--- /dev/null
+
+/*-------------------------------------------------------------*/
+/*--- Block sorting machinery ---*/
+/*--- blocksort.c ---*/
+/*-------------------------------------------------------------*/
+
+/* ------------------------------------------------------------------
+ This file is part of bzip2/libbzip2, a program and library for
+ lossless, block-sorting data compression.
+
+ bzip2/libbzip2 version 1.0.5 of 10 December 2007
+ Copyright (C) 1996-2007 Julian Seward <jseward@bzip.org>
+
+ Please read the WARNING, DISCLAIMER and PATENTS sections in the
+ README file.
+
+ This program is released under the terms of the license contained
+ in the file LICENSE.
+ ------------------------------------------------------------------ */
+
+
+#include "bzlib_private.h"
+
+/*---------------------------------------------*/
+/*--- Fallback O(N log(N)^2) sorting ---*/
+/*--- algorithm, for repetitive blocks ---*/
+/*---------------------------------------------*/
+
+/*---------------------------------------------*/
+static
+__inline__
+void fallbackSimpleSort ( UInt32* fmap,
+ UInt32* eclass,
+ Int32 lo,
+ Int32 hi )
+{
+ Int32 i, j, tmp;
+ UInt32 ec_tmp;
+
+ if (lo == hi) return;
+
+ if (hi - lo > 3) {
+ for ( i = hi-4; i >= lo; i-- ) {
+ tmp = fmap[i];
+ ec_tmp = eclass[tmp];
+ for ( j = i+4; j <= hi && ec_tmp > eclass[fmap[j]]; j += 4 )
+ fmap[j-4] = fmap[j];
+ fmap[j-4] = tmp;
+ }
+ }
+
+ for ( i = hi-1; i >= lo; i-- ) {
+ tmp = fmap[i];
+ ec_tmp = eclass[tmp];
+ for ( j = i+1; j <= hi && ec_tmp > eclass[fmap[j]]; j++ )
+ fmap[j-1] = fmap[j];
+ fmap[j-1] = tmp;
+ }
+}
+
+
+/*---------------------------------------------*/
+#define fswap(zz1, zz2) \
+ { Int32 zztmp = zz1; zz1 = zz2; zz2 = zztmp; }
+
+#define fvswap(zzp1, zzp2, zzn) \
+{ \
+ Int32 yyp1 = (zzp1); \
+ Int32 yyp2 = (zzp2); \
+ Int32 yyn = (zzn); \
+ while (yyn > 0) { \
+ fswap(fmap[yyp1], fmap[yyp2]); \
+ yyp1++; yyp2++; yyn--; \
+ } \
+}
+
+
+#define fmin(a,b) ((a) < (b)) ? (a) : (b)
+
+#define fpush(lz,hz) { stackLo[sp] = lz; \
+ stackHi[sp] = hz; \
+ sp++; }
+
+#define fpop(lz,hz) { sp--; \
+ lz = stackLo[sp]; \
+ hz = stackHi[sp]; }
+
+#define FALLBACK_QSORT_SMALL_THRESH 10
+#define FALLBACK_QSORT_STACK_SIZE 100
+
+
+static
+void fallbackQSort3 ( UInt32* fmap,
+ UInt32* eclass,
+ Int32 loSt,
+ Int32 hiSt )
+{
+ Int32 unLo, unHi, ltLo, gtHi, n, m;
+ Int32 sp, lo, hi;
+ UInt32 med, r, r3;
+ Int32 stackLo[FALLBACK_QSORT_STACK_SIZE];
+ Int32 stackHi[FALLBACK_QSORT_STACK_SIZE];
+
+ r = 0;
+
+ sp = 0;
+ fpush ( loSt, hiSt );
+
+ while (sp > 0) {
+
+ AssertH ( sp < FALLBACK_QSORT_STACK_SIZE - 1, 1004 );
+
+ fpop ( lo, hi );
+ if (hi - lo < FALLBACK_QSORT_SMALL_THRESH) {
+ fallbackSimpleSort ( fmap, eclass, lo, hi );
+ continue;
+ }
+
+ /* Random partitioning. Median of 3 sometimes fails to
+ avoid bad cases. Median of 9 seems to help but
+ looks rather expensive. This too seems to work but
+ is cheaper. Guidance for the magic constants
+ 7621 and 32768 is taken from Sedgewick's algorithms
+ book, chapter 35.
+ */
+ r = ((r * 7621) + 1) % 32768;
+ r3 = r % 3;
+ if (r3 == 0) med = eclass[fmap[lo]]; else
+ if (r3 == 1) med = eclass[fmap[(lo+hi)>>1]]; else
+ med = eclass[fmap[hi]];
+
+ unLo = ltLo = lo;
+ unHi = gtHi = hi;
+
+ while (1) {
+ while (1) {
+ if (unLo > unHi) break;
+ n = (Int32)eclass[fmap[unLo]] - (Int32)med;
+ if (n == 0) {
+ fswap(fmap[unLo], fmap[ltLo]);
+ ltLo++; unLo++;
+ continue;
+ };
+ if (n > 0) break;
+ unLo++;
+ }
+ while (1) {
+ if (unLo > unHi) break;
+ n = (Int32)eclass[fmap[unHi]] - (Int32)med;
+ if (n == 0) {
+ fswap(fmap[unHi], fmap[gtHi]);
+ gtHi--; unHi--;
+ continue;
+ };
+ if (n < 0) break;
+ unHi--;
+ }
+ if (unLo > unHi) break;
+ fswap(fmap[unLo], fmap[unHi]); unLo++; unHi--;
+ }
+
+ AssertD ( unHi == unLo-1, "fallbackQSort3(2)" );
+
+ if (gtHi < ltLo) continue;
+
+ n = fmin(ltLo-lo, unLo-ltLo); fvswap(lo, unLo-n, n);
+ m = fmin(hi-gtHi, gtHi-unHi); fvswap(unLo, hi-m+1, m);
+
+ n = lo + unLo - ltLo - 1;
+ m = hi - (gtHi - unHi) + 1;
+
+ if (n - lo > hi - m) {
+ fpush ( lo, n );
+ fpush ( m, hi );
+ } else {
+ fpush ( m, hi );
+ fpush ( lo, n );
+ }
+ }
+}
+
+#undef fmin
+#undef fpush
+#undef fpop
+#undef fswap
+#undef fvswap
+#undef FALLBACK_QSORT_SMALL_THRESH
+#undef FALLBACK_QSORT_STACK_SIZE
+
+
+/*---------------------------------------------*/
+/* Pre:
+ nblock > 0
+ eclass exists for [0 .. nblock-1]
+ ((UChar*)eclass) [0 .. nblock-1] holds block
+ ptr exists for [0 .. nblock-1]
+
+ Post:
+ ((UChar*)eclass) [0 .. nblock-1] holds block
+ All other areas of eclass destroyed
+ fmap [0 .. nblock-1] holds sorted order
+ bhtab [ 0 .. 2+(nblock/32) ] destroyed
+*/
+
+#define SET_BH(zz) bhtab[(zz) >> 5] |= (1 << ((zz) & 31))
+#define CLEAR_BH(zz) bhtab[(zz) >> 5] &= ~(1 << ((zz) & 31))
+#define ISSET_BH(zz) (bhtab[(zz) >> 5] & (1 << ((zz) & 31)))
+#define WORD_BH(zz) bhtab[(zz) >> 5]
+#define UNALIGNED_BH(zz) ((zz) & 0x01f)
+
+static
+void fallbackSort ( UInt32* fmap,
+ UInt32* eclass,
+ UInt32* bhtab,
+ Int32 nblock,
+ Int32 verb )
+{
+ Int32 ftab[257];
+ Int32 ftabCopy[256];
+ Int32 H, i, j, k, l, r, cc, cc1;
+ Int32 nNotDone;
+ Int32 nBhtab;
+ UChar* eclass8 = (UChar*)eclass;
+
+ /*--
+ Initial 1-char radix sort to generate
+ initial fmap and initial BH bits.
+ --*/
+ if (verb >= 4)
+ VPrintf0 ( " bucket sorting ...\n" );
+ for (i = 0; i < 257; i++) ftab[i] = 0;
+ for (i = 0; i < nblock; i++) ftab[eclass8[i]]++;
+ for (i = 0; i < 256; i++) ftabCopy[i] = ftab[i];
+ for (i = 1; i < 257; i++) ftab[i] += ftab[i-1];
+
+ for (i = 0; i < nblock; i++) {
+ j = eclass8[i];
+ k = ftab[j] - 1;
+ ftab[j] = k;
+ fmap[k] = i;
+ }
+
+ nBhtab = 2 + (nblock / 32);
+ for (i = 0; i < nBhtab; i++) bhtab[i] = 0;
+ for (i = 0; i < 256; i++) SET_BH(ftab[i]);
+
+ /*--
+ Inductively refine the buckets. Kind-of an
+ "exponential radix sort" (!), inspired by the
+ Manber-Myers suffix array construction algorithm.
+ --*/
+
+ /*-- set sentinel bits for block-end detection --*/
+ for (i = 0; i < 32; i++) {
+ SET_BH(nblock + 2*i);
+ CLEAR_BH(nblock + 2*i + 1);
+ }
+
+ /*-- the log(N) loop --*/
+ H = 1;
+ while (1) {
+
+ if (verb >= 4)
+ VPrintf1 ( " depth %6d has ", H );
+
+ j = 0;
+ for (i = 0; i < nblock; i++) {
+ if (ISSET_BH(i)) j = i;
+ k = fmap[i] - H; if (k < 0) k += nblock;
+ eclass[k] = j;
+ }
+
+ nNotDone = 0;
+ r = -1;
+ while (1) {
+
+ /*-- find the next non-singleton bucket --*/
+ k = r + 1;
+ while (ISSET_BH(k) && UNALIGNED_BH(k)) k++;
+ if (ISSET_BH(k)) {
+ while (WORD_BH(k) == 0xffffffff) k += 32;
+ while (ISSET_BH(k)) k++;
+ }
+ l = k - 1;
+ if (l >= nblock) break;
+ while (!ISSET_BH(k) && UNALIGNED_BH(k)) k++;
+ if (!ISSET_BH(k)) {
+ while (WORD_BH(k) == 0x00000000) k += 32;
+ while (!ISSET_BH(k)) k++;
+ }
+ r = k - 1;
+ if (r >= nblock) break;
+
+ /*-- now [l, r] bracket current bucket --*/
+ if (r > l) {
+ nNotDone += (r - l + 1);
+ fallbackQSort3 ( fmap, eclass, l, r );
+
+ /*-- scan bucket and generate header bits-- */
+ cc = -1;
+ for (i = l; i <= r; i++) {
+ cc1 = eclass[fmap[i]];
+ if (cc != cc1) { SET_BH(i); cc = cc1; };
+ }
+ }
+ }
+
+ if (verb >= 4)
+ VPrintf1 ( "%6d unresolved strings\n", nNotDone );
+
+ H *= 2;
+ if (H > nblock || nNotDone == 0) break;
+ }
+
+ /*--
+ Reconstruct the original block in
+ eclass8 [0 .. nblock-1], since the
+ previous phase destroyed it.
+ --*/
+ if (verb >= 4)
+ VPrintf0 ( " reconstructing block ...\n" );
+ j = 0;
+ for (i = 0; i < nblock; i++) {
+ while (ftabCopy[j] == 0) j++;
+ ftabCopy[j]--;
+ eclass8[fmap[i]] = (UChar)j;
+ }
+ AssertH ( j < 256, 1005 );
+}
+
+#undef SET_BH
+#undef CLEAR_BH
+#undef ISSET_BH
+#undef WORD_BH
+#undef UNALIGNED_BH
+
+
+/*---------------------------------------------*/
+/*--- The main, O(N^2 log(N)) sorting ---*/
+/*--- algorithm. Faster for "normal" ---*/
+/*--- non-repetitive blocks. ---*/
+/*---------------------------------------------*/
+
+/*---------------------------------------------*/
+static
+__inline__
+Bool mainGtU ( UInt32 i1,
+ UInt32 i2,
+ UChar* block,
+ UInt16* quadrant,
+ UInt32 nblock,
+ Int32* budget )
+{
+ Int32 k;
+ UChar c1, c2;
+ UInt16 s1, s2;
+
+ AssertD ( i1 != i2, "mainGtU" );
+ /* 1 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 2 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 3 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 4 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 5 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 6 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 7 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 8 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 9 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 10 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 11 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+ /* 12 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ i1++; i2++;
+
+ k = nblock + 8;
+
+ do {
+ /* 1 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ s1 = quadrant[i1]; s2 = quadrant[i2];
+ if (s1 != s2) return (s1 > s2);
+ i1++; i2++;
+ /* 2 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ s1 = quadrant[i1]; s2 = quadrant[i2];
+ if (s1 != s2) return (s1 > s2);
+ i1++; i2++;
+ /* 3 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ s1 = quadrant[i1]; s2 = quadrant[i2];
+ if (s1 != s2) return (s1 > s2);
+ i1++; i2++;
+ /* 4 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ s1 = quadrant[i1]; s2 = quadrant[i2];
+ if (s1 != s2) return (s1 > s2);
+ i1++; i2++;
+ /* 5 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ s1 = quadrant[i1]; s2 = quadrant[i2];
+ if (s1 != s2) return (s1 > s2);
+ i1++; i2++;
+ /* 6 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ s1 = quadrant[i1]; s2 = quadrant[i2];
+ if (s1 != s2) return (s1 > s2);
+ i1++; i2++;
+ /* 7 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ s1 = quadrant[i1]; s2 = quadrant[i2];
+ if (s1 != s2) return (s1 > s2);
+ i1++; i2++;
+ /* 8 */
+ c1 = block[i1]; c2 = block[i2];
+ if (c1 != c2) return (c1 > c2);
+ s1 = quadrant[i1]; s2 = quadrant[i2];
+ if (s1 != s2) return (s1 > s2);
+ i1++; i2++;
+
+ if (i1 >= nblock) i1 -= nblock;
+ if (i2 >= nblock) i2 -= nblock;
+
+ k -= 8;
+ (*budget)--;
+ }
+ while (k >= 0);
+
+ return False;
+}
+
+
+/*---------------------------------------------*/
+/*--
+ Knuth's increments seem to work better
+ than Incerpi-Sedgewick here. Possibly
+ because the number of elems to sort is
+ usually small, typically <= 20.
+--*/
+static
+Int32 incs[14] = { 1, 4, 13, 40, 121, 364, 1093, 3280,
+ 9841, 29524, 88573, 265720,
+ 797161, 2391484 };
+
+static
+void mainSimpleSort ( UInt32* ptr,
+ UChar* block,
+ UInt16* quadrant,
+ Int32 nblock,
+ Int32 lo,
+ Int32 hi,
+ Int32 d,
+ Int32* budget )
+{
+ Int32 i, j, h, bigN, hp;
+ UInt32 v;
+
+ bigN = hi - lo + 1;
+ if (bigN < 2) return;
+
+ hp = 0;
+ while (incs[hp] < bigN) hp++;
+ hp--;
+
+ for (; hp >= 0; hp--) {
+ h = incs[hp];
+
+ i = lo + h;
+ while (True) {
+
+ /*-- copy 1 --*/
+ if (i > hi) break;
+ v = ptr[i];
+ j = i;
+ while ( mainGtU (
+ ptr[j-h]+d, v+d, block, quadrant, nblock, budget
+ ) ) {
+ ptr[j] = ptr[j-h];
+ j = j - h;
+ if (j <= (lo + h - 1)) break;
+ }
+ ptr[j] = v;
+ i++;
+
+ /*-- copy 2 --*/
+ if (i > hi) break;
+ v = ptr[i];
+ j = i;
+ while ( mainGtU (
+ ptr[j-h]+d, v+d, block, quadrant, nblock, budget
+ ) ) {
+ ptr[j] = ptr[j-h];
+ j = j - h;
+ if (j <= (lo + h - 1)) break;
+ }
+ ptr[j] = v;
+ i++;
+
+ /*-- copy 3 --*/
+ if (i > hi) break;
+ v = ptr[i];
+ j = i;
+ while ( mainGtU (
+ ptr[j-h]+d, v+d, block, quadrant, nblock, budget
+ ) ) {
+ ptr[j] = ptr[j-h];
+ j = j - h;
+ if (j <= (lo + h - 1)) break;
+ }
+ ptr[j] = v;
+ i++;
+
+ if (*budget < 0) return;
+ }
+ }
+}
+
+
+/*---------------------------------------------*/
+/*--
+ The following is an implementation of
+ an elegant 3-way quicksort for strings,
+ described in a paper "Fast Algorithms for
+ Sorting and Searching Strings", by Robert
+ Sedgewick and Jon L. Bentley.
+--*/
+
+#define mswap(zz1, zz2) \
+ { Int32 zztmp = zz1; zz1 = zz2; zz2 = zztmp; }
+
+#define mvswap(zzp1, zzp2, zzn) \
+{ \
+ Int32 yyp1 = (zzp1); \
+ Int32 yyp2 = (zzp2); \
+ Int32 yyn = (zzn); \
+ while (yyn > 0) { \
+ mswap(ptr[yyp1], ptr[yyp2]); \
+ yyp1++; yyp2++; yyn--; \
+ } \
+}
+
+static
+__inline__
+UChar mmed3 ( UChar a, UChar b, UChar c )
+{
+ UChar t;
+ if (a > b) { t = a; a = b; b = t; };
+ if (b > c) {
+ b = c;
+ if (a > b) b = a;
+ }
+ return b;
+}
+
+#define mmin(a,b) ((a) < (b)) ? (a) : (b)
+
+#define mpush(lz,hz,dz) { stackLo[sp] = lz; \
+ stackHi[sp] = hz; \
+ stackD [sp] = dz; \
+ sp++; }
+
+#define mpop(lz,hz,dz) { sp--; \
+ lz = stackLo[sp]; \
+ hz = stackHi[sp]; \
+ dz = stackD [sp]; }
+
+
+#define mnextsize(az) (nextHi[az]-nextLo[az])
+
+#define mnextswap(az,bz) \
+ { Int32 tz; \
+ tz = nextLo[az]; nextLo[az] = nextLo[bz]; nextLo[bz] = tz; \
+ tz = nextHi[az]; nextHi[az] = nextHi[bz]; nextHi[bz] = tz; \
+ tz = nextD [az]; nextD [az] = nextD [bz]; nextD [bz] = tz; }
+
+
+#define MAIN_QSORT_SMALL_THRESH 20
+#define MAIN_QSORT_DEPTH_THRESH (BZ_N_RADIX + BZ_N_QSORT)
+#define MAIN_QSORT_STACK_SIZE 100
+
+static
+void mainQSort3 ( UInt32* ptr,
+ UChar* block,
+ UInt16* quadrant,
+ Int32 nblock,
+ Int32 loSt,
+ Int32 hiSt,
+ Int32 dSt,
+ Int32* budget )
+{
+ Int32 unLo, unHi, ltLo, gtHi, n, m, med;
+ Int32 sp, lo, hi, d;
+
+ Int32 stackLo[MAIN_QSORT_STACK_SIZE];
+ Int32 stackHi[MAIN_QSORT_STACK_SIZE];
+ Int32 stackD [MAIN_QSORT_STACK_SIZE];
+
+ Int32 nextLo[3];
+ Int32 nextHi[3];
+ Int32 nextD [3];
+
+ sp = 0;
+ mpush ( loSt, hiSt, dSt );
+
+ while (sp > 0) {
+
+ AssertH ( sp < MAIN_QSORT_STACK_SIZE - 2, 1001 );
+
+ mpop ( lo, hi, d );
+ if (hi - lo < MAIN_QSORT_SMALL_THRESH ||
+ d > MAIN_QSORT_DEPTH_THRESH) {
+ mainSimpleSort ( ptr, block, quadrant, nblock, lo, hi, d, budget );
+ if (*budget < 0) return;
+ continue;
+ }
+
+ med = (Int32)
+ mmed3 ( block[ptr[ lo ]+d],
+ block[ptr[ hi ]+d],
+ block[ptr[ (lo+hi)>>1 ]+d] );
+
+ unLo = ltLo = lo;
+ unHi = gtHi = hi;
+
+ while (True) {
+ while (True) {
+ if (unLo > unHi) break;
+ n = ((Int32)block[ptr[unLo]+d]) - med;
+ if (n == 0) {
+ mswap(ptr[unLo], ptr[ltLo]);
+ ltLo++; unLo++; continue;
+ };
+ if (n > 0) break;
+ unLo++;
+ }
+ while (True) {
+ if (unLo > unHi) break;
+ n = ((Int32)block[ptr[unHi]+d]) - med;
+ if (n == 0) {
+ mswap(ptr[unHi], ptr[gtHi]);
+ gtHi--; unHi--; continue;
+ };
+ if (n < 0) break;
+ unHi--;
+ }
+ if (unLo > unHi) break;
+ mswap(ptr[unLo], ptr[unHi]); unLo++; unHi--;
+ }
+
+ AssertD ( unHi == unLo-1, "mainQSort3(2)" );
+
+ if (gtHi < ltLo) {
+ mpush(lo, hi, d+1 );
+ continue;
+ }
+
+ n = mmin(ltLo-lo, unLo-ltLo); mvswap(lo, unLo-n, n);
+ m = mmin(hi-gtHi, gtHi-unHi); mvswap(unLo, hi-m+1, m);
+
+ n = lo + unLo - ltLo - 1;
+ m = hi - (gtHi - unHi) + 1;
+
+ nextLo[0] = lo; nextHi[0] = n; nextD[0] = d;
+ nextLo[1] = m; nextHi[1] = hi; nextD[1] = d;
+ nextLo[2] = n+1; nextHi[2] = m-1; nextD[2] = d+1;
+
+ if (mnextsize(0) < mnextsize(1)) mnextswap(0,1);
+ if (mnextsize(1) < mnextsize(2)) mnextswap(1,2);
+ if (mnextsize(0) < mnextsize(1)) mnextswap(0,1);
+
+ AssertD (mnextsize(0) >= mnextsize(1), "mainQSort3(8)" );
+ AssertD (mnextsize(1) >= mnextsize(2), "mainQSort3(9)" );
+
+ mpush (nextLo[0], nextHi[0], nextD[0]);
+ mpush (nextLo[1], nextHi[1], nextD[1]);
+ mpush (nextLo[2], nextHi[2], nextD[2]);
+ }
+}
+
+#undef mswap
+#undef mvswap
+#undef mpush
+#undef mpop
+#undef mmin
+#undef mnextsize
+#undef mnextswap
+#undef MAIN_QSORT_SMALL_THRESH
+#undef MAIN_QSORT_DEPTH_THRESH
+#undef MAIN_QSORT_STACK_SIZE
+
+
+/*---------------------------------------------*/
+/* Pre:
+ nblock > N_OVERSHOOT
+ block32 exists for [0 .. nblock-1 +N_OVERSHOOT]
+ ((UChar*)block32) [0 .. nblock-1] holds block
+ ptr exists for [0 .. nblock-1]
+
+ Post:
+ ((UChar*)block32) [0 .. nblock-1] holds block
+ All other areas of block32 destroyed
+ ftab [0 .. 65536 ] destroyed
+ ptr [0 .. nblock-1] holds sorted order
+ if (*budget < 0), sorting was abandoned
+*/
+
+#define BIGFREQ(b) (ftab[((b)+1) << 8] - ftab[(b) << 8])
+#define SETMASK (1 << 21)
+#define CLEARMASK (~(SETMASK))
+
+static
+void mainSort ( UInt32* ptr,
+ UChar* block,
+ UInt16* quadrant,
+ UInt32* ftab,
+ Int32 nblock,
+ Int32 verb,
+ Int32* budget )
+{
+ Int32 i, j, k, ss, sb;
+ Int32 runningOrder[256];
+ Bool bigDone[256];
+ Int32 copyStart[256];
+ Int32 copyEnd [256];
+ UChar c1;
+ Int32 numQSorted;
+ UInt16 s;
+ if (verb >= 4) VPrintf0 ( " main sort initialise ...\n" );
+
+ /*-- set up the 2-byte frequency table --*/
+ for (i = 65536; i >= 0; i--) ftab[i] = 0;
+
+ j = block[0] << 8;
+ i = nblock-1;
+ for (; i >= 3; i -= 4) {
+ quadrant[i] = 0;
+ j = (j >> 8) | ( ((UInt16)block[i]) << 8);
+ ftab[j]++;
+ quadrant[i-1] = 0;
+ j = (j >> 8) | ( ((UInt16)block[i-1]) << 8);
+ ftab[j]++;
+ quadrant[i-2] = 0;
+ j = (j >> 8) | ( ((UInt16)block[i-2]) << 8);
+ ftab[j]++;
+ quadrant[i-3] = 0;
+ j = (j >> 8) | ( ((UInt16)block[i-3]) << 8);
+ ftab[j]++;
+ }
+ for (; i >= 0; i--) {
+ quadrant[i] = 0;
+ j = (j >> 8) | ( ((UInt16)block[i]) << 8);
+ ftab[j]++;
+ }
+
+ /*-- (emphasises close relationship of block & quadrant) --*/
+ for (i = 0; i < BZ_N_OVERSHOOT; i++) {
+ block [nblock+i] = block[i];
+ quadrant[nblock+i] = 0;
+ }
+
+ if (verb >= 4) VPrintf0 ( " bucket sorting ...\n" );
+
+ /*-- Complete the initial radix sort --*/
+ for (i = 1; i <= 65536; i++) ftab[i] += ftab[i-1];
+
+ s = block[0] << 8;
+ i = nblock-1;
+ for (; i >= 3; i -= 4) {
+ s = (s >> 8) | (block[i] << 8);
+ j = ftab[s] -1;
+ ftab[s] = j;
+ ptr[j] = i;
+ s = (s >> 8) | (block[i-1] << 8);
+ j = ftab[s] -1;
+ ftab[s] = j;
+ ptr[j] = i-1;
+ s = (s >> 8) | (block[i-2] << 8);
+ j = ftab[s] -1;
+ ftab[s] = j;
+ ptr[j] = i-2;
+ s = (s >> 8) | (block[i-3] << 8);
+ j = ftab[s] -1;
+ ftab[s] = j;
+ ptr[j] = i-3;
+ }
+ for (; i >= 0; i--) {
+ s = (s >> 8) | (block[i] << 8);
+ j = ftab[s] -1;
+ ftab[s] = j;
+ ptr[j] = i;
+ }
+
+ /*--
+ Now ftab contains the first loc of every small bucket.
+ Calculate the running order, from smallest to largest
+ big bucket.
+ --*/
+ for (i = 0; i <= 255; i++) {
+ bigDone [i] = False;
+ runningOrder[i] = i;
+ }
+
+ {
+ Int32 vv;
+ Int32 h = 1;
+ do h = 3 * h + 1; while (h <= 256);
+ do {
+ h = h / 3;
+ for (i = h; i <= 255; i++) {
+ vv = runningOrder[i];
+ j = i;
+ while ( BIGFREQ(runningOrder[j-h]) > BIGFREQ(vv) ) {
+ runningOrder[j] = runningOrder[j-h];
+ j = j - h;
+ if (j <= (h - 1)) goto zero;
+ }
+ zero:
+ runningOrder[j] = vv;
+ }
+ } while (h != 1);
+ }
+
+ /*--
+ The main sorting loop.
+ --*/
+
+ numQSorted = 0;
+
+ for (i = 0; i <= 255; i++) {
+
+ /*--
+ Process big buckets, starting with the least full.
+ Basically this is a 3-step process in which we call
+ mainQSort3 to sort the small buckets [ss, j], but
+ also make a big effort to avoid the calls if we can.
+ --*/
+ ss = runningOrder[i];
+
+ /*--
+ Step 1:
+ Complete the big bucket [ss] by quicksorting
+ any unsorted small buckets [ss, j], for j != ss.
+ Hopefully previous pointer-scanning phases have already
+ completed many of the small buckets [ss, j], so
+ we don't have to sort them at all.
+ --*/
+ for (j = 0; j <= 255; j++) {
+ if (j != ss) {
+ sb = (ss << 8) + j;
+ if ( ! (ftab[sb] & SETMASK) ) {
+ Int32 lo = ftab[sb] & CLEARMASK;
+ Int32 hi = (ftab[sb+1] & CLEARMASK) - 1;
+ if (hi > lo) {
+ if (verb >= 4)
+ VPrintf4 ( " qsort [0x%x, 0x%x] "
+ "done %d this %d\n",
+ ss, j, numQSorted, hi - lo + 1 );
+ mainQSort3 (
+ ptr, block, quadrant, nblock,
+ lo, hi, BZ_N_RADIX, budget
+ );
+ numQSorted += (hi - lo + 1);
+ if (*budget < 0) return;
+ }
+ }
+ ftab[sb] |= SETMASK;
+ }
+ }
+
+ AssertH ( !bigDone[ss], 1006 );
+
+ /*--
+ Step 2:
+ Now scan this big bucket [ss] so as to synthesise the
+ sorted order for small buckets [t, ss] for all t,
+ including, magically, the bucket [ss,ss] too.
+ This will avoid doing Real Work in subsequent Step 1's.
