Git conversion: Make reactos the root directory, move rosapps, rostests, wallpapers...
[reactos.git] / reactos / dll / win32 / itss / lzx.c
diff --git a/reactos/dll/win32/itss/lzx.c b/reactos/dll/win32/itss/lzx.c
deleted file mode 100644 (file)
index 68b1e23..0000000
+++ /dev/null
@@ -1,827 +0,0 @@
-/***************************************************************************
- *                        lzx.c - LZX decompression routines               *
- *                           -------------------                           *
- *                                                                         *
- *  maintainer: Jed Wing <jedwin@ugcs.caltech.edu>                         *
- *  source:     modified lzx.c from cabextract v0.5                        *
- *  notes:      This file was taken from cabextract v0.5, which was,       *
- *              itself, a modified version of the lzx decompression code   *
- *              from unlzx.                                                *
- *                                                                         *
- *  platforms:  In its current incarnation, this file has been tested on   *
- *              two different Linux platforms (one, redhat-based, with a   *
- *              2.1.2 glibc and gcc 2.95.x, and the other, Debian, with    *
- *              2.2.4 glibc and both gcc 2.95.4 and gcc 3.0.2).  Both were *
- *              Intel x86 compatible machines.                             *
- ***************************************************************************/
-
-/***************************************************************************
- *
- *   Copyright(C) Stuart Caie
- *
- * This library is free software; you can redistribute it and/or
- * modify it under the terms of the GNU Lesser General Public
- * License as published by the Free Software Foundation; either
- * version 2.1 of the License, or (at your option) any later version.
- *
- * This library is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
- * Lesser General Public License for more details.
- *
- * You should have received a copy of the GNU Lesser General Public
- * License along with this library; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
- *
- ***************************************************************************/
-
-#include "precomp.h"
-
-/* sized types */
-typedef unsigned char  UBYTE; /* 8 bits exactly    */
-typedef unsigned short UWORD; /* 16 bits (or more) */
-
-/* some constants defined by the LZX specification */
-#define LZX_MIN_MATCH                (2)
-#define LZX_MAX_MATCH                (257)
-#define LZX_NUM_CHARS                (256)
-#define LZX_BLOCKTYPE_INVALID        (0)   /* also blocktypes 4-7 invalid */
-#define LZX_BLOCKTYPE_VERBATIM       (1)
-#define LZX_BLOCKTYPE_ALIGNED        (2)
-#define LZX_BLOCKTYPE_UNCOMPRESSED   (3)
-#define LZX_PRETREE_NUM_ELEMENTS     (20)
-#define LZX_ALIGNED_NUM_ELEMENTS     (8)   /* aligned offset tree #elements */
-#define LZX_NUM_PRIMARY_LENGTHS      (7)   /* this one missing from spec! */
-#define LZX_NUM_SECONDARY_LENGTHS    (249) /* length tree #elements */
-
-/* LZX huffman defines: tweak tablebits as desired */
-#define LZX_PRETREE_MAXSYMBOLS  (LZX_PRETREE_NUM_ELEMENTS)
-#define LZX_PRETREE_TABLEBITS   (6)
-#define LZX_MAINTREE_MAXSYMBOLS (LZX_NUM_CHARS + 50*8)
-#define LZX_MAINTREE_TABLEBITS  (12)
-#define LZX_LENGTH_MAXSYMBOLS   (LZX_NUM_SECONDARY_LENGTHS+1)
-#define LZX_LENGTH_TABLEBITS    (12)
-#define LZX_ALIGNED_MAXSYMBOLS  (LZX_ALIGNED_NUM_ELEMENTS)
-#define LZX_ALIGNED_TABLEBITS   (7)
-
-#define LZX_LENTABLE_SAFETY (64) /* we allow length table decoding overruns */
-
-#define LZX_DECLARE_TABLE(tbl) \
-  UWORD tbl##_table[(1<<LZX_##tbl##_TABLEBITS) + (LZX_##tbl##_MAXSYMBOLS<<1)];\
-  UBYTE tbl##_len  [LZX_##tbl##_MAXSYMBOLS + LZX_LENTABLE_SAFETY]
-
-struct LZXstate
-{
-    UBYTE *window;         /* the actual decoding window              */
-    ULONG window_size;     /* window size (32Kb through 2Mb)          */
-    ULONG actual_size;     /* window size when it was first allocated */
-    ULONG window_posn;     /* current offset within the window        */
-    ULONG R0, R1, R2;      /* for the LRU offset system               */
-    UWORD main_elements;   /* number of main tree elements            */
-    int   header_read;     /* have we started decoding at all yet?    */
-    UWORD block_type;      /* type of this block                      */
-    ULONG block_length;    /* uncompressed length of this block       */
-    ULONG block_remaining; /* uncompressed bytes still left to decode */
-    ULONG frames_read;     /* the number of CFDATA blocks processed   */
-    LONG  intel_filesize;  /* magic header value used for transform   */
-    LONG  intel_curpos;    /* current offset in transform space       */
-    int   intel_started;   /* have we seen any translatable data yet? */
-
-    LZX_DECLARE_TABLE(PRETREE);
-    LZX_DECLARE_TABLE(MAINTREE);
-    LZX_DECLARE_TABLE(LENGTH);
-    LZX_DECLARE_TABLE(ALIGNED);
-};
-
-/* LZX decruncher */
-
-/* Microsoft's LZX document and their implementation of the
- * com.ms.util.cab Java package do not concur.
