2003-10-18 01:58:35 +00:00
|
|
|
/* vi: set sw=4 ts=4: */
|
2006-04-14 19:25:01 +00:00
|
|
|
/* Small bzip2 deflate implementation, by Rob Landley (rob@landley.net).
|
2002-11-03 07:28:38 +00:00
|
|
|
|
2006-04-14 19:25:01 +00:00
|
|
|
Based on bzip2 decompression code by Julian R Seward (jseward@acm.org),
|
|
|
|
which also acknowledges contributions by Mike Burrows, David Wheeler,
|
|
|
|
Peter Fenwick, Alistair Moffat, Radford Neal, Ian H. Witten,
|
|
|
|
Robert Sedgewick, and Jon L. Bentley.
|
2002-11-03 07:28:38 +00:00
|
|
|
|
2006-04-14 19:25:01 +00:00
|
|
|
Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
|
2003-10-18 01:58:35 +00:00
|
|
|
*/
|
2002-11-03 07:28:38 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/*
|
|
|
|
Size and speed optimizations by Manuel Novoa III (mjn3@codepoet.org).
|
|
|
|
|
2004-04-14 17:51:38 +00:00
|
|
|
More efficient reading of Huffman codes, a streamlined read_bunzip()
|
2003-10-23 06:52:01 +00:00
|
|
|
function, and various other tweaks. In (limited) tests, approximately
|
|
|
|
20% faster than bzcat on x86 and about 10% faster on arm.
|
|
|
|
|
|
|
|
Note that about 2/3 of the time is spent in read_unzip() reversing
|
|
|
|
the Burrows-Wheeler transformation. Much of that time is delay
|
|
|
|
resulting from cache misses.
|
|
|
|
|
|
|
|
I would ask that anyone benefiting from this work, especially those
|
|
|
|
using it in commercial products, consider making a donation to my local
|
2006-02-17 03:43:49 +00:00
|
|
|
non-profit hospice organization (www.hospiceacadiana.com) in the name of
|
2006-04-02 21:50:01 +00:00
|
|
|
the woman I loved, Toni W. Hagan, who passed away Feb. 12, 2003.
|
2003-10-23 06:52:01 +00:00
|
|
|
|
|
|
|
Manuel
|
|
|
|
*/
|
|
|
|
|
2003-10-28 23:32:12 +00:00
|
|
|
#include "libbb.h"
|
2006-04-02 21:50:01 +00:00
|
|
|
#include "unarchive.h"
|
|
|
|
|
2004-04-14 17:51:38 +00:00
|
|
|
/* Constants for Huffman coding */
|
2006-12-26 10:42:51 +00:00
|
|
|
#define MAX_GROUPS 6
|
|
|
|
#define GROUP_SIZE 50 /* 64 would have been more efficient */
|
|
|
|
#define MAX_HUFCODE_BITS 20 /* Longest Huffman code allowed */
|
|
|
|
#define MAX_SYMBOLS 258 /* 256 literals + RUNA + RUNB */
|
|
|
|
#define SYMBOL_RUNA 0
|
|
|
|
#define SYMBOL_RUNB 1
|
2003-10-18 01:58:35 +00:00
|
|
|
|
|
|
|
/* Status return values */
|
2006-12-26 10:42:51 +00:00
|
|
|
#define RETVAL_OK 0
|
|
|
|
#define RETVAL_LAST_BLOCK (-1)
|
|
|
|
#define RETVAL_NOT_BZIP_DATA (-2)
|
|
|
|
#define RETVAL_UNEXPECTED_INPUT_EOF (-3)
|
|
|
|
#define RETVAL_UNEXPECTED_OUTPUT_EOF (-4)
|
|
|
|
#define RETVAL_DATA_ERROR (-5)
|
|
|
|
#define RETVAL_OUT_OF_MEMORY (-6)
|
|
|
|
#define RETVAL_OBSOLETE_INPUT (-7)
|
2003-10-18 01:58:35 +00:00
|
|
|
|
|
|
|
/* Other housekeeping constants */
|
2006-12-26 10:42:51 +00:00
|
|
|
#define IOBUF_SIZE 4096
|
2003-10-18 01:58:35 +00:00
|
|
|
|
2004-04-14 17:51:38 +00:00
|
|
|
/* This is what we know about each Huffman coding group */
|
2003-10-18 01:58:35 +00:00
|
|
|
struct group_data {
|
2003-10-18 01:59:46 +00:00
|
|
|
/* We have an extra slot at the end of limit[] for a sentinal value. */
|
2007-04-10 17:16:33 +00:00
|
|
|
int limit[MAX_HUFCODE_BITS+1], base[MAX_HUFCODE_BITS], permute[MAX_SYMBOLS];
|
2003-10-23 06:52:01 +00:00
|
|
|
int minLen, maxLen;
|
2002-11-03 07:28:38 +00:00
|
|
|
};
|
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Structure holding all the housekeeping data, including IO buffers and
|
|
|
|
memory that persists between calls to bunzip */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 21:40:19 +00:00
|
|
|
struct bunzip_data {
|
2003-10-23 06:52:01 +00:00
|
|
|
/* State for interrupting output loop */
|
2007-04-10 17:16:33 +00:00
|
|
|
int writeCopies, writePos, writeRunCountdown, writeCount, writeCurrent;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* I/O tracking data (file handles, buffers, positions, etc.) */
|
2007-04-10 17:16:33 +00:00
|
|
|
int in_fd, out_fd, inbufCount, inbufPos /*, outbufPos*/;
|
2003-10-23 06:52:01 +00:00
|
|
|
unsigned char *inbuf /*,*outbuf*/;
|
2007-04-10 17:16:33 +00:00
|
|
|
unsigned inbufBitCount, inbufBits;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* The CRC values stored in the block header and calculated from the data */