+ --*/
+ {
+ for (j = 0; j <= 255; j++) {
+ copyStart[j] = ftab[(j << 8) + ss] & CLEARMASK;
+ copyEnd [j] = (ftab[(j << 8) + ss + 1] & CLEARMASK) - 1;
+ }
+ for (j = ftab[ss << 8] & CLEARMASK; j < copyStart[ss]; j++) {
+ k = ptr[j]-1; if (k < 0) k += nblock;
+ c1 = block[k];
+ if (!bigDone[c1])
+ ptr[ copyStart[c1]++ ] = k;
+ }
+ for (j = (ftab[(ss+1) << 8] & CLEARMASK) - 1; j > copyEnd[ss]; j--) {
+ k = ptr[j]-1; if (k < 0) k += nblock;
+ c1 = block[k];
+ if (!bigDone[c1])
+ ptr[ copyEnd[c1]-- ] = k;
+ }
+ }
+
+ AssertH ( (copyStart[ss]-1 == copyEnd[ss])
+ ||
+ /* Extremely rare case missing in bzip2-1.0.0 and 1.0.1.
+ Necessity for this case is demonstrated by compressing
+ a sequence of approximately 48.5 million of character
+ 251; 1.0.0/1.0.1 will then die here. */
+ (copyStart[ss] == 0 && copyEnd[ss] == nblock-1),
+ 1007 )
+
+ for (j = 0; j <= 255; j++) ftab[(j << 8) + ss] |= SETMASK;
+
+ /*--
+ Step 3:
+ The [ss] big bucket is now done. Record this fact,
+ and update the quadrant descriptors. Remember to
+ update quadrants in the overshoot area too, if
+ necessary. The "if (i < 255)" test merely skips
+ this updating for the last bucket processed, since
+ updating for the last bucket is pointless.
+
+ The quadrant array provides a way to incrementally
+ cache sort orderings, as they appear, so as to
+ make subsequent comparisons in fullGtU() complete
+ faster. For repetitive blocks this makes a big
+ difference (but not big enough to be able to avoid
+ the fallback sorting mechanism, exponential radix sort).
+
+ The precise meaning is: at all times:
+
+ for 0 <= i < nblock and 0 <= j <= nblock
+
+ if block[i] != block[j],
+
+ then the relative values of quadrant[i] and
+ quadrant[j] are meaningless.
+
+ else {
+ if quadrant[i] < quadrant[j]
+ then the string starting at i lexicographically
+ precedes the string starting at j
+
+ else if quadrant[i] > quadrant[j]
+ then the string starting at j lexicographically
+ precedes the string starting at i
+
+ else
+ the relative ordering of the strings starting
+ at i and j has not yet been determined.
+ }
+ --*/
+ bigDone[ss] = True;
+
+ if (i < 255) {
+ Int32 bbStart = ftab[ss << 8] & CLEARMASK;
+ Int32 bbSize = (ftab[(ss+1) << 8] & CLEARMASK) - bbStart;
+ Int32 shifts = 0;
+
+ while ((bbSize >> shifts) > 65534) shifts++;
+
+ for (j = bbSize-1; j >= 0; j--) {
+ Int32 a2update = ptr[bbStart + j];
+ UInt16 qVal = (UInt16)(j >> shifts);
+ quadrant[a2update] = qVal;
+ if (a2update < BZ_N_OVERSHOOT)
+ quadrant[a2update + nblock] = qVal;
+ }
+ AssertH ( ((bbSize-1) >> shifts) <= 65535, 1002 );
+ }
+
+ }
+
+ if (verb >= 4)
+ VPrintf3 ( " %d pointers, %d sorted, %d scanned\n",
+ nblock, numQSorted, nblock - numQSorted );
+}
+
+#undef BIGFREQ
+#undef SETMASK
+#undef CLEARMASK
+
+
+/*---------------------------------------------*/
+/* Pre:
+ nblock > 0
+ arr2 exists for [0 .. nblock-1 +N_OVERSHOOT]
+ ((UChar*)arr2) [0 .. nblock-1] holds block
+ arr1 exists for [0 .. nblock-1]
+
+ Post:
+ ((UChar*)arr2) [0 .. nblock-1] holds block
+ All other areas of block destroyed
+ ftab [ 0 .. 65536 ] destroyed
+ arr1 [0 .. nblock-1] holds sorted order
+*/
+void BZ2_blockSort ( EState* s )
+{
+ UInt32* ptr = s->ptr;
+ UChar* block = s->block;
+ UInt32* ftab = s->ftab;
+ Int32 nblock = s->nblock;
+ Int32 verb = s->verbosity;
+ Int32 wfact = s->workFactor;
+ UInt16* quadrant;
+ Int32 budget;
+ Int32 budgetInit;
+ Int32 i;
+
+ if (nblock < 10000) {
+ fallbackSort ( s->arr1, s->arr2, ftab, nblock, verb );
+ } else {
+ /* Calculate the location for quadrant, remembering to get
+ the alignment right. Assumes that &(block[0]) is at least
+ 2-byte aligned -- this should be ok since block is really
+ the first section of arr2.
+ */
+ i = nblock+BZ_N_OVERSHOOT;
+ if (i & 1) i++;
+ quadrant = (UInt16*)(&(block[i]));
+
+ /* (wfact-1) / 3 puts the default-factor-30
+ transition point at very roughly the same place as
+ with v0.1 and v0.9.0.
+ Not that it particularly matters any more, since the
+ resulting compressed stream is now the same regardless
+ of whether or not we use the main sort or fallback sort.
+ */
+ if (wfact < 1 ) wfact = 1;
+ if (wfact > 100) wfact = 100;
+ budgetInit = nblock * ((wfact-1) / 3);
+ budget = budgetInit;
+
+ mainSort ( ptr, block, quadrant, ftab, nblock, verb, &budget );
+ if (verb >= 3)
+ VPrintf3 ( " %d work, %d block, ratio %5.2f\n",
+ budgetInit - budget,
+ nblock,
+ (float)(budgetInit - budget) /
+ (float)(nblock==0 ? 1 : nblock) );
+ if (budget < 0) {
+ if (verb >= 2)
+ VPrintf0 ( " too repetitive; using fallback"
+ " sorting algorithm\n" );
+ fallbackSort ( s->arr1, s->arr2, ftab, nblock, verb );
+ }
+ }
+
+ s->origPtr = -1;
+ for (i = 0; i < s->nblock; i++)
+ if (ptr[i] == 0)
+ { s->origPtr = i; break; };
+
+ AssertH( s->origPtr != -1, 1003 );
+}
+
+
+/*-------------------------------------------------------------*/
+/*--- end blocksort.c ---*/
+/*-------------------------------------------------------------*/
--- /dev/null
+/*--
+ This file is a part of bzip2 and/or libbzip2, a program and
+ library for lossless, block-sorting data compression.
+
+ Copyright (C) 1996-1999 Julian R Seward. All rights reserved.
+
+ Redistribution and use in source and binary forms, with or without
+ modification, are permitted provided that the following conditions
+ are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. The origin of this software must not be misrepresented; you must
+ not claim that you wrote the original software. If you use this
+ software in a product, an acknowledgment in the product
+ documentation would be appreciated but is not required.
+
+ 3. Altered source versions must be plainly marked as such, and must
+ not be misrepresented as being the original software.
+
+ 4. The name of the author may not be used to endorse or promote
+ products derived from this software without specific prior written
+ permission.
+
+ THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS
+ OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
+ DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
+ GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+ NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ Julian Seward, Cambridge, UK.
+ jseward@acm.org
+ bzip2/libbzip2 version 0.9.5 of 24 May 1999
+
+ (Now heavily hacked to be part of zgz; decompression and
+ portability ripped out.)
+
+ This program is based on (at least) the work of:
+ Mike Burrows
+ David Wheeler
+ Peter Fenwick
+ Alistair Moffat
+ Radford Neal
+ Ian H. Witten
+ Robert Sedgewick
+ Jon L. Bentley
+
+ For more information on these sources, see the manual.
+--*/
--- /dev/null
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <signal.h>
+#include <math.h>
+#include <errno.h>
+#include <ctype.h>
+#include "bzlib.h"
+
+#define ERROR_IF_EOF(i) { if ((i) == EOF) ioError(); }
+#define ERROR_IF_NOT_ZERO(i) { if ((i) != 0) ioError(); }
+#define ERROR_IF_MINUS_ONE(i) { if ((i) == (-1)) ioError(); }
+
+#include <sys/types.h>
+#include <utime.h>
+#include <unistd.h>
+#include <sys/stat.h>
+#include <sys/times.h>
+
+typedef char Char;
+typedef unsigned char Bool;
+typedef unsigned char UChar;
+typedef int Int32;
+typedef unsigned int UInt32;
+typedef short Int16;
+typedef unsigned short UInt16;
+
+#define True ((Bool)1)
+#define False ((Bool)0)
+
+/*--
+ IntNative is your platform's `native' int size.
+ Only here to avoid probs with 64-bit platforms.
+--*/
+typedef int IntNative;
+
+static Int32 verbosity;
+static Int32 blockSize100k;
+static Int32 workFactor;
+
+static
+void panic (char *msg) {
+ perror("oops");
+ fprintf(stderr, "%s\n", msg);
+ exit(1);
+}
+
+static
+void compressStream ( FILE *stream, FILE *zStream )
+{
+ BZFILE* bzf = NULL;
+ UChar ibuf[5000];
+ Int32 nIbuf;
+ UInt32 nbytes_in, nbytes_out;
+ Int32 bzerr, bzerr_dummy, ret;
+
+ if (ferror(stream)) goto errhandler_io;
+ if (ferror(zStream)) goto errhandler_io;
+
+ bzf = bzWriteOpen ( &bzerr, zStream,
+ blockSize100k, verbosity, workFactor );
+ if (bzerr != BZ_OK) goto errhandler;
+
+ if (verbosity >= 2) fprintf ( stderr, "\n" );
+
+ while (True) {
+
+ if (myfeof(stream)) break;
+ nIbuf = fread ( ibuf, sizeof(UChar), 5000, stream );
+ if (ferror(stream)) goto errhandler_io;
+ if (nIbuf > 0) bzWrite ( &bzerr, bzf, (void*)ibuf, nIbuf );
+ if (bzerr != BZ_OK) goto errhandler;
+
+ }
+
+ bzWriteClose ( &bzerr, bzf, 0, &nbytes_in, &nbytes_out );
+ if (bzerr != BZ_OK) goto errhandler;
+
+ if (ferror(zStream)) goto errhandler_io;
+ ret = fflush ( zStream );
+ if (ret == EOF) goto errhandler_io;
+ if (zStream != stdout) {
+ ret = fclose ( zStream );
+ if (ret == EOF) goto errhandler_io;
+ }
+ if (ferror(stream)) goto errhandler_io;
+ ret = fclose ( stream );
+ if (ret == EOF) goto errhandler_io;
+
+ if (nbytes_in == 0) nbytes_in = 1;
+
+ return;
+
+ errhandler:
+ bzWriteClose ( &bzerr_dummy, bzf, 1, &nbytes_in, &nbytes_out );
+ switch (bzerr) {
+ case BZ_MEM_ERROR:
+ panic ( "out of memory" );
+ case BZ_IO_ERROR:
+ errhandler_io:
+ panic ( "io error" );
+ default:
+ panic ( "compress:unexpected error" );
+ }
+
+ panic ( "compress:end" );
+ /*notreached*/
+}
+
+void suse_bzip2(int level) {
+ workFactor = 30;
+ blockSize100k = level;
+
+ compressStream(stdin, stdout);
+}
+/*-------------------------------------------------------------*/
+/*--- Library top-level functions. ---*/
+/*--- bzlib.c ---*/
+/*-------------------------------------------------------------*/
+
+/* ------------------------------------------------------------------
+ This file is part of bzip2/libbzip2, a program and library for
+ lossless, block-sorting data compression.
+
+ bzip2/libbzip2 version 1.0.5 of 10 December 2007
+ Copyright (C) 1996-2007 Julian Seward <jseward@bzip.org>
+
+ Please read the WARNING, DISCLAIMER and PATENTS sections in the
+ README file.
+
+ This program is released under the terms of the license contained
+ in the file LICENSE.
+ ------------------------------------------------------------------ */
+
+/* CHANGES
+ 0.9.0 -- original version.
+ 0.9.0a/b -- no changes in this file.
+ 0.9.0c -- made zero-length BZ_FLUSH work correctly in bzCompress().
+ fixed bzWrite/bzRead to ignore zero-length requests.
+ fixed bzread to correctly handle read requests after EOF.
+ wrong parameter order in call to bzDecompressInit in
+ bzBuffToBuffDecompress. Fixed.
+*/
+
+#include "bzlib_private.h"
+
+
+/*---------------------------------------------------*/
+/*--- Compression stuff ---*/
+/*---------------------------------------------------*/
+
+
+/*---------------------------------------------------*/
+#ifndef BZ_NO_STDIO
+void BZ2_bz__AssertH__fail ( int errcode )
+{
+ fprintf(stderr,
+ "\n\nbzip2/libbzip2: internal error number %d.\n"
+ "This is a bug in bzip2/libbzip2, %s.\n"
+ "Please report it to me at: jseward@bzip.org. If this happened\n"
+ "when you were using some program which uses libbzip2 as a\n"
+ "component, you should also report this bug to the author(s)\n"
+ "of that program. Please make an effort to report this bug;\n"
+ "timely and accurate bug reports eventually lead to higher\n"
+ "quality software. Thanks. Julian Seward, 10 December 2007.\n\n",
+ errcode,
+ BZ2_bzlibVersion()
+ );
+
+ if (errcode == 1007) {
+ fprintf(stderr,
+ "\n*** A special note about internal error number 1007 ***\n"
+ "\n"
+ "Experience suggests that a common cause of i.e. 1007\n"
+ "is unreliable memory or other hardware. The 1007 assertion\n"
+ "just happens to cross-check the results of huge numbers of\n"
+ "memory reads/writes, and so acts (unintendedly) as a stress\n"
+ "test of your memory system.\n"
+ "\n"
+ "I suggest the following: try compressing the file again,\n"
+ "possibly monitoring progress in detail with the -vv flag.\n"
+ "\n"
+ "* If the error cannot be reproduced, and/or happens at different\n"
+ " points in compression, you may have a flaky memory system.\n"
+ " Try a memory-test program. I have used Memtest86\n"
+ " (www.memtest86.com). At the time of writing it is free (GPLd).\n"
+ " Memtest86 tests memory much more thorougly than your BIOSs\n"
+ " power-on test, and may find failures that the BIOS doesn't.\n"
+ "\n"
+ "* If the error can be repeatably reproduced, this is a bug in\n"
+ " bzip2, and I would very much like to hear about it. Please\n"
+ " let me know, and, ideally, save a copy of the file causing the\n"
+ " problem -- without which I will be unable to investigate it.\n"
+ "\n"
+ );
+ }
+
+ exit(3);
+}
+#endif
+
+
+/*---------------------------------------------------*/
+static
+int bz_config_ok ( void )
+{
+ if (sizeof(int) != 4) return 0;
+ if (sizeof(short) != 2) return 0;
+ if (sizeof(char) != 1) return 0;
+ return 1;
+}
+
+
+/*---------------------------------------------------*/
+static
+void* default_bzalloc ( void* opaque, Int32 items, Int32 size )
+{
+ void* v = malloc ( items * size );
+ return v;
+}
+
+static
+void default_bzfree ( void* opaque, void* addr )
+{
+ if (addr != NULL) free ( addr );
+}
+
+
+/*---------------------------------------------------*/
+static
+void prepare_new_block ( EState* s )
+{
+ Int32 i;
+ s->nblock = 0;
+ s->numZ = 0;
+ s->state_out_pos = 0;
+ BZ_INITIALISE_CRC ( s->blockCRC );
+ for (i = 0; i < 256; i++) s->inUse[i] = False;
+ s->blockNo++;
+}
+
+
+/*---------------------------------------------------*/
+static
+void init_RL ( EState* s )
+{
+ s->state_in_ch = 256;
+ s->state_in_len = 0;
+}
+
+
+static
+Bool isempty_RL ( EState* s )
+{
+ if (s->state_in_ch < 256 && s->state_in_len > 0)
+ return False; else
+ return True;
+}
+
+
+/*---------------------------------------------------*/
+int BZ_API(BZ2_bzCompressInit)
+ ( bz_stream* strm,
+ int blockSize100k,
+ int verbosity,
+ int workFactor )
+{
+ Int32 n;
+ EState* s;
+
+ if (!bz_config_ok()) return BZ_CONFIG_ERROR;
+
+ if (strm == NULL ||
+ blockSize100k < 1 || blockSize100k > 9 ||
+ workFactor < 0 || workFactor > 250)
+ return BZ_PARAM_ERROR;
+
+ if (workFactor == 0) workFactor = 30;
+ if (strm->bzalloc == NULL) strm->bzalloc = default_bzalloc;
+ if (strm->bzfree == NULL) strm->bzfree = default_bzfree;
+
+ s = BZALLOC( sizeof(EState) );
+ if (s == NULL) return BZ_MEM_ERROR;
+ s->strm = strm;
+
+ s->arr1 = NULL;
+ s->arr2 = NULL;
+ s->ftab = NULL;
+
+ n = 100000 * blockSize100k;
+ s->arr1 = BZALLOC( n * sizeof(UInt32) );
+ s->arr2 = BZALLOC( (n+BZ_N_OVERSHOOT) * sizeof(UInt32) );
+ s->ftab = BZALLOC( 65537 * sizeof(UInt32) );
+
+ if (s->arr1 == NULL || s->arr2 == NULL || s->ftab == NULL) {
+ if (s->arr1 != NULL) BZFREE(s->arr1);
+ if (s->arr2 != NULL) BZFREE(s->arr2);
+ if (s->ftab != NULL) BZFREE(s->ftab);
+ if (s != NULL) BZFREE(s);
+ return BZ_MEM_ERROR;
+ }
+
+ s->blockNo = 0;
+ s->state = BZ_S_INPUT;
+ s->mode = BZ_M_RUNNING;
+ s->combinedCRC = 0;
+ s->blockSize100k = blockSize100k;
+ s->nblockMAX = 100000 * blockSize100k - 19;
+ s->verbosity = verbosity;
+ s->workFactor = workFactor;
+
+ s->block = (UChar*)s->arr2;
+ s->mtfv = (UInt16*)s->arr1;
+ s->zbits = NULL;
+ s->ptr = (UInt32*)s->arr1;
+
+ strm->state = s;
+ strm->total_in_lo32 = 0;
+ strm->total_in_hi32 = 0;
+ strm->total_out_lo32 = 0;
+ strm->total_out_hi32 = 0;
+ init_RL ( s );
+ prepare_new_block ( s );
+ return BZ_OK;
+}
+
+
+/*---------------------------------------------------*/
+static
+void add_pair_to_block ( EState* s )
+{
+ Int32 i;
+ UChar ch = (UChar)(s->state_in_ch);
+ for (i = 0; i < s->state_in_len; i++) {
+ BZ_UPDATE_CRC( s->blockCRC, ch );
+ }
+ s->inUse[s->state_in_ch] = True;
+ switch (s->state_in_len) {
+ case 1:
+ s->block[s->nblock] = (UChar)ch; s->nblock++;
+ break;
+ case 2:
+ s->block[s->nblock] = (UChar)ch; s->nblock++;
+ s->block[s->nblock] = (UChar)ch; s->nblock++;
+ break;
+ case 3:
+ s->block[s->nblock] = (UChar)ch; s->nblock++;
+ s->block[s->nblock] = (UChar)ch; s->nblock++;
+ s->block[s->nblock] = (UChar)ch; s->nblock++;
+ break;
+ default:
+ s->inUse[s->state_in_len-4] = True;
+ s->block[s->nblock] = (UChar)ch; s->nblock++;
+ s->block[s->nblock] = (UChar)ch; s->nblock++;
+ s->block[s->nblock] = (UChar)ch; s->nblock++;
+ s->block[s->nblock] = (UChar)ch; s->nblock++;
+ s->block[s->nblock] = ((UChar)(s->state_in_len-4));
+ s->nblock++;
+ break;
+ }
+}
+
+
+/*---------------------------------------------------*/
+static
+void flush_RL ( EState* s )
+{
+ if (s->state_in_ch < 256) add_pair_to_block ( s );
+ init_RL ( s );
+}
+
+
+/*---------------------------------------------------*/
+#define ADD_CHAR_TO_BLOCK(zs,zchh0) \
+{ \
+ UInt32 zchh = (UInt32)(zchh0); \
+ /*-- fast track the common case --*/ \
+ if (zchh != zs->state_in_ch && \
+ zs->state_in_len == 1) { \
+ UChar ch = (UChar)(zs->state_in_ch); \
+ BZ_UPDATE_CRC( zs->blockCRC, ch ); \
+ zs->inUse[zs->state_in_ch] = True; \
+ zs->block[zs->nblock] = (UChar)ch; \
+ zs->nblock++; \
+ zs->state_in_ch = zchh; \
+ } \
+ else \
+ /*-- general, uncommon cases --*/ \
+ if (zchh != zs->state_in_ch || \
+ zs->state_in_len == 255) { \
+ if (zs->state_in_ch < 256) \
+ add_pair_to_block ( zs ); \
+ zs->state_in_ch = zchh; \
+ zs->state_in_len = 1; \
+ } else { \
+ zs->state_in_len++; \
+ } \
+}
+
+
+/*---------------------------------------------------*/
+static
+Bool copy_input_until_stop ( EState* s )
+{
+ Bool progress_in = False;
+
+ if (s->mode == BZ_M_RUNNING) {
+
+ /*-- fast track the common case --*/
+ while (True) {
+ /*-- block full? --*/
+ if (s->nblock >= s->nblockMAX) break;
+ /*-- no input? --*/
+ if (s->strm->avail_in == 0) break;
+ progress_in = True;
+ ADD_CHAR_TO_BLOCK ( s, (UInt32)(*((UChar*)(s->strm->next_in))) );
+ s->strm->next_in++;
+ s->strm->avail_in--;
+ s->strm->total_in_lo32++;
+ if (s->strm->total_in_lo32 == 0) s->strm->total_in_hi32++;
+ }
+
+ } else {
+
+ /*-- general, uncommon case --*/
+ while (True) {
+ /*-- block full? --*/
+ if (s->nblock >= s->nblockMAX) break;
+ /*-- no input? --*/
+ if (s->strm->avail_in == 0) break;
+ /*-- flush/finish end? --*/
+ if (s->avail_in_expect == 0) break;
+ progress_in = True;
+ ADD_CHAR_TO_BLOCK ( s, (UInt32)(*((UChar*)(s->strm->next_in))) );
+ s->strm->next_in++;
+ s->strm->avail_in--;
+ s->strm->total_in_lo32++;
+ if (s->strm->total_in_lo32 == 0) s->strm->total_in_hi32++;
+ s->avail_in_expect--;
+ }
+ }
+ return progress_in;
+}
+
+
+/*---------------------------------------------------*/
+static
+Bool copy_output_until_stop ( EState* s )
+{
+ Bool progress_out = False;
+
+ while (True) {
+
+ /*-- no output space? --*/
+ if (s->strm->avail_out == 0) break;
+
+ /*-- block done? --*/
+ if (s->state_out_pos >= s->numZ) break;
+
+ progress_out = True;
+ *(s->strm->next_out) = s->zbits[s->state_out_pos];
+ s->state_out_pos++;
+ s->strm->avail_out--;
+ s->strm->next_out++;
+ s->strm->total_out_lo32++;
+ if (s->strm->total_out_lo32 == 0) s->strm->total_out_hi32++;
+ }
+
+ return progress_out;
+}
+
+
+/*---------------------------------------------------*/
+static
+Bool handle_compress ( bz_stream* strm )
+{
+ Bool progress_in = False;
+ Bool progress_out = False;
+ EState* s = strm->state;
+
+ while (True) {
+
+ if (s->state == BZ_S_OUTPUT) {
+ progress_out |= copy_output_until_stop ( s );
+ if (s->state_out_pos < s->numZ) break;
+ if (s->mode == BZ_M_FINISHING &&
+ s->avail_in_expect == 0 &&
+ isempty_RL(s)) break;
+ prepare_new_block ( s );
+ s->state = BZ_S_INPUT;
+ if (s->mode == BZ_M_FLUSHING &&
+ s->avail_in_expect == 0 &&
+ isempty_RL(s)) break;
+ }
+
+ if (s->state == BZ_S_INPUT) {
+ progress_in |= copy_input_until_stop ( s );
+ if (s->mode != BZ_M_RUNNING && s->avail_in_expect == 0) {
+ flush_RL ( s );
+ BZ2_compressBlock ( s, (Bool)(s->mode == BZ_M_FINISHING) );
+ s->state = BZ_S_OUTPUT;
+ }
+ else
+ if (s->nblock >= s->nblockMAX) {
+ BZ2_compressBlock ( s, False );
+ s->state = BZ_S_OUTPUT;
+ }
+ else
+ if (s->strm->avail_in == 0) {
+ break;
+ }
+ }
+
+ }
+
+ return progress_in || progress_out;
+}
+
+
+/*---------------------------------------------------*/
+int BZ_API(BZ2_bzCompress) ( bz_stream *strm, int action )
+{
+ Bool progress;
+ EState* s;
+ if (strm == NULL) return BZ_PARAM_ERROR;
+ s = strm->state;
+ if (s == NULL) return BZ_PARAM_ERROR;
+ if (s->strm != strm) return BZ_PARAM_ERROR;
+
+ preswitch:
+ switch (s->mode) {
+
+ case BZ_M_IDLE:
+ return BZ_SEQUENCE_ERROR;
+
+ case BZ_M_RUNNING:
+ if (action == BZ_RUN) {
+ progress = handle_compress ( strm );
+ return progress ? BZ_RUN_OK : BZ_PARAM_ERROR;
+ }
+ else
+ if (action == BZ_FLUSH) {
+ s->avail_in_expect = strm->avail_in;
+ s->mode = BZ_M_FLUSHING;
+ goto preswitch;
+ }
+ else
+ if (action == BZ_FINISH) {
+ s->avail_in_expect = strm->avail_in;
+ s->mode = BZ_M_FINISHING;
+ goto preswitch;
+ }
+ else
+ return BZ_PARAM_ERROR;
+
+ case BZ_M_FLUSHING:
+ if (action != BZ_FLUSH) return BZ_SEQUENCE_ERROR;
+ if (s->avail_in_expect != s->strm->avail_in)
+ return BZ_SEQUENCE_ERROR;
+ progress = handle_compress ( strm );
+ if (s->avail_in_expect > 0 || !isempty_RL(s) ||
+ s->state_out_pos < s->numZ) return BZ_FLUSH_OK;
+ s->mode = BZ_M_RUNNING;
+ return BZ_RUN_OK;
+
+ case BZ_M_FINISHING:
+ if (action != BZ_FINISH) return BZ_SEQUENCE_ERROR;
+ if (s->avail_in_expect != s->strm->avail_in)
+ return BZ_SEQUENCE_ERROR;
+ progress = handle_compress ( strm );
+ if (!progress) return BZ_SEQUENCE_ERROR;
+ if (s->avail_in_expect > 0 || !isempty_RL(s) ||
+ s->state_out_pos < s->numZ) return BZ_FINISH_OK;
+ s->mode = BZ_M_IDLE;
+ return BZ_STREAM_END;
+ }
+ return BZ_OK; /*--not reached--*/
+}
+
+
+/*---------------------------------------------------*/
+int BZ_API(BZ2_bzCompressEnd) ( bz_stream *strm )
+{
+ EState* s;
+ if (strm == NULL) return BZ_PARAM_ERROR;
+ s = strm->state;
+ if (s == NULL) return BZ_PARAM_ERROR;
+ if (s->strm != strm) return BZ_PARAM_ERROR;
+
+ if (s->arr1 != NULL) BZFREE(s->arr1);
+ if (s->arr2 != NULL) BZFREE(s->arr2);
+ if (s->ftab != NULL) BZFREE(s->ftab);
+ BZFREE(strm->state);
+
+ strm->state = NULL;
+
+ return BZ_OK;
+}
+
+#ifndef BZ_NO_STDIO
+/*---------------------------------------------------*/
+/*--- File I/O stuff ---*/
+/*---------------------------------------------------*/
+
+#define BZ_SETERR(eee) \
+{ \
+ if (bzerror != NULL) *bzerror = eee; \
+ if (bzf != NULL) bzf->lastErr = eee; \
+}
+
+typedef
+ struct {
+ FILE* handle;
+ Char buf[BZ_MAX_UNUSED];
+ Int32 bufN;
+ Bool writing;
+ bz_stream strm;
+ Int32 lastErr;
+ Bool initialisedOk;
+ }
+ bzFile;
+
+
+/*---------------------------------------------*/
+static Bool myfeof ( FILE* f )
+{
+ Int32 c = fgetc ( f );
+ if (c == EOF) return True;
+ ungetc ( c, f );
+ return False;
+}
+
+
+/*---------------------------------------------------*/
+BZFILE* BZ_API(BZ2_bzWriteOpen)
+ ( int* bzerror,
+ FILE* f,
+ int blockSize100k,
+ int verbosity,
+ int workFactor )
+{
+ Int32 ret;
+ bzFile* bzf = NULL;
+
+ BZ_SETERR(BZ_OK);
+
+ if (f == NULL ||
+ (blockSize100k < 1 || blockSize100k > 9) ||
+ (workFactor < 0 || workFactor > 250) ||
+ (verbosity < 0 || verbosity > 4))
+ { BZ_SETERR(BZ_PARAM_ERROR); return NULL; };
+
+ if (ferror(f))
+ { BZ_SETERR(BZ_IO_ERROR); return NULL; };
+
+ bzf = malloc ( sizeof(bzFile) );
+ if (bzf == NULL)
+ { BZ_SETERR(BZ_MEM_ERROR); return NULL; };
+
+ BZ_SETERR(BZ_OK);
+ bzf->initialisedOk = False;
+ bzf->bufN = 0;
+ bzf->handle = f;
+ bzf->writing = True;
+ bzf->strm.bzalloc = NULL;
+ bzf->strm.bzfree = NULL;
+ bzf->strm.opaque = NULL;
+
+ if (workFactor == 0) workFactor = 30;
+ ret = BZ2_bzCompressInit ( &(bzf->strm), blockSize100k,
+ verbosity, workFactor );
+ if (ret != BZ_OK)
+ { BZ_SETERR(ret); free(bzf); return NULL; };
+
+ bzf->strm.avail_in = 0;
+ bzf->initialisedOk = True;
+ return bzf;
+}
+
+
+
+/*---------------------------------------------------*/
+void BZ_API(BZ2_bzWrite)
+ ( int* bzerror,
+ BZFILE* b,
+ void* buf,
+ int len )
+{
+ Int32 n, n2, ret;
+ bzFile* bzf = (bzFile*)b;
+
+ BZ_SETERR(BZ_OK);
+ if (bzf == NULL || buf == NULL || len < 0)
+ { BZ_SETERR(BZ_PARAM_ERROR); return; };
+ if (!(bzf->writing))
+ { BZ_SETERR(BZ_SEQUENCE_ERROR); return; };
+ if (ferror(bzf->handle))
+ { BZ_SETERR(BZ_IO_ERROR); return; };
+
+ if (len == 0)
+ { BZ_SETERR(BZ_OK); return; };
+
+ bzf->strm.avail_in = len;
+ bzf->strm.next_in = buf;
+
+ while (True) {
+ bzf->strm.avail_out = BZ_MAX_UNUSED;
+ bzf->strm.next_out = bzf->buf;
+ ret = BZ2_bzCompress ( &(bzf->strm), BZ_RUN );
+ if (ret != BZ_RUN_OK)
+ { BZ_SETERR(ret); return; };
+
+ if (bzf->strm.avail_out < BZ_MAX_UNUSED) {
+ n = BZ_MAX_UNUSED - bzf->strm.avail_out;
+ n2 = fwrite ( (void*)(bzf->buf), sizeof(UChar),
+ n, bzf->handle );
+ if (n != n2 || ferror(bzf->handle))
+ { BZ_SETERR(BZ_IO_ERROR); return; };
+ }
+
+ if (bzf->strm.avail_in == 0)
+ { BZ_SETERR(BZ_OK); return; };
+ }
+}
+
+
+/*---------------------------------------------------*/
+void BZ_API(BZ2_bzWriteClose)
+ ( int* bzerror,
+ BZFILE* b,
+ int abandon,
+ unsigned int* nbytes_in,
+ unsigned int* nbytes_out )
+{
+ BZ2_bzWriteClose64 ( bzerror, b, abandon,
+ nbytes_in, NULL, nbytes_out, NULL );
+}
+
+
+void BZ_API(BZ2_bzWriteClose64)
+ ( int* bzerror,
+ BZFILE* b,
+ int abandon,
+ unsigned int* nbytes_in_lo32,
+ unsigned int* nbytes_in_hi32,
+ unsigned int* nbytes_out_lo32,
+ unsigned int* nbytes_out_hi32 )
+{
+ Int32 n, n2, ret;
+ bzFile* bzf = (bzFile*)b;
+
+ if (bzf == NULL)
+ { BZ_SETERR(BZ_OK); return; };
+ if (!(bzf->writing))
+ { BZ_SETERR(BZ_SEQUENCE_ERROR); return; };
+ if (ferror(bzf->handle))
+ { BZ_SETERR(BZ_IO_ERROR); return; };
+
+ if (nbytes_in_lo32 != NULL) *nbytes_in_lo32 = 0;
+ if (nbytes_in_hi32 != NULL) *nbytes_in_hi32 = 0;
+ if (nbytes_out_lo32 != NULL) *nbytes_out_lo32 = 0;
+ if (nbytes_out_hi32 != NULL) *nbytes_out_hi32 = 0;
+
+ if ((!abandon) && bzf->lastErr == BZ_OK) {
+ while (True) {
+ bzf->strm.avail_out = BZ_MAX_UNUSED;
+ bzf->strm.next_out = bzf->buf;
+ ret = BZ2_bzCompress ( &(bzf->strm), BZ_FINISH );
+ if (ret != BZ_FINISH_OK && ret != BZ_STREAM_END)
+ { BZ_SETERR(ret); return; };
+
+ if (bzf->strm.avail_out < BZ_MAX_UNUSED) {
+ n = BZ_MAX_UNUSED - bzf->strm.avail_out;
+ n2 = fwrite ( (void*)(bzf->buf), sizeof(UChar),
+ n, bzf->handle );
+ if (n != n2 || ferror(bzf->handle))
+ { BZ_SETERR(BZ_IO_ERROR); return; };
+ }
+
+ if (ret == BZ_STREAM_END) break;
+ }
+ }
+
+ if ( !abandon && !ferror ( bzf->handle ) ) {
+ fflush ( bzf->handle );
+ if (ferror(bzf->handle))
+ { BZ_SETERR(BZ_IO_ERROR); return; };
+ }
+
+ if (nbytes_in_lo32 != NULL)
+ *nbytes_in_lo32 = bzf->strm.total_in_lo32;
+ if (nbytes_in_hi32 != NULL)
+ *nbytes_in_hi32 = bzf->strm.total_in_hi32;
+ if (nbytes_out_lo32 != NULL)
+ *nbytes_out_lo32 = bzf->strm.total_out_lo32;
+ if (nbytes_out_hi32 != NULL)
+ *nbytes_out_hi32 = bzf->strm.total_out_hi32;
+
+ BZ_SETERR(BZ_OK);
+ BZ2_bzCompressEnd ( &(bzf->strm) );
+ free ( bzf );
+}
+
+#endif
+
+
+/*---------------------------------------------------*/
+/*--- Misc convenience stuff ---*/
+/*---------------------------------------------------*/
+
+/*---------------------------------------------------*/
+int BZ_API(BZ2_bzBuffToBuffCompress)
+ ( char* dest,
+ unsigned int* destLen,
+ char* source,
+ unsigned int sourceLen,
+ int blockSize100k,
+ int verbosity,
+ int workFactor )
+{
+ bz_stream strm;
+ int ret;
+
+ if (dest == NULL || destLen == NULL ||
+ source == NULL ||
+ blockSize100k < 1 || blockSize100k > 9 ||
+ verbosity < 0 || verbosity > 4 ||
+ workFactor < 0 || workFactor > 250)
+ return BZ_PARAM_ERROR;
+
+ if (workFactor == 0) workFactor = 30;
+ strm.bzalloc = NULL;
+ strm.bzfree = NULL;
+ strm.opaque = NULL;
+ ret = BZ2_bzCompressInit ( &strm, blockSize100k,
+ verbosity, workFactor );
+ if (ret != BZ_OK) return ret;
+
+ strm.next_in = source;
+ strm.next_out = dest;
+ strm.avail_in = sourceLen;
+ strm.avail_out = *destLen;
+
+ ret = BZ2_bzCompress ( &strm, BZ_FINISH );
+ if (ret == BZ_FINISH_OK) goto output_overflow;
+ if (ret != BZ_STREAM_END) goto errhandler;
+
+ /* normal termination */
+ *destLen -= strm.avail_out;
+ BZ2_bzCompressEnd ( &strm );
+ return BZ_OK;
+
+ output_overflow:
+ BZ2_bzCompressEnd ( &strm );
+ return BZ_OUTBUFF_FULL;
+
+ errhandler:
+ BZ2_bzCompressEnd ( &strm );
+ return ret;
+}
+
+/*---------------------------------------------------*/
+/*--
+ Code contributed by Yoshioka Tsuneo (tsuneo@rr.iij4u.or.jp)
+ to support better zlib compatibility.