- *
- * In the LZX document, there is a table showing the correlation between
- * window size and the number of position slots. It states that the 1MB
- * window = 40 slots and the 2MB window = 42 slots. In the implementation,
- * 1MB = 42 slots, 2MB = 50 slots. The actual calculation is 'find the
- * first slot whose position base is equal to or more than the required
- * window size'. This would explain why other tables in the document refer
- * to 50 slots rather than 42.
- *
- * The constant NUM_PRIMARY_LENGTHS used in the decompression pseudocode
- * is not defined in the specification.
- *
- * The LZX document does not state the uncompressed block has an
- * uncompressed length field. Where does this length field come from, so
- * we can know how large the block is? The implementation has it as the 24
- * bits following after the 3 blocktype bits, before the alignment
- * padding.
- *
- * The LZX document states that aligned offset blocks have their aligned
- * offset huffman tree AFTER the main and length trees. The implementation
- * suggests that the aligned offset tree is BEFORE the main and length
- * trees.
- *
- * The LZX document decoding algorithm states that, in an aligned offset
- * block, if an extra_bits value is 1, 2 or 3, then that number of bits
- * should be read and the result added to the match offset. This is
- * correct for 1 and 2, but not 3, where just a huffman symbol (using the
- * aligned tree) should be read.
- *
- * Regarding the E8 preprocessing, the LZX document states 'No translation
- * may be performed on the last 6 bytes of the input block'. This is
- * correct.  However, the pseudocode provided checks for the *E8 leader*
- * up to the last 6 bytes. If the leader appears between -10 and -7 bytes
- * from the end, this would cause the next four bytes to be modified, at
- * least one of which would be in the last 6 bytes, which is not allowed
- * according to the spec.
- *
- * The specification states that the huffman trees must always contain at
- * least one element. However, many CAB files contain blocks where the
- * length tree is completely empty (because there are no matches), and
- * this is expected to succeed.
- */
-
-
-/* LZX uses what it calls 'position slots' to represent match offsets.
- * What this means is that a small 'position slot' number and a small
- * offset from that slot are encoded instead of one large offset for
- * every match.
- * - position_base is an index to the position slot bases
- * - extra_bits states how many bits of offset-from-base data is needed.