|
2006-04-10 17:07:15 +00:00
|
|
|
uint32_t headerCRC, totalCRC, writeCRC;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 21:40:19 +00:00
|
|
|
/* Intermediate buffer and its size (in bytes) */
|
2007-04-10 17:16:33 +00:00
|
|
|
unsigned *dbuf, dbufSize;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 21:40:19 +00:00
|
|
|
/* For I/O error handling */
|
|
|
|
jmp_buf jmpbuf;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 21:40:19 +00:00
|
|
|
/* Big things go last (register-relative addressing can be larger for big offsets */
|
|
|
|
uint32_t crc32Table[256];
|
2003-10-18 01:58:35 +00:00
|
|
|
unsigned char selectors[32768]; /* nSelectors=15 bits */
|
2004-04-14 17:51:38 +00:00
|
|
|
struct group_data groups[MAX_GROUPS]; /* Huffman coding tables */
|
2007-04-10 21:40:19 +00:00
|
|
|
};
|
|
|
|
/* typedef struct bunzip_data bunzip_data; -- done in .h file */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
|
|
|
|
/* Return the next nnn bits of input. All reads from the compressed input
|
|
|
|
are done through this function. All reads are big endian */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
static unsigned get_bits(bunzip_data *bd, char bits_wanted)
|
2002-11-03 07:28:38 +00:00
|
|
|
{
|
2007-04-10 17:16:33 +00:00
|
|
|
unsigned bits = 0;
|
2003-10-18 01:58:35 +00:00
|
|
|
|
|
|
|
/* If we need to get more data from the byte buffer, do so. (Loop getting
|
|
|
|
one byte at a time to enforce endianness and avoid unaligned access.) */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
while (bd->inbufBitCount < bits_wanted) {
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* If we need to read more data from file into byte buffer, do so */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (bd->inbufPos == bd->inbufCount) {
|
2007-04-10 21:40:19 +00:00
|
|
|
/* if "no input fd" case: in_fd == -1, read fails, we jump */
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->inbufCount = read(bd->in_fd, bd->inbuf, IOBUF_SIZE);
|
|
|
|
if (bd->inbufCount <= 0)
|
|
|
|
longjmp(bd->jmpbuf, RETVAL_UNEXPECTED_INPUT_EOF);
|
|
|
|
bd->inbufPos = 0;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Avoid 32-bit overflow (dump bit buffer to top of output) */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (bd->inbufBitCount >= 24) {
|
|
|
|
bits = bd->inbufBits & ((1 << bd->inbufBitCount) - 1);
|
|
|
|
bits_wanted -= bd->inbufBitCount;
|
|
|
|
bits <<= bits_wanted;
|
|
|
|
bd->inbufBitCount = 0;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Grab next 8 bits of input from buffer. */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->inbufBits = (bd->inbufBits<<8) | bd->inbuf[bd->inbufPos++];
|
|
|
|
bd->inbufBitCount += 8;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Calculate result */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->inbufBitCount -= bits_wanted;
|
|
|
|
bits |= (bd->inbufBits >> bd->inbufBitCount) & ((1 << bits_wanted) - 1);
|
2002-11-03 07:28:38 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
return bits;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Unpacks the next block and sets up for the inverse burrows-wheeler step. */
|
2002-11-03 07:28:38 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
static int get_next_block(bunzip_data *bd)
|
2002-11-03 07:28:38 +00:00
|
|
|
{
|
2003-10-18 01:58:35 +00:00
|
|
|
struct group_data *hufGroup;
|
2007-04-10 17:16:33 +00:00
|
|
|
int dbufCount, nextSym, dbufSize, groupCount, *base, *limit, selector,
|
|
|
|
i, j, k, t, runPos, symCount, symTotal, nSelectors, byteCount[256];
|
2003-10-18 01:58:35 +00:00
|
|
|
unsigned char uc, symToByte[256], mtfSymbol[256], *selectors;
|
2007-04-10 17:16:33 +00:00
|
|
|
unsigned *dbuf, origPtr;
|
2003-10-18 01:58:35 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
dbuf = bd->dbuf;
|
|
|
|
dbufSize = bd->dbufSize;
|
|
|
|
selectors = bd->selectors;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Reset longjmp I/O error handling */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
i = setjmp(bd->jmpbuf);
|
2006-12-26 10:42:51 +00:00
|
|
|
if (i) return i;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Read in header signature and CRC, then validate signature.
