 16 bits in size. In this case, the READ_HUFFSYM() macro used
539 * in building the tables will exhaust the buffer, so we should
540 * allow for this, but not allow those accidentally read bits to
541 * be used (so we check that there are at least 16 bits
542 * remaining - in this boundary case they aren't really part of
543 * the compressed data)
545 if (inpos > (endinp+2) || bitsleft < 16) return DECR_ILLEGALDATA;
548 while ((this_run = pState->block_remaining) > 0 && togo > 0) {
549 if (this_run > togo) this_run = togo;
551 pState->block_remaining -= this_run;
553 /* apply 2^x-1 mask */
554 window_posn &= window_size - 1;
555 /* runs can't straddle the window wraparound */
556 if ((window_posn + this_run) > window_size)
557 return DECR_DATAFORMAT;
559 switch (pState->block_type) {
561 case LZX_BLOCKTYPE_VERBATIM:
562 while (this_run > 0) {
563 READ_HUFFSYM(MAINTREE, main_element);
565 if (main_element < LZX_NUM_CHARS) {
566 /* literal: 0 to LZX_NUM_CHARS-1 */
567 window[window_posn++] = main_element;
571 /* match: LZX_NUM_CHARS + ((slot<<3) | length_header (3 bits)) */
572 main_element -= LZX_NUM_CHARS;
574 match_length = main_element & LZX_NUM_PRIMARY_LENGTHS;
575 if (match_length == LZX_NUM_PRIMARY_LENGTHS) {
576 READ_HUFFSYM(LENGTH, length_footer);
577 match_length += length_footer;
579 match_length += LZX_MIN_MATCH;
581 match_offset = main_element >> 3;
583 if (match_offset > 2) {
584 /* not repeated offset */
585 if (match_offset != 3) {
586 extra = extra_bits[match_offset];
587 READ_BITS(verbatim_bits, extra);
588 match_offset = position_base[match_offset] - 2 + verbatim_bits;
594 /* update repeated offset LRU queue */
595 R2 = R1; R1 = R0; R0 = match_offset;
597 else if (match_offset == 0) {
600 else if (match_offset == 1) {
602 R1 = R0; R0 = match_offset;
604 else /* match_offset == 2 */ {
606 R2 = R0; R0 = match_offset;
609 rundest = window + window_posn;
610 this_run -= match_length;
612 /* copy any wrapped around source data */
613 if (window_posn >= match_offset) {
615 runsrc = rundest - match_offset;
617 runsrc = rundest + (window_size - match_offset);
618 copy_length = match_offset - window_posn;
619 if (copy_length < match_length) {
620 match_length -= copy_length;
621 window_posn += copy_length;
622 while (copy_length-- > 0) *rundest++ = *runsrc++;
626 window_posn += match_length;
628 /* copy match data - no worries about destination wraps */
629 while (match_length-- > 0) *rundest++ = *runsrc++;
635 case LZX_BLOCKTYPE_ALIGNED:
636 while (this_run > 0) {
637 READ_HUFFSYM(MAINTREE, main_element);
639 if (main_element < LZX_NUM_CHARS) {
640 /* literal: 0 to LZX_NUM_CHARS-1 */
641 window[window_posn++] = main_element;
645 /* match: LZX_NUM_CHARS + ((slot<<3) | length_header (3 bits)) */
646 main_element -= LZX_NUM_CHARS;
648 match_length = main_element & LZX_NUM_PRIMARY_LENGTHS;
649 if (match_length == LZX_NUM_PRIMARY_LENGTHS) {
650 READ_HUFFSYM(LENGTH, length_footer);
651 match_length += length_footer;
653 match_length += LZX_MIN_MATCH;
655 match_offset = main_element >> 3;
657 if (match_offset > 2) {
658 /* not repeated offset */