- */
-static const UBYTE extra_bits[51] = {
-     0,  0,  0,  0,  1,  1,  2,  2,  3,  3,  4,  4,  5,  5,  6,  6,
-     7,  7,  8,  8,  9,  9, 10, 10, 11, 11, 12, 12, 13, 13, 14, 14,
-    15, 15, 16, 16, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17,
-    17, 17, 17
-};
-
-static const ULONG position_base[51] = {
-          0,       1,       2,      3,      4,      6,      8,     12,     16,     24,     32,       48,      64,      96,     128,     192,
-        256,     384,     512,    768,   1024,   1536,   2048,   3072,   4096,   6144,   8192,    12288,   16384,   24576,   32768,   49152,
-      65536,   98304,  131072, 196608, 262144, 393216, 524288, 655360, 786432, 917504, 1048576, 1179648, 1310720, 1441792, 1572864, 1703936,
-    1835008, 1966080, 2097152
-};
-
-struct LZXstate *LZXinit(int window)
-{
-    struct LZXstate *pState=NULL;
-    ULONG wndsize = 1 << window;
-    int i, posn_slots;
-
-    /* LZX supports window sizes of 2^15 (32Kb) through 2^21 (2Mb) */
-    /* if a previously allocated window is big enough, keep it     */
-    if (window < 15 || window > 21) return NULL;
-
-    /* allocate state and associated window */
-    pState = HeapAlloc(GetProcessHeap(), 0, sizeof(struct LZXstate));
-    if (!(pState->window = HeapAlloc(GetProcessHeap(), 0, wndsize)))
-    {
-        HeapFree(GetProcessHeap(), 0, pState);
-        return NULL;
-    }
-    pState->actual_size = wndsize;
-    pState->window_size = wndsize;
-
-    /* calculate required position slots */
-    if (window == 20) posn_slots = 42;
-    else if (window == 21) posn_slots = 50;
-    else posn_slots = window << 1;
-
-    /** alternatively **/
-    /* posn_slots=i=0; while (i < wndsize) i += 1 << extra_bits[posn_slots++]; */
-
-    /* initialize other state */
-    pState->R0  =  pState->R1  = pState->R2 = 1;
-    pState->main_elements   = LZX_NUM_CHARS + (posn_slots << 3);
-    pState->header_read     = 0;
-    pState->frames_read     = 0;
-    pState->block_remaining = 0;
-    pState->block_type      = LZX_BLOCKTYPE_INVALID;
-    pState->intel_curpos    = 0;
-    pState->intel_started   = 0;
-    pState->window_posn     = 0;
-
-    /* initialise tables to 0 (because deltas will be applied to them) */
-    for (i = 0; i < LZX_MAINTREE_MAXSYMBOLS; i++) pState->MAINTREE_len[i] = 0;
-    for (i = 0; i < LZX_LENGTH_MAXSYMBOLS; i++)   pState->LENGTH_len[i]   = 0;
-
-    return pState;
-}
-
-void LZXteardown(struct LZXstate *pState)
-{
-    if (pState)
-    {
-        HeapFree(GetProcessHeap(), 0, pState->window);
-        HeapFree(GetProcessHeap(), 0, pState);
-    }
-}
-
-int LZXreset(struct LZXstate *pState)
-{
-    int i;
-
-    pState->R0  =  pState->R1  = pState->R2 = 1;
-    pState->header_read     = 0;
-    pState->frames_read     = 0;
-    pState->block_remaining = 0;
-    pState->block_type      = LZX_BLOCKTYPE_INVALID;
-    pState->intel_curpos    = 0;
-    pState->intel_started   = 0;
-    pState->window_posn     = 0;
-
-    for (i = 0; i < LZX_MAINTREE_MAXSYMBOLS + LZX_LENTABLE_SAFETY; i++) pState->MAINTREE_len[i] = 0;
-    for (i = 0; i < LZX_LENGTH_MAXSYMBOLS + LZX_LENTABLE_SAFETY; i++)   pState->LENGTH_len[i]   = 0;
-
-    return DECR_OK;
-}
-
-
-/* Bitstream reading macros:
- *
- * INIT_BITSTREAM    should be used first to set up the system
- * READ_BITS(var,n)  takes N bits from the buffer and puts them in var
- *
- * ENSURE_BITS(n)    ensures there are at least N bits in the bit buffer
- * PEEK_BITS(n)      extracts (without removing) N bits from the bit buffer
- * REMOVE_BITS(n)    removes N bits from the bit buffer
- *
- * These bit access routines work by using the area beyond the MSB and the
- * LSB as a free source of zeroes. This avoids having to mask any bits.
- * So we have to know the bit width of the bitbuffer variable. This is
- * sizeof(ULONG) * 8, also defined as ULONG_BITS
- */
-
-/* number of bits in ULONG. Note: This must be at multiple of 16, and at
- * least 32 for the bitbuffer code to work (ie, it must be able to ensure
- * up to 17 bits - that's adding 16 bits when there's one bit left, or
- * adding 32 bits when there are no bits left. The code should work fine
- * for machines where ULONG >= 32 bits.