|
|
|
|
(last block signature means CRC is for whole file, return now) */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
i = get_bits(bd, 24);
|
|
|
|
j = get_bits(bd, 24);
|
|
|
|
bd->headerCRC = get_bits(bd, 32);
|
2003-10-23 06:52:01 +00:00
|
|
|
if ((i == 0x177245) && (j == 0x385090)) return RETVAL_LAST_BLOCK;
|
|
|
|
if ((i != 0x314159) || (j != 0x265359)) return RETVAL_NOT_BZIP_DATA;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* We can add support for blockRandomised if anybody complains. There was
|
|
|
|
some code for this in busybox 1.0.0-pre3, but nobody ever noticed that
|
|
|
|
it didn't actually work. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (get_bits(bd, 1)) return RETVAL_OBSOLETE_INPUT;
|
|
|
|
origPtr = get_bits(bd, 24);
|
|
|
|
if (origPtr > dbufSize) return RETVAL_DATA_ERROR;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* mapping table: if some byte values are never used (encoding things
|
|
|
|
like ascii text), the compression code removes the gaps to have fewer
|
|
|
|
symbols to deal with, and writes a sparse bitfield indicating which
|
|
|
|
values were present. We make a translation table to convert the symbols
|
|
|
|
back to the corresponding bytes. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
t = get_bits(bd, 16);
|
|
|
|
symTotal = 0;
|
|
|
|
for (i = 0; i < 16; i++) {
|
|
|
|
if (t & (1 << (15-i))) {
|
|
|
|
k = get_bits(bd, 16);
|
|
|
|
for (j = 0; j < 16; j++)
|
|
|
|
if (k & (1 << (15-j)))
|
|
|
|
symToByte[symTotal++] = (16*i) + j;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2004-04-14 17:51:38 +00:00
|
|
|
/* How many different Huffman coding groups does this block use? */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
groupCount = get_bits(bd, 3);
|
|
|
|
if (groupCount < 2 || groupCount > MAX_GROUPS)
|
|
|
|
return RETVAL_DATA_ERROR;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2004-04-14 17:51:38 +00:00
|
|
|
/* nSelectors: Every GROUP_SIZE many symbols we select a new Huffman coding
|
2003-10-18 01:58:35 +00:00
|
|
|
group. Read in the group selector list, which is stored as MTF encoded
|
2003-10-23 06:52:01 +00:00
|
|
|
bit runs. (MTF=Move To Front, as each value is used it's moved to the
|
|
|
|
start of the list.) */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
nSelectors = get_bits(bd, 15);
|
|
|
|
if (!nSelectors) return RETVAL_DATA_ERROR;
|
|
|
|
for (i = 0; i < groupCount; i++) mtfSymbol[i] = i;
|
|
|
|
for (i = 0; i < nSelectors; i++) {
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Get next value */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
for (j = 0; get_bits(bd, 1); j++)
|
|
|
|
if (j>=groupCount) return RETVAL_DATA_ERROR;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Decode MTF to get the next selector */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
uc = mtfSymbol[j];
|
2006-12-26 10:42:51 +00:00
|
|
|
for (;j;j--) mtfSymbol[j] = mtfSymbol[j-1];
|
2007-04-10 17:16:33 +00:00
|
|
|
mtfSymbol[0] = selectors[i] = uc;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2004-04-14 17:51:38 +00:00
|
|
|
/* Read the Huffman coding tables for each group, which code for symTotal
|
2003-10-18 01:58:35 +00:00
|
|
|
literal symbols, plus two run symbols (RUNA, RUNB) */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
symCount = symTotal + 2;
|
|
|
|
for (j = 0; j < groupCount; j++) {
|
|
|
|
unsigned char length[MAX_SYMBOLS], temp[MAX_HUFCODE_BITS+1];
|
|
|
|
int minLen, maxLen, pp;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2004-04-14 17:51:38 +00:00
|
|
|
/* Read Huffman code lengths for each symbol. They're stored in
|
2003-10-23 06:52:01 +00:00
|
|
|
a way similar to mtf; record a starting value for the first symbol,
|
|
|
|
and an offset from the previous value for everys symbol after that.
|
|
|
|
(Subtracting 1 before the loop and then adding it back at the end is
|
|
|
|
an optimization that makes the test inside the loop simpler: symbol
|
|
|
|
length 0 becomes negative, so an unsigned inequality catches it.) */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
t = get_bits(bd, 5) - 1;
|
2003-10-18 01:58:35 +00:00
|
|
|
for (i = 0; i < symCount; i++) {
|
2006-12-26 10:42:51 +00:00
|
|
|
for (;;) {
|
2007-04-10 17:16:33 +00:00
|
|
|
if ((unsigned)t > (MAX_HUFCODE_BITS-1))
|
2003-10-23 06:52:01 +00:00
|
|
|
return RETVAL_DATA_ERROR;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* If first bit is 0, stop. Else second bit indicates whether
|
|
|
|
to increment or decrement the value. Optimization: grab 2
|
|
|
|
bits and unget the second if the first was 0. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
k = get_bits(bd, 2);
|
2003-10-23 06:52:01 +00:00
|
|
|
if (k < 2) {
|
|
|
|
bd->inbufBitCount++;
|
|
|
|
break;
|
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Add one if second bit 1, else subtract 1. Avoids if/else */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
t += (((k+1) & 2) - 1);
|
2003-10-18 01:58:35 +00:00
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Correct for the initial -1, to get the final symbol length */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
length[i] = t + 1;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Find largest and smallest lengths in this group */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
minLen = maxLen = length[0];
|
2006-12-26 10:42:51 +00:00
|
|
|
for (i = 1; i < symCount; i++) {
|
2007-04-10 17:16:33 +00:00
|
|
|
if (length[i] > maxLen) maxLen = length[i];
|
|
|
|
else if (length[i] < minLen) minLen = length[i];
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Calculate permute[], base[], and limit[] tables from length[].
|
|
|
|
*
|
2004-04-14 17:51:38 +00:00
|
|
|
* permute[] is the lookup table for converting Huffman coded symbols
|
2003-10-18 01:58:35 +00:00
|
|
|
* into decoded symbols. base[] is the amount to subtract from the
|
2004-04-14 17:51:38 +00:00
|
|
|
* value of a Huffman symbol of a given length when using permute[].