+ This code is not _officially_ part of libbzip2 (yet);
+ I haven't tested it, documented it, or considered the
+ threading-safeness of it.
+ If this code breaks, please contact both Yoshioka and me.
+--*/
+/*---------------------------------------------------*/
+
+/*---------------------------------------------------*/
+/*--
+ return version like "0.9.5d, 4-Sept-1999".
+--*/
+const char * BZ_API(BZ2_bzlibVersion)(void)
+{
+ return BZ_VERSION;
+}
+
+
+#ifndef BZ_NO_STDIO
+/*---------------------------------------------------*/
+
+#if defined(_WIN32) || defined(OS2) || defined(MSDOS)
+# include <fcntl.h>
+# include <io.h>
+# define SET_BINARY_MODE(file) setmode(fileno(file),O_BINARY)
+#else
+# define SET_BINARY_MODE(file)
+#endif
+static
+BZFILE * bzopen_or_bzdopen
+ ( const char *path, /* no use when bzdopen */
+ int fd, /* no use when bzdopen */
+ const char *mode,
+ int open_mode) /* bzopen: 0, bzdopen:1 */
+{
+ int bzerr;
+ char unused[BZ_MAX_UNUSED];
+ int blockSize100k = 9;
+ int writing = 0;
+ char mode2[10] = "";
+ FILE *fp = NULL;
+ BZFILE *bzfp = NULL;
+ int verbosity = 0;
+ int workFactor = 30;
+ int smallMode = 0;
+ int nUnused = 0;
+
+ if (mode == NULL) return NULL;
+ while (*mode) {
+ switch (*mode) {
+ case 'r':
+ writing = 0; break;
+ case 'w':
+ writing = 1; break;
+ case 's':
+ smallMode = 1; break;
+ default:
+ if (isdigit((int)(*mode))) {
+ blockSize100k = *mode-BZ_HDR_0;
+ }
+ }
+ mode++;
+ }
+ strcat(mode2, writing ? "w" : "r" );
+ strcat(mode2,"b"); /* binary mode */
+
+ if (open_mode==0) {
+ if (path==NULL || strcmp(path,"")==0) {
+ fp = (writing ? stdout : stdin);
+ SET_BINARY_MODE(fp);
+ } else {
+ fp = fopen(path,mode2);
+ }
+ } else {
+#ifdef BZ_STRICT_ANSI
+ fp = NULL;
+#else
+ fp = fdopen(fd,mode2);
+#endif
+ }
+ if (fp == NULL) return NULL;
+
+ if (writing) {
+ /* Guard against total chaos and anarchy -- JRS */
+ if (blockSize100k < 1) blockSize100k = 1;
+ if (blockSize100k > 9) blockSize100k = 9;
+ bzfp = BZ2_bzWriteOpen(&bzerr,fp,blockSize100k,
+ verbosity,workFactor);
+ }
+ if (bzfp == NULL) {
+ if (fp != stdin && fp != stdout) fclose(fp);
+ return NULL;
+ }
+ return bzfp;
+}
+
+
+/*---------------------------------------------------*/
+/*--
+ open file for read or write.
+ ex) bzopen("file","w9")
+ case path="" or NULL => use stdin or stdout.
+--*/
+BZFILE * BZ_API(BZ2_bzopen)
+ ( const char *path,
+ const char *mode )
+{
+ return bzopen_or_bzdopen(path,-1,mode,/*bzopen*/0);
+}
+
+
+/*---------------------------------------------------*/
+BZFILE * BZ_API(BZ2_bzdopen)
+ ( int fd,
+ const char *mode )
+{
+ return bzopen_or_bzdopen(NULL,fd,mode,/*bzdopen*/1);
+}
+
+/*---------------------------------------------------*/
+int BZ_API(BZ2_bzwrite) (BZFILE* b, void* buf, int len )
+{
+ int bzerr;
+
+ BZ2_bzWrite(&bzerr,b,buf,len);
+ if(bzerr == BZ_OK){
+ return len;
+ }else{
+ return -1;
+ }
+}
+
+
+/*---------------------------------------------------*/
+int BZ_API(BZ2_bzflush) (BZFILE *b)
+{
+ /* do nothing now... */
+ return 0;
+}
+
+
+/*---------------------------------------------------*/
+void BZ_API(BZ2_bzclose) (BZFILE* b)
+{
+ int bzerr;
+ FILE *fp;
+
+ if (b==NULL) {return;}
+ fp = ((bzFile *)b)->handle;
+ if(((bzFile*)b)->writing){
+ BZ2_bzWriteClose(&bzerr,b,0,NULL,NULL);
+ if(bzerr != BZ_OK){
+ BZ2_bzWriteClose(NULL,b,1,NULL,NULL);
+ }
+ }
+ if(fp!=stdin && fp!=stdout){
+ fclose(fp);
+ }
+}
+
+
+/*---------------------------------------------------*/
+/*--
+ return last error code
+--*/
+static const char *bzerrorstrings[] = {
+ "OK"
+ ,"SEQUENCE_ERROR"
+ ,"PARAM_ERROR"
+ ,"MEM_ERROR"
+ ,"DATA_ERROR"
+ ,"DATA_ERROR_MAGIC"
+ ,"IO_ERROR"
+ ,"UNEXPECTED_EOF"
+ ,"OUTBUFF_FULL"
+ ,"CONFIG_ERROR"
+ ,"???" /* for future */
+ ,"???" /* for future */
+ ,"???" /* for future */
+ ,"???" /* for future */
+ ,"???" /* for future */
+ ,"???" /* for future */
+};
+
+
+const char * BZ_API(BZ2_bzerror) (BZFILE *b, int *errnum)
+{
+ int err = ((bzFile *)b)->lastErr;
+
+ if(err>0) err = 0;
+ *errnum = err;
+ return bzerrorstrings[err*-1];
+}
+#endif
+
+/*-------------------------------------------------------------*/
+/*--- end bzlib.c ---*/
+/*-------------------------------------------------------------*/
--- /dev/null
+
+/*-------------------------------------------------------------*/
+/*--- Private header file for the library. ---*/
+/*--- bzlib_private.h ---*/
+/*-------------------------------------------------------------*/
+
+/* ------------------------------------------------------------------
+ This file is part of bzip2/libbzip2, a program and library for
+ lossless, block-sorting data compression.
+
+ bzip2/libbzip2 version 1.0.5 of 10 December 2007
+ Copyright (C) 1996-2007 Julian Seward <jseward@bzip.org>
+
+ Please read the WARNING, DISCLAIMER and PATENTS sections in the
+ README file.
+
+ This program is released under the terms of the license contained
+ in the file LICENSE.
+ ------------------------------------------------------------------ */
+
+
+#ifndef _BZLIB_PRIVATE_H
+#define _BZLIB_PRIVATE_H
+
+#include <stdlib.h>
+
+#ifndef BZ_NO_STDIO
+#include <stdio.h>
+#include <ctype.h>
+#include <string.h>
+#endif
+
+#include "bzlib.h"
+
+
+
+/*-- General stuff. --*/
+
+#define BZ_VERSION "1.0.5, 10-Dec-2007"
+
+typedef char Char;
+typedef unsigned char Bool;
+typedef unsigned char UChar;
+typedef int Int32;
+typedef unsigned int UInt32;
+typedef short Int16;
+typedef unsigned short UInt16;
+
+#define True ((Bool)1)
+#define False ((Bool)0)
+
+#ifndef __GNUC__
+#define __inline__ /* */
+#endif
+
+#ifndef BZ_NO_STDIO
+
+extern void BZ2_bz__AssertH__fail ( int errcode );
+#define AssertH(cond,errcode) \
+ { if (!(cond)) BZ2_bz__AssertH__fail ( errcode ); }
+
+#if BZ_DEBUG
+#define AssertD(cond,msg) \
+ { if (!(cond)) { \
+ fprintf ( stderr, \
+ "\n\nlibbzip2(debug build): internal error\n\t%s\n", msg );\
+ exit(1); \
+ }}
+#else
+#define AssertD(cond,msg) /* */
+#endif
+
+#define VPrintf0(zf) \
+ fprintf(stderr,zf)
+#define VPrintf1(zf,za1) \
+ fprintf(stderr,zf,za1)
+#define VPrintf2(zf,za1,za2) \
+ fprintf(stderr,zf,za1,za2)
+#define VPrintf3(zf,za1,za2,za3) \
+ fprintf(stderr,zf,za1,za2,za3)
+#define VPrintf4(zf,za1,za2,za3,za4) \
+ fprintf(stderr,zf,za1,za2,za3,za4)
+#define VPrintf5(zf,za1,za2,za3,za4,za5) \
+ fprintf(stderr,zf,za1,za2,za3,za4,za5)
+
+#else
+
+extern void bz_internal_error ( int errcode );
+#define AssertH(cond,errcode) \
+ { if (!(cond)) bz_internal_error ( errcode ); }
+#define AssertD(cond,msg) do { } while (0)
+#define VPrintf0(zf) do { } while (0)
+#define VPrintf1(zf,za1) do { } while (0)
+#define VPrintf2(zf,za1,za2) do { } while (0)
+#define VPrintf3(zf,za1,za2,za3) do { } while (0)
+#define VPrintf4(zf,za1,za2,za3,za4) do { } while (0)
+#define VPrintf5(zf,za1,za2,za3,za4,za5) do { } while (0)
+
+#endif
+
+
+#define BZALLOC(nnn) (strm->bzalloc)(strm->opaque,(nnn),1)
+#define BZFREE(ppp) (strm->bzfree)(strm->opaque,(ppp))
+
+
+/*-- Header bytes. --*/
+
+#define BZ_HDR_B 0x42 /* 'B' */
+#define BZ_HDR_Z 0x5a /* 'Z' */
+#define BZ_HDR_h 0x68 /* 'h' */
+#define BZ_HDR_0 0x30 /* '0' */
+
+/*-- Constants for the back end. --*/
+
+#define BZ_MAX_ALPHA_SIZE 258
+#define BZ_MAX_CODE_LEN 23
+
+#define BZ_RUNA 0
+#define BZ_RUNB 1
+
+#define BZ_N_GROUPS 6
+#define BZ_G_SIZE 50
+#define BZ_N_ITERS 4
+
+#define BZ_MAX_SELECTORS (2 + (900000 / BZ_G_SIZE))
+
+
+
+/*-- Stuff for randomising repetitive blocks. --*/
+
+extern Int32 rNums[512];
+
+#define BZ_RAND_DECLS \
+ Int32 rNToGo; \
+ Int32 rTPos \
+
+#define BZ_RAND_INIT_MASK \
+ s->rNToGo = 0; \
+ s->rTPos = 0 \
+
+#define BZ_RAND_MASK ((s->rNToGo == 1) ? 1 : 0)
+
+#define BZ_RAND_UPD_MASK \
+ if (s->rNToGo == 0) { \
+ s->rNToGo = rNums[s->rTPos]; \
+ s->rTPos++; \
+ if (s->rTPos == 512) s->rTPos = 0; \
+ } \
+ s->rNToGo--;
+
+
+
+/*-- Stuff for doing CRCs. --*/
+
+extern UInt32 crc32Table[256];
+
+#define BZ_INITIALISE_CRC(crcVar) \
+{ \
+ crcVar = 0xffffffffL; \
+}
+
+#define BZ_FINALISE_CRC(crcVar) \
+{ \
+ crcVar = ~(crcVar); \
+}
+
+#define BZ_UPDATE_CRC(crcVar,cha) \
+{ \
+ crcVar = (crcVar << 8) ^ \
+ crc32Table[(crcVar >> 24) ^ \
+ ((UChar)cha)]; \
+}
+
+
+
+/*-- States and modes for compression. --*/
+
+#define BZ_M_IDLE 1
+#define BZ_M_RUNNING 2
+#define BZ_M_FLUSHING 3
+#define BZ_M_FINISHING 4
+
+#define BZ_S_OUTPUT 1
+#define BZ_S_INPUT 2
+
+#define BZ_N_RADIX 2
+#define BZ_N_QSORT 12
+#define BZ_N_SHELL 18
+#define BZ_N_OVERSHOOT (BZ_N_RADIX + BZ_N_QSORT + BZ_N_SHELL + 2)
+
+
+
+
+/*-- Structure holding all the compression-side stuff. --*/
+
+typedef
+ struct {
+ /* pointer back to the struct bz_stream */
+ bz_stream* strm;
+
+ /* mode this stream is in, and whether inputting */
+ /* or outputting data */
+ Int32 mode;
+ Int32 state;
+
+ /* remembers avail_in when flush/finish requested */
+ UInt32 avail_in_expect;
+
+ /* for doing the block sorting */
+ UInt32* arr1;
+ UInt32* arr2;
+ UInt32* ftab;
+ Int32 origPtr;
+
+ /* aliases for arr1 and arr2 */
+ UInt32* ptr;
+ UChar* block;
+ UInt16* mtfv;
+ UChar* zbits;
+
+ /* for deciding when to use the fallback sorting algorithm */
+ Int32 workFactor;
+
+ /* run-length-encoding of the input */
+ UInt32 state_in_ch;
+ Int32 state_in_len;
+ BZ_RAND_DECLS;
+
+ /* input and output limits and current posns */
+ Int32 nblock;
+ Int32 nblockMAX;
+ Int32 numZ;
+ Int32 state_out_pos;
+
+ /* map of bytes used in block */
+ Int32 nInUse;
+ Bool inUse[256];
+ UChar unseqToSeq[256];
+
+ /* the buffer for bit stream creation */
+ UInt32 bsBuff;
+ Int32 bsLive;
+
+ /* block and combined CRCs */
+ UInt32 blockCRC;
+ UInt32 combinedCRC;
+
+ /* misc administratium */
+ Int32 verbosity;
+ Int32 blockNo;
+ Int32 blockSize100k;
+
+ /* stuff for coding the MTF values */
+ Int32 nMTF;
+ Int32 mtfFreq [BZ_MAX_ALPHA_SIZE];
+ UChar selector [BZ_MAX_SELECTORS];
+ UChar selectorMtf[BZ_MAX_SELECTORS];
+
+ UChar len [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
+ Int32 code [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
+ Int32 rfreq [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
+ /* second dimension: only 3 needed; 4 makes index calculations faster */
+ UInt32 len_pack[BZ_MAX_ALPHA_SIZE][4];
+
+ }
+ EState;
+
+
+
+/*-- externs for compression. --*/
+
+extern void
+BZ2_blockSort ( EState* );
+
+extern void
+BZ2_compressBlock ( EState*, Bool );
+
+extern void
+BZ2_bsInitWrite ( EState* );
+
+extern void
+BZ2_hbAssignCodes ( Int32*, UChar*, Int32, Int32, Int32 );
+
+extern void
+BZ2_hbMakeCodeLengths ( UChar*, Int32*, Int32, Int32 );
+
+
+
+/*-- states for decompression. --*/
+
+#define BZ_X_IDLE 1
+#define BZ_X_OUTPUT 2
+
+#define BZ_X_MAGIC_1 10
+#define BZ_X_MAGIC_2 11
+#define BZ_X_MAGIC_3 12
+#define BZ_X_MAGIC_4 13
+#define BZ_X_BLKHDR_1 14
+#define BZ_X_BLKHDR_2 15
+#define BZ_X_BLKHDR_3 16
+#define BZ_X_BLKHDR_4 17
+#define BZ_X_BLKHDR_5 18
+#define BZ_X_BLKHDR_6 19
+#define BZ_X_BCRC_1 20
+#define BZ_X_BCRC_2 21
+#define BZ_X_BCRC_3 22
+#define BZ_X_BCRC_4 23
+#define BZ_X_RANDBIT 24
+#define BZ_X_ORIGPTR_1 25
+#define BZ_X_ORIGPTR_2 26
+#define BZ_X_ORIGPTR_3 27
+#define BZ_X_MAPPING_1 28
+#define BZ_X_MAPPING_2 29
+#define BZ_X_SELECTOR_1 30
+#define BZ_X_SELECTOR_2 31
+#define BZ_X_SELECTOR_3 32
+#define BZ_X_CODING_1 33
+#define BZ_X_CODING_2 34
+#define BZ_X_CODING_3 35
+#define BZ_X_MTF_1 36
+#define BZ_X_MTF_2 37
+#define BZ_X_MTF_3 38
+#define BZ_X_MTF_4 39
+#define BZ_X_MTF_5 40
+#define BZ_X_MTF_6 41
+#define BZ_X_ENDHDR_2 42
+#define BZ_X_ENDHDR_3 43
+#define BZ_X_ENDHDR_4 44
+#define BZ_X_ENDHDR_5 45
+#define BZ_X_ENDHDR_6 46
+#define BZ_X_CCRC_1 47
+#define BZ_X_CCRC_2 48
+#define BZ_X_CCRC_3 49
+#define BZ_X_CCRC_4 50
+
+
+
+/*-- Constants for the fast MTF decoder. --*/
+
+#define MTFA_SIZE 4096
+#define MTFL_SIZE 16
+
+
+
+/*-- Structure holding all the decompression-side stuff. --*/
+
+typedef
+ struct {
+ /* pointer back to the struct bz_stream */
+ bz_stream* strm;
+
+ /* state indicator for this stream */
+ Int32 state;
+
+ /* for doing the final run-length decoding */
+ UChar state_out_ch;
+ Int32 state_out_len;
+ Bool blockRandomised;
+ BZ_RAND_DECLS;
+
+ /* the buffer for bit stream reading */
+ UInt32 bsBuff;
+ Int32 bsLive;
+
+ /* misc administratium */
+ Int32 blockSize100k;
+ Bool smallDecompress;
+ Int32 currBlockNo;
+ Int32 verbosity;
+
+ /* for undoing the Burrows-Wheeler transform */
+ Int32 origPtr;
+ UInt32 tPos;
+ Int32 k0;
+ Int32 unzftab[256];
+ Int32 nblock_used;
+ Int32 cftab[257];
+ Int32 cftabCopy[257];
+
+ /* for undoing the Burrows-Wheeler transform (FAST) */
+ UInt32 *tt;
+
+ /* for undoing the Burrows-Wheeler transform (SMALL) */
+ UInt16 *ll16;
+ UChar *ll4;
+
+ /* stored and calculated CRCs */
+ UInt32 storedBlockCRC;
+ UInt32 storedCombinedCRC;
+ UInt32 calculatedBlockCRC;
+ UInt32 calculatedCombinedCRC;
+
+ /* map of bytes used in block */
+ Int32 nInUse;
+ Bool inUse[256];
+ Bool inUse16[16];
+ UChar seqToUnseq[256];
+
+ /* for decoding the MTF values */
+ UChar mtfa [MTFA_SIZE];
+ Int32 mtfbase[256 / MTFL_SIZE];
+ UChar selector [BZ_MAX_SELECTORS];
+ UChar selectorMtf[BZ_MAX_SELECTORS];
+ UChar len [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
+
+ Int32 limit [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
+ Int32 base [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
+ Int32 perm [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
+ Int32 minLens[BZ_N_GROUPS];
+
+ /* save area for scalars in the main decompress code */
+ Int32 save_i;
+ Int32 save_j;
+ Int32 save_t;
+ Int32 save_alphaSize;
+ Int32 save_nGroups;
+ Int32 save_nSelectors;
+ Int32 save_EOB;
+ Int32 save_groupNo;
+ Int32 save_groupPos;
+ Int32 save_nextSym;
+ Int32 save_nblockMAX;
+ Int32 save_nblock;
+ Int32 save_es;
+ Int32 save_N;
+ Int32 save_curr;
+ Int32 save_zt;
+ Int32 save_zn;
+ Int32 save_zvec;
+ Int32 save_zj;
+ Int32 save_gSel;
+ Int32 save_gMinlen;
+ Int32* save_gLimit;
+ Int32* save_gBase;
+ Int32* save_gPerm;
+
+ }
+ DState;
+
+
+
+/*-- Macros for decompression. --*/
+
+#define BZ_GET_FAST(cccc) \
+ /* c_tPos is unsigned, hence test < 0 is pointless. */ \
+ if (s->tPos >= (UInt32)100000 * (UInt32)s->blockSize100k) return True; \
+ s->tPos = s->tt[s->tPos]; \
+ cccc = (UChar)(s->tPos & 0xff); \
+ s->tPos >>= 8;
+
+#define BZ_GET_FAST_C(cccc) \
+ /* c_tPos is unsigned, hence test < 0 is pointless. */ \
+ if (c_tPos >= (UInt32)100000 * (UInt32)ro_blockSize100k) return True; \
+ c_tPos = c_tt[c_tPos]; \
+ cccc = (UChar)(c_tPos & 0xff); \
+ c_tPos >>= 8;
+
+#define SET_LL4(i,n) \
+ { if (((i) & 0x1) == 0) \
+ s->ll4[(i) >> 1] = (s->ll4[(i) >> 1] & 0xf0) | (n); else \
+ s->ll4[(i) >> 1] = (s->ll4[(i) >> 1] & 0x0f) | ((n) << 4); \
+ }
+
+#define GET_LL4(i) \
+ ((((UInt32)(s->ll4[(i) >> 1])) >> (((i) << 2) & 0x4)) & 0xF)
+
+#define SET_LL(i,n) \
+ { s->ll16[i] = (UInt16)(n & 0x0000ffff); \
+ SET_LL4(i, n >> 16); \
+ }
+
+#define GET_LL(i) \
+ (((UInt32)s->ll16[i]) | (GET_LL4(i) << 16))
+
+#define BZ_GET_SMALL(cccc) \
+ /* c_tPos is unsigned, hence test < 0 is pointless. */ \
+ if (s->tPos >= (UInt32)100000 * (UInt32)s->blockSize100k) return True; \
+ cccc = BZ2_indexIntoF ( s->tPos, s->cftab ); \
+ s->tPos = GET_LL(s->tPos);
+
+
+/*-- externs for decompression. --*/
+
+extern Int32
+BZ2_indexIntoF ( Int32, Int32* );
+
+extern void
+BZ2_hbCreateDecodeTables ( Int32*, Int32*, Int32*, UChar*,
+ Int32, Int32, Int32 );
+
+
+#endif
+
+
+/*-- BZ_NO_STDIO seems to make NULL disappear on some platforms. --*/
+
+#ifdef BZ_NO_STDIO
+#ifndef NULL
+#define NULL 0
+#endif
+#endif
+
+
+/*-------------------------------------------------------------*/
+/*--- end bzlib_private.h ---*/
+/*-------------------------------------------------------------*/
--- /dev/null
+
+/*-------------------------------------------------------------*/
+/*--- Compression machinery (not incl block sorting) ---*/
+/*--- compress.c ---*/
+/*-------------------------------------------------------------*/
+
+/* ------------------------------------------------------------------
+ This file is part of bzip2/libbzip2, a program and library for
+ lossless, block-sorting data compression.