- */
-#define ULONG_BITS (sizeof(ULONG)<<3)
-
-#define INIT_BITSTREAM do { bitsleft = 0; bitbuf = 0; } while (0)
-
-#define ENSURE_BITS(n)                                                 \
-  while (bitsleft < (n)) {                                             \
-    bitbuf |= ((inpos[1]<<8)|inpos[0]) << (ULONG_BITS-16 - bitsleft);  \
-    bitsleft += 16; inpos+=2;                                          \
-  }
-
-#define PEEK_BITS(n)   (bitbuf >> (ULONG_BITS - (n)))
-#define REMOVE_BITS(n) ((bitbuf <<= (n)), (bitsleft -= (n)))
-
-#define READ_BITS(v,n) do {                                            \
-  ENSURE_BITS(n);                                                      \
-  (v) = PEEK_BITS(n);                                                  \
-  REMOVE_BITS(n);                                                      \
-} while (0)
-
-
-/* Huffman macros */
-
-#define TABLEBITS(tbl)   (LZX_##tbl##_TABLEBITS)
-#define MAXSYMBOLS(tbl)  (LZX_##tbl##_MAXSYMBOLS)
-#define SYMTABLE(tbl)    (pState->tbl##_table)
-#define LENTABLE(tbl)    (pState->tbl##_len)
-
-/* BUILD_TABLE(tablename) builds a huffman lookup table from code lengths.
- * In reality, it just calls make_decode_table() with the appropriate
- * values - they're all fixed by some #defines anyway, so there's no point
- * writing each call out in full by hand.
- */
-#define BUILD_TABLE(tbl)                                               \
-  if (make_decode_table(                                               \
-    MAXSYMBOLS(tbl), TABLEBITS(tbl), LENTABLE(tbl), SYMTABLE(tbl)      \
-  )) { return DECR_ILLEGALDATA; }
-
-
-/* READ_HUFFSYM(tablename, var) decodes one huffman symbol from the
- * bitstream using the stated table and puts it in var.
- */
-#define READ_HUFFSYM(tbl,var) do {                                     \
-  ENSURE_BITS(16);                                                     \
-  hufftbl = SYMTABLE(tbl);                                             \
-  if ((i = hufftbl[PEEK_BITS(TABLEBITS(tbl))]) >= MAXSYMBOLS(tbl)) {   \
-    j = 1 << (ULONG_BITS - TABLEBITS(tbl));                            \
-    do {                                                               \
-      j >>= 1; i <<= 1; i |= (bitbuf & j) ? 1 : 0;                     \
-      if (!j) { return DECR_ILLEGALDATA; }                             \
-    } while ((i = hufftbl[i]) >= MAXSYMBOLS(tbl));                     \
-  }                                                                    \
-  j = LENTABLE(tbl)[(var) = i];                                                \
-  REMOVE_BITS(j);                                                      \
-} while (0)
-
-
-/* READ_LENGTHS(tablename, first, last) reads in code lengths for symbols
- * first to last in the given table. The code lengths are stored in their
- * own special LZX way.
- */
-#define READ_LENGTHS(tbl,first,last) do { \
-  lb.bb = bitbuf; lb.bl = bitsleft; lb.ip = inpos; \
-  if (lzx_read_lens(pState, LENTABLE(tbl),(first),(last),&lb)) { \
-    return DECR_ILLEGALDATA; \
-  } \
-  bitbuf = lb.bb; bitsleft = lb.bl; inpos = lb.ip; \
-} while (0)
-
-
-/* make_decode_table(nsyms, nbits, length[], table[])
- *
- * This function was coded by David Tritscher. It builds a fast huffman
- * decoding table out of just a canonical huffman code lengths table.
- *
- * nsyms  = total number of symbols in this huffman tree.
- * nbits  = any symbols with a code length of nbits or less can be decoded
- *          in one lookup of the table.
- * length = A table to get code lengths from [0 to syms-1]
- * table  = The table to fill up with decoded symbols and pointers.