|
2003-10-18 01:58:35 +00:00
|
|
|
*
|
|
|
|
* limit[] indicates the largest numerical value a symbol with a given
|
2004-04-14 17:51:38 +00:00
|
|
|
* number of bits can have. This is how the Huffman codes can vary in
|
2003-10-23 06:52:01 +00:00
|
|
|
* length: each code with a value>limit[length] needs another bit.
|
2003-10-18 01:58:35 +00:00
|
|
|
*/
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
hufGroup = bd->groups + j;
|
2003-10-18 01:58:35 +00:00
|
|
|
hufGroup->minLen = minLen;
|
|
|
|
hufGroup->maxLen = maxLen;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Note that minLen can't be smaller than 1, so we adjust the base
|
|
|
|
and limit array pointers so we're not always wasting the first
|
|
|
|
entry. We do this again when using them (during symbol decoding).*/
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
base = hufGroup->base - 1;
|
|
|
|
limit = hufGroup->limit - 1;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Calculate permute[]. Concurently, initialize temp[] and limit[]. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
pp = 0;
|
|
|
|
for (i = minLen; i <= maxLen; i++) {
|
|
|
|
temp[i] = limit[i] = 0;
|
|
|
|
for (t = 0; t < symCount; t++)
|
|
|
|
if (length[t] == i)
|
|
|
|
hufGroup->permute[pp++] = t;
|
2003-10-23 06:52:01 +00:00
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Count symbols coded for at each bit length */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
for (i = 0; i < symCount; i++) temp[length[i]]++;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Calculate limit[] (the largest symbol-coding value at each bit
|
|
|
|
* length, which is (previous limit<<1)+symbols at this level), and
|
|
|
|
* base[] (number of symbols to ignore at each bit length, which is
|
2003-10-23 06:52:01 +00:00
|
|
|
* limit minus the cumulative count of symbols coded for already). */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
pp = t = 0;
|
|
|
|
for (i = minLen; i < maxLen; i++) {
|
|
|
|
pp += temp[i];
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* We read the largest possible symbol size and then unget bits
|
|
|
|
after determining how many we need, and those extra bits could
|
|
|
|
be set to anything. (They're noise from future symbols.) At
|
|
|
|
each level we're really only interested in the first few bits,
|
|
|
|
so here we set all the trailing to-be-ignored bits to 1 so they
|
|
|
|
don't affect the value>limit[length] comparison. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
limit[i] = (pp << (maxLen - i)) - 1;
|
|
|
|
pp <<= 1;
|
|
|
|
t += temp[i];
|
|
|
|
base[i+1] = pp - t;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2003-10-18 01:59:46 +00:00
|
|
|
limit[maxLen+1] = INT_MAX; /* Sentinal value for reading next sym. */
|
2007-04-10 17:16:33 +00:00
|
|
|
limit[maxLen] = pp + temp[maxLen] - 1;
|
|
|
|
base[minLen] = 0;
|
2003-10-18 01:58:35 +00:00
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* We've finished reading and digesting the block header. Now read this
|
2004-04-14 17:51:38 +00:00
|
|
|
block's Huffman coded symbols from the file and undo the Huffman coding
|
2003-10-18 01:58:35 +00:00
|
|
|
and run length encoding, saving the result into dbuf[dbufCount++]=uc */
|
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Initialize symbol occurrence counters and symbol Move To Front table */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
for (i = 0; i < 256; i++) {
|
2003-10-23 06:52:01 +00:00
|
|
|
byteCount[i] = 0;
|
2007-04-10 17:16:33 +00:00
|
|
|
mtfSymbol[i] = (unsigned char)i;
|
2003-10-23 06:52:01 +00:00
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Loop through compressed symbols. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
runPos = dbufCount = selector = 0;
|
2006-12-26 10:42:51 +00:00
|
|
|
for (;;) {
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2004-11-16 12:07:04 +00:00
|
|
|
/* fetch next Huffman coding group from list. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
symCount = GROUP_SIZE - 1;
|
|
|
|
if (selector >= nSelectors) return RETVAL_DATA_ERROR;
|
|
|
|
hufGroup = bd->groups + selectors[selector++];
|
|
|
|
base = hufGroup->base - 1;
|
|
|
|
limit = hufGroup->limit - 1;
|
|
|
|
continue_this_group:
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2004-04-14 17:51:38 +00:00
|
|
|
/* Read next Huffman-coded symbol. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Note: It is far cheaper to read maxLen bits and back up than it is
|
|
|
|
to read minLen bits and then an additional bit at a time, testing
|
|
|
|
as we go. Because there is a trailing last block (with file CRC),
|
|
|
|
there is no danger of the overread causing an unexpected EOF for a
|
|
|
|
valid compressed file. As a further optimization, we do the read
|
|
|
|
inline (falling back to a call to get_bits if the buffer runs
|
|
|
|
dry). The following (up to got_huff_bits:) is equivalent to
|
2007-04-10 17:16:33 +00:00
|
|
|
j = get_bits(bd, hufGroup->maxLen);
|
2003-10-23 06:52:01 +00:00
|
|
|
*/
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
while (bd->inbufBitCount < hufGroup->maxLen) {
|
|
|
|
if (bd->inbufPos == bd->inbufCount) {
|
|
|
|
j = get_bits(bd, hufGroup->maxLen);
|
2003-10-23 06:52:01 +00:00
|
|
|
goto got_huff_bits;
|
|
|
|
}
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->inbufBits = (bd->inbufBits << 8) | bd->inbuf[bd->inbufPos++];
|
|
|
|
bd->inbufBitCount += 8;
|
2003-10-23 06:52:01 +00:00
|
|
|
};