+
+ bzip2/libbzip2 version 1.0.5 of 10 December 2007
+ Copyright (C) 1996-2007 Julian Seward <jseward@bzip.org>
+
+ Please read the WARNING, DISCLAIMER and PATENTS sections in the
+ README file.
+
+ This program is released under the terms of the license contained
+ in the file LICENSE.
+ ------------------------------------------------------------------ */
+
+
+/* CHANGES
+ 0.9.0 -- original version.
+ 0.9.0a/b -- no changes in this file.
+ 0.9.0c -- changed setting of nGroups in sendMTFValues()
+ so as to do a bit better on small files
+*/
+
+#include "bzlib_private.h"
+
+
+/*---------------------------------------------------*/
+/*--- Bit stream I/O ---*/
+/*---------------------------------------------------*/
+
+/*---------------------------------------------------*/
+void BZ2_bsInitWrite ( EState* s )
+{
+ s->bsLive = 0;
+ s->bsBuff = 0;
+}
+
+
+/*---------------------------------------------------*/
+static
+void bsFinishWrite ( EState* s )
+{
+ while (s->bsLive > 0) {
+ s->zbits[s->numZ] = (UChar)(s->bsBuff >> 24);
+ s->numZ++;
+ s->bsBuff <<= 8;
+ s->bsLive -= 8;
+ }
+}
+
+
+/*---------------------------------------------------*/
+#define bsNEEDW(nz) \
+{ \
+ while (s->bsLive >= 8) { \
+ s->zbits[s->numZ] \
+ = (UChar)(s->bsBuff >> 24); \
+ s->numZ++; \
+ s->bsBuff <<= 8; \
+ s->bsLive -= 8; \
+ } \
+}
+
+
+/*---------------------------------------------------*/
+static
+__inline__
+void bsW ( EState* s, Int32 n, UInt32 v )
+{
+ bsNEEDW ( n );
+ s->bsBuff |= (v << (32 - s->bsLive - n));
+ s->bsLive += n;
+}
+
+
+/*---------------------------------------------------*/
+static
+void bsPutUInt32 ( EState* s, UInt32 u )
+{
+ bsW ( s, 8, (u >> 24) & 0xffL );
+ bsW ( s, 8, (u >> 16) & 0xffL );
+ bsW ( s, 8, (u >> 8) & 0xffL );
+ bsW ( s, 8, u & 0xffL );
+}
+
+
+/*---------------------------------------------------*/
+static
+void bsPutUChar ( EState* s, UChar c )
+{
+ bsW( s, 8, (UInt32)c );
+}
+
+
+/*---------------------------------------------------*/
+/*--- The back end proper ---*/
+/*---------------------------------------------------*/
+
+/*---------------------------------------------------*/
+static
+void makeMaps_e ( EState* s )
+{
+ Int32 i;
+ s->nInUse = 0;
+ for (i = 0; i < 256; i++)
+ if (s->inUse[i]) {
+ s->unseqToSeq[i] = s->nInUse;
+ s->nInUse++;
+ }
+}
+
+
+/*---------------------------------------------------*/
+static
+void generateMTFValues ( EState* s )
+{
+ UChar yy[256];
+ Int32 i, j;
+ Int32 zPend;
+ Int32 wr;
+ Int32 EOB;
+
+ /*
+ After sorting (eg, here),
+ s->arr1 [ 0 .. s->nblock-1 ] holds sorted order,
+ and
+ ((UChar*)s->arr2) [ 0 .. s->nblock-1 ]
+ holds the original block data.
+
+ The first thing to do is generate the MTF values,
+ and put them in
+ ((UInt16*)s->arr1) [ 0 .. s->nblock-1 ].
+ Because there are strictly fewer or equal MTF values
+ than block values, ptr values in this area are overwritten
+ with MTF values only when they are no longer needed.
+
+ The final compressed bitstream is generated into the
+ area starting at
+ (UChar*) (&((UChar*)s->arr2)[s->nblock])
+
+ These storage aliases are set up in bzCompressInit(),
+ except for the last one, which is arranged in
+ compressBlock().
+ */
+ UInt32* ptr = s->ptr;
+ UChar* block = s->block;
+ UInt16* mtfv = s->mtfv;
+
+ makeMaps_e ( s );
+ EOB = s->nInUse+1;
+
+ for (i = 0; i <= EOB; i++) s->mtfFreq[i] = 0;
+
+ wr = 0;
+ zPend = 0;
+ for (i = 0; i < s->nInUse; i++) yy[i] = (UChar) i;
+
+ for (i = 0; i < s->nblock; i++) {
+ UChar ll_i;
+ AssertD ( wr <= i, "generateMTFValues(1)" );
+ j = ptr[i]-1; if (j < 0) j += s->nblock;
+ ll_i = s->unseqToSeq[block[j]];
+ AssertD ( ll_i < s->nInUse, "generateMTFValues(2a)" );
+
+ if (yy[0] == ll_i) {
+ zPend++;
+ } else {
+
+ if (zPend > 0) {
+ zPend--;
+ while (True) {
+ if (zPend & 1) {
+ mtfv[wr] = BZ_RUNB; wr++;
+ s->mtfFreq[BZ_RUNB]++;
+ } else {
+ mtfv[wr] = BZ_RUNA; wr++;
+ s->mtfFreq[BZ_RUNA]++;
+ }
+ if (zPend < 2) break;
+ zPend = (zPend - 2) / 2;
+ };
+ zPend = 0;
+ }
+ {
+ register UChar rtmp;
+ register UChar* ryy_j;
+ register UChar rll_i;
+ rtmp = yy[1];
+ yy[1] = yy[0];
+ ryy_j = &(yy[1]);
+ rll_i = ll_i;
+ while ( rll_i != rtmp ) {
+ register UChar rtmp2;
+ ryy_j++;
+ rtmp2 = rtmp;
+ rtmp = *ryy_j;
+ *ryy_j = rtmp2;
+ };
+ yy[0] = rtmp;
+ j = ryy_j - &(yy[0]);
+ mtfv[wr] = j+1; wr++; s->mtfFreq[j+1]++;
+ }
+
+ }
+ }
+
+ if (zPend > 0) {
+ zPend--;
+ while (True) {
+ if (zPend & 1) {
+ mtfv[wr] = BZ_RUNB; wr++;
+ s->mtfFreq[BZ_RUNB]++;
+ } else {
+ mtfv[wr] = BZ_RUNA; wr++;
+ s->mtfFreq[BZ_RUNA]++;
+ }
+ if (zPend < 2) break;
+ zPend = (zPend - 2) / 2;
+ };
+ zPend = 0;
+ }
+
+ mtfv[wr] = EOB; wr++; s->mtfFreq[EOB]++;
+
+ s->nMTF = wr;
+}
+
+
+/*---------------------------------------------------*/
+#define BZ_LESSER_ICOST 0
+#define BZ_GREATER_ICOST 15
+
+static
+void sendMTFValues ( EState* s )
+{
+ Int32 v, t, i, j, gs, ge, totc, bt, bc, iter;
+ Int32 nSelectors, alphaSize, minLen, maxLen, selCtr;
+ Int32 nGroups, nBytes;
+
+ /*--
+ UChar len [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
+ is a global since the decoder also needs it.
+
+ Int32 code[BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
+ Int32 rfreq[BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
+ are also globals only used in this proc.
+ Made global to keep stack frame size small.
+ --*/
+
+
+ UInt16 cost[BZ_N_GROUPS];
+ Int32 fave[BZ_N_GROUPS];
+
+ UInt16* mtfv = s->mtfv;
+
+ if (s->verbosity >= 3)
+ VPrintf3( " %d in block, %d after MTF & 1-2 coding, "
+ "%d+2 syms in use\n",
+ s->nblock, s->nMTF, s->nInUse );
+
+ alphaSize = s->nInUse+2;
+ for (t = 0; t < BZ_N_GROUPS; t++)
+ for (v = 0; v < alphaSize; v++)
+ s->len[t][v] = BZ_GREATER_ICOST;
+
+ /*--- Decide how many coding tables to use ---*/
+ AssertH ( s->nMTF > 0, 3001 );
+ if (s->nMTF < 200) nGroups = 2; else
+ if (s->nMTF < 600) nGroups = 3; else
+ if (s->nMTF < 1200) nGroups = 4; else
+ if (s->nMTF < 2400) nGroups = 5; else
+ nGroups = 6;
+
+ /*--- Generate an initial set of coding tables ---*/
+ {
+ Int32 nPart, remF, tFreq, aFreq;
+
+ nPart = nGroups;
+ remF = s->nMTF;
+ gs = 0;
+ while (nPart > 0) {
+ tFreq = remF / nPart;
+ ge = gs-1;
+ aFreq = 0;
+ while (aFreq < tFreq && ge < alphaSize-1) {
+ ge++;
+ aFreq += s->mtfFreq[ge];
+ }
+
+ if (ge > gs
+ && nPart != nGroups && nPart != 1
+ && ((nGroups-nPart) % 2 == 1)) {
+ aFreq -= s->mtfFreq[ge];
+ ge--;
+ }
+
+ if (s->verbosity >= 3)
+ VPrintf5( " initial group %d, [%d .. %d], "
+ "has %d syms (%4.1f%%)\n",
+ nPart, gs, ge, aFreq,
+ (100.0 * (float)aFreq) / (float)(s->nMTF) );
+
+ for (v = 0; v < alphaSize; v++)
+ if (v >= gs && v <= ge)
+ s->len[nPart-1][v] = BZ_LESSER_ICOST; else
+ s->len[nPart-1][v] = BZ_GREATER_ICOST;
+
+ nPart--;
+ gs = ge+1;
+ remF -= aFreq;
+ }
+ }
+
+ /*---
+ Iterate up to BZ_N_ITERS times to improve the tables.
+ ---*/
+ for (iter = 0; iter < BZ_N_ITERS; iter++) {
+
+ for (t = 0; t < nGroups; t++) fave[t] = 0;
+
+ for (t = 0; t < nGroups; t++)
+ for (v = 0; v < alphaSize; v++)
+ s->rfreq[t][v] = 0;
+
+ /*---
+ Set up an auxiliary length table which is used to fast-track
+ the common case (nGroups == 6).
+ ---*/
+ if (nGroups == 6) {
+ for (v = 0; v < alphaSize; v++) {
+ s->len_pack[v][0] = (s->len[1][v] << 16) | s->len[0][v];
+ s->len_pack[v][1] = (s->len[3][v] << 16) | s->len[2][v];
+ s->len_pack[v][2] = (s->len[5][v] << 16) | s->len[4][v];
+ }
+ }
+
+ nSelectors = 0;
+ totc = 0;
+ gs = 0;
+ while (True) {
+
+ /*--- Set group start & end marks. --*/
+ if (gs >= s->nMTF) break;
+ ge = gs + BZ_G_SIZE - 1;
+ if (ge >= s->nMTF) ge = s->nMTF-1;
+
+ /*--
+ Calculate the cost of this group as coded
+ by each of the coding tables.
+ --*/
+ for (t = 0; t < nGroups; t++) cost[t] = 0;
+
+ if (nGroups == 6 && 50 == ge-gs+1) {
+ /*--- fast track the common case ---*/
+ register UInt32 cost01, cost23, cost45;
+ register UInt16 icv;
+ cost01 = cost23 = cost45 = 0;
+
+# define BZ_ITER(nn) \
+ icv = mtfv[gs+(nn)]; \
+ cost01 += s->len_pack[icv][0]; \
+ cost23 += s->len_pack[icv][1]; \
+ cost45 += s->len_pack[icv][2]; \
+
+ BZ_ITER(0); BZ_ITER(1); BZ_ITER(2); BZ_ITER(3); BZ_ITER(4);
+ BZ_ITER(5); BZ_ITER(6); BZ_ITER(7); BZ_ITER(8); BZ_ITER(9);
+ BZ_ITER(10); BZ_ITER(11); BZ_ITER(12); BZ_ITER(13); BZ_ITER(14);
+ BZ_ITER(15); BZ_ITER(16); BZ_ITER(17); BZ_ITER(18); BZ_ITER(19);
+ BZ_ITER(20); BZ_ITER(21); BZ_ITER(22); BZ_ITER(23); BZ_ITER(24);
+ BZ_ITER(25); BZ_ITER(26); BZ_ITER(27); BZ_ITER(28); BZ_ITER(29);
+ BZ_ITER(30); BZ_ITER(31); BZ_ITER(32); BZ_ITER(33); BZ_ITER(34);
+ BZ_ITER(35); BZ_ITER(36); BZ_ITER(37); BZ_ITER(38); BZ_ITER(39);
+ BZ_ITER(40); BZ_ITER(41); BZ_ITER(42); BZ_ITER(43); BZ_ITER(44);
+ BZ_ITER(45); BZ_ITER(46); BZ_ITER(47); BZ_ITER(48); BZ_ITER(49);
+
+# undef BZ_ITER
+
+ cost[0] = cost01 & 0xffff; cost[1] = cost01 >> 16;
+ cost[2] = cost23 & 0xffff; cost[3] = cost23 >> 16;
+ cost[4] = cost45 & 0xffff; cost[5] = cost45 >> 16;
+
+ } else {
+ /*--- slow version which correctly handles all situations ---*/
+ for (i = gs; i <= ge; i++) {
+ UInt16 icv = mtfv[i];
+ for (t = 0; t < nGroups; t++) cost[t] += s->len[t][icv];
+ }
+ }
+
+ /*--
+ Find the coding table which is best for this group,
+ and record its identity in the selector table.
+ --*/
+ bc = 999999999; bt = -1;
+ for (t = 0; t < nGroups; t++)
+ if (cost[t] < bc) { bc = cost[t]; bt = t; };
+ totc += bc;
+ fave[bt]++;
+ s->selector[nSelectors] = bt;
+ nSelectors++;
+
+ /*--
+ Increment the symbol frequencies for the selected table.
+ --*/
+ if (nGroups == 6 && 50 == ge-gs+1) {
+ /*--- fast track the common case ---*/
+
+# define BZ_ITUR(nn) s->rfreq[bt][ mtfv[gs+(nn)] ]++
+
+ BZ_ITUR(0); BZ_ITUR(1); BZ_ITUR(2); BZ_ITUR(3); BZ_ITUR(4);
+ BZ_ITUR(5); BZ_ITUR(6); BZ_ITUR(7); BZ_ITUR(8); BZ_ITUR(9);
+ BZ_ITUR(10); BZ_ITUR(11); BZ_ITUR(12); BZ_ITUR(13); BZ_ITUR(14);
+ BZ_ITUR(15); BZ_ITUR(16); BZ_ITUR(17); BZ_ITUR(18); BZ_ITUR(19);
+ BZ_ITUR(20); BZ_ITUR(21); BZ_ITUR(22); BZ_ITUR(23); BZ_ITUR(24);
+ BZ_ITUR(25); BZ_ITUR(26); BZ_ITUR(27); BZ_ITUR(28); BZ_ITUR(29);
+ BZ_ITUR(30); BZ_ITUR(31); BZ_ITUR(32); BZ_ITUR(33); BZ_ITUR(34);
+ BZ_ITUR(35); BZ_ITUR(36); BZ_ITUR(37); BZ_ITUR(38); BZ_ITUR(39);
+ BZ_ITUR(40); BZ_ITUR(41); BZ_ITUR(42); BZ_ITUR(43); BZ_ITUR(44);
+ BZ_ITUR(45); BZ_ITUR(46); BZ_ITUR(47); BZ_ITUR(48); BZ_ITUR(49);
+
+# undef BZ_ITUR
+
+ } else {
+ /*--- slow version which correctly handles all situations ---*/
+ for (i = gs; i <= ge; i++)
+ s->rfreq[bt][ mtfv[i] ]++;
+ }
+
+ gs = ge+1;
+ }
+ if (s->verbosity >= 3) {
+ VPrintf2 ( " pass %d: size is %d, grp uses are ",
+ iter+1, totc/8 );
+ for (t = 0; t < nGroups; t++)
+ VPrintf1 ( "%d ", fave[t] );
+ VPrintf0 ( "\n" );
+ }
+
+ /*--
+ Recompute the tables based on the accumulated frequencies.
+ --*/
+ /* maxLen was changed from 20 to 17 in bzip2-1.0.3. See
+ comment in huffman.c for details. */
+ for (t = 0; t < nGroups; t++)
+ BZ2_hbMakeCodeLengths ( &(s->len[t][0]), &(s->rfreq[t][0]),
+ alphaSize, 17 /*20*/ );
+ }
+
+
+ AssertH( nGroups < 8, 3002 );
+ AssertH( nSelectors < 32768 &&
+ nSelectors <= (2 + (900000 / BZ_G_SIZE)),
+ 3003 );
+
+
+ /*--- Compute MTF values for the selectors. ---*/
+ {
+ UChar pos[BZ_N_GROUPS], ll_i, tmp2, tmp;
+ for (i = 0; i < nGroups; i++) pos[i] = i;
+ for (i = 0; i < nSelectors; i++) {
+ ll_i = s->selector[i];
+ j = 0;
+ tmp = pos[j];
+ while ( ll_i != tmp ) {
+ j++;
+ tmp2 = tmp;
+ tmp = pos[j];
+ pos[j] = tmp2;
+ };
+ pos[0] = tmp;
+ s->selectorMtf[i] = j;
+ }
+ };
+
+ /*--- Assign actual codes for the tables. --*/
+ for (t = 0; t < nGroups; t++) {
+ minLen = 32;
+ maxLen = 0;
+ for (i = 0; i < alphaSize; i++) {
+ if (s->len[t][i] > maxLen) maxLen = s->len[t][i];
+ if (s->len[t][i] < minLen) minLen = s->len[t][i];
+ }
+ AssertH ( !(maxLen > 17 /*20*/ ), 3004 );
+ AssertH ( !(minLen < 1), 3005 );
+ BZ2_hbAssignCodes ( &(s->code[t][0]), &(s->len[t][0]),
+ minLen, maxLen, alphaSize );
+ }
+
+ /*--- Transmit the mapping table. ---*/
+ {
+ Bool inUse16[16];
+ for (i = 0; i < 16; i++) {
+ inUse16[i] = False;
+ for (j = 0; j < 16; j++)
+ if (s->inUse[i * 16 + j]) inUse16[i] = True;
+ }
+
+ nBytes = s->numZ;
+ for (i = 0; i < 16; i++)
+ if (inUse16[i]) bsW(s,1,1); else bsW(s,1,0);
+
+ for (i = 0; i < 16; i++)
+ if (inUse16[i])
+ for (j = 0; j < 16; j++) {
+ if (s->inUse[i * 16 + j]) bsW(s,1,1); else bsW(s,1,0);
+ }
+
+ if (s->verbosity >= 3)
+ VPrintf1( " bytes: mapping %d, ", s->numZ-nBytes );
+ }
+
+ /*--- Now the selectors. ---*/
+ nBytes = s->numZ;
+ bsW ( s, 3, nGroups );
+ bsW ( s, 15, nSelectors );
+ for (i = 0; i < nSelectors; i++) {
+ for (j = 0; j < s->selectorMtf[i]; j++) bsW(s,1,1);
+ bsW(s,1,0);
+ }
+ if (s->verbosity >= 3)
+ VPrintf1( "selectors %d, ", s->numZ-nBytes );
+
+ /*--- Now the coding tables. ---*/
+ nBytes = s->numZ;
+
+ for (t = 0; t < nGroups; t++) {
+ Int32 curr = s->len[t][0];
+ bsW ( s, 5, curr );
+ for (i = 0; i < alphaSize; i++) {
+ while (curr < s->len[t][i]) { bsW(s,2,2); curr++; /* 10 */ };
+ while (curr > s->len[t][i]) { bsW(s,2,3); curr--; /* 11 */ };
+ bsW ( s, 1, 0 );
+ }
+ }
+
+ if (s->verbosity >= 3)
+ VPrintf1 ( "code lengths %d, ", s->numZ-nBytes );
+
+ /*--- And finally, the block data proper ---*/
+ nBytes = s->numZ;
+ selCtr = 0;
+ gs = 0;
+ while (True) {
+ if (gs >= s->nMTF) break;
+ ge = gs + BZ_G_SIZE - 1;
+ if (ge >= s->nMTF) ge = s->nMTF-1;
+ AssertH ( s->selector[selCtr] < nGroups, 3006 );
+
+ if (nGroups == 6 && 50 == ge-gs+1) {
+ /*--- fast track the common case ---*/
+ UInt16 mtfv_i;
+ UChar* s_len_sel_selCtr
+ = &(s->len[s->selector[selCtr]][0]);
+ Int32* s_code_sel_selCtr
+ = &(s->code[s->selector[selCtr]][0]);
+
+# define BZ_ITAH(nn) \
+ mtfv_i = mtfv[gs+(nn)]; \
+ bsW ( s, \
+ s_len_sel_selCtr[mtfv_i], \
+ s_code_sel_selCtr[mtfv_i] )
+
+ BZ_ITAH(0); BZ_ITAH(1); BZ_ITAH(2); BZ_ITAH(3); BZ_ITAH(4);
+ BZ_ITAH(5); BZ_ITAH(6); BZ_ITAH(7); BZ_ITAH(8); BZ_ITAH(9);
+ BZ_ITAH(10); BZ_ITAH(11); BZ_ITAH(12); BZ_ITAH(13); BZ_ITAH(14);
+ BZ_ITAH(15); BZ_ITAH(16); BZ_ITAH(17); BZ_ITAH(18); BZ_ITAH(19);
+ BZ_ITAH(20); BZ_ITAH(21); BZ_ITAH(22); BZ_ITAH(23); BZ_ITAH(24);
+ BZ_ITAH(25); BZ_ITAH(26); BZ_ITAH(27); BZ_ITAH(28); BZ_ITAH(29);
+ BZ_ITAH(30); BZ_ITAH(31); BZ_ITAH(32); BZ_ITAH(33); BZ_ITAH(34);
+ BZ_ITAH(35); BZ_ITAH(36); BZ_ITAH(37); BZ_ITAH(38); BZ_ITAH(39);
+ BZ_ITAH(40); BZ_ITAH(41); BZ_ITAH(42); BZ_ITAH(43); BZ_ITAH(44);
+ BZ_ITAH(45); BZ_ITAH(46); BZ_ITAH(47); BZ_ITAH(48); BZ_ITAH(49);
+
+# undef BZ_ITAH
+
+ } else {
+ /*--- slow version which correctly handles all situations ---*/
+ for (i = gs; i <= ge; i++) {
+ bsW ( s,
+ s->len [s->selector[selCtr]] [mtfv[i]],
+ s->code [s->selector[selCtr]] [mtfv[i]] );
+ }
+ }
+
+
+ gs = ge+1;
+ selCtr++;
+ }
+ AssertH( selCtr == nSelectors, 3007 );
+
+ if (s->verbosity >= 3)
+ VPrintf1( "codes %d\n", s->numZ-nBytes );
+}
+
+
+/*---------------------------------------------------*/
+void BZ2_compressBlock ( EState* s, Bool is_last_block )
+{
+ if (s->nblock > 0) {
+
+ BZ_FINALISE_CRC ( s->blockCRC );
+ s->combinedCRC = (s->combinedCRC << 1) | (s->combinedCRC >> 31);
+ s->combinedCRC ^= s->blockCRC;
+ if (s->blockNo > 1) s->numZ = 0;
+
+ if (s->verbosity >= 2)
+ VPrintf4( " block %d: crc = 0x%08x, "
+ "combined CRC = 0x%08x, size = %d\n",
+ s->blockNo, s->blockCRC, s->combinedCRC, s->nblock );
+
+ BZ2_blockSort ( s );
+ }
+
+ s->zbits = (UChar*) (&((UChar*)s->arr2)[s->nblock]);
+
+ /*-- If this is the first block, create the stream header. --*/
+ if (s->blockNo == 1) {
+ BZ2_bsInitWrite ( s );
+ bsPutUChar ( s, BZ_HDR_B );
+ bsPutUChar ( s, BZ_HDR_Z );
+ bsPutUChar ( s, BZ_HDR_h );
+ bsPutUChar ( s, (UChar)(BZ_HDR_0 + s->blockSize100k) );
+ }
+
+ if (s->nblock > 0) {
+
+ bsPutUChar ( s, 0x31 ); bsPutUChar ( s, 0x41 );
+ bsPutUChar ( s, 0x59 ); bsPutUChar ( s, 0x26 );
+ bsPutUChar ( s, 0x53 ); bsPutUChar ( s, 0x59 );
+
+ /*-- Now the block's CRC, so it is in a known place. --*/
+ bsPutUInt32 ( s, s->blockCRC );
+
+ /*--
+ Now a single bit indicating (non-)randomisation.
+ As of version 0.9.5, we use a better sorting algorithm
+ which makes randomisation unnecessary. So always set
+ the randomised bit to 'no'. Of course, the decoder
+ still needs to be able to handle randomised blocks
+ so as to maintain backwards compatibility with
+ older versions of bzip2.
+ --*/
+ bsW(s,1,0);
+
+ bsW ( s, 24, s->origPtr );
+ generateMTFValues ( s );
+ sendMTFValues ( s );
+ }
+
+
+ /*-- If this is the last block, add the stream trailer. --*/
+ if (is_last_block) {
+
+ bsPutUChar ( s, 0x17 ); bsPutUChar ( s, 0x72 );
+ bsPutUChar ( s, 0x45 ); bsPutUChar ( s, 0x38 );
+ bsPutUChar ( s, 0x50 ); bsPutUChar ( s, 0x90 );
+ bsPutUInt32 ( s, s->combinedCRC );
+ if (s->verbosity >= 2)
+ VPrintf1( " final combined CRC = 0x%08x\n ", s->combinedCRC );
+ bsFinishWrite ( s );
+ }
+}
+
+
+/*-------------------------------------------------------------*/
+/*--- end compress.c ---*/
+/*-------------------------------------------------------------*/
--- /dev/null
+
+/*-------------------------------------------------------------*/
+/*--- Huffman coding low-level stuff ---*/
+/*--- huffman.c ---*/
+/*-------------------------------------------------------------*/
+
+/* ------------------------------------------------------------------
+ This file is part of bzip2/libbzip2, a program and library for
+ lossless, block-sorting data compression.
+
+ bzip2/libbzip2 version 1.0.5 of 10 December 2007
+ Copyright (C) 1996-2007 Julian Seward <jseward@bzip.org>
+
+ Please read the WARNING, DISCLAIMER and PATENTS sections in the
+ README file.
+
+ This program is released under the terms of the license contained
+ in the file LICENSE.
+ ------------------------------------------------------------------ */
+
+
+#include "bzlib_private.h"
+
+/*---------------------------------------------------*/
+#define WEIGHTOF(zz0) ((zz0) & 0xffffff00)
+#define DEPTHOF(zz1) ((zz1) & 0x000000ff)
+#define MYMAX(zz2,zz3) ((zz2) > (zz3) ? (zz2) : (zz3))
+
+#define ADDWEIGHTS(zw1,zw2) \
+ (WEIGHTOF(zw1)+WEIGHTOF(zw2)) | \
+ (1 + MYMAX(DEPTHOF(zw1),DEPTHOF(zw2)))
+
+#define UPHEAP(z) \
+{ \
+ Int32 zz, tmp; \
+ zz = z; tmp = heap[zz]; \
+ while (weight[tmp] < weight[heap[zz >> 1]]) { \
+ heap[zz] = heap[zz >> 1]; \
+ zz >>= 1; \
+ } \
+ heap[zz] = tmp; \
+}
+
+#define DOWNHEAP(z) \
+{ \
+ Int32 zz, yy, tmp; \
+ zz = z; tmp = heap[zz]; \
+ while (True) { \
+ yy = zz << 1; \
+ if (yy > nHeap) break; \
+ if (yy < nHeap && \
+ weight[heap[yy+1]] < weight[heap[yy]]) \
+ yy++; \
+ if (weight[tmp] < weight[heap[yy]]) break; \
+ heap[zz] = heap[yy]; \
+ zz = yy; \
+ } \
+ heap[zz] = tmp; \
+}
+
+
+/*---------------------------------------------------*/
+void BZ2_hbMakeCodeLengths ( UChar *len,
+ Int32 *freq,
+ Int32 alphaSize,
+ Int32 maxLen )
+{
+ /*--
+ Nodes and heap entries run from 1. Entry 0
+ for both the heap and nodes is a sentinel.