- *
- * Returns 0 for OK or 1 for error
- */
-
-static int make_decode_table(ULONG nsyms, ULONG nbits, UBYTE *length, UWORD *table) {
-    register UWORD sym;
-    register ULONG leaf;
-    register UBYTE bit_num = 1;
-    ULONG fill;
-    ULONG pos         = 0; /* the current position in the decode table */
-    ULONG table_mask  = 1 << nbits;
-    ULONG bit_mask    = table_mask >> 1; /* don't do 0 length codes */
-    ULONG next_symbol = bit_mask; /* base of allocation for long codes */
-
-    /* fill entries for codes short enough for a direct mapping */
-    while (bit_num <= nbits) {
-        for (sym = 0; sym < nsyms; sym++) {
-            if (length[sym] == bit_num) {
-                leaf = pos;
-
-                if((pos += bit_mask) > table_mask) return 1; /* table overrun */
-
-                /* fill all possible lookups of this symbol with the symbol itself */
-                fill = bit_mask;
-                while (fill-- > 0) table[leaf++] = sym;
-            }
-        }
-        bit_mask >>= 1;
-        bit_num++;
-    }
-
-    /* if there are any codes longer than nbits */
-    if (pos != table_mask) {
-        /* clear the remainder of the table */
-        for (sym = pos; sym < table_mask; sym++) table[sym] = 0;
-
-        /* give ourselves room for codes to grow by up to 16 more bits */
-        pos <<= 16;
-        table_mask <<= 16;
-        bit_mask = 1 << 15;
-
-        while (bit_num <= 16) {
-            for (sym = 0; sym < nsyms; sym++) {
-                if (length[sym] == bit_num) {
-                    leaf = pos >> 16;
-                    for (fill = 0; fill < bit_num - nbits; fill++) {
-                        /* if this path hasn't been taken yet, 'allocate' two entries */
-                        if (table[leaf] == 0) {
-                            table[(next_symbol << 1)] = 0;
-                            table[(next_symbol << 1) + 1] = 0;
-                            table[leaf] = next_symbol++;
-                        }
-                        /* follow the path and select either left or right for next bit */
-                        leaf = table[leaf] << 1;
-                        if ((pos >> (15-fill)) & 1) leaf++;
-                    }
-                    table[leaf] = sym;
-
-                    if ((pos += bit_mask) > table_mask) return 1; /* table overflow */
-                }
-            }
-            bit_mask >>= 1;
-            bit_num++;
-        }
-    }
-
-    /* full table? */
-    if (pos == table_mask) return 0;
-
-    /* either erroneous table, or all elements are 0 - let's find out. */
-    for (sym = 0; sym < nsyms; sym++) if (length[sym]) return 1;
-    return 0;
-}
-
-struct lzx_bits {
-  ULONG bb;
-  int bl;
-  UBYTE *ip;
-};
-
-static int lzx_read_lens(struct LZXstate *pState, UBYTE *lens, ULONG first, ULONG last, struct lzx_bits *lb) {
-    ULONG i,j, x,y;
-    int z;
-
-    register ULONG bitbuf = lb->bb;
-    register int bitsleft = lb->bl;
-    UBYTE *inpos = lb->ip;
-    UWORD *hufftbl;
-
-    for (x = 0; x < 20; x++) {
-        READ_BITS(y, 4);
-        LENTABLE(PRETREE)[x] = y;
-    }
-    BUILD_TABLE(PRETREE);
-
-    for (x = first; x < last; ) {
-        READ_HUFFSYM(PRETREE, z);
-        if (z == 17) {
-            READ_BITS(y, 4); y += 4;
-            while (y--) lens[x++] = 0;
-        }
-        else if (z == 18) {
-            READ_BITS(y, 5); y += 20;
-            while (y--) lens[x++] = 0;
-        }
-        else if (z == 19) {
-            READ_BITS(y, 1); y += 4;
-            READ_HUFFSYM(PRETREE, z);
-            z = lens[x] - z; if (z < 0) z += 17;
-            while (y--) lens[x++] = z;
-        }
-        else {
-            z = lens[x] - z; if (z < 0) z += 17;
-            lens[x++] = z;
-        }
-    }
-
-    lb->bb = bitbuf;
-    lb->bl = bitsleft;
-    lb->ip = inpos;
-    return 0;
-}
-
-int LZXdecompress(struct LZXstate *pState, unsigned char *inpos, unsigned char *outpos, int inlen, int outlen) {
-    UBYTE *endinp = inpos + inlen;
-    UBYTE *window = pState->window;
-    UBYTE *runsrc, *rundest;
-    UWORD *hufftbl; /* used in READ_HUFFSYM macro as chosen decoding table */
-
-    ULONG window_posn = pState->window_posn;
-    ULONG window_size = pState->window_size;
-    ULONG R0 = pState->R0;
-    ULONG R1 = pState->R1;
-    ULONG R2 = pState->R2;
-
-    register ULONG bitbuf;
-    register int bitsleft;
-    ULONG match_offset, i,j,k; /* ijk used in READ_HUFFSYM macro */
-    struct lzx_bits lb; /* used in READ_LENGTHS macro */
-
-    int togo = outlen, this_run, main_element, aligned_bits;
-    int match_length, length_footer, extra, verbatim_bits;
-    int copy_length;
-
-    INIT_BITSTREAM;
-
-    /* read header if necessary */
-    if (!pState->header_read) {
-        i = j = 0;
-        READ_BITS(k, 1); if (k) { READ_BITS(i,16); READ_BITS(j,16); }