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->inbufBitCount -= hufGroup->maxLen;
|
|
|
|
j = (bd->inbufBits >> bd->inbufBitCount) & ((1 << hufGroup->maxLen) - 1);
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
got_huff_bits:
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Figure how how many bits are in next symbol and unget extras */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
i = hufGroup->minLen;
|
|
|
|
while (j > limit[i]) ++i;
|
2003-10-23 06:52:01 +00:00
|
|
|
bd->inbufBitCount += (hufGroup->maxLen - i);
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Huffman decode value to get nextSym (with bounds checking) */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (i > hufGroup->maxLen)
|
|
|
|
return RETVAL_DATA_ERROR;
|
|
|
|
j = (j >> (hufGroup->maxLen - i)) - base[i];
|
|
|
|
if ((unsigned)j >= MAX_SYMBOLS)
|
2003-10-23 06:52:01 +00:00
|
|
|
return RETVAL_DATA_ERROR;
|
2003-10-18 01:58:35 +00:00
|
|
|
nextSym = hufGroup->permute[j];
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* We have now decoded the symbol, which indicates either a new literal
|
|
|
|
byte, or a repeated run of the most recent literal byte. First,
|
|
|
|
check if nextSym indicates a repeated run, and if so loop collecting
|
|
|
|
how many times to repeat the last literal. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if ((unsigned)nextSym <= SYMBOL_RUNB) { /* RUNA or RUNB */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* If this is the start of a new run, zero out counter */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (!runPos) {
|
2003-10-18 01:58:35 +00:00
|
|
|
runPos = 1;
|
|
|
|
t = 0;
|
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Neat trick that saves 1 symbol: instead of or-ing 0 or 1 at
|
|
|
|
each bit position, add 1 or 2 instead. For example,
|
|
|
|
1011 is 1<<0 + 1<<1 + 2<<2. 1010 is 2<<0 + 2<<1 + 1<<2.
|
|
|
|
You can make any bit pattern that way using 1 less symbol than
|
|
|
|
the basic or 0/1 method (except all bits 0, which would use no
|
|
|
|
symbols, but a run of length 0 doesn't mean anything in this
|
|
|
|
context). Thus space is saved. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:59:46 +00:00
|
|
|
t += (runPos << nextSym); /* +runPos if RUNA; +2*runPos if RUNB */
|
2007-04-10 17:16:33 +00:00
|
|
|
if (runPos < dbufSize) runPos <<= 1;
|
2004-11-16 12:07:04 +00:00
|
|
|
goto end_of_huffman_loop;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* When we hit the first non-run symbol after a run, we now know
|
|
|
|
how many times to repeat the last literal, so append that many
|
|
|
|
copies to our buffer of decoded symbols (dbuf) now. (The last
|
|
|
|
literal used is the one at the head of the mtfSymbol array.) */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (runPos) {
|
|
|
|
runPos = 0;
|
|
|
|
if (dbufCount + t >= dbufSize) return RETVAL_DATA_ERROR;
|
2003-10-18 01:58:35 +00:00
|
|
|
|
|
|
|
uc = symToByte[mtfSymbol[0]];
|
|
|
|
byteCount[uc] += t;
|
2007-04-10 17:16:33 +00:00
|
|
|
while (t--) dbuf[dbufCount++] = uc;
|
2003-10-18 01:58:35 +00:00
|
|
|
}
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Is this the terminating symbol? */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (nextSym > symTotal) break;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* At this point, nextSym indicates a new literal character. Subtract
|
|
|
|
one to get the position in the MTF array at which this literal is
|
|
|
|
currently to be found. (Note that the result can't be -1 or 0,
|
|
|
|
because 0 and 1 are RUNA and RUNB. But another instance of the
|
|
|
|
first symbol in the mtf array, position 0, would have been handled
|
|
|
|
as part of a run above. Therefore 1 unused mtf position minus
|
|
|
|
2 non-literal nextSym values equals -1.) */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (dbufCount >= dbufSize) return RETVAL_DATA_ERROR;
|
2003-10-18 01:58:35 +00:00
|
|
|
i = nextSym - 1;
|
|
|
|
uc = mtfSymbol[i];
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Adjust the MTF array. Since we typically expect to move only a
|
|
|
|
* small number of symbols, and are bound by 256 in any case, using
|
|
|
|
* memmove here would typically be bigger and slower due to function
|
|
|
|
* call overhead and other assorted setup costs. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:59:46 +00:00
|
|
|
do {
|
|
|
|
mtfSymbol[i] = mtfSymbol[i-1];
|
|
|
|
} while (--i);
|
2003-10-18 01:58:35 +00:00
|
|
|
mtfSymbol[0] = uc;
|
2007-04-10 17:16:33 +00:00
|
|
|
uc = symToByte[uc];
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* We have our literal byte. Save it into dbuf. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
byteCount[uc]++;
|
2007-04-10 17:16:33 +00:00
|
|
|
dbuf[dbufCount++] = (unsigned)uc;
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2006-02-17 03:43:49 +00:00
|
|
|
/* Skip group initialization if we're not done with this group. Done
|
|
|
|
* this way to avoid compiler warning. */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
end_of_huffman_loop:
|
|
|
|
if (symCount--) goto continue_this_group;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2004-04-14 17:51:38 +00:00
|
|
|
/* At this point, we've read all the Huffman-coded symbols (and repeated
|
2007-04-10 17:18:12 +00:00
|
|
|
runs) for this block from the input stream, and decoded them into the
|
2003-10-23 06:52:01 +00:00
|
|
|
intermediate buffer. There are dbufCount many decoded bytes in dbuf[].