+ --*/
+ Int32 nNodes, nHeap, n1, n2, i, j, k;
+ Bool tooLong;
+
+ Int32 heap [ BZ_MAX_ALPHA_SIZE + 2 ];
+ Int32 weight [ BZ_MAX_ALPHA_SIZE * 2 ];
+ Int32 parent [ BZ_MAX_ALPHA_SIZE * 2 ];
+
+ for (i = 0; i < alphaSize; i++)
+ weight[i+1] = (freq[i] == 0 ? 1 : freq[i]) << 8;
+
+ while (True) {
+
+ nNodes = alphaSize;
+ nHeap = 0;
+
+ heap[0] = 0;
+ weight[0] = 0;
+ parent[0] = -2;
+
+ for (i = 1; i <= alphaSize; i++) {
+ parent[i] = -1;
+ nHeap++;
+ heap[nHeap] = i;
+ UPHEAP(nHeap);
+ }
+
+ AssertH( nHeap < (BZ_MAX_ALPHA_SIZE+2), 2001 );
+
+ while (nHeap > 1) {
+ n1 = heap[1]; heap[1] = heap[nHeap]; nHeap--; DOWNHEAP(1);
+ n2 = heap[1]; heap[1] = heap[nHeap]; nHeap--; DOWNHEAP(1);
+ nNodes++;
+ parent[n1] = parent[n2] = nNodes;
+ weight[nNodes] = ADDWEIGHTS(weight[n1], weight[n2]);
+ parent[nNodes] = -1;
+ nHeap++;
+ heap[nHeap] = nNodes;
+ UPHEAP(nHeap);
+ }
+
+ AssertH( nNodes < (BZ_MAX_ALPHA_SIZE * 2), 2002 );
+
+ tooLong = False;
+ for (i = 1; i <= alphaSize; i++) {
+ j = 0;
+ k = i;
+ while (parent[k] >= 0) { k = parent[k]; j++; }
+ len[i-1] = j;
+ if (j > maxLen) tooLong = True;
+ }
+
+ if (! tooLong) break;
+
+ /* 17 Oct 04: keep-going condition for the following loop used
+ to be 'i < alphaSize', which missed the last element,
+ theoretically leading to the possibility of the compressor
+ looping. However, this count-scaling step is only needed if
+ one of the generated Huffman code words is longer than
+ maxLen, which up to and including version 1.0.2 was 20 bits,
+ which is extremely unlikely. In version 1.0.3 maxLen was
+ changed to 17 bits, which has minimal effect on compression
+ ratio, but does mean this scaling step is used from time to
+ time, enough to verify that it works.
+
+ This means that bzip2-1.0.3 and later will only produce
+ Huffman codes with a maximum length of 17 bits. However, in
+ order to preserve backwards compatibility with bitstreams
+ produced by versions pre-1.0.3, the decompressor must still
+ handle lengths of up to 20. */
+
+ for (i = 1; i <= alphaSize; i++) {
+ j = weight[i] >> 8;
+ j = 1 + (j / 2);
+ weight[i] = j << 8;
+ }
+ }
+}
+
+
+/*---------------------------------------------------*/
+void BZ2_hbAssignCodes ( Int32 *code,
+ UChar *length,
+ Int32 minLen,
+ Int32 maxLen,
+ Int32 alphaSize )
+{
+ Int32 n, vec, i;
+
+ vec = 0;
+ for (n = minLen; n <= maxLen; n++) {
+ for (i = 0; i < alphaSize; i++)
+ if (length[i] == n) { code[i] = vec; vec++; };
+ vec <<= 1;
+ }
+}
+
+
+/*---------------------------------------------------*/
+void BZ2_hbCreateDecodeTables ( Int32 *limit,
+ Int32 *base,
+ Int32 *perm,
+ UChar *length,
+ Int32 minLen,
+ Int32 maxLen,
+ Int32 alphaSize )
+{
+ Int32 pp, i, j, vec;
+
+ pp = 0;
+ for (i = minLen; i <= maxLen; i++)
+ for (j = 0; j < alphaSize; j++)
+ if (length[j] == i) { perm[pp] = j; pp++; };
+
+ for (i = 0; i < BZ_MAX_CODE_LEN; i++) base[i] = 0;
+ for (i = 0; i < alphaSize; i++) base[length[i]+1]++;
+
+ for (i = 1; i < BZ_MAX_CODE_LEN; i++) base[i] += base[i-1];
+
+ for (i = 0; i < BZ_MAX_CODE_LEN; i++) limit[i] = 0;
+ vec = 0;
+
+ for (i = minLen; i <= maxLen; i++) {
+ vec += (base[i+1] - base[i]);
+ limit[i] = vec-1;
+ vec <<= 1;
+ }
+ for (i = minLen + 1; i <= maxLen; i++)
+ base[i] = ((limit[i-1] + 1) << 1) - base[i];
+}
+
+
+/*-------------------------------------------------------------*/
+/*--- end huffman.c ---*/
+/*-------------------------------------------------------------*/
+++ /dev/null
-
-/*-------------------------------------------------------------*/
-/*--- Block sorting machinery ---*/
-/*--- blocksort.c ---*/
-/*-------------------------------------------------------------*/
-
-/* ------------------------------------------------------------------
- This file is part of bzip2/libbzip2, a program and library for
- lossless, block-sorting data compression.
-
- bzip2/libbzip2 version 1.0.5 of 10 December 2007
- Copyright (C) 1996-2007 Julian Seward <jseward@bzip.org>
-
- Please read the WARNING, DISCLAIMER and PATENTS sections in the
- README file.
-
- This program is released under the terms of the license contained
- in the file LICENSE.
- ------------------------------------------------------------------ */
-
-
-#include "bzlib_private.h"
-
-/*---------------------------------------------*/
-/*--- Fallback O(N log(N)^2) sorting ---*/
-/*--- algorithm, for repetitive blocks ---*/
-/*---------------------------------------------*/
-
-/*---------------------------------------------*/
-static
-__inline__
-void fallbackSimpleSort ( UInt32* fmap,
- UInt32* eclass,
- Int32 lo,
- Int32 hi )
-{
- Int32 i, j, tmp;
- UInt32 ec_tmp;
-
- if (lo == hi) return;
-
- if (hi - lo > 3) {
- for ( i = hi-4; i >= lo; i-- ) {
- tmp = fmap[i];
- ec_tmp = eclass[tmp];
- for ( j = i+4; j <= hi && ec_tmp > eclass[fmap[j]]; j += 4 )
- fmap[j-4] = fmap[j];
- fmap[j-4] = tmp;
- }
- }
-
- for ( i = hi-1; i >= lo; i-- ) {
- tmp = fmap[i];
- ec_tmp = eclass[tmp];
- for ( j = i+1; j <= hi && ec_tmp > eclass[fmap[j]]; j++ )
- fmap[j-1] = fmap[j];
- fmap[j-1] = tmp;
- }
-}
-
-
-/*---------------------------------------------*/
-#define fswap(zz1, zz2) \
- { Int32 zztmp = zz1; zz1 = zz2; zz2 = zztmp; }
-
-#define fvswap(zzp1, zzp2, zzn) \
-{ \
- Int32 yyp1 = (zzp1); \
- Int32 yyp2 = (zzp2); \
- Int32 yyn = (zzn); \
- while (yyn > 0) { \
- fswap(fmap[yyp1], fmap[yyp2]); \
- yyp1++; yyp2++; yyn--; \
- } \
-}
-
-
-#define fmin(a,b) ((a) < (b)) ? (a) : (b)
-
-#define fpush(lz,hz) { stackLo[sp] = lz; \
- stackHi[sp] = hz; \
- sp++; }
-
-#define fpop(lz,hz) { sp--; \
- lz = stackLo[sp]; \
- hz = stackHi[sp]; }
-
-#define FALLBACK_QSORT_SMALL_THRESH 10
-#define FALLBACK_QSORT_STACK_SIZE 100
-
-
-static
-void fallbackQSort3 ( UInt32* fmap,
- UInt32* eclass,
- Int32 loSt,
- Int32 hiSt )
-{
- Int32 unLo, unHi, ltLo, gtHi, n, m;
- Int32 sp, lo, hi;
- UInt32 med, r, r3;
- Int32 stackLo[FALLBACK_QSORT_STACK_SIZE];
- Int32 stackHi[FALLBACK_QSORT_STACK_SIZE];
-
- r = 0;
-
- sp = 0;
- fpush ( loSt, hiSt );
-
- while (sp > 0) {
-
- AssertH ( sp < FALLBACK_QSORT_STACK_SIZE - 1, 1004 );
-
- fpop ( lo, hi );
- if (hi - lo < FALLBACK_QSORT_SMALL_THRESH) {
- fallbackSimpleSort ( fmap, eclass, lo, hi );
- continue;
- }
-
- /* Random partitioning. Median of 3 sometimes fails to
- avoid bad cases. Median of 9 seems to help but
- looks rather expensive. This too seems to work but
- is cheaper. Guidance for the magic constants
- 7621 and 32768 is taken from Sedgewick's algorithms
- book, chapter 35.
- */
- r = ((r * 7621) + 1) % 32768;
- r3 = r % 3;
- if (r3 == 0) med = eclass[fmap[lo]]; else
- if (r3 == 1) med = eclass[fmap[(lo+hi)>>1]]; else
- med = eclass[fmap[hi]];
-
- unLo = ltLo = lo;
- unHi = gtHi = hi;
-
- while (1) {
- while (1) {
- if (unLo > unHi) break;
- n = (Int32)eclass[fmap[unLo]] - (Int32)med;
- if (n == 0) {
- fswap(fmap[unLo], fmap[ltLo]);
- ltLo++; unLo++;
- continue;
- };
- if (n > 0) break;
- unLo++;
- }
- while (1) {
- if (unLo > unHi) break;
- n = (Int32)eclass[fmap[unHi]] - (Int32)med;
- if (n == 0) {
- fswap(fmap[unHi], fmap[gtHi]);
- gtHi--; unHi--;
- continue;
- };
- if (n < 0) break;
- unHi--;
- }
- if (unLo > unHi) break;
- fswap(fmap[unLo], fmap[unHi]); unLo++; unHi--;
- }
-
- AssertD ( unHi == unLo-1, "fallbackQSort3(2)" );
-
- if (gtHi < ltLo) continue;
-
- n = fmin(ltLo-lo, unLo-ltLo); fvswap(lo, unLo-n, n);
- m = fmin(hi-gtHi, gtHi-unHi); fvswap(unLo, hi-m+1, m);
-
- n = lo + unLo - ltLo - 1;
- m = hi - (gtHi - unHi) + 1;
-
- if (n - lo > hi - m) {
- fpush ( lo, n );
- fpush ( m, hi );
- } else {
- fpush ( m, hi );
- fpush ( lo, n );
- }
- }
-}
-
-#undef fmin
-#undef fpush
-#undef fpop
-#undef fswap
-#undef fvswap
-#undef FALLBACK_QSORT_SMALL_THRESH
-#undef FALLBACK_QSORT_STACK_SIZE
-
-
-/*---------------------------------------------*/
-/* Pre:
- nblock > 0
- eclass exists for [0 .. nblock-1]
- ((UChar*)eclass) [0 .. nblock-1] holds block
- ptr exists for [0 .. nblock-1]
-
- Post:
- ((UChar*)eclass) [0 .. nblock-1] holds block
- All other areas of eclass destroyed
- fmap [0 .. nblock-1] holds sorted order
- bhtab [ 0 .. 2+(nblock/32) ] destroyed
-*/
-
-#define SET_BH(zz) bhtab[(zz) >> 5] |= (1 << ((zz) & 31))
-#define CLEAR_BH(zz) bhtab[(zz) >> 5] &= ~(1 << ((zz) & 31))
-#define ISSET_BH(zz) (bhtab[(zz) >> 5] & (1 << ((zz) & 31)))
-#define WORD_BH(zz) bhtab[(zz) >> 5]
-#define UNALIGNED_BH(zz) ((zz) & 0x01f)
-
-static
-void fallbackSort ( UInt32* fmap,
- UInt32* eclass,
- UInt32* bhtab,
- Int32 nblock,
- Int32 verb )
-{
- Int32 ftab[257];
- Int32 ftabCopy[256];
- Int32 H, i, j, k, l, r, cc, cc1;
- Int32 nNotDone;
- Int32 nBhtab;
- UChar* eclass8 = (UChar*)eclass;
-
- /*--
- Initial 1-char radix sort to generate
- initial fmap and initial BH bits.
- --*/
- if (verb >= 4)
- VPrintf0 ( " bucket sorting ...\n" );
- for (i = 0; i < 257; i++) ftab[i] = 0;
- for (i = 0; i < nblock; i++) ftab[eclass8[i]]++;
- for (i = 0; i < 256; i++) ftabCopy[i] = ftab[i];
- for (i = 1; i < 257; i++) ftab[i] += ftab[i-1];
-
- for (i = 0; i < nblock; i++) {
- j = eclass8[i];
- k = ftab[j] - 1;
- ftab[j] = k;
- fmap[k] = i;
- }
-
- nBhtab = 2 + (nblock / 32);
- for (i = 0; i < nBhtab; i++) bhtab[i] = 0;
- for (i = 0; i < 256; i++) SET_BH(ftab[i]);
-
- /*--
- Inductively refine the buckets. Kind-of an
- "exponential radix sort" (!), inspired by the
- Manber-Myers suffix array construction algorithm.
- --*/
-
- /*-- set sentinel bits for block-end detection --*/
- for (i = 0; i < 32; i++) {
- SET_BH(nblock + 2*i);
- CLEAR_BH(nblock + 2*i + 1);
- }
-
- /*-- the log(N) loop --*/
- H = 1;
- while (1) {
-
- if (verb >= 4)
- VPrintf1 ( " depth %6d has ", H );
-
- j = 0;
- for (i = 0; i < nblock; i++) {
- if (ISSET_BH(i)) j = i;
- k = fmap[i] - H; if (k < 0) k += nblock;
- eclass[k] = j;
- }
-
- nNotDone = 0;
- r = -1;
- while (1) {
-
- /*-- find the next non-singleton bucket --*/
- k = r + 1;
- while (ISSET_BH(k) && UNALIGNED_BH(k)) k++;
- if (ISSET_BH(k)) {
- while (WORD_BH(k) == 0xffffffff) k += 32;
- while (ISSET_BH(k)) k++;
- }
- l = k - 1;
- if (l >= nblock) break;
- while (!ISSET_BH(k) && UNALIGNED_BH(k)) k++;
- if (!ISSET_BH(k)) {
- while (WORD_BH(k) == 0x00000000) k += 32;
- while (!ISSET_BH(k)) k++;
- }
- r = k - 1;
- if (r >= nblock) break;
-
- /*-- now [l, r] bracket current bucket --*/
- if (r > l) {
- nNotDone += (r - l + 1);
- fallbackQSort3 ( fmap, eclass, l, r );
-
- /*-- scan bucket and generate header bits-- */
- cc = -1;
- for (i = l; i <= r; i++) {
- cc1 = eclass[fmap[i]];
- if (cc != cc1) { SET_BH(i); cc = cc1; };
- }
- }
- }
-
- if (verb >= 4)
- VPrintf1 ( "%6d unresolved strings\n", nNotDone );
-
- H *= 2;
- if (H > nblock || nNotDone == 0) break;
- }
-
- /*--
- Reconstruct the original block in
- eclass8 [0 .. nblock-1], since the
- previous phase destroyed it.
- --*/
- if (verb >= 4)
- VPrintf0 ( " reconstructing block ...\n" );
- j = 0;
- for (i = 0; i < nblock; i++) {
- while (ftabCopy[j] == 0) j++;
- ftabCopy[j]--;
- eclass8[fmap[i]] = (UChar)j;
- }
- AssertH ( j < 256, 1005 );
-}
-
-#undef SET_BH
-#undef CLEAR_BH
-#undef ISSET_BH
-#undef WORD_BH
-#undef UNALIGNED_BH
-
-
-/*---------------------------------------------*/
-/*--- The main, O(N^2 log(N)) sorting ---*/
-/*--- algorithm. Faster for "normal" ---*/
-/*--- non-repetitive blocks. ---*/
-/*---------------------------------------------*/
-
-/*---------------------------------------------*/
-static
-__inline__
-Bool mainGtU ( UInt32 i1,
- UInt32 i2,
- UChar* block,
- UInt16* quadrant,
- UInt32 nblock,
- Int32* budget )
-{
- Int32 k;
- UChar c1, c2;
- UInt16 s1, s2;
-
- AssertD ( i1 != i2, "mainGtU" );
- /* 1 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 2 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 3 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 4 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 5 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 6 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 7 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 8 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 9 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 10 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 11 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
- /* 12 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- i1++; i2++;
-
- k = nblock + 8;
-
- do {
- /* 1 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- s1 = quadrant[i1]; s2 = quadrant[i2];
- if (s1 != s2) return (s1 > s2);
- i1++; i2++;
- /* 2 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- s1 = quadrant[i1]; s2 = quadrant[i2];
- if (s1 != s2) return (s1 > s2);
- i1++; i2++;
- /* 3 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- s1 = quadrant[i1]; s2 = quadrant[i2];
- if (s1 != s2) return (s1 > s2);
- i1++; i2++;
- /* 4 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- s1 = quadrant[i1]; s2 = quadrant[i2];
- if (s1 != s2) return (s1 > s2);
- i1++; i2++;
- /* 5 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- s1 = quadrant[i1]; s2 = quadrant[i2];
- if (s1 != s2) return (s1 > s2);
- i1++; i2++;
- /* 6 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- s1 = quadrant[i1]; s2 = quadrant[i2];
- if (s1 != s2) return (s1 > s2);
- i1++; i2++;
- /* 7 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- s1 = quadrant[i1]; s2 = quadrant[i2];
- if (s1 != s2) return (s1 > s2);
- i1++; i2++;
- /* 8 */
- c1 = block[i1]; c2 = block[i2];
- if (c1 != c2) return (c1 > c2);
- s1 = quadrant[i1]; s2 = quadrant[i2];
- if (s1 != s2) return (s1 > s2);
- i1++; i2++;
-
- if (i1 >= nblock) i1 -= nblock;
- if (i2 >= nblock) i2 -= nblock;
-
- k -= 8;
- (*budget)--;
- }
- while (k >= 0);
-
- return False;
-}
-
-
-/*---------------------------------------------*/
-/*--
- Knuth's increments seem to work better
- than Incerpi-Sedgewick here. Possibly
- because the number of elems to sort is
- usually small, typically <= 20.
---*/
-static
-Int32 incs[14] = { 1, 4, 13, 40, 121, 364, 1093, 3280,
- 9841, 29524, 88573, 265720,
- 797161, 2391484 };
-
-static
-void mainSimpleSort ( UInt32* ptr,
- UChar* block,
- UInt16* quadrant,
- Int32 nblock,
- Int32 lo,
- Int32 hi,
- Int32 d,
- Int32* budget )
-{
- Int32 i, j, h, bigN, hp;
- UInt32 v;
-
- bigN = hi - lo + 1;
- if (bigN < 2) return;
-
- hp = 0;
- while (incs[hp] < bigN) hp++;
- hp--;
-
- for (; hp >= 0; hp--) {
- h = incs[hp];
-
- i = lo + h;
- while (True) {
-
- /*-- copy 1 --*/
- if (i > hi) break;
- v = ptr[i];
- j = i;
- while ( mainGtU (
- ptr[j-h]+d, v+d, block, quadrant, nblock, budget
- ) ) {
- ptr[j] = ptr[j-h];
- j = j - h;
- if (j <= (lo + h - 1)) break;
- }
- ptr[j] = v;
- i++;
-
- /*-- copy 2 --*/
- if (i > hi) break;
- v = ptr[i];
- j = i;
- while ( mainGtU (
- ptr[j-h]+d, v+d, block, quadrant, nblock, budget
- ) ) {
- ptr[j] = ptr[j-h];
- j = j - h;
- if (j <= (lo + h - 1)) break;
- }
- ptr[j] = v;
- i++;
-
- /*-- copy 3 --*/
- if (i > hi) break;
- v = ptr[i];
- j = i;
- while ( mainGtU (
- ptr[j-h]+d, v+d, block, quadrant, nblock, budget
- ) ) {
- ptr[j] = ptr[j-h];
- j = j - h;
- if (j <= (lo + h - 1)) break;
- }
- ptr[j] = v;
- i++;
-
- if (*budget < 0) return;
- }
- }
-}
-
-
-/*---------------------------------------------*/
-/*--
- The following is an implementation of
- an elegant 3-way quicksort for strings,
- described in a paper "Fast Algorithms for
- Sorting and Searching Strings", by Robert
- Sedgewick and Jon L. Bentley.
---*/
-
-#define mswap(zz1, zz2) \
- { Int32 zztmp = zz1; zz1 = zz2; zz2 = zztmp; }
-
-#define mvswap(zzp1, zzp2, zzn) \
-{ \
- Int32 yyp1 = (zzp1); \
- Int32 yyp2 = (zzp2); \
- Int32 yyn = (zzn); \
- while (yyn > 0) { \
- mswap(ptr[yyp1], ptr[yyp2]); \
- yyp1++; yyp2++; yyn--; \
- } \
-}
-
-static
-__inline__
-UChar mmed3 ( UChar a, UChar b, UChar c )
-{
- UChar t;
- if (a > b) { t = a; a = b; b = t; };
- if (b > c) {
- b = c;
- if (a > b) b = a;
- }
- return b;
-}
-
-#define mmin(a,b) ((a) < (b)) ? (a) : (b)
-
-#define mpush(lz,hz,dz) { stackLo[sp] = lz; \
- stackHi[sp] = hz; \
- stackD [sp] = dz; \
- sp++; }
-
-#define mpop(lz,hz,dz) { sp--; \
- lz = stackLo[sp]; \
- hz = stackHi[sp]; \
- dz = stackD [sp]; }
-
-
-#define mnextsize(az) (nextHi[az]-nextLo[az])
-
-#define mnextswap(az,bz) \
- { Int32 tz; \
- tz = nextLo[az]; nextLo[az] = nextLo[bz]; nextLo[bz] = tz; \
- tz = nextHi[az]; nextHi[az] = nextHi[bz]; nextHi[bz] = tz; \
- tz = nextD [az]; nextD [az] = nextD [bz]; nextD [bz] = tz; }
-
-
-#define MAIN_QSORT_SMALL_THRESH 20
-#define MAIN_QSORT_DEPTH_THRESH (BZ_N_RADIX + BZ_N_QSORT)
-#define MAIN_QSORT_STACK_SIZE 100
-
-static
-void mainQSort3 ( UInt32* ptr,
- UChar* block,
- UInt16* quadrant,
- Int32 nblock,
- Int32 loSt,
- Int32 hiSt,
- Int32 dSt,
- Int32* budget )
-{
- Int32 unLo, unHi, ltLo, gtHi, n, m, med;
- Int32 sp, lo, hi, d;
-
- Int32 stackLo[MAIN_QSORT_STACK_SIZE];
- Int32 stackHi[MAIN_QSORT_STACK_SIZE];
- Int32 stackD [MAIN_QSORT_STACK_SIZE];
-
- Int32 nextLo[3];
- Int32 nextHi[3];
- Int32 nextD [3];
-
- sp = 0;
- mpush ( loSt, hiSt, dSt );
-
- while (sp > 0) {
-
- AssertH ( sp < MAIN_QSORT_STACK_SIZE - 2, 1001 );
-
- mpop ( lo, hi, d );
- if (hi - lo < MAIN_QSORT_SMALL_THRESH ||
- d > MAIN_QSORT_DEPTH_THRESH) {
- mainSimpleSort ( ptr, block, quadrant, nblock, lo, hi, d, budget );
- if (*budget < 0) return;
- continue;
- }
-
- med = (Int32)
- mmed3 ( block[ptr[ lo ]+d],
- block[ptr[ hi ]+d],
- block[ptr[ (lo+hi)>>1 ]+d] );
-
- unLo = ltLo = lo;
- unHi = gtHi = hi;
-
- while (True) {
- while (True) {
- if (unLo > unHi) break;
- n = ((Int32)block[ptr[unLo]+d]) - med;
- if (n == 0) {
- mswap(ptr[unLo], ptr[ltLo]);
- ltLo++; unLo++; continue;
- };
- if (n > 0) break;
- unLo++;
- }
- while (True) {
- if (unLo > unHi) break;
- n = ((Int32)block[ptr[unHi]+d]) - med;
- if (n == 0) {
- mswap(ptr[unHi], ptr[gtHi]);
- gtHi--; unHi--; continue;
- };
- if (n < 0) break;
- unHi--;
- }
- if (unLo > unHi) break;
- mswap(ptr[unLo], ptr[unHi]); unLo++; unHi--;
- }
-
- AssertD ( unHi == unLo-1, "mainQSort3(2)" );
-
- if (gtHi < ltLo) {
- mpush(lo, hi, d+1 );
- continue;
- }
-
- n = mmin(ltLo-lo, unLo-ltLo); mvswap(lo, unLo-n, n);
- m = mmin(hi-gtHi, gtHi-unHi); mvswap(unLo, hi-m+1, m);
-
- n = lo + unLo - ltLo - 1;
- m = hi - (gtHi - unHi) + 1;
-
- nextLo[0] = lo; nextHi[0] = n; nextD[0] = d;
- nextLo[1] = m; nextHi[1] = hi; nextD[1] = d;
- nextLo[2] = n+1; nextHi[2] = m-1; nextD[2] = d+1;
-
- if (mnextsize(0) < mnextsize(1)) mnextswap(0,1);
- if (mnextsize(1) < mnextsize(2)) mnextswap(1,2);
- if (mnextsize(0) < mnextsize(1)) mnextswap(0,1);
-
- AssertD (mnextsize(0) >= mnextsize(1), "mainQSort3(8)" );
- AssertD (mnextsize(1) >= mnextsize(2), "mainQSort3(9)" );
-
- mpush (nextLo[0], nextHi[0], nextD[0]);
- mpush (nextLo[1], nextHi[1], nextD[1]);
- mpush (nextLo[2], nextHi[2], nextD[2]);
- }
-}
-
-#undef mswap
-#undef mvswap
-#undef mpush
-#undef mpop
-#undef mmin
-#undef mnextsize
-#undef mnextswap
-#undef MAIN_QSORT_SMALL_THRESH
-#undef MAIN_QSORT_DEPTH_THRESH
-#undef MAIN_QSORT_STACK_SIZE
-
-
-/*---------------------------------------------*/
-/* Pre:
- nblock > N_OVERSHOOT
- block32 exists for [0 .. nblock-1 +N_OVERSHOOT]
- ((UChar*)block32) [0 .. nblock-1] holds block
- ptr exists for [0 .. nblock-1]
-
- Post:
- ((UChar*)block32) [0 .. nblock-1] holds block
- All other areas of block32 destroyed
- ftab [0 .. 65536 ] destroyed
- ptr [0 .. nblock-1] holds sorted order
- if (*budget < 0), sorting was abandoned
-*/
-
-#define BIGFREQ(b) (ftab[((b)+1) << 8] - ftab[(b) << 8])
-#define SETMASK (1 << 21)
-#define CLEARMASK (~(SETMASK))
-
-static
-void mainSort ( UInt32* ptr,
- UChar* block,
- UInt16* quadrant,
- UInt32* ftab,
- Int32 nblock,
- Int32 verb,
- Int32* budget )
-{
- Int32 i, j, k, ss, sb;
- Int32 runningOrder[256];
- Bool bigDone[256];
- Int32 copyStart[256];
- Int32 copyEnd [256];
- UChar c1;
- Int32 numQSorted;
- UInt16 s;
- if (verb >= 4) VPrintf0 ( " main sort initialise ...\n" );
-
- /*-- set up the 2-byte frequency table --*/
- for (i = 65536; i >= 0; i--) ftab[i] = 0;
-
- j = block[0] << 8;
- i = nblock-1;
- for (; i >= 3; i -= 4) {
- quadrant[i] = 0;
- j = (j >> 8) | ( ((UInt16)block[i]) << 8);
- ftab[j]++;
- quadrant[i-1] = 0;
- j = (j >> 8) | ( ((UInt16)block[i-1]) << 8);
- ftab[j]++;
- quadrant[i-2] = 0;
- j = (j >> 8) | ( ((UInt16)block[i-2]) << 8);
- ftab[j]++;
- quadrant[i-3] = 0;
- j = (j >> 8) | ( ((UInt16)block[i-3]) << 8);
- ftab[j]++;
- }
- for (; i >= 0; i--) {
- quadrant[i] = 0;
- j = (j >> 8) | ( ((UInt16)block[i]) << 8);
- ftab[j]++;
- }
-
- /*-- (emphasises close relationship of block & quadrant) --*/
- for (i = 0; i < BZ_N_OVERSHOOT; i++) {
- block [nblock+i] = block[i];
- quadrant[nblock+i] = 0;
- }
-
- if (verb >= 4) VPrintf0 ( " bucket sorting ...\n" );
-
- /*-- Complete the initial radix sort --*/
- for (i = 1; i <= 65536; i++) ftab[i] += ftab[i-1];
-
- s = block[0] << 8;
- i = nblock-1;
- for (; i >= 3; i -= 4) {
- s = (s >> 8) | (block[i] << 8);
- j = ftab[s] -1;
- ftab[s] = j;
- ptr[j] = i;
- s = (s >> 8) | (block[i-1] << 8);
- j = ftab[s] -1;
- ftab[s] = j;
- ptr[j] = i-1;
- s = (s >> 8) | (block[i-2] << 8);
- j = ftab[s] -1;
- ftab[s] = j;
- ptr[j] = i-2;
- s = (s >> 8) | (block[i-3] << 8);
- j = ftab[s] -1;
- ftab[s] = j;
- ptr[j] = i-3;
- }
- for (; i >= 0; i--) {
- s = (s >> 8) | (block[i] << 8);
- j = ftab[s] -1;
- ftab[s] = j;
- ptr[j] = i;
- }
-
- /*--
- Now ftab contains the first loc of every small bucket.