-        pState->intel_filesize = (i << 16) | j; /* or 0 if not encoded */
-        pState->header_read = 1;
-    }
-
-    /* main decoding loop */
-    while (togo > 0) {
-        /* last block finished, new block expected */
-        if (pState->block_remaining == 0) {
-            if (pState->block_type == LZX_BLOCKTYPE_UNCOMPRESSED) {
-                if (pState->block_length & 1) inpos++; /* realign bitstream to word */
-                INIT_BITSTREAM;
-            }
-
-            READ_BITS(pState->block_type, 3);
-            READ_BITS(i, 16);
-            READ_BITS(j, 8);
-            pState->block_remaining = pState->block_length = (i << 8) | j;
-
-            switch (pState->block_type) {
-                case LZX_BLOCKTYPE_ALIGNED:
-                    for (i = 0; i < 8; i++) { READ_BITS(j, 3); LENTABLE(ALIGNED)[i] = j; }
-                    BUILD_TABLE(ALIGNED);
-                    /* rest of aligned header is same as verbatim */
-
-                case LZX_BLOCKTYPE_VERBATIM:
-                    READ_LENGTHS(MAINTREE, 0, 256);
-                    READ_LENGTHS(MAINTREE, 256, pState->main_elements);
-                    BUILD_TABLE(MAINTREE);
-                    if (LENTABLE(MAINTREE)[0xE8] != 0) pState->intel_started = 1;
-
-                    READ_LENGTHS(LENGTH, 0, LZX_NUM_SECONDARY_LENGTHS);
-                    BUILD_TABLE(LENGTH);
-                    break;
-
-                case LZX_BLOCKTYPE_UNCOMPRESSED:
-                    pState->intel_started = 1; /* because we can't assume otherwise */
-                    ENSURE_BITS(16); /* get up to 16 pad bits into the buffer */
-                    if (bitsleft > 16) inpos -= 2; /* and align the bitstream! */
-                    R0 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
-                    R1 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
-                    R2 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
-                    break;
-
-                default:
-                    return DECR_ILLEGALDATA;
-            }
-        }
-
-        /* buffer exhaustion check */
-        if (inpos > endinp) {
-            /* it's possible to have a file where the next run is less than
-             * 16 bits in size. In this case, the READ_HUFFSYM() macro used
-             * in building the tables will exhaust the buffer, so we should
-             * allow for this, but not allow those accidentally read bits to
-             * be used (so we check that there are at least 16 bits
-             * remaining - in this boundary case they aren't really part of
-             * the compressed data)
-             */
-            if (inpos > (endinp+2) || bitsleft < 16) return DECR_ILLEGALDATA;
-        }
-
-        while ((this_run = pState->block_remaining) > 0 && togo > 0) {
-            if (this_run > togo) this_run = togo;
-            togo -= this_run;
-            pState->block_remaining -= this_run;
-
-            /* apply 2^x-1 mask */
-            window_posn &= window_size - 1;
-            /* runs can't straddle the window wraparound */
-            if ((window_posn + this_run) > window_size)
-                return DECR_DATAFORMAT;
-
-            switch (pState->block_type) {
-
-                case LZX_BLOCKTYPE_VERBATIM:
-                    while (this_run > 0) {
-                        READ_HUFFSYM(MAINTREE, main_element);
-
-                        if (main_element < LZX_NUM_CHARS) {
-                            /* literal: 0 to LZX_NUM_CHARS-1 */
-                            window[window_posn++] = main_element;
-                            this_run--;
-                        }
-                        else {
-                            /* match: LZX_NUM_CHARS + ((slot<<3) | length_header (3 bits)) */
-                            main_element -= LZX_NUM_CHARS;
-
-                            match_length = main_element & LZX_NUM_PRIMARY_LENGTHS;
-                            if (match_length == LZX_NUM_PRIMARY_LENGTHS) {
-                                READ_HUFFSYM(LENGTH, length_footer);
-                                match_length += length_footer;
-                            }
-                            match_length += LZX_MIN_MATCH;
-
-                            match_offset = main_element >> 3;
-
-                            if (match_offset > 2) {
-                                /* not repeated offset */
-                                if (match_offset != 3) {
-                                    extra = extra_bits[match_offset];
-                                    READ_BITS(verbatim_bits, extra);
-                                    match_offset = position_base[match_offset] - 2 + verbatim_bits;
-                                }
-                                else {
-                                    match_offset = 1;