|
|
|
|
Now undo the Burrows-Wheeler transform on dbuf.
|
2003-10-18 01:58:35 +00:00
|
|
|
See http://dogma.net/markn/articles/bwt/bwt.htm
|
|
|
|
*/
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Turn byteCount into cumulative occurrence counts of 0 to n-1. */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
j = 0;
|
|
|
|
for (i = 0; i < 256; i++) {
|
|
|
|
k = j + byteCount[i];
|
2003-10-18 01:58:35 +00:00
|
|
|
byteCount[i] = j;
|
2007-04-10 17:16:33 +00:00
|
|
|
j = k;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Figure out what order dbuf would be in if we sorted it. */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
for (i = 0; i < dbufCount; i++) {
|
|
|
|
uc = (unsigned char)(dbuf[i] & 0xff);
|
2003-10-18 01:58:35 +00:00
|
|
|
dbuf[byteCount[uc]] |= (i << 8);
|
|
|
|
byteCount[uc]++;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Decode first byte by hand to initialize "previous" byte. Note that it
|
|
|
|
doesn't get output, and if the first three characters are identical
|
2003-10-23 06:52:01 +00:00
|
|
|
it doesn't qualify as a run (hence writeRunCountdown=5). */
|
2006-02-17 05:12:03 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (dbufCount) {
|
|
|
|
if (origPtr >= dbufCount) return RETVAL_DATA_ERROR;
|
|
|
|
bd->writePos = dbuf[origPtr];
|
|
|
|
bd->writeCurrent = (unsigned char)(bd->writePos & 0xff);
|
|
|
|
bd->writePos >>= 8;
|
|
|
|
bd->writeRunCountdown = 5;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->writeCount = dbufCount;
|
2002-11-03 07:28:38 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
return RETVAL_OK;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Undo burrows-wheeler transform on intermediate buffer to produce output.
|
2003-10-23 06:52:01 +00:00
|
|
|
If start_bunzip was initialized with out_fd=-1, then up to len bytes of
|
|
|
|
data are written to outbuf. Return value is number of bytes written or
|
|
|
|
error (all errors are negative numbers). If out_fd!=-1, outbuf and len
|
|
|
|
are ignored, data is written to out_fd and return is RETVAL_OK or error.
|
|
|
|
*/
|
|
|
|
|
2007-04-10 21:40:19 +00:00
|
|
|
int read_bunzip(bunzip_data *bd, char *outbuf, int len)
|
2003-10-18 01:58:35 +00:00
|
|
|
{
|
2007-04-10 17:16:33 +00:00
|
|
|
const unsigned *dbuf;
|
|
|
|
int pos, current, previous, gotcount;
|
2003-10-18 01:58:35 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* If last read was short due to end of file, return last block now */
|
2007-04-10 17:16:33 +00:00
|
|
|
if (bd->writeCount < 0) return bd->writeCount;
|
2003-10-23 06:52:01 +00:00
|
|
|
|
|
|
|
gotcount = 0;
|
2007-04-10 17:16:33 +00:00
|
|
|
dbuf = bd->dbuf;
|
|
|
|
pos = bd->writePos;
|
|
|
|
current = bd->writeCurrent;
|
2003-10-23 06:52:01 +00:00
|
|
|
|
|
|
|
/* We will always have pending decoded data to write into the output
|
|
|
|
buffer unless this is the very first call (in which case we haven't
|
2004-04-14 17:51:38 +00:00
|
|
|
Huffman-decoded a block into the intermediate buffer yet). */
|
2003-10-23 06:52:01 +00:00
|
|
|
|
|
|
|
if (bd->writeCopies) {
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Inside the loop, writeCopies means extra copies (beyond 1) */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
--bd->writeCopies;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Loop outputting bytes */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2006-12-26 10:42:51 +00:00
|
|
|
for (;;) {
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* If the output buffer is full, snapshot state and return */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (gotcount >= len) {
|
|
|
|
bd->writePos =pos;
|
|
|
|
bd->writeCurrent = current;
|
2003-10-23 06:52:01 +00:00
|
|
|
bd->writeCopies++;
|
|
|
|
return len;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Write next byte into output buffer, updating CRC */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
outbuf[gotcount++] = current;
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->writeCRC = (bd->writeCRC << 8)
|
|
|
|
^ bd->crc32Table[(bd->writeCRC >> 24) ^ current];
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Loop now if we're outputting multiple copies of this byte */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
if (bd->writeCopies) {
|
|
|
|
--bd->writeCopies;
|
|
|
|
continue;
|
|
|
|
}
|
2007-04-10 17:16:33 +00:00
|
|
|
decode_next_byte:
|
2003-10-23 06:52:01 +00:00
|
|
|
if (!bd->writeCount--) break;
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Follow sequence vector to undo Burrows-Wheeler transform */
|