- Calculate the running order, from smallest to largest
- big bucket.
- --*/
- for (i = 0; i <= 255; i++) {
- bigDone [i] = False;
- runningOrder[i] = i;
- }
-
- {
- Int32 vv;
- Int32 h = 1;
- do h = 3 * h + 1; while (h <= 256);
- do {
- h = h / 3;
- for (i = h; i <= 255; i++) {
- vv = runningOrder[i];
- j = i;
- while ( BIGFREQ(runningOrder[j-h]) > BIGFREQ(vv) ) {
- runningOrder[j] = runningOrder[j-h];
- j = j - h;
- if (j <= (h - 1)) goto zero;
- }
- zero:
- runningOrder[j] = vv;
- }
- } while (h != 1);
- }
-
- /*--
- The main sorting loop.
- --*/
-
- numQSorted = 0;
-
- for (i = 0; i <= 255; i++) {
-
- /*--
- Process big buckets, starting with the least full.
- Basically this is a 3-step process in which we call
- mainQSort3 to sort the small buckets [ss, j], but
- also make a big effort to avoid the calls if we can.
- --*/
- ss = runningOrder[i];
-
- /*--
- Step 1:
- Complete the big bucket [ss] by quicksorting
- any unsorted small buckets [ss, j], for j != ss.
- Hopefully previous pointer-scanning phases have already
- completed many of the small buckets [ss, j], so
- we don't have to sort them at all.
- --*/
- for (j = 0; j <= 255; j++) {
- if (j != ss) {
- sb = (ss << 8) + j;
- if ( ! (ftab[sb] & SETMASK) ) {
- Int32 lo = ftab[sb] & CLEARMASK;
- Int32 hi = (ftab[sb+1] & CLEARMASK) - 1;
- if (hi > lo) {
- if (verb >= 4)
- VPrintf4 ( " qsort [0x%x, 0x%x] "
- "done %d this %d\n",
- ss, j, numQSorted, hi - lo + 1 );
- mainQSort3 (
- ptr, block, quadrant, nblock,
- lo, hi, BZ_N_RADIX, budget
- );
- numQSorted += (hi - lo + 1);
- if (*budget < 0) return;
- }
- }
- ftab[sb] |= SETMASK;
- }
- }
-
- AssertH ( !bigDone[ss], 1006 );
-
- /*--
- Step 2:
- Now scan this big bucket [ss] so as to synthesise the
- sorted order for small buckets [t, ss] for all t,
- including, magically, the bucket [ss,ss] too.
- This will avoid doing Real Work in subsequent Step 1's.
- --*/
- {
- for (j = 0; j <= 255; j++) {
- copyStart[j] = ftab[(j << 8) + ss] & CLEARMASK;
- copyEnd [j] = (ftab[(j << 8) + ss + 1] & CLEARMASK) - 1;
- }
- for (j = ftab[ss << 8] & CLEARMASK; j < copyStart[ss]; j++) {
- k = ptr[j]-1; if (k < 0) k += nblock;
- c1 = block[k];
- if (!bigDone[c1])
- ptr[ copyStart[c1]++ ] = k;
- }
- for (j = (ftab[(ss+1) << 8] & CLEARMASK) - 1; j > copyEnd[ss]; j--) {
- k = ptr[j]-1; if (k < 0) k += nblock;
- c1 = block[k];
- if (!bigDone[c1])
- ptr[ copyEnd[c1]-- ] = k;
- }
- }
-
- AssertH ( (copyStart[ss]-1 == copyEnd[ss])
- ||
- /* Extremely rare case missing in bzip2-1.0.0 and 1.0.1.
- Necessity for this case is demonstrated by compressing
- a sequence of approximately 48.5 million of character
- 251; 1.0.0/1.0.1 will then die here. */
- (copyStart[ss] == 0 && copyEnd[ss] == nblock-1),
- 1007 )
-
- for (j = 0; j <= 255; j++) ftab[(j << 8) + ss] |= SETMASK;
-
- /*--
- Step 3:
- The [ss] big bucket is now done. Record this fact,
- and update the quadrant descriptors. Remember to
- update quadrants in the overshoot area too, if
- necessary. The "if (i < 255)" test merely skips
- this updating for the last bucket processed, since
- updating for the last bucket is pointless.
-
- The quadrant array provides a way to incrementally
- cache sort orderings, as they appear, so as to
- make subsequent comparisons in fullGtU() complete
- faster. For repetitive blocks this makes a big
- difference (but not big enough to be able to avoid
- the fallback sorting mechanism, exponential radix sort).
-
- The precise meaning is: at all times:
-
- for 0 <= i < nblock and 0 <= j <= nblock
-
- if block[i] != block[j],
-
- then the relative values of quadrant[i] and
- quadrant[j] are meaningless.
-
- else {
- if quadrant[i] < quadrant[j]
- then the string starting at i lexicographically
- precedes the string starting at j
-
- else if quadrant[i] > quadrant[j]
- then the string starting at j lexicographically
- precedes the string starting at i
-
- else
- the relative ordering of the strings starting
- at i and j has not yet been determined.
- }
- --*/
- bigDone[ss] = True;
-
- if (i < 255) {
- Int32 bbStart = ftab[ss << 8] & CLEARMASK;
- Int32 bbSize = (ftab[(ss+1) << 8] & CLEARMASK) - bbStart;
- Int32 shifts = 0;
-
- while ((bbSize >> shifts) > 65534) shifts++;
-
- for (j = bbSize-1; j >= 0; j--) {
- Int32 a2update = ptr[bbStart + j];
- UInt16 qVal = (UInt16)(j >> shifts);
- quadrant[a2update] = qVal;
- if (a2update < BZ_N_OVERSHOOT)
- quadrant[a2update + nblock] = qVal;
- }
- AssertH ( ((bbSize-1) >> shifts) <= 65535, 1002 );
- }
-
- }
-
- if (verb >= 4)
- VPrintf3 ( " %d pointers, %d sorted, %d scanned\n",
- nblock, numQSorted, nblock - numQSorted );
-}
-
-#undef BIGFREQ
-#undef SETMASK
-#undef CLEARMASK
-
-
-/*---------------------------------------------*/
-/* Pre:
- nblock > 0
- arr2 exists for [0 .. nblock-1 +N_OVERSHOOT]
- ((UChar*)arr2) [0 .. nblock-1] holds block
- arr1 exists for [0 .. nblock-1]
-
- Post:
- ((UChar*)arr2) [0 .. nblock-1] holds block
- All other areas of block destroyed
- ftab [ 0 .. 65536 ] destroyed
- arr1 [0 .. nblock-1] holds sorted order
-*/
-void BZ2_blockSort ( EState* s )
-{
- UInt32* ptr = s->ptr;
- UChar* block = s->block;
- UInt32* ftab = s->ftab;
- Int32 nblock = s->nblock;
- Int32 verb = s->verbosity;
- Int32 wfact = s->workFactor;
- UInt16* quadrant;
- Int32 budget;
- Int32 budgetInit;
- Int32 i;
-
- if (nblock < 10000) {
- fallbackSort ( s->arr1, s->arr2, ftab, nblock, verb );
- } else {
- /* Calculate the location for quadrant, remembering to get
- the alignment right. Assumes that &(block[0]) is at least
- 2-byte aligned -- this should be ok since block is really
- the first section of arr2.
- */
- i = nblock+BZ_N_OVERSHOOT;
- if (i & 1) i++;
- quadrant = (UInt16*)(&(block[i]));
-
- /* (wfact-1) / 3 puts the default-factor-30
- transition point at very roughly the same place as
- with v0.1 and v0.9.0.
- Not that it particularly matters any more, since the
- resulting compressed stream is now the same regardless
- of whether or not we use the main sort or fallback sort.
- */
- if (wfact < 1 ) wfact = 1;
- if (wfact > 100) wfact = 100;
- budgetInit = nblock * ((wfact-1) / 3);
- budget = budgetInit;
-
- mainSort ( ptr, block, quadrant, ftab, nblock, verb, &budget );
- if (verb >= 3)
- VPrintf3 ( " %d work, %d block, ratio %5.2f\n",
- budgetInit - budget,
- nblock,
- (float)(budgetInit - budget) /
- (float)(nblock==0 ? 1 : nblock) );
- if (budget < 0) {
- if (verb >= 2)
- VPrintf0 ( " too repetitive; using fallback"
- " sorting algorithm\n" );
- fallbackSort ( s->arr1, s->arr2, ftab, nblock, verb );
- }
- }
-
- s->origPtr = -1;
- for (i = 0; i < s->nblock; i++)
- if (ptr[i] == 0)
- { s->origPtr = i; break; };
-
- AssertH( s->origPtr != -1, 1003 );
-}
-
-
-/*-------------------------------------------------------------*/
-/*--- end blocksort.c ---*/
-/*-------------------------------------------------------------*/
+++ /dev/null
-/*--
- This file is a part of bzip2 and/or libbzip2, a program and
- library for lossless, block-sorting data compression.
-
- Copyright (C) 1996-1999 Julian R Seward. All rights reserved.
-
- Redistribution and use in source and binary forms, with or without
- modification, are permitted provided that the following conditions
- are met:
-
- 1. Redistributions of source code must retain the above copyright
- notice, this list of conditions and the following disclaimer.
-
- 2. The origin of this software must not be misrepresented; you must
- not claim that you wrote the original software. If you use this
- software in a product, an acknowledgment in the product
- documentation would be appreciated but is not required.
-
- 3. Altered source versions must be plainly marked as such, and must
- not be misrepresented as being the original software.
-
- 4. The name of the author may not be used to endorse or promote
- products derived from this software without specific prior written
- permission.
-
- THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS
- OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
- DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
- DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
- GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
- WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
- NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-
- Julian Seward, Cambridge, UK.
- jseward@acm.org
- bzip2/libbzip2 version 0.9.5 of 24 May 1999
-
- (Now heavily hacked to be part of zgz; decompression and
- portability ripped out.)
-
- This program is based on (at least) the work of:
- Mike Burrows
- David Wheeler
- Peter Fenwick
- Alistair Moffat
- Radford Neal
- Ian H. Witten
- Robert Sedgewick
- Jon L. Bentley
-
- For more information on these sources, see the manual.
---*/
+++ /dev/null
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <signal.h>
-#include <math.h>
-#include <errno.h>
-#include <ctype.h>
-#include "bzlib.h"
-
-#define ERROR_IF_EOF(i) { if ((i) == EOF) ioError(); }
-#define ERROR_IF_NOT_ZERO(i) { if ((i) != 0) ioError(); }
-#define ERROR_IF_MINUS_ONE(i) { if ((i) == (-1)) ioError(); }
-
-#include <sys/types.h>
-#include <utime.h>
-#include <unistd.h>
-#include <sys/stat.h>
-#include <sys/times.h>
-
-typedef char Char;
-typedef unsigned char Bool;
-typedef unsigned char UChar;
-typedef int Int32;
-typedef unsigned int UInt32;
-typedef short Int16;
-typedef unsigned short UInt16;
-
-#define True ((Bool)1)
-#define False ((Bool)0)
-
-/*--
- IntNative is your platform's `native' int size.
- Only here to avoid probs with 64-bit platforms.
---*/
-typedef int IntNative;
-
-static Int32 verbosity;
-static Int32 blockSize100k;
-static Int32 workFactor;
-
-static
-void panic (char *msg) {
- perror("oops");
- fprintf(stderr, "%s\n", msg);
- exit(1);
-}
-
-static
-void compressStream ( FILE *stream, FILE *zStream )
-{
- BZFILE* bzf = NULL;
- UChar ibuf[5000];
- Int32 nIbuf;
- UInt32 nbytes_in, nbytes_out;
- Int32 bzerr, bzerr_dummy, ret;
-
- if (ferror(stream)) goto errhandler_io;
- if (ferror(zStream)) goto errhandler_io;
-
- bzf = bzWriteOpen ( &bzerr, zStream,
- blockSize100k, verbosity, workFactor );
- if (bzerr != BZ_OK) goto errhandler;
-
- if (verbosity >= 2) fprintf ( stderr, "\n" );
-
- while (True) {
-
- if (myfeof(stream)) break;
- nIbuf = fread ( ibuf, sizeof(UChar), 5000, stream );
- if (ferror(stream)) goto errhandler_io;
- if (nIbuf > 0) bzWrite ( &bzerr, bzf, (void*)ibuf, nIbuf );
- if (bzerr != BZ_OK) goto errhandler;
-
- }
-
- bzWriteClose ( &bzerr, bzf, 0, &nbytes_in, &nbytes_out );
- if (bzerr != BZ_OK) goto errhandler;
-
- if (ferror(zStream)) goto errhandler_io;
- ret = fflush ( zStream );
- if (ret == EOF) goto errhandler_io;
- if (zStream != stdout) {
- ret = fclose ( zStream );
- if (ret == EOF) goto errhandler_io;
- }
- if (ferror(stream)) goto errhandler_io;
- ret = fclose ( stream );
- if (ret == EOF) goto errhandler_io;
-
- if (nbytes_in == 0) nbytes_in = 1;
-
- return;
-
- errhandler:
- bzWriteClose ( &bzerr_dummy, bzf, 1, &nbytes_in, &nbytes_out );
- switch (bzerr) {
- case BZ_MEM_ERROR:
- panic ( "out of memory" );
- case BZ_IO_ERROR:
- errhandler_io:
- panic ( "io error" );
- default:
- panic ( "compress:unexpected error" );
- }
-
- panic ( "compress:end" );
- /*notreached*/
-}
-
-void suse_bzip2(int level) {
- workFactor = 30;
- blockSize100k = level;
-
- compressStream(stdin, stdout);
-}
-/*-------------------------------------------------------------*/
-/*--- Library top-level functions. ---*/
-/*--- bzlib.c ---*/
-/*-------------------------------------------------------------*/
-
-/* ------------------------------------------------------------------
- This file is part of bzip2/libbzip2, a program and library for
- lossless, block-sorting data compression.
-
- bzip2/libbzip2 version 1.0.5 of 10 December 2007
- Copyright (C) 1996-2007 Julian Seward <jseward@bzip.org>
-
- Please read the WARNING, DISCLAIMER and PATENTS sections in the
- README file.
-
- This program is released under the terms of the license contained
- in the file LICENSE.
- ------------------------------------------------------------------ */
-
-/* CHANGES
- 0.9.0 -- original version.
- 0.9.0a/b -- no changes in this file.
- 0.9.0c -- made zero-length BZ_FLUSH work correctly in bzCompress().
- fixed bzWrite/bzRead to ignore zero-length requests.
- fixed bzread to correctly handle read requests after EOF.
- wrong parameter order in call to bzDecompressInit in
- bzBuffToBuffDecompress. Fixed.
-*/
-
-#include "bzlib_private.h"
-
-
-/*---------------------------------------------------*/
-/*--- Compression stuff ---*/
-/*---------------------------------------------------*/
-
-
-/*---------------------------------------------------*/
-#ifndef BZ_NO_STDIO
-void BZ2_bz__AssertH__fail ( int errcode )
-{
- fprintf(stderr,
- "\n\nbzip2/libbzip2: internal error number %d.\n"
- "This is a bug in bzip2/libbzip2, %s.\n"
- "Please report it to me at: jseward@bzip.org. If this happened\n"
- "when you were using some program which uses libbzip2 as a\n"
- "component, you should also report this bug to the author(s)\n"
- "of that program. Please make an effort to report this bug;\n"
- "timely and accurate bug reports eventually lead to higher\n"
- "quality software. Thanks. Julian Seward, 10 December 2007.\n\n",
- errcode,
- BZ2_bzlibVersion()
- );
-
- if (errcode == 1007) {
- fprintf(stderr,
- "\n*** A special note about internal error number 1007 ***\n"
- "\n"
- "Experience suggests that a common cause of i.e. 1007\n"
- "is unreliable memory or other hardware. The 1007 assertion\n"
- "just happens to cross-check the results of huge numbers of\n"
- "memory reads/writes, and so acts (unintendedly) as a stress\n"
- "test of your memory system.\n"
- "\n"
- "I suggest the following: try compressing the file again,\n"
- "possibly monitoring progress in detail with the -vv flag.\n"
- "\n"
- "* If the error cannot be reproduced, and/or happens at different\n"
- " points in compression, you may have a flaky memory system.\n"
- " Try a memory-test program. I have used Memtest86\n"
- " (www.memtest86.com). At the time of writing it is free (GPLd).\n"
- " Memtest86 tests memory much more thorougly than your BIOSs\n"
- " power-on test, and may find failures that the BIOS doesn't.\n"
- "\n"
- "* If the error can be repeatably reproduced, this is a bug in\n"
- " bzip2, and I would very much like to hear about it. Please\n"
- " let me know, and, ideally, save a copy of the file causing the\n"
- " problem -- without which I will be unable to investigate it.\n"
- "\n"
- );
- }
-
- exit(3);
-}
-#endif
-
-
-/*---------------------------------------------------*/
-static
-int bz_config_ok ( void )
-{
- if (sizeof(int) != 4) return 0;
- if (sizeof(short) != 2) return 0;
- if (sizeof(char) != 1) return 0;
- return 1;
-}
-
-
-/*---------------------------------------------------*/
-static
-void* default_bzalloc ( void* opaque, Int32 items, Int32 size )
-{
- void* v = malloc ( items * size );
- return v;
-}
-
-static
-void default_bzfree ( void* opaque, void* addr )
-{
- if (addr != NULL) free ( addr );
-}
-
-
-/*---------------------------------------------------*/
-static
-void prepare_new_block ( EState* s )
-{
- Int32 i;
- s->nblock = 0;
- s->numZ = 0;
- s->state_out_pos = 0;
- BZ_INITIALISE_CRC ( s->blockCRC );
- for (i = 0; i < 256; i++) s->inUse[i] = False;
- s->blockNo++;
-}
-
-
-/*---------------------------------------------------*/
-static
-void init_RL ( EState* s )
-{
- s->state_in_ch = 256;
- s->state_in_len = 0;
-}
-
-
-static
-Bool isempty_RL ( EState* s )
-{
- if (s->state_in_ch < 256 && s->state_in_len > 0)
- return False; else
- return True;
-}
-
-
-/*---------------------------------------------------*/
-int BZ_API(BZ2_bzCompressInit)
- ( bz_stream* strm,
- int blockSize100k,
- int verbosity,
- int workFactor )
-{
- Int32 n;
- EState* s;
-
- if (!bz_config_ok()) return BZ_CONFIG_ERROR;
-
- if (strm == NULL ||
- blockSize100k < 1 || blockSize100k > 9 ||
- workFactor < 0 || workFactor > 250)
- return BZ_PARAM_ERROR;
-
- if (workFactor == 0) workFactor = 30;
- if (strm->bzalloc == NULL) strm->bzalloc = default_bzalloc;
- if (strm->bzfree == NULL) strm->bzfree = default_bzfree;
-
- s = BZALLOC( sizeof(EState) );
- if (s == NULL) return BZ_MEM_ERROR;
- s->strm = strm;
-
- s->arr1 = NULL;
- s->arr2 = NULL;
- s->ftab = NULL;
-
- n = 100000 * blockSize100k;
- s->arr1 = BZALLOC( n * sizeof(UInt32) );
- s->arr2 = BZALLOC( (n+BZ_N_OVERSHOOT) * sizeof(UInt32) );
- s->ftab = BZALLOC( 65537 * sizeof(UInt32) );
-
- if (s->arr1 == NULL || s->arr2 == NULL || s->ftab == NULL) {
- if (s->arr1 != NULL) BZFREE(s->arr1);
- if (s->arr2 != NULL) BZFREE(s->arr2);
- if (s->ftab != NULL) BZFREE(s->ftab);
- if (s != NULL) BZFREE(s);
- return BZ_MEM_ERROR;
- }
-
- s->blockNo = 0;
- s->state = BZ_S_INPUT;
- s->mode = BZ_M_RUNNING;
- s->combinedCRC = 0;
- s->blockSize100k = blockSize100k;
- s->nblockMAX = 100000 * blockSize100k - 19;
- s->verbosity = verbosity;
- s->workFactor = workFactor;
-
- s->block = (UChar*)s->arr2;
- s->mtfv = (UInt16*)s->arr1;
- s->zbits = NULL;
- s->ptr = (UInt32*)s->arr1;
-
- strm->state = s;
- strm->total_in_lo32 = 0;
- strm->total_in_hi32 = 0;
- strm->total_out_lo32 = 0;
- strm->total_out_hi32 = 0;
- init_RL ( s );
- prepare_new_block ( s );
- return BZ_OK;
-}
-
-
-/*---------------------------------------------------*/
-static
-void add_pair_to_block ( EState* s )
-{
- Int32 i;
- UChar ch = (UChar)(s->state_in_ch);
- for (i = 0; i < s->state_in_len; i++) {
- BZ_UPDATE_CRC( s->blockCRC, ch );
- }
- s->inUse[s->state_in_ch] = True;
- switch (s->state_in_len) {
- case 1:
- s->block[s->nblock] = (UChar)ch; s->nblock++;
- break;
- case 2:
- s->block[s->nblock] = (UChar)ch; s->nblock++;
- s->block[s->nblock] = (UChar)ch; s->nblock++;
- break;
- case 3:
- s->block[s->nblock] = (UChar)ch; s->nblock++;
- s->block[s->nblock] = (UChar)ch; s->nblock++;
- s->block[s->nblock] = (UChar)ch; s->nblock++;
- break;
- default:
- s->inUse[s->state_in_len-4] = True;
- s->block[s->nblock] = (UChar)ch; s->nblock++;
- s->block[s->nblock] = (UChar)ch; s->nblock++;
- s->block[s->nblock] = (UChar)ch; s->nblock++;
- s->block[s->nblock] = (UChar)ch; s->nblock++;
- s->block[s->nblock] = ((UChar)(s->state_in_len-4));
- s->nblock++;
- break;
- }
-}
-
-
-/*---------------------------------------------------*/
-static
-void flush_RL ( EState* s )
-{
- if (s->state_in_ch < 256) add_pair_to_block ( s );
- init_RL ( s );
-}
-
-
-/*---------------------------------------------------*/
-#define ADD_CHAR_TO_BLOCK(zs,zchh0) \
-{ \
- UInt32 zchh = (UInt32)(zchh0); \
- /*-- fast track the common case --*/ \
- if (zchh != zs->state_in_ch && \
- zs->state_in_len == 1) { \
- UChar ch = (UChar)(zs->state_in_ch); \
- BZ_UPDATE_CRC( zs->blockCRC, ch ); \
- zs->inUse[zs->state_in_ch] = True; \
- zs->block[zs->nblock] = (UChar)ch; \
- zs->nblock++; \
- zs->state_in_ch = zchh; \
- } \
- else \
- /*-- general, uncommon cases --*/ \
- if (zchh != zs->state_in_ch || \
- zs->state_in_len == 255) { \
- if (zs->state_in_ch < 256) \
- add_pair_to_block ( zs ); \
- zs->state_in_ch = zchh; \
- zs->state_in_len = 1; \
- } else { \
- zs->state_in_len++; \
- } \
-}
-
-
-/*---------------------------------------------------*/
-static
-Bool copy_input_until_stop ( EState* s )
-{
- Bool progress_in = False;
-
- if (s->mode == BZ_M_RUNNING) {
-
- /*-- fast track the common case --*/
- while (True) {
- /*-- block full? --*/
- if (s->nblock >= s->nblockMAX) break;
- /*-- no input? --*/
- if (s->strm->avail_in == 0) break;
- progress_in = True;
- ADD_CHAR_TO_BLOCK ( s, (UInt32)(*((UChar*)(s->strm->next_in))) );
- s->strm->next_in++;
- s->strm->avail_in--;
- s->strm->total_in_lo32++;
- if (s->strm->total_in_lo32 == 0) s->strm->total_in_hi32++;
- }
-
- } else {
-
- /*-- general, uncommon case --*/
- while (True) {
- /*-- block full? --*/
- if (s->nblock >= s->nblockMAX) break;
- /*-- no input? --*/
- if (s->strm->avail_in == 0) break;
- /*-- flush/finish end? --*/
- if (s->avail_in_expect == 0) break;
- progress_in = True;
- ADD_CHAR_TO_BLOCK ( s, (UInt32)(*((UChar*)(s->strm->next_in))) );
- s->strm->next_in++;
- s->strm->avail_in--;
- s->strm->total_in_lo32++;
- if (s->strm->total_in_lo32 == 0) s->strm->total_in_hi32++;
- s->avail_in_expect--;
- }
- }
- return progress_in;
-}
-
-
-/*---------------------------------------------------*/
-static
-Bool copy_output_until_stop ( EState* s )
-{
- Bool progress_out = False;
-
- while (True) {
-
- /*-- no output space? --*/
- if (s->strm->avail_out == 0) break;
-
- /*-- block done? --*/
- if (s->state_out_pos >= s->numZ) break;
-
- progress_out = True;
- *(s->strm->next_out) = s->zbits[s->state_out_pos];
- s->state_out_pos++;
- s->strm->avail_out--;
- s->strm->next_out++;
- s->strm->total_out_lo32++;
- if (s->strm->total_out_lo32 == 0) s->strm->total_out_hi32++;
- }
-
- return progress_out;
-}
-
-
-/*---------------------------------------------------*/
-static
-Bool handle_compress ( bz_stream* strm )
-{
- Bool progress_in = False;
- Bool progress_out = False;
- EState* s = strm->state;
-
- while (True) {
-
- if (s->state == BZ_S_OUTPUT) {
- progress_out |= copy_output_until_stop ( s );
- if (s->state_out_pos < s->numZ) break;
- if (s->mode == BZ_M_FINISHING &&
- s->avail_in_expect == 0 &&
- isempty_RL(s)) break;
- prepare_new_block ( s );
- s->state = BZ_S_INPUT;
- if (s->mode == BZ_M_FLUSHING &&
- s->avail_in_expect == 0 &&
- isempty_RL(s)) break;
- }
-
- if (s->state == BZ_S_INPUT) {
- progress_in |= copy_input_until_stop ( s );
- if (s->mode != BZ_M_RUNNING && s->avail_in_expect == 0) {
- flush_RL ( s );
- BZ2_compressBlock ( s, (Bool)(s->mode == BZ_M_FINISHING) );
- s->state = BZ_S_OUTPUT;
- }
- else
- if (s->nblock >= s->nblockMAX) {
- BZ2_compressBlock ( s, False );
- s->state = BZ_S_OUTPUT;
- }
- else
- if (s->strm->avail_in == 0) {
- break;
- }
- }
-
- }
-
- return progress_in || progress_out;
-}
-
-
-/*---------------------------------------------------*/
-int BZ_API(BZ2_bzCompress) ( bz_stream *strm, int action )
-{
- Bool progress;
- EState* s;
- if (strm == NULL) return BZ_PARAM_ERROR;
- s = strm->state;
- if (s == NULL) return BZ_PARAM_ERROR;
- if (s->strm != strm) return BZ_PARAM_ERROR;
-
- preswitch:
- switch (s->mode) {
-
- case BZ_M_IDLE:
- return BZ_SEQUENCE_ERROR;
-
- case BZ_M_RUNNING:
- if (action == BZ_RUN) {
- progress = handle_compress ( strm );
- return progress ? BZ_RUN_OK : BZ_PARAM_ERROR;
- }
- else
- if (action == BZ_FLUSH) {
- s->avail_in_expect = strm->avail_in;
- s->mode = BZ_M_FLUSHING;
- goto preswitch;
- }
- else
- if (action == BZ_FINISH) {
- s->avail_in_expect = strm->avail_in;
- s->mode = BZ_M_FINISHING;
- goto preswitch;
- }
- else
- return BZ_PARAM_ERROR;
-
- case BZ_M_FLUSHING:
- if (action != BZ_FLUSH) return BZ_SEQUENCE_ERROR;
- if (s->avail_in_expect != s->strm->avail_in)
- return BZ_SEQUENCE_ERROR;
- progress = handle_compress ( strm );
- if (s->avail_in_expect > 0 || !isempty_RL(s) ||
- s->state_out_pos < s->numZ) return BZ_FLUSH_OK;
- s->mode = BZ_M_RUNNING;
- return BZ_RUN_OK;
-
- case BZ_M_FINISHING:
- if (action != BZ_FINISH) return BZ_SEQUENCE_ERROR;
- if (s->avail_in_expect != s->strm->avail_in)