659 extra = extra_bits[match_offset];
660 match_offset = position_base[match_offset] - 2;
662 /* verbatim and aligned bits */
664 READ_BITS(verbatim_bits, extra);
665 match_offset += (verbatim_bits << 3);
666 READ_HUFFSYM(ALIGNED, aligned_bits);
667 match_offset += aligned_bits;
669 else if (extra == 3) {
670 /* aligned bits only */
671 READ_HUFFSYM(ALIGNED, aligned_bits);
672 match_offset += aligned_bits;
674 else if (extra > 0) { /* extra==1, extra==2 */
675 /* verbatim bits only */
676 READ_BITS(verbatim_bits, extra);
677 match_offset += verbatim_bits;
679 else /* extra == 0 */ {
684 /* update repeated offset LRU queue */
685 R2 = R1; R1 = R0; R0 = match_offset;
687 else if (match_offset == 0) {
690 else if (match_offset == 1) {
692 R1 = R0; R0 = match_offset;
694 else /* match_offset == 2 */ {
696 R2 = R0; R0 = match_offset;
699 rundest = window + window_posn;
700 this_run -= match_length;
702 /* copy any wrapped around source data */
703 if (window_posn >= match_offset) {
705 runsrc = rundest - match_offset;
707 runsrc = rundest + (window_size - match_offset);
708 copy_length = match_offset - window_posn;
709 if (copy_length < match_length) {
710 match_length -= copy_length;
711 window_posn += copy_length;
712 while (copy_length-- > 0) *rundest++ = *runsrc++;
716 window_posn += match_length;
718 /* copy match data - no worries about destination wraps */
719 while (match_length-- > 0) *rundest++ = *runsrc++;
725 case LZX_BLOCKTYPE_UNCOMPRESSED:
726 if ((inpos + this_run) > endinp) return DECR_ILLEGALDATA;
727 memcpy(window + window_posn, inpos, (size_t) this_run);
728 inpos += this_run; window_posn += this_run;
732 return DECR_ILLEGALDATA; /* might as well */
738 if (togo != 0) return DECR_ILLEGALDATA;
739 memcpy(outpos, window + ((!window_posn) ? window_size : window_posn) - outlen, (size_t) outlen);
741 pState->window_posn = window_posn;
746 /* intel E8 decoding */
747 if ((pState->frames_read++ < 32768) && pState->intel_filesize != 0) {
748 if (outlen <= 6 || !pState->intel_started) {
749 pState->intel_curpos += outlen;
752 UBYTE *data = outpos;
753 UBYTE *dataend = data + outlen - 10;
754 LONG curpos = pState->intel_curpos;
755 LONG filesize = pState->intel_filesize;
756 LONG abs_off, rel_off;
758 pState->intel_curpos = curpos + outlen;
760 while (data < dataend) {
761 if (*data++ != 0xE8) { curpos++; continue; }
762 abs_off = data[0] | (data[1]<<8) | (data[2]<<16) | (data[3]<<24);
763 if ((abs_off >= -curpos) && (abs_off < filesize)) {
764 rel_off = (abs_off >= 0) ? abs_off - curpos : abs_off + filesize;
765 data[0] = (UBYTE) rel_off;
766 data[1] = (UBYTE) (rel_off >> 8);
767 data[2] = (UBYTE) (rel_off >> 16);
768 data[3] = (UBYTE) (rel_off >> 24);
778 #ifdef LZX_CHM_TESTDRIVER
779 int main(int c, char **v)
782 struct LZXstate state;
791 fout = fopen(v[2], "wb");
794 fin = fopen(v[i], "rb");
795 ilen = fread(ibuf, 1, 16384, fin);
796 status = LZXdecompress(&state, ibuf, obuf, ilen, 32768);
801 fwrite(obuf, 1, 32768, fout);
803 case DECR_DATAFORMAT:
804 printf("bad format\n");
806 case DECR_ILLEGALDATA:
807 printf("illegal data\n");
810 printf("no memory\n");