2007-04-10 17:16:33 +00:00
|
|
|
previous = current;
|
|
|
|
pos = dbuf[pos];
|
|
|
|
current = pos & 0xff;
|
|
|
|
pos >>= 8;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* After 3 consecutive copies of the same byte, the 4th is a repeat
|
|
|
|
count. We count down from 4 instead
|
|
|
|
* of counting up because testing for non-zero is faster */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (--bd->writeRunCountdown) {
|
|
|
|
if (current != previous)
|
|
|
|
bd->writeRunCountdown = 4;
|
2003-10-18 01:58:35 +00:00
|
|
|
} else {
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* We have a repeated run, this byte indicates the count */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->writeCopies = current;
|
|
|
|
current = previous;
|
|
|
|
bd->writeRunCountdown = 5;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Sometimes there are just 3 bytes (run length 0) */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (!bd->writeCopies) goto decode_next_byte;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Subtract the 1 copy we'd output anyway to get extras */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
--bd->writeCopies;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
|
|
|
}
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Decompression of this block completed successfully */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->writeCRC = ~bd->writeCRC;
|
|
|
|
bd->totalCRC = ((bd->totalCRC << 1) | (bd->totalCRC >> 31)) ^ bd->writeCRC;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* If this block had a CRC error, force file level CRC error. */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (bd->writeCRC != bd->headerCRC) {
|
|
|
|
bd->totalCRC = bd->headerCRC+1;
|
2003-10-18 01:58:35 +00:00
|
|
|
return RETVAL_LAST_BLOCK;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
|
|
|
}
|
2003-10-23 06:52:01 +00:00
|
|
|
|
2004-04-14 17:51:38 +00:00
|
|
|
/* Refill the intermediate buffer by Huffman-decoding next block of input */
|
2003-10-23 06:52:01 +00:00
|
|
|
/* (previous is just a convenient unused temp variable here) */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
previous = get_next_block(bd);
|
|
|
|
if (previous) {
|
|
|
|
bd->writeCount = previous;
|
|
|
|
return (previous != RETVAL_LAST_BLOCK) ? previous : gotcount;
|
2003-10-23 06:52:01 +00:00
|
|
|
}
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->writeCRC = ~0;
|
|
|
|
pos = bd->writePos;
|
|
|
|
current = bd->writeCurrent;
|
2003-10-23 06:52:01 +00:00
|
|
|
goto decode_next_byte;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
|
|
|
|
2007-04-10 21:40:19 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Allocate the structure, read file header. If in_fd==-1, inbuf must contain
|
|
|
|
a complete bunzip file (len bytes long). If in_fd!=-1, inbuf and len are
|
|
|
|
ignored, and data is read from file handle into temporary buffer. */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 21:40:19 +00:00
|
|
|
/* Because bunzip2 is used for help text unpacking, and because bb_show_usage()
|
|
|
|
should work for NOFORK applets too, we must be extremely careful to not leak
|
|
|
|
any allocations! */
|
|
|
|
|
|
|
|
int start_bunzip(bunzip_data **bdp, int in_fd, const unsigned char *inbuf,
|
2005-09-25 03:12:26 +00:00
|
|
|
int len)
|
2002-11-03 07:28:38 +00:00
|
|
|
{
|
2003-10-18 01:58:35 +00:00
|
|
|
bunzip_data *bd;
|
2007-04-10 17:16:33 +00:00
|
|
|
unsigned i;
|
|
|
|
enum {
|
|
|
|
BZh0 = ('B' << 24) + ('Z' << 16) + ('h' << 8) + '0'
|
|
|
|
};
|
2003-10-18 01:58:35 +00:00
|
|
|
|
|
|
|
/* Figure out how much data to allocate */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
i = sizeof(bunzip_data);
|
|
|
|
if (in_fd != -1) i += IOBUF_SIZE;
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Allocate bunzip_data. Most fields initialize to zero. */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
bd = *bdp = xzalloc(i);
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Setup input buffer */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->in_fd = in_fd;
|
|
|
|
if (-1 == in_fd) {
|
2007-04-10 21:40:19 +00:00
|
|
|
/* in this case, bd->inbuf is read-only */
|
|
|
|
bd->inbuf = (void*)inbuf; /* cast away const-ness */
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->inbufCount = len;
|
|
|
|
} else
|
|
|
|
bd->inbuf = (unsigned char *)(bd + 1);
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Init the CRC32 table (big endian) */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 21:40:19 +00:00
|
|
|
crc32_filltable(bd->crc32Table, 1);