- return BZ_SEQUENCE_ERROR;
- progress = handle_compress ( strm );
- if (!progress) return BZ_SEQUENCE_ERROR;
- if (s->avail_in_expect > 0 || !isempty_RL(s) ||
- s->state_out_pos < s->numZ) return BZ_FINISH_OK;
- s->mode = BZ_M_IDLE;
- return BZ_STREAM_END;
- }
- return BZ_OK; /*--not reached--*/
-}
-
-
-/*---------------------------------------------------*/
-int BZ_API(BZ2_bzCompressEnd) ( bz_stream *strm )
-{
- EState* s;
- if (strm == NULL) return BZ_PARAM_ERROR;
- s = strm->state;
- if (s == NULL) return BZ_PARAM_ERROR;
- if (s->strm != strm) return BZ_PARAM_ERROR;
-
- if (s->arr1 != NULL) BZFREE(s->arr1);
- if (s->arr2 != NULL) BZFREE(s->arr2);
- if (s->ftab != NULL) BZFREE(s->ftab);
- BZFREE(strm->state);
-
- strm->state = NULL;
-
- return BZ_OK;
-}
-
-#ifndef BZ_NO_STDIO
-/*---------------------------------------------------*/
-/*--- File I/O stuff ---*/
-/*---------------------------------------------------*/
-
-#define BZ_SETERR(eee) \
-{ \
- if (bzerror != NULL) *bzerror = eee; \
- if (bzf != NULL) bzf->lastErr = eee; \
-}
-
-typedef
- struct {
- FILE* handle;
- Char buf[BZ_MAX_UNUSED];
- Int32 bufN;
- Bool writing;
- bz_stream strm;
- Int32 lastErr;
- Bool initialisedOk;
- }
- bzFile;
-
-
-/*---------------------------------------------*/
-static Bool myfeof ( FILE* f )
-{
- Int32 c = fgetc ( f );
- if (c == EOF) return True;
- ungetc ( c, f );
- return False;
-}
-
-
-/*---------------------------------------------------*/
-BZFILE* BZ_API(BZ2_bzWriteOpen)
- ( int* bzerror,
- FILE* f,
- int blockSize100k,
- int verbosity,
- int workFactor )
-{
- Int32 ret;
- bzFile* bzf = NULL;
-
- BZ_SETERR(BZ_OK);
-
- if (f == NULL ||
- (blockSize100k < 1 || blockSize100k > 9) ||
- (workFactor < 0 || workFactor > 250) ||
- (verbosity < 0 || verbosity > 4))
- { BZ_SETERR(BZ_PARAM_ERROR); return NULL; };
-
- if (ferror(f))
- { BZ_SETERR(BZ_IO_ERROR); return NULL; };
-
- bzf = malloc ( sizeof(bzFile) );
- if (bzf == NULL)
- { BZ_SETERR(BZ_MEM_ERROR); return NULL; };
-
- BZ_SETERR(BZ_OK);
- bzf->initialisedOk = False;
- bzf->bufN = 0;
- bzf->handle = f;
- bzf->writing = True;
- bzf->strm.bzalloc = NULL;
- bzf->strm.bzfree = NULL;
- bzf->strm.opaque = NULL;
-
- if (workFactor == 0) workFactor = 30;
- ret = BZ2_bzCompressInit ( &(bzf->strm), blockSize100k,
- verbosity, workFactor );
- if (ret != BZ_OK)
- { BZ_SETERR(ret); free(bzf); return NULL; };
-
- bzf->strm.avail_in = 0;
- bzf->initialisedOk = True;
- return bzf;
-}
-
-
-
-/*---------------------------------------------------*/
-void BZ_API(BZ2_bzWrite)
- ( int* bzerror,
- BZFILE* b,
- void* buf,
- int len )
-{
- Int32 n, n2, ret;
- bzFile* bzf = (bzFile*)b;
-
- BZ_SETERR(BZ_OK);
- if (bzf == NULL || buf == NULL || len < 0)
- { BZ_SETERR(BZ_PARAM_ERROR); return; };
- if (!(bzf->writing))
- { BZ_SETERR(BZ_SEQUENCE_ERROR); return; };
- if (ferror(bzf->handle))
- { BZ_SETERR(BZ_IO_ERROR); return; };
-
- if (len == 0)
- { BZ_SETERR(BZ_OK); return; };
-
- bzf->strm.avail_in = len;
- bzf->strm.next_in = buf;
-
- while (True) {
- bzf->strm.avail_out = BZ_MAX_UNUSED;
- bzf->strm.next_out = bzf->buf;
- ret = BZ2_bzCompress ( &(bzf->strm), BZ_RUN );
- if (ret != BZ_RUN_OK)
- { BZ_SETERR(ret); return; };
-
- if (bzf->strm.avail_out < BZ_MAX_UNUSED) {
- n = BZ_MAX_UNUSED - bzf->strm.avail_out;
- n2 = fwrite ( (void*)(bzf->buf), sizeof(UChar),
- n, bzf->handle );
- if (n != n2 || ferror(bzf->handle))
- { BZ_SETERR(BZ_IO_ERROR); return; };
- }
-
- if (bzf->strm.avail_in == 0)
- { BZ_SETERR(BZ_OK); return; };
- }
-}
-
-
-/*---------------------------------------------------*/
-void BZ_API(BZ2_bzWriteClose)
- ( int* bzerror,
- BZFILE* b,
- int abandon,
- unsigned int* nbytes_in,
- unsigned int* nbytes_out )
-{
- BZ2_bzWriteClose64 ( bzerror, b, abandon,
- nbytes_in, NULL, nbytes_out, NULL );
-}
-
-
-void BZ_API(BZ2_bzWriteClose64)
- ( int* bzerror,
- BZFILE* b,
- int abandon,
- unsigned int* nbytes_in_lo32,
- unsigned int* nbytes_in_hi32,
- unsigned int* nbytes_out_lo32,
- unsigned int* nbytes_out_hi32 )
-{
- Int32 n, n2, ret;
- bzFile* bzf = (bzFile*)b;
-
- if (bzf == NULL)
- { BZ_SETERR(BZ_OK); return; };
- if (!(bzf->writing))
- { BZ_SETERR(BZ_SEQUENCE_ERROR); return; };
- if (ferror(bzf->handle))
- { BZ_SETERR(BZ_IO_ERROR); return; };
-
- if (nbytes_in_lo32 != NULL) *nbytes_in_lo32 = 0;
- if (nbytes_in_hi32 != NULL) *nbytes_in_hi32 = 0;
- if (nbytes_out_lo32 != NULL) *nbytes_out_lo32 = 0;
- if (nbytes_out_hi32 != NULL) *nbytes_out_hi32 = 0;
-
- if ((!abandon) && bzf->lastErr == BZ_OK) {
- while (True) {
- bzf->strm.avail_out = BZ_MAX_UNUSED;
- bzf->strm.next_out = bzf->buf;
- ret = BZ2_bzCompress ( &(bzf->strm), BZ_FINISH );
- if (ret != BZ_FINISH_OK && ret != BZ_STREAM_END)
- { BZ_SETERR(ret); return; };
-
- if (bzf->strm.avail_out < BZ_MAX_UNUSED) {
- n = BZ_MAX_UNUSED - bzf->strm.avail_out;
- n2 = fwrite ( (void*)(bzf->buf), sizeof(UChar),
- n, bzf->handle );
- if (n != n2 || ferror(bzf->handle))
- { BZ_SETERR(BZ_IO_ERROR); return; };
- }
-
- if (ret == BZ_STREAM_END) break;
- }
- }
-
- if ( !abandon && !ferror ( bzf->handle ) ) {
- fflush ( bzf->handle );
- if (ferror(bzf->handle))
- { BZ_SETERR(BZ_IO_ERROR); return; };
- }
-
- if (nbytes_in_lo32 != NULL)
- *nbytes_in_lo32 = bzf->strm.total_in_lo32;
- if (nbytes_in_hi32 != NULL)
- *nbytes_in_hi32 = bzf->strm.total_in_hi32;
- if (nbytes_out_lo32 != NULL)
- *nbytes_out_lo32 = bzf->strm.total_out_lo32;
- if (nbytes_out_hi32 != NULL)
- *nbytes_out_hi32 = bzf->strm.total_out_hi32;
-
- BZ_SETERR(BZ_OK);
- BZ2_bzCompressEnd ( &(bzf->strm) );
- free ( bzf );
-}
-
-#endif
-
-
-/*---------------------------------------------------*/
-/*--- Misc convenience stuff ---*/
-/*---------------------------------------------------*/
-
-/*---------------------------------------------------*/
-int BZ_API(BZ2_bzBuffToBuffCompress)
- ( char* dest,
- unsigned int* destLen,
- char* source,
- unsigned int sourceLen,
- int blockSize100k,
- int verbosity,
- int workFactor )
-{
- bz_stream strm;
- int ret;
-
- if (dest == NULL || destLen == NULL ||
- source == NULL ||
- blockSize100k < 1 || blockSize100k > 9 ||
- verbosity < 0 || verbosity > 4 ||
- workFactor < 0 || workFactor > 250)
- return BZ_PARAM_ERROR;
-
- if (workFactor == 0) workFactor = 30;
- strm.bzalloc = NULL;
- strm.bzfree = NULL;
- strm.opaque = NULL;
- ret = BZ2_bzCompressInit ( &strm, blockSize100k,
- verbosity, workFactor );
- if (ret != BZ_OK) return ret;
-
- strm.next_in = source;
- strm.next_out = dest;
- strm.avail_in = sourceLen;
- strm.avail_out = *destLen;
-
- ret = BZ2_bzCompress ( &strm, BZ_FINISH );
- if (ret == BZ_FINISH_OK) goto output_overflow;
- if (ret != BZ_STREAM_END) goto errhandler;
-
- /* normal termination */
- *destLen -= strm.avail_out;
- BZ2_bzCompressEnd ( &strm );
- return BZ_OK;
-
- output_overflow:
- BZ2_bzCompressEnd ( &strm );
- return BZ_OUTBUFF_FULL;
-
- errhandler:
- BZ2_bzCompressEnd ( &strm );
- return ret;
-}
-
-/*---------------------------------------------------*/
-/*--
- Code contributed by Yoshioka Tsuneo (tsuneo@rr.iij4u.or.jp)
- to support better zlib compatibility.
- This code is not _officially_ part of libbzip2 (yet);
- I haven't tested it, documented it, or considered the
- threading-safeness of it.
- If this code breaks, please contact both Yoshioka and me.
---*/
-/*---------------------------------------------------*/
-
-/*---------------------------------------------------*/
-/*--
- return version like "0.9.5d, 4-Sept-1999".
---*/
-const char * BZ_API(BZ2_bzlibVersion)(void)
-{
- return BZ_VERSION;
-}
-
-
-#ifndef BZ_NO_STDIO
-/*---------------------------------------------------*/
-
-#if defined(_WIN32) || defined(OS2) || defined(MSDOS)
-# include <fcntl.h>
-# include <io.h>
-# define SET_BINARY_MODE(file) setmode(fileno(file),O_BINARY)
-#else
-# define SET_BINARY_MODE(file)
-#endif
-static
-BZFILE * bzopen_or_bzdopen
- ( const char *path, /* no use when bzdopen */
- int fd, /* no use when bzdopen */
- const char *mode,
- int open_mode) /* bzopen: 0, bzdopen:1 */
-{
- int bzerr;
- char unused[BZ_MAX_UNUSED];
- int blockSize100k = 9;
- int writing = 0;
- char mode2[10] = "";
- FILE *fp = NULL;
- BZFILE *bzfp = NULL;
- int verbosity = 0;
- int workFactor = 30;
- int smallMode = 0;
- int nUnused = 0;
-
- if (mode == NULL) return NULL;
- while (*mode) {
- switch (*mode) {
- case 'r':
- writing = 0; break;
- case 'w':
- writing = 1; break;
- case 's':
- smallMode = 1; break;
- default:
- if (isdigit((int)(*mode))) {
- blockSize100k = *mode-BZ_HDR_0;
- }
- }
- mode++;
- }
- strcat(mode2, writing ? "w" : "r" );
- strcat(mode2,"b"); /* binary mode */
-
- if (open_mode==0) {
- if (path==NULL || strcmp(path,"")==0) {
- fp = (writing ? stdout : stdin);
- SET_BINARY_MODE(fp);
- } else {
- fp = fopen(path,mode2);
- }
- } else {
-#ifdef BZ_STRICT_ANSI
- fp = NULL;
-#else
- fp = fdopen(fd,mode2);
-#endif
- }
- if (fp == NULL) return NULL;
-
- if (writing) {
- /* Guard against total chaos and anarchy -- JRS */
- if (blockSize100k < 1) blockSize100k = 1;
- if (blockSize100k > 9) blockSize100k = 9;
- bzfp = BZ2_bzWriteOpen(&bzerr,fp,blockSize100k,
- verbosity,workFactor);
- }
- if (bzfp == NULL) {
- if (fp != stdin && fp != stdout) fclose(fp);
- return NULL;
- }
- return bzfp;
-}
-
-
-/*---------------------------------------------------*/
-/*--
- open file for read or write.
- ex) bzopen("file","w9")
- case path="" or NULL => use stdin or stdout.
---*/
-BZFILE * BZ_API(BZ2_bzopen)
- ( const char *path,
- const char *mode )
-{
- return bzopen_or_bzdopen(path,-1,mode,/*bzopen*/0);
-}
-
-
-/*---------------------------------------------------*/
-BZFILE * BZ_API(BZ2_bzdopen)
- ( int fd,
- const char *mode )
-{
- return bzopen_or_bzdopen(NULL,fd,mode,/*bzdopen*/1);
-}
-
-/*---------------------------------------------------*/
-int BZ_API(BZ2_bzwrite) (BZFILE* b, void* buf, int len )
-{
- int bzerr;
-
- BZ2_bzWrite(&bzerr,b,buf,len);
- if(bzerr == BZ_OK){
- return len;
- }else{
- return -1;
- }
-}
-
-
-/*---------------------------------------------------*/
-int BZ_API(BZ2_bzflush) (BZFILE *b)
-{
- /* do nothing now... */
- return 0;
-}
-
-
-/*---------------------------------------------------*/
-void BZ_API(BZ2_bzclose) (BZFILE* b)
-{
- int bzerr;
- FILE *fp;
-
- if (b==NULL) {return;}
- fp = ((bzFile *)b)->handle;
- if(((bzFile*)b)->writing){
- BZ2_bzWriteClose(&bzerr,b,0,NULL,NULL);
- if(bzerr != BZ_OK){
- BZ2_bzWriteClose(NULL,b,1,NULL,NULL);
- }
- }
- if(fp!=stdin && fp!=stdout){
- fclose(fp);
- }
-}
-
-
-/*---------------------------------------------------*/
-/*--
- return last error code
---*/
-static const char *bzerrorstrings[] = {
- "OK"
- ,"SEQUENCE_ERROR"
- ,"PARAM_ERROR"
- ,"MEM_ERROR"
- ,"DATA_ERROR"
- ,"DATA_ERROR_MAGIC"
- ,"IO_ERROR"
- ,"UNEXPECTED_EOF"
- ,"OUTBUFF_FULL"
- ,"CONFIG_ERROR"
- ,"???" /* for future */
- ,"???" /* for future */
- ,"???" /* for future */
- ,"???" /* for future */
- ,"???" /* for future */
- ,"???" /* for future */
-};
-
-
-const char * BZ_API(BZ2_bzerror) (BZFILE *b, int *errnum)
-{
- int err = ((bzFile *)b)->lastErr;
-
- if(err>0) err = 0;
- *errnum = err;
- return bzerrorstrings[err*-1];
-}
-#endif
-
-/*-------------------------------------------------------------*/
-/*--- end bzlib.c ---*/
-/*-------------------------------------------------------------*/
+++ /dev/null
-
-/*-------------------------------------------------------------*/
-/*--- Private header file for the library. ---*/
-/*--- bzlib_private.h ---*/
-/*-------------------------------------------------------------*/
-
-/* ------------------------------------------------------------------
- This file is part of bzip2/libbzip2, a program and library for
- lossless, block-sorting data compression.
-
- bzip2/libbzip2 version 1.0.5 of 10 December 2007
- Copyright (C) 1996-2007 Julian Seward <jseward@bzip.org>
-
- Please read the WARNING, DISCLAIMER and PATENTS sections in the
- README file.
-
- This program is released under the terms of the license contained
- in the file LICENSE.
- ------------------------------------------------------------------ */
-
-
-#ifndef _BZLIB_PRIVATE_H
-#define _BZLIB_PRIVATE_H
-
-#include <stdlib.h>
-
-#ifndef BZ_NO_STDIO
-#include <stdio.h>
-#include <ctype.h>
-#include <string.h>
-#endif
-
-#include "bzlib.h"
-
-
-
-/*-- General stuff. --*/
-
-#define BZ_VERSION "1.0.5, 10-Dec-2007"
-
-typedef char Char;
-typedef unsigned char Bool;
-typedef unsigned char UChar;
-typedef int Int32;
-typedef unsigned int UInt32;
-typedef short Int16;
-typedef unsigned short UInt16;
-
-#define True ((Bool)1)
-#define False ((Bool)0)
-
-#ifndef __GNUC__
-#define __inline__ /* */
-#endif
-
-#ifndef BZ_NO_STDIO
-
-extern void BZ2_bz__AssertH__fail ( int errcode );
-#define AssertH(cond,errcode) \
- { if (!(cond)) BZ2_bz__AssertH__fail ( errcode ); }
-
-#if BZ_DEBUG
-#define AssertD(cond,msg) \
- { if (!(cond)) { \
- fprintf ( stderr, \
- "\n\nlibbzip2(debug build): internal error\n\t%s\n", msg );\
- exit(1); \
- }}
-#else
-#define AssertD(cond,msg) /* */
-#endif
-
-#define VPrintf0(zf) \
- fprintf(stderr,zf)
-#define VPrintf1(zf,za1) \
- fprintf(stderr,zf,za1)
-#define VPrintf2(zf,za1,za2) \
- fprintf(stderr,zf,za1,za2)
-#define VPrintf3(zf,za1,za2,za3) \
- fprintf(stderr,zf,za1,za2,za3)
-#define VPrintf4(zf,za1,za2,za3,za4) \
- fprintf(stderr,zf,za1,za2,za3,za4)
-#define VPrintf5(zf,za1,za2,za3,za4,za5) \
- fprintf(stderr,zf,za1,za2,za3,za4,za5)
-
-#else
-
-extern void bz_internal_error ( int errcode );
-#define AssertH(cond,errcode) \
- { if (!(cond)) bz_internal_error ( errcode ); }
-#define AssertD(cond,msg) do { } while (0)
-#define VPrintf0(zf) do { } while (0)
-#define VPrintf1(zf,za1) do { } while (0)
-#define VPrintf2(zf,za1,za2) do { } while (0)
-#define VPrintf3(zf,za1,za2,za3) do { } while (0)
-#define VPrintf4(zf,za1,za2,za3,za4) do { } while (0)
-#define VPrintf5(zf,za1,za2,za3,za4,za5) do { } while (0)
-
-#endif
-
-
-#define BZALLOC(nnn) (strm->bzalloc)(strm->opaque,(nnn),1)
-#define BZFREE(ppp) (strm->bzfree)(strm->opaque,(ppp))
-
-
-/*-- Header bytes. --*/
-
-#define BZ_HDR_B 0x42 /* 'B' */
-#define BZ_HDR_Z 0x5a /* 'Z' */
-#define BZ_HDR_h 0x68 /* 'h' */
-#define BZ_HDR_0 0x30 /* '0' */
-
-/*-- Constants for the back end. --*/
-
-#define BZ_MAX_ALPHA_SIZE 258
-#define BZ_MAX_CODE_LEN 23
-
-#define BZ_RUNA 0
-#define BZ_RUNB 1
-
-#define BZ_N_GROUPS 6
-#define BZ_G_SIZE 50
-#define BZ_N_ITERS 4
-
-#define BZ_MAX_SELECTORS (2 + (900000 / BZ_G_SIZE))
-
-
-
-/*-- Stuff for randomising repetitive blocks. --*/
-
-extern Int32 rNums[512];
-
-#define BZ_RAND_DECLS \
- Int32 rNToGo; \
- Int32 rTPos \
-
-#define BZ_RAND_INIT_MASK \
- s->rNToGo = 0; \
- s->rTPos = 0 \
-
-#define BZ_RAND_MASK ((s->rNToGo == 1) ? 1 : 0)
-
-#define BZ_RAND_UPD_MASK \
- if (s->rNToGo == 0) { \
- s->rNToGo = rNums[s->rTPos]; \
- s->rTPos++; \
- if (s->rTPos == 512) s->rTPos = 0; \
- } \
- s->rNToGo--;
-
-
-
-/*-- Stuff for doing CRCs. --*/
-
-extern UInt32 crc32Table[256];
-
-#define BZ_INITIALISE_CRC(crcVar) \
-{ \
- crcVar = 0xffffffffL; \
-}
-
-#define BZ_FINALISE_CRC(crcVar) \
-{ \
- crcVar = ~(crcVar); \
-}
-
-#define BZ_UPDATE_CRC(crcVar,cha) \
-{ \
- crcVar = (crcVar << 8) ^ \
- crc32Table[(crcVar >> 24) ^ \
- ((UChar)cha)]; \
-}
-
-
-
-/*-- States and modes for compression. --*/
-
-#define BZ_M_IDLE 1
-#define BZ_M_RUNNING 2
-#define BZ_M_FLUSHING 3
-#define BZ_M_FINISHING 4
-
-#define BZ_S_OUTPUT 1
-#define BZ_S_INPUT 2
-
-#define BZ_N_RADIX 2
-#define BZ_N_QSORT 12
-#define BZ_N_SHELL 18
-#define BZ_N_OVERSHOOT (BZ_N_RADIX + BZ_N_QSORT + BZ_N_SHELL + 2)
-
-
-
-
-/*-- Structure holding all the compression-side stuff. --*/
-
-typedef
- struct {
- /* pointer back to the struct bz_stream */
- bz_stream* strm;
-
- /* mode this stream is in, and whether inputting */
- /* or outputting data */
- Int32 mode;
- Int32 state;
-
- /* remembers avail_in when flush/finish requested */
- UInt32 avail_in_expect;
-
- /* for doing the block sorting */
- UInt32* arr1;
- UInt32* arr2;
- UInt32* ftab;
- Int32 origPtr;
-
- /* aliases for arr1 and arr2 */
- UInt32* ptr;
- UChar* block;
- UInt16* mtfv;
- UChar* zbits;
-
- /* for deciding when to use the fallback sorting algorithm */
- Int32 workFactor;
-
- /* run-length-encoding of the input */
- UInt32 state_in_ch;
- Int32 state_in_len;
- BZ_RAND_DECLS;
-
- /* input and output limits and current posns */
- Int32 nblock;
- Int32 nblockMAX;
- Int32 numZ;
- Int32 state_out_pos;
-
- /* map of bytes used in block */
- Int32 nInUse;
- Bool inUse[256];
- UChar unseqToSeq[256];
-
- /* the buffer for bit stream creation */
- UInt32 bsBuff;
- Int32 bsLive;
-
- /* block and combined CRCs */
- UInt32 blockCRC;
- UInt32 combinedCRC;
-
- /* misc administratium */
- Int32 verbosity;
- Int32 blockNo;
- Int32 blockSize100k;
-
- /* stuff for coding the MTF values */
- Int32 nMTF;
- Int32 mtfFreq [BZ_MAX_ALPHA_SIZE];
- UChar selector [BZ_MAX_SELECTORS];
- UChar selectorMtf[BZ_MAX_SELECTORS];
-
- UChar len [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
- Int32 code [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
- Int32 rfreq [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
- /* second dimension: only 3 needed; 4 makes index calculations faster */
- UInt32 len_pack[BZ_MAX_ALPHA_SIZE][4];
-
- }
- EState;
-
-
-
-/*-- externs for compression. --*/
-
-extern void
-BZ2_blockSort ( EState* );
-
-extern void
-BZ2_compressBlock ( EState*, Bool );
-
-extern void
-BZ2_bsInitWrite ( EState* );
-
-extern void
-BZ2_hbAssignCodes ( Int32*, UChar*, Int32, Int32, Int32 );
-
-extern void
-BZ2_hbMakeCodeLengths ( UChar*, Int32*, Int32, Int32 );
-
-
-
-/*-- states for decompression. --*/
-
-#define BZ_X_IDLE 1
-#define BZ_X_OUTPUT 2
-
-#define BZ_X_MAGIC_1 10
-#define BZ_X_MAGIC_2 11
-#define BZ_X_MAGIC_3 12
-#define BZ_X_MAGIC_4 13
-#define BZ_X_BLKHDR_1 14
-#define BZ_X_BLKHDR_2 15
-#define BZ_X_BLKHDR_3 16
-#define BZ_X_BLKHDR_4 17
-#define BZ_X_BLKHDR_5 18
-#define BZ_X_BLKHDR_6 19
-#define BZ_X_BCRC_1 20
-#define BZ_X_BCRC_2 21
-#define BZ_X_BCRC_3 22
-#define BZ_X_BCRC_4 23
-#define BZ_X_RANDBIT 24
-#define BZ_X_ORIGPTR_1 25
-#define BZ_X_ORIGPTR_2 26
-#define BZ_X_ORIGPTR_3 27
-#define BZ_X_MAPPING_1 28
-#define BZ_X_MAPPING_2 29
-#define BZ_X_SELECTOR_1 30
-#define BZ_X_SELECTOR_2 31
-#define BZ_X_SELECTOR_3 32
-#define BZ_X_CODING_1 33
-#define BZ_X_CODING_2 34
-#define BZ_X_CODING_3 35
-#define BZ_X_MTF_1 36
-#define BZ_X_MTF_2 37
-#define BZ_X_MTF_3 38
-#define BZ_X_MTF_4 39
-#define BZ_X_MTF_5 40
-#define BZ_X_MTF_6 41
-#define BZ_X_ENDHDR_2 42
-#define BZ_X_ENDHDR_3 43
-#define BZ_X_ENDHDR_4 44
-#define BZ_X_ENDHDR_5 45
-#define BZ_X_ENDHDR_6 46
-#define BZ_X_CCRC_1 47
-#define BZ_X_CCRC_2 48
-#define BZ_X_CCRC_3 49
-#define BZ_X_CCRC_4 50
-
-
-
-/*-- Constants for the fast MTF decoder. --*/
-
-#define MTFA_SIZE 4096
-#define MTFL_SIZE 16
-
-
-
-/*-- Structure holding all the decompression-side stuff. --*/
-
-typedef
- struct {
- /* pointer back to the struct bz_stream */
- bz_stream* strm;
-
- /* state indicator for this stream */
- Int32 state;
-
- /* for doing the final run-length decoding */
- UChar state_out_ch;
- Int32 state_out_len;
- Bool blockRandomised;
- BZ_RAND_DECLS;
-
- /* the buffer for bit stream reading */
- UInt32 bsBuff;
- Int32 bsLive;
-
- /* misc administratium */
- Int32 blockSize100k;
- Bool smallDecompress;
- Int32 currBlockNo;
- Int32 verbosity;
-
- /* for undoing the Burrows-Wheeler transform */
- Int32 origPtr;
- UInt32 tPos;
- Int32 k0;
- Int32 unzftab[256];
- Int32 nblock_used;
- Int32 cftab[257];
- Int32 cftabCopy[257];
-
- /* for undoing the Burrows-Wheeler transform (FAST) */
- UInt32 *tt;
-
- /* for undoing the Burrows-Wheeler transform (SMALL) */
- UInt16 *ll16;
- UChar *ll4;
-
- /* stored and calculated CRCs */
- UInt32 storedBlockCRC;
- UInt32 storedCombinedCRC;
- UInt32 calculatedBlockCRC;
- UInt32 calculatedCombinedCRC;
-
- /* map of bytes used in block */
- Int32 nInUse;
- Bool inUse[256];
- Bool inUse16[16];
- UChar seqToUnseq[256];
-
- /* for decoding the MTF values */
- UChar mtfa [MTFA_SIZE];
- Int32 mtfbase[256 / MTFL_SIZE];
- UChar selector [BZ_MAX_SELECTORS];
- UChar selectorMtf[BZ_MAX_SELECTORS];
- UChar len [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
-
- Int32 limit [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
- Int32 base [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
- Int32 perm [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
- Int32 minLens[BZ_N_GROUPS];
-
- /* save area for scalars in the main decompress code */
- Int32 save_i;
- Int32 save_j;
- Int32 save_t;
- Int32 save_alphaSize;
- Int32 save_nGroups;
- Int32 save_nSelectors;
- Int32 save_EOB;
- Int32 save_groupNo;
- Int32 save_groupPos;
- Int32 save_nextSym;
- Int32 save_nblockMAX;
- Int32 save_nblock;
- Int32 save_es;
- Int32 save_N;
- Int32 save_curr;
- Int32 save_zt;
- Int32 save_zn;
- Int32 save_zvec;
- Int32 save_zj;
- Int32 save_gSel;
- Int32 save_gMinlen;
- Int32* save_gLimit;
- Int32* save_gBase;
- Int32* save_gPerm;
-
- }
- DState;
-
-
-
-/*-- Macros for decompression. --*/
-
-#define BZ_GET_FAST(cccc) \
- /* c_tPos is unsigned, hence test < 0 is pointless. */ \
- if (s->tPos >= (UInt32)100000 * (UInt32)s->blockSize100k) return True; \
- s->tPos = s->tt[s->tPos]; \
- cccc = (UChar)(s->tPos & 0xff); \
- s->tPos >>= 8;
-
-#define BZ_GET_FAST_C(cccc) \
- /* c_tPos is unsigned, hence test < 0 is pointless. */ \
- if (c_tPos >= (UInt32)100000 * (UInt32)ro_blockSize100k) return True; \
- c_tPos = c_tt[c_tPos]; \
- cccc = (UChar)(c_tPos & 0xff); \
- c_tPos >>= 8;
-
-#define SET_LL4(i,n) \
- { if (((i) & 0x1) == 0) \
- s->ll4[(i) >> 1] = (s->ll4[(i) >> 1] & 0xf0) | (n); else \
- s->ll4[(i) >> 1] = (s->ll4[(i) >> 1] & 0x0f) | ((n) << 4); \
- }
-
-#define GET_LL4(i) \
- ((((UInt32)(s->ll4[(i) >> 1])) >> (((i) << 2) & 0x4)) & 0xF)
-
-#define SET_LL(i,n) \
- { s->ll16[i] = (UInt16)(n & 0x0000ffff); \
- SET_LL4(i, n >> 16); \
- }
-
-#define GET_LL(i) \
- (((UInt32)s->ll16[i]) | (GET_LL4(i) << 16))
-
-#define BZ_GET_SMALL(cccc) \
- /* c_tPos is unsigned, hence test < 0 is pointless. */ \
- if (s->tPos >= (UInt32)100000 * (UInt32)s->blockSize100k) return True; \
- cccc = BZ2_indexIntoF ( s->tPos, s->cftab ); \
- s->tPos = GET_LL(s->tPos);
-
-
-/*-- externs for decompression. --*/
-
-extern Int32
-BZ2_indexIntoF ( Int32, Int32* );
-
-extern void
-BZ2_hbCreateDecodeTables ( Int32*, Int32*, Int32*, UChar*,
- Int32, Int32, Int32 );
-
-
-#endif
-
-
-/*-- BZ_NO_STDIO seems to make NULL disappear on some platforms. --*/
-
-#ifdef BZ_NO_STDIO
-#ifndef NULL
-#define NULL 0
-#endif
-#endif
-
-
-/*-------------------------------------------------------------*/
-/*--- end bzlib_private.h ---*/
-/*-------------------------------------------------------------*/
+++ /dev/null
-
-/*-------------------------------------------------------------*/
-/*--- Compression machinery (not incl block sorting) ---*/
-/*--- compress.c ---*/
-/*-------------------------------------------------------------*/
-
-/* ------------------------------------------------------------------
- This file is part of bzip2/libbzip2, a program and library for
- lossless, block-sorting data compression.