-                                }
-
-                                /* update repeated offset LRU queue */
-                                R2 = R1; R1 = R0; R0 = match_offset;
-                            }
-                            else if (match_offset == 0) {
-                                match_offset = R0;
-                            }
-                            else if (match_offset == 1) {
-                                match_offset = R1;
-                                R1 = R0; R0 = match_offset;
-                            }
-                            else /* match_offset == 2 */ {
-                                match_offset = R2;
-                                R2 = R0; R0 = match_offset;
-                            }
-
-                            rundest = window + window_posn;
-                            this_run -= match_length;
-
-                            /* copy any wrapped around source data */
-                            if (window_posn >= match_offset) {
-                                /* no wrap */
-                                 runsrc = rundest - match_offset;
-                            } else {
-                                runsrc = rundest + (window_size - match_offset);
-                                copy_length = match_offset - window_posn;
-                                if (copy_length < match_length) {
-                                     match_length -= copy_length;
-                                     window_posn += copy_length;
-                                     while (copy_length-- > 0) *rundest++ = *runsrc++;
-                                     runsrc = window;
-                                }
-                            }
-                            window_posn += match_length;
-                            /* copy match data - no worries about destination wraps */
-                            while (match_length-- > 0) *rundest++ = *runsrc++;
-
-                        }
-                    }
-                    break;
-
-                case LZX_BLOCKTYPE_ALIGNED:
-                    while (this_run > 0) {
-                        READ_HUFFSYM(MAINTREE, main_element);
-
-                        if (main_element < LZX_NUM_CHARS) {
-                            /* literal: 0 to LZX_NUM_CHARS-1 */
-                            window[window_posn++] = main_element;
-                            this_run--;
-                        }
-                        else {
-                            /* match: LZX_NUM_CHARS + ((slot<<3) | length_header (3 bits)) */
-                            main_element -= LZX_NUM_CHARS;
-
-                            match_length = main_element & LZX_NUM_PRIMARY_LENGTHS;
-                            if (match_length == LZX_NUM_PRIMARY_LENGTHS) {
-                                READ_HUFFSYM(LENGTH, length_footer);
-                                match_length += length_footer;
-                            }
-                            match_length += LZX_MIN_MATCH;
-
-                            match_offset = main_element >> 3;
-
-                            if (match_offset > 2) {
-                                /* not repeated offset */
-                                extra = extra_bits[match_offset];
-                                match_offset = position_base[match_offset] - 2;
-                                if (extra > 3) {
-                                    /* verbatim and aligned bits */
-                                    extra -= 3;
-                                    READ_BITS(verbatim_bits, extra);
-                                    match_offset += (verbatim_bits << 3);
-                                    READ_HUFFSYM(ALIGNED, aligned_bits);
-                                    match_offset += aligned_bits;
-                                }
-                                else if (extra == 3) {
-                                    /* aligned bits only */
-                                    READ_HUFFSYM(ALIGNED, aligned_bits);
-                                    match_offset += aligned_bits;
-                                }
-                                else if (extra > 0) { /* extra==1, extra==2 */
-                                    /* verbatim bits only */
-                                    READ_BITS(verbatim_bits, extra);
-                                    match_offset += verbatim_bits;
-                                }
-                                else /* extra == 0 */ {
-                                    /* ??? */
-                                    match_offset = 1;
-                                }
-
-                                /* update repeated offset LRU queue */
-                                R2 = R1; R1 = R0; R0 = match_offset;
-                            }
-                            else if (match_offset == 0) {