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Setup for I/O error handling via longjmp */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
i = setjmp(bd->jmpbuf);
|
|
|
|
if (i) return i;
|
2003-10-23 06:52:01 +00:00
|
|
|
|
|
|
|
/* Ensure that file starts with "BZh['1'-'9']." */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
i = get_bits(bd, 32);
|
2007-04-10 21:40:19 +00:00
|
|
|
if ((unsigned)(i - BZh0 - 1) >= 9) return RETVAL_NOT_BZIP_DATA;
|
2003-10-23 06:52:01 +00:00
|
|
|
|
|
|
|
/* Fourth byte (ascii '1'-'9'), indicates block size in units of 100k of
|
2003-10-18 01:58:35 +00:00
|
|
|
uncompressed data. Allocate intermediate buffer for block. */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
bd->dbufSize = 100000 * (i - BZh0);
|
2003-10-23 06:52:01 +00:00
|
|
|
|
2007-04-10 21:40:19 +00:00
|
|
|
/* Cannot use xmalloc - may leak bd in NOFORK case! */
|
|
|
|
bd->dbuf = malloc_or_warn(bd->dbufSize * sizeof(int));
|
|
|
|
if (!bd->dbuf) {
|
|
|
|
free(bd);
|
|
|
|
xfunc_die();
|
|
|
|
}
|
2003-10-18 01:58:35 +00:00
|
|
|
return RETVAL_OK;
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
|
|
|
|
2007-04-10 21:40:19 +00:00
|
|
|
void dealloc_bunzip(bunzip_data *bd)
|
|
|
|
{
|
2007-05-30 00:29:55 +00:00
|
|
|
free(bd->dbuf);
|
|
|
|
free(bd);
|
2007-04-10 21:40:19 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/* Decompress src_fd to dst_fd. Stops at end of bzip data, not end of file. */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2006-10-01 15:55:11 +00:00
|
|
|
USE_DESKTOP(long long) int
|
2007-06-08 13:05:39 +00:00
|
|
|
unpack_bz2_stream(int src_fd, int dst_fd)
|
2002-11-03 07:28:38 +00:00
|
|
|
{
|
2006-10-01 15:55:11 +00:00
|
|
|
USE_DESKTOP(long long total_written = 0;)
|
2003-10-23 06:52:01 +00:00
|
|
|
char *outbuf;
|
2003-10-18 01:58:35 +00:00
|
|
|
bunzip_data *bd;
|
|
|
|
int i;
|
2002-11-03 07:28:38 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
outbuf = xmalloc(IOBUF_SIZE);
|
2007-04-10 21:40:19 +00:00
|
|
|
i = start_bunzip(&bd, src_fd, NULL, 0);
|
2007-04-10 17:16:33 +00:00
|
|
|
if (!i) {
|
2006-12-26 10:42:51 +00:00
|
|
|
for (;;) {
|
2007-04-10 17:16:33 +00:00
|
|
|
i = read_bunzip(bd, outbuf, IOBUF_SIZE);
|
|
|
|
if (i <= 0) break;
|
|
|
|
if (i != safe_write(dst_fd, outbuf, i)) {
|
|
|
|
i = RETVAL_UNEXPECTED_OUTPUT_EOF;
|
2003-10-23 06:52:01 +00:00
|
|
|
break;
|
|
|
|
}
|
2006-10-01 15:55:11 +00:00
|
|
|
USE_DESKTOP(total_written += i;)
|
2003-10-23 06:52:01 +00:00
|
|
|
}
|
2002-11-03 07:28:38 +00:00
|
|
|
}
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
/* Check CRC and release memory */
|
2006-02-17 03:43:49 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (i == RETVAL_LAST_BLOCK) {
|
|
|
|
if (bd->headerCRC != bd->totalCRC) {
|
2007-09-28 23:45:56 +00:00
|
|
|
bb_error_msg("CRC error");
|
2003-10-28 23:32:12 +00:00
|
|
|
} else {
|
2007-04-10 17:16:33 +00:00
|
|
|
i = RETVAL_OK;
|
2003-10-28 23:32:12 +00:00
|
|
|
}
|
2007-04-10 17:16:33 +00:00
|
|
|
} else if (i == RETVAL_UNEXPECTED_OUTPUT_EOF) {
|
2007-09-28 23:45:56 +00:00
|
|
|
bb_error_msg("unexpected EOF");
|
2003-10-28 23:32:12 +00:00
|
|
|
} else {
|
2007-09-28 23:45:56 +00:00
|
|
|
bb_error_msg("bunzip error %d", i);
|
2003-10-28 23:32:12 +00:00
|
|
|
}
|
2007-04-10 21:40:19 +00:00
|
|
|
dealloc_bunzip(bd);
|
2003-10-23 06:52:01 +00:00
|
|
|
free(outbuf);
|
2003-10-28 23:32:12 +00:00
|
|
|
|
2006-10-01 15:55:11 +00:00
|
|
|
return i ? i : USE_DESKTOP(total_written) + 0;
|
2003-10-18 01:58:35 +00:00
|
|
|
}
|
2002-11-03 07:28:38 +00:00
|
|
|
|
2003-10-23 06:52:01 +00:00
|
|
|
#ifdef TESTING
|
2002-11-03 14:05:15 +00:00
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
static char *const bunzip_errors[] = {
|
|
|
|
NULL, "Bad file checksum", "Not bzip data",
|
|
|
|
"Unexpected input EOF", "Unexpected output EOF", "Data error",
|
|
|
|
"Out of memory", "Obsolete (pre 0.9.5) bzip format not supported"
|
|
|
|
};
|
2002-11-03 14:05:15 +00:00
|
|
|
|
2003-10-18 01:58:35 +00:00
|
|
|
/* Dumb little test thing, decompress stdin to stdout */
|
2007-04-04 20:52:03 +00:00
|
|
|
int main(int argc, char **argv)
|
2003-10-18 01:58:35 +00:00
|
|
|
{
|
2007-06-08 13:05:39 +00:00
|
|
|
int i = unpack_bz2_stream(0, 1);
|
2003-10-23 06:52:01 +00:00
|
|
|
char c;
|
|
|
|
|
2007-04-10 17:16:33 +00:00
|
|
|
if (i < 0)
|
|
|
|
fprintf(stderr,"%s\n", bunzip_errors[-i]);
|
|
|
|
else if (read(0, &c, 1))
|
|
|
|
fprintf(stderr,"Trailing garbage ignored\n");
|
2003-10-23 06:52:01 +00:00
|
|
|
return -i;
|
2003-10-18 01:58:35 +00:00
|
|
|
}
|
|
|
|
#endif
|