-                                match_offset = R0;
-                            }
-                            else if (match_offset == 1) {
-                                match_offset = R1;
-                                R1 = R0; R0 = match_offset;
-                            }
-                            else /* match_offset == 2 */ {
-                                match_offset = R2;
-                                R2 = R0; R0 = match_offset;
-                            }
-
-                            rundest = window + window_posn;
-                            this_run -= match_length;
-
-                            /* copy any wrapped around source data */
-                            if (window_posn >= match_offset) {
-                                /* no wrap */
-                                 runsrc = rundest - match_offset;
-                            } else {
-                                runsrc = rundest + (window_size - match_offset);
-                                copy_length = match_offset - window_posn;
-                                if (copy_length < match_length) {
-                                     match_length -= copy_length;
-                                     window_posn += copy_length;
-                                     while (copy_length-- > 0) *rundest++ = *runsrc++;
-                                     runsrc = window;
-                                }
-                            }
-                            window_posn += match_length;
-                            /* copy match data - no worries about destination wraps */
-                            while (match_length-- > 0) *rundest++ = *runsrc++;
-
-                        }
-                    }
-                    break;
-
-                case LZX_BLOCKTYPE_UNCOMPRESSED:
-                    if ((inpos + this_run) > endinp) return DECR_ILLEGALDATA;
-                    memcpy(window + window_posn, inpos, (size_t) this_run);
-                    inpos += this_run; window_posn += this_run;
-                    break;
-
-                default:
-                    return DECR_ILLEGALDATA; /* might as well */
-            }
-
-        }
-    }
-
-    if (togo != 0) return DECR_ILLEGALDATA;
-    memcpy(outpos, window + ((!window_posn) ? window_size : window_posn) - outlen, (size_t) outlen);
-
-    pState->window_posn = window_posn;
-    pState->R0 = R0;
-    pState->R1 = R1;
-    pState->R2 = R2;
-
-    /* intel E8 decoding */
-    if ((pState->frames_read++ < 32768) && pState->intel_filesize != 0) {
-        if (outlen <= 6 || !pState->intel_started) {
-            pState->intel_curpos += outlen;
-        }
-        else {
-            UBYTE *data    = outpos;
-            UBYTE *dataend = data + outlen - 10;
-            LONG curpos    = pState->intel_curpos;
-            LONG filesize  = pState->intel_filesize;
-            LONG abs_off, rel_off;
-
-            pState->intel_curpos = curpos + outlen;
-
-            while (data < dataend) {
-                if (*data++ != 0xE8) { curpos++; continue; }
-                abs_off = data[0] | (data[1]<<8) | (data[2]<<16) | (data[3]<<24);
-                if ((abs_off >= -curpos) && (abs_off < filesize)) {
-                    rel_off = (abs_off >= 0) ? abs_off - curpos : abs_off + filesize;
-                    data[0] = (UBYTE) rel_off;
-                    data[1] = (UBYTE) (rel_off >> 8);
-                    data[2] = (UBYTE) (rel_off >> 16);
-                    data[3] = (UBYTE) (rel_off >> 24);
-                }
-                data += 4;
-                curpos += 5;
-            }
-        }
-    }
-    return DECR_OK;
-}
-
-#ifdef LZX_CHM_TESTDRIVER
-int main(int c, char **v)
-{
-    FILE *fin, *fout;
-    struct LZXstate state;
-    UBYTE ibuf[16384];
-    UBYTE obuf[32768];
-    int ilen, olen;
-    int status;
-    int i;
-    int count=0;
-    int w = atoi(v[1]);
-    LZXinit(&state, w);
-    fout = fopen(v[2], "wb");
-    for (i=3; i<c; i++)
-    {
-        fin = fopen(v[i], "rb");
-        ilen = fread(ibuf, 1, 16384, fin);
-        status = LZXdecompress(&state, ibuf, obuf, ilen, 32768);
-        switch (status)
-        {
-            case DECR_OK:
-                printf("ok\n");
-                fwrite(obuf, 1, 32768, fout);
-                break;
-            case DECR_DATAFORMAT:
-                printf("bad format\n");
-                break;
-            case DECR_ILLEGALDATA:
-                printf("illegal data\n");
-                break;
-            case DECR_NOMEMORY:
-                printf("no memory\n");
-                break;
-            default:
-                break;
-        }
-        fclose(fin);
-        if (++count == 2)
-        {
-            count = 0;
-            LZXreset(&state);
-        }
-    }
-    fclose(fout);
-}
-#endif