-
- bzip2/libbzip2 version 1.0.5 of 10 December 2007
- Copyright (C) 1996-2007 Julian Seward <jseward@bzip.org>
-
- Please read the WARNING, DISCLAIMER and PATENTS sections in the
- README file.
-
- This program is released under the terms of the license contained
- in the file LICENSE.
- ------------------------------------------------------------------ */
-
-
-/* CHANGES
- 0.9.0 -- original version.
- 0.9.0a/b -- no changes in this file.
- 0.9.0c -- changed setting of nGroups in sendMTFValues()
- so as to do a bit better on small files
-*/
-
-#include "bzlib_private.h"
-
-
-/*---------------------------------------------------*/
-/*--- Bit stream I/O ---*/
-/*---------------------------------------------------*/
-
-/*---------------------------------------------------*/
-void BZ2_bsInitWrite ( EState* s )
-{
- s->bsLive = 0;
- s->bsBuff = 0;
-}
-
-
-/*---------------------------------------------------*/
-static
-void bsFinishWrite ( EState* s )
-{
- while (s->bsLive > 0) {
- s->zbits[s->numZ] = (UChar)(s->bsBuff >> 24);
- s->numZ++;
- s->bsBuff <<= 8;
- s->bsLive -= 8;
- }
-}
-
-
-/*---------------------------------------------------*/
-#define bsNEEDW(nz) \
-{ \
- while (s->bsLive >= 8) { \
- s->zbits[s->numZ] \
- = (UChar)(s->bsBuff >> 24); \
- s->numZ++; \
- s->bsBuff <<= 8; \
- s->bsLive -= 8; \
- } \
-}
-
-
-/*---------------------------------------------------*/
-static
-__inline__
-void bsW ( EState* s, Int32 n, UInt32 v )
-{
- bsNEEDW ( n );
- s->bsBuff |= (v << (32 - s->bsLive - n));
- s->bsLive += n;
-}
-
-
-/*---------------------------------------------------*/
-static
-void bsPutUInt32 ( EState* s, UInt32 u )
-{
- bsW ( s, 8, (u >> 24) & 0xffL );
- bsW ( s, 8, (u >> 16) & 0xffL );
- bsW ( s, 8, (u >> 8) & 0xffL );
- bsW ( s, 8, u & 0xffL );
-}
-
-
-/*---------------------------------------------------*/
-static
-void bsPutUChar ( EState* s, UChar c )
-{
- bsW( s, 8, (UInt32)c );
-}
-
-
-/*---------------------------------------------------*/
-/*--- The back end proper ---*/
-/*---------------------------------------------------*/
-
-/*---------------------------------------------------*/
-static
-void makeMaps_e ( EState* s )
-{
- Int32 i;
- s->nInUse = 0;
- for (i = 0; i < 256; i++)
- if (s->inUse[i]) {
- s->unseqToSeq[i] = s->nInUse;
- s->nInUse++;
- }
-}
-
-
-/*---------------------------------------------------*/
-static
-void generateMTFValues ( EState* s )
-{
- UChar yy[256];
- Int32 i, j;
- Int32 zPend;
- Int32 wr;
- Int32 EOB;
-
- /*
- After sorting (eg, here),
- s->arr1 [ 0 .. s->nblock-1 ] holds sorted order,
- and
- ((UChar*)s->arr2) [ 0 .. s->nblock-1 ]
- holds the original block data.
-
- The first thing to do is generate the MTF values,
- and put them in
- ((UInt16*)s->arr1) [ 0 .. s->nblock-1 ].
- Because there are strictly fewer or equal MTF values
- than block values, ptr values in this area are overwritten
- with MTF values only when they are no longer needed.
-
- The final compressed bitstream is generated into the
- area starting at
- (UChar*) (&((UChar*)s->arr2)[s->nblock])
-
- These storage aliases are set up in bzCompressInit(),
- except for the last one, which is arranged in
- compressBlock().
- */
- UInt32* ptr = s->ptr;
- UChar* block = s->block;
- UInt16* mtfv = s->mtfv;
-
- makeMaps_e ( s );
- EOB = s->nInUse+1;
-
- for (i = 0; i <= EOB; i++) s->mtfFreq[i] = 0;
-
- wr = 0;
- zPend = 0;
- for (i = 0; i < s->nInUse; i++) yy[i] = (UChar) i;
-
- for (i = 0; i < s->nblock; i++) {
- UChar ll_i;
- AssertD ( wr <= i, "generateMTFValues(1)" );
- j = ptr[i]-1; if (j < 0) j += s->nblock;
- ll_i = s->unseqToSeq[block[j]];
- AssertD ( ll_i < s->nInUse, "generateMTFValues(2a)" );
-
- if (yy[0] == ll_i) {
- zPend++;
- } else {
-
- if (zPend > 0) {
- zPend--;
- while (True) {
- if (zPend & 1) {
- mtfv[wr] = BZ_RUNB; wr++;
- s->mtfFreq[BZ_RUNB]++;
- } else {
- mtfv[wr] = BZ_RUNA; wr++;
- s->mtfFreq[BZ_RUNA]++;
- }
- if (zPend < 2) break;
- zPend = (zPend - 2) / 2;
- };
- zPend = 0;
- }
- {
- register UChar rtmp;
- register UChar* ryy_j;
- register UChar rll_i;
- rtmp = yy[1];
- yy[1] = yy[0];
- ryy_j = &(yy[1]);
- rll_i = ll_i;
- while ( rll_i != rtmp ) {
- register UChar rtmp2;
- ryy_j++;
- rtmp2 = rtmp;
- rtmp = *ryy_j;
- *ryy_j = rtmp2;
- };
- yy[0] = rtmp;
- j = ryy_j - &(yy[0]);
- mtfv[wr] = j+1; wr++; s->mtfFreq[j+1]++;
- }
-
- }
- }
-
- if (zPend > 0) {
- zPend--;
- while (True) {
- if (zPend & 1) {
- mtfv[wr] = BZ_RUNB; wr++;
- s->mtfFreq[BZ_RUNB]++;
- } else {
- mtfv[wr] = BZ_RUNA; wr++;
- s->mtfFreq[BZ_RUNA]++;
- }
- if (zPend < 2) break;
- zPend = (zPend - 2) / 2;
- };
- zPend = 0;
- }
-
- mtfv[wr] = EOB; wr++; s->mtfFreq[EOB]++;
-
- s->nMTF = wr;
-}
-
-
-/*---------------------------------------------------*/
-#define BZ_LESSER_ICOST 0
-#define BZ_GREATER_ICOST 15
-
-static
-void sendMTFValues ( EState* s )
-{
- Int32 v, t, i, j, gs, ge, totc, bt, bc, iter;
- Int32 nSelectors, alphaSize, minLen, maxLen, selCtr;
- Int32 nGroups, nBytes;
-
- /*--
- UChar len [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
- is a global since the decoder also needs it.
-
- Int32 code[BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
- Int32 rfreq[BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE];
- are also globals only used in this proc.
- Made global to keep stack frame size small.
- --*/
-
-
- UInt16 cost[BZ_N_GROUPS];
- Int32 fave[BZ_N_GROUPS];
-
- UInt16* mtfv = s->mtfv;
-
- if (s->verbosity >= 3)
- VPrintf3( " %d in block, %d after MTF & 1-2 coding, "
- "%d+2 syms in use\n",
- s->nblock, s->nMTF, s->nInUse );
-
- alphaSize = s->nInUse+2;
- for (t = 0; t < BZ_N_GROUPS; t++)
- for (v = 0; v < alphaSize; v++)
- s->len[t][v] = BZ_GREATER_ICOST;
-
- /*--- Decide how many coding tables to use ---*/
- AssertH ( s->nMTF > 0, 3001 );
- if (s->nMTF < 200) nGroups = 2; else
- if (s->nMTF < 600) nGroups = 3; else
- if (s->nMTF < 1200) nGroups = 4; else
- if (s->nMTF < 2400) nGroups = 5; else
- nGroups = 6;
-
- /*--- Generate an initial set of coding tables ---*/
- {
- Int32 nPart, remF, tFreq, aFreq;
-
- nPart = nGroups;
- remF = s->nMTF;
- gs = 0;
- while (nPart > 0) {
- tFreq = remF / nPart;
- ge = gs-1;
- aFreq = 0;
- while (aFreq < tFreq && ge < alphaSize-1) {
- ge++;
- aFreq += s->mtfFreq[ge];
- }
-
- if (ge > gs
- && nPart != nGroups && nPart != 1
- && ((nGroups-nPart) % 2 == 1)) {
- aFreq -= s->mtfFreq[ge];
- ge--;
- }
-
- if (s->verbosity >= 3)
- VPrintf5( " initial group %d, [%d .. %d], "
- "has %d syms (%4.1f%%)\n",
- nPart, gs, ge, aFreq,
- (100.0 * (float)aFreq) / (float)(s->nMTF) );
-
- for (v = 0; v < alphaSize; v++)
- if (v >= gs && v <= ge)
- s->len[nPart-1][v] = BZ_LESSER_ICOST; else
- s->len[nPart-1][v] = BZ_GREATER_ICOST;
-
- nPart--;
- gs = ge+1;
- remF -= aFreq;
- }
- }
-
- /*---
- Iterate up to BZ_N_ITERS times to improve the tables.
- ---*/
- for (iter = 0; iter < BZ_N_ITERS; iter++) {
-
- for (t = 0; t < nGroups; t++) fave[t] = 0;
-
- for (t = 0; t < nGroups; t++)
- for (v = 0; v < alphaSize; v++)
- s->rfreq[t][v] = 0;
-
- /*---
- Set up an auxiliary length table which is used to fast-track
- the common case (nGroups == 6).
- ---*/
- if (nGroups == 6) {
- for (v = 0; v < alphaSize; v++) {
- s->len_pack[v][0] = (s->len[1][v] << 16) | s->len[0][v];
- s->len_pack[v][1] = (s->len[3][v] << 16) | s->len[2][v];
- s->len_pack[v][2] = (s->len[5][v] << 16) | s->len[4][v];
- }
- }
-
- nSelectors = 0;
- totc = 0;
- gs = 0;
- while (True) {
-
- /*--- Set group start & end marks. --*/
- if (gs >= s->nMTF) break;
- ge = gs + BZ_G_SIZE - 1;
- if (ge >= s->nMTF) ge = s->nMTF-1;
-
- /*--
- Calculate the cost of this group as coded
- by each of the coding tables.
- --*/
- for (t = 0; t < nGroups; t++) cost[t] = 0;
-
- if (nGroups == 6 && 50 == ge-gs+1) {
- /*--- fast track the common case ---*/
- register UInt32 cost01, cost23, cost45;
- register UInt16 icv;
- cost01 = cost23 = cost45 = 0;
-
-# define BZ_ITER(nn) \
- icv = mtfv[gs+(nn)]; \
- cost01 += s->len_pack[icv][0]; \
- cost23 += s->len_pack[icv][1]; \
- cost45 += s->len_pack[icv][2]; \
-
- BZ_ITER(0); BZ_ITER(1); BZ_ITER(2); BZ_ITER(3); BZ_ITER(4);
- BZ_ITER(5); BZ_ITER(6); BZ_ITER(7); BZ_ITER(8); BZ_ITER(9);
- BZ_ITER(10); BZ_ITER(11); BZ_ITER(12); BZ_ITER(13); BZ_ITER(14);
- BZ_ITER(15); BZ_ITER(16); BZ_ITER(17); BZ_ITER(18); BZ_ITER(19);
- BZ_ITER(20); BZ_ITER(21); BZ_ITER(22); BZ_ITER(23); BZ_ITER(24);
- BZ_ITER(25); BZ_ITER(26); BZ_ITER(27); BZ_ITER(28); BZ_ITER(29);
- BZ_ITER(30); BZ_ITER(31); BZ_ITER(32); BZ_ITER(33); BZ_ITER(34);
- BZ_ITER(35); BZ_ITER(36); BZ_ITER(37); BZ_ITER(38); BZ_ITER(39);
- BZ_ITER(40); BZ_ITER(41); BZ_ITER(42); BZ_ITER(43); BZ_ITER(44);
- BZ_ITER(45); BZ_ITER(46); BZ_ITER(47); BZ_ITER(48); BZ_ITER(49);
-
-# undef BZ_ITER
-
- cost[0] = cost01 & 0xffff; cost[1] = cost01 >> 16;
- cost[2] = cost23 & 0xffff; cost[3] = cost23 >> 16;
- cost[4] = cost45 & 0xffff; cost[5] = cost45 >> 16;
-
- } else {
- /*--- slow version which correctly handles all situations ---*/
- for (i = gs; i <= ge; i++) {
- UInt16 icv = mtfv[i];
- for (t = 0; t < nGroups; t++) cost[t] += s->len[t][icv];
- }
- }
-
- /*--
- Find the coding table which is best for this group,
- and record its identity in the selector table.
- --*/
- bc = 999999999; bt = -1;
- for (t = 0; t < nGroups; t++)
- if (cost[t] < bc) { bc = cost[t]; bt = t; };
- totc += bc;
- fave[bt]++;
- s->selector[nSelectors] = bt;
- nSelectors++;
-
- /*--
- Increment the symbol frequencies for the selected table.
- --*/
- if (nGroups == 6 && 50 == ge-gs+1) {
- /*--- fast track the common case ---*/
-
-# define BZ_ITUR(nn) s->rfreq[bt][ mtfv[gs+(nn)] ]++
-
- BZ_ITUR(0); BZ_ITUR(1); BZ_ITUR(2); BZ_ITUR(3); BZ_ITUR(4);
- BZ_ITUR(5); BZ_ITUR(6); BZ_ITUR(7); BZ_ITUR(8); BZ_ITUR(9);
- BZ_ITUR(10); BZ_ITUR(11); BZ_ITUR(12); BZ_ITUR(13); BZ_ITUR(14);
- BZ_ITUR(15); BZ_ITUR(16); BZ_ITUR(17); BZ_ITUR(18); BZ_ITUR(19);
- BZ_ITUR(20); BZ_ITUR(21); BZ_ITUR(22); BZ_ITUR(23); BZ_ITUR(24);
- BZ_ITUR(25); BZ_ITUR(26); BZ_ITUR(27); BZ_ITUR(28); BZ_ITUR(29);
- BZ_ITUR(30); BZ_ITUR(31); BZ_ITUR(32); BZ_ITUR(33); BZ_ITUR(34);
- BZ_ITUR(35); BZ_ITUR(36); BZ_ITUR(37); BZ_ITUR(38); BZ_ITUR(39);
- BZ_ITUR(40); BZ_ITUR(41); BZ_ITUR(42); BZ_ITUR(43); BZ_ITUR(44);
- BZ_ITUR(45); BZ_ITUR(46); BZ_ITUR(47); BZ_ITUR(48); BZ_ITUR(49);
-
-# undef BZ_ITUR
-
- } else {
- /*--- slow version which correctly handles all situations ---*/
- for (i = gs; i <= ge; i++)
- s->rfreq[bt][ mtfv[i] ]++;
- }
-
- gs = ge+1;
- }
- if (s->verbosity >= 3) {
- VPrintf2 ( " pass %d: size is %d, grp uses are ",
- iter+1, totc/8 );
- for (t = 0; t < nGroups; t++)
- VPrintf1 ( "%d ", fave[t] );
- VPrintf0 ( "\n" );
- }
-
- /*--
- Recompute the tables based on the accumulated frequencies.
- --*/
- /* maxLen was changed from 20 to 17 in bzip2-1.0.3. See
- comment in huffman.c for details. */
- for (t = 0; t < nGroups; t++)
- BZ2_hbMakeCodeLengths ( &(s->len[t][0]), &(s->rfreq[t][0]),
- alphaSize, 17 /*20*/ );
- }
-
-
- AssertH( nGroups < 8, 3002 );
- AssertH( nSelectors < 32768 &&
- nSelectors <= (2 + (900000 / BZ_G_SIZE)),
- 3003 );
-
-
- /*--- Compute MTF values for the selectors. ---*/
- {
- UChar pos[BZ_N_GROUPS], ll_i, tmp2, tmp;
- for (i = 0; i < nGroups; i++) pos[i] = i;
- for (i = 0; i < nSelectors; i++) {
- ll_i = s->selector[i];
- j = 0;
- tmp = pos[j];
- while ( ll_i != tmp ) {
- j++;
- tmp2 = tmp;
- tmp = pos[j];
- pos[j] = tmp2;
- };
- pos[0] = tmp;
- s->selectorMtf[i] = j;
- }
- };
-
- /*--- Assign actual codes for the tables. --*/
- for (t = 0; t < nGroups; t++) {
- minLen = 32;
- maxLen = 0;
- for (i = 0; i < alphaSize; i++) {
- if (s->len[t][i] > maxLen) maxLen = s->len[t][i];
- if (s->len[t][i] < minLen) minLen = s->len[t][i];
- }
- AssertH ( !(maxLen > 17 /*20*/ ), 3004 );
- AssertH ( !(minLen < 1), 3005 );
- BZ2_hbAssignCodes ( &(s->code[t][0]), &(s->len[t][0]),
- minLen, maxLen, alphaSize );
- }
-
- /*--- Transmit the mapping table. ---*/
- {
- Bool inUse16[16];
- for (i = 0; i < 16; i++) {
- inUse16[i] = False;
- for (j = 0; j < 16; j++)
- if (s->inUse[i * 16 + j]) inUse16[i] = True;
- }
-
- nBytes = s->numZ;
- for (i = 0; i < 16; i++)
- if (inUse16[i]) bsW(s,1,1); else bsW(s,1,0);
-
- for (i = 0; i < 16; i++)
- if (inUse16[i])
- for (j = 0; j < 16; j++) {
- if (s->inUse[i * 16 + j]) bsW(s,1,1); else bsW(s,1,0);
- }
-
- if (s->verbosity >= 3)
- VPrintf1( " bytes: mapping %d, ", s->numZ-nBytes );
- }
-
- /*--- Now the selectors. ---*/
- nBytes = s->numZ;
- bsW ( s, 3, nGroups );
- bsW ( s, 15, nSelectors );
- for (i = 0; i < nSelectors; i++) {
- for (j = 0; j < s->selectorMtf[i]; j++) bsW(s,1,1);
- bsW(s,1,0);
- }
- if (s->verbosity >= 3)
- VPrintf1( "selectors %d, ", s->numZ-nBytes );
-
- /*--- Now the coding tables. ---*/
- nBytes = s->numZ;
-
- for (t = 0; t < nGroups; t++) {
- Int32 curr = s->len[t][0];
- bsW ( s, 5, curr );
- for (i = 0; i < alphaSize; i++) {
- while (curr < s->len[t][i]) { bsW(s,2,2); curr++; /* 10 */ };
- while (curr > s->len[t][i]) { bsW(s,2,3); curr--; /* 11 */ };
- bsW ( s, 1, 0 );
- }
- }
-
- if (s->verbosity >= 3)
- VPrintf1 ( "code lengths %d, ", s->numZ-nBytes );
-
- /*--- And finally, the block data proper ---*/
- nBytes = s->numZ;
- selCtr = 0;
- gs = 0;
- while (True) {
- if (gs >= s->nMTF) break;
- ge = gs + BZ_G_SIZE - 1;
- if (ge >= s->nMTF) ge = s->nMTF-1;
- AssertH ( s->selector[selCtr] < nGroups, 3006 );
-
- if (nGroups == 6 && 50 == ge-gs+1) {
- /*--- fast track the common case ---*/
- UInt16 mtfv_i;
- UChar* s_len_sel_selCtr
- = &(s->len[s->selector[selCtr]][0]);
- Int32* s_code_sel_selCtr
- = &(s->code[s->selector[selCtr]][0]);
-
-# define BZ_ITAH(nn) \
- mtfv_i = mtfv[gs+(nn)]; \
- bsW ( s, \
- s_len_sel_selCtr[mtfv_i], \
- s_code_sel_selCtr[mtfv_i] )
-
- BZ_ITAH(0); BZ_ITAH(1); BZ_ITAH(2); BZ_ITAH(3); BZ_ITAH(4);
- BZ_ITAH(5); BZ_ITAH(6); BZ_ITAH(7); BZ_ITAH(8); BZ_ITAH(9);
- BZ_ITAH(10); BZ_ITAH(11); BZ_ITAH(12); BZ_ITAH(13); BZ_ITAH(14);
- BZ_ITAH(15); BZ_ITAH(16); BZ_ITAH(17); BZ_ITAH(18); BZ_ITAH(19);
- BZ_ITAH(20); BZ_ITAH(21); BZ_ITAH(22); BZ_ITAH(23); BZ_ITAH(24);
- BZ_ITAH(25); BZ_ITAH(26); BZ_ITAH(27); BZ_ITAH(28); BZ_ITAH(29);
- BZ_ITAH(30); BZ_ITAH(31); BZ_ITAH(32); BZ_ITAH(33); BZ_ITAH(34);
- BZ_ITAH(35); BZ_ITAH(36); BZ_ITAH(37); BZ_ITAH(38); BZ_ITAH(39);
- BZ_ITAH(40); BZ_ITAH(41); BZ_ITAH(42); BZ_ITAH(43); BZ_ITAH(44);
- BZ_ITAH(45); BZ_ITAH(46); BZ_ITAH(47); BZ_ITAH(48); BZ_ITAH(49);
-
-# undef BZ_ITAH
-
- } else {
- /*--- slow version which correctly handles all situations ---*/
- for (i = gs; i <= ge; i++) {
- bsW ( s,
- s->len [s->selector[selCtr]] [mtfv[i]],
- s->code [s->selector[selCtr]] [mtfv[i]] );
- }
- }
-
-
- gs = ge+1;
- selCtr++;
- }
- AssertH( selCtr == nSelectors, 3007 );
-
- if (s->verbosity >= 3)
- VPrintf1( "codes %d\n", s->numZ-nBytes );
-}
-
-
-/*---------------------------------------------------*/
-void BZ2_compressBlock ( EState* s, Bool is_last_block )
-{
- if (s->nblock > 0) {
-
- BZ_FINALISE_CRC ( s->blockCRC );
- s->combinedCRC = (s->combinedCRC << 1) | (s->combinedCRC >> 31);
- s->combinedCRC ^= s->blockCRC;
- if (s->blockNo > 1) s->numZ = 0;
-
- if (s->verbosity >= 2)
- VPrintf4( " block %d: crc = 0x%08x, "
- "combined CRC = 0x%08x, size = %d\n",
- s->blockNo, s->blockCRC, s->combinedCRC, s->nblock );
-
- BZ2_blockSort ( s );
- }
-
- s->zbits = (UChar*) (&((UChar*)s->arr2)[s->nblock]);
-
- /*-- If this is the first block, create the stream header. --*/
- if (s->blockNo == 1) {
- BZ2_bsInitWrite ( s );
- bsPutUChar ( s, BZ_HDR_B );
- bsPutUChar ( s, BZ_HDR_Z );
- bsPutUChar ( s, BZ_HDR_h );
- bsPutUChar ( s, (UChar)(BZ_HDR_0 + s->blockSize100k) );
- }
-
- if (s->nblock > 0) {
-
- bsPutUChar ( s, 0x31 ); bsPutUChar ( s, 0x41 );
- bsPutUChar ( s, 0x59 ); bsPutUChar ( s, 0x26 );
- bsPutUChar ( s, 0x53 ); bsPutUChar ( s, 0x59 );
-
- /*-- Now the block's CRC, so it is in a known place. --*/
- bsPutUInt32 ( s, s->blockCRC );
-
- /*--
- Now a single bit indicating (non-)randomisation.
- As of version 0.9.5, we use a better sorting algorithm
- which makes randomisation unnecessary. So always set
- the randomised bit to 'no'. Of course, the decoder
- still needs to be able to handle randomised blocks
- so as to maintain backwards compatibility with
- older versions of bzip2.
- --*/
- bsW(s,1,0);
-
- bsW ( s, 24, s->origPtr );
- generateMTFValues ( s );
- sendMTFValues ( s );
- }
-
-
- /*-- If this is the last block, add the stream trailer. --*/
- if (is_last_block) {
-
- bsPutUChar ( s, 0x17 ); bsPutUChar ( s, 0x72 );
- bsPutUChar ( s, 0x45 ); bsPutUChar ( s, 0x38 );
- bsPutUChar ( s, 0x50 ); bsPutUChar ( s, 0x90 );
- bsPutUInt32 ( s, s->combinedCRC );
- if (s->verbosity >= 2)
- VPrintf1( " final combined CRC = 0x%08x\n ", s->combinedCRC );
- bsFinishWrite ( s );
- }
-}
-
-
-/*-------------------------------------------------------------*/
-/*--- end compress.c ---*/
-/*-------------------------------------------------------------*/
+++ /dev/null
-
-/*-------------------------------------------------------------*/
-/*--- Huffman coding low-level stuff ---*/
-/*--- huffman.c ---*/
-/*-------------------------------------------------------------*/
-
-/* ------------------------------------------------------------------
- This file is part of bzip2/libbzip2, a program and library for
- lossless, block-sorting data compression.
-
- bzip2/libbzip2 version 1.0.5 of 10 December 2007
- Copyright (C) 1996-2007 Julian Seward <jseward@bzip.org>
-
- Please read the WARNING, DISCLAIMER and PATENTS sections in the
- README file.
-
- This program is released under the terms of the license contained
- in the file LICENSE.
- ------------------------------------------------------------------ */
-
-
-#include "bzlib_private.h"
-
-/*---------------------------------------------------*/
-#define WEIGHTOF(zz0) ((zz0) & 0xffffff00)
-#define DEPTHOF(zz1) ((zz1) & 0x000000ff)
-#define MYMAX(zz2,zz3) ((zz2) > (zz3) ? (zz2) : (zz3))
-
-#define ADDWEIGHTS(zw1,zw2) \
- (WEIGHTOF(zw1)+WEIGHTOF(zw2)) | \
- (1 + MYMAX(DEPTHOF(zw1),DEPTHOF(zw2)))
-
-#define UPHEAP(z) \
-{ \
- Int32 zz, tmp; \
- zz = z; tmp = heap[zz]; \
- while (weight[tmp] < weight[heap[zz >> 1]]) { \
- heap[zz] = heap[zz >> 1]; \
- zz >>= 1; \
- } \
- heap[zz] = tmp; \
-}
-
-#define DOWNHEAP(z) \
-{ \
- Int32 zz, yy, tmp; \
- zz = z; tmp = heap[zz]; \
- while (True) { \
- yy = zz << 1; \
- if (yy > nHeap) break; \
- if (yy < nHeap && \
- weight[heap[yy+1]] < weight[heap[yy]]) \
- yy++; \
- if (weight[tmp] < weight[heap[yy]]) break; \
- heap[zz] = heap[yy]; \
- zz = yy; \
- } \
- heap[zz] = tmp; \
-}
-
-
-/*---------------------------------------------------*/
-void BZ2_hbMakeCodeLengths ( UChar *len,
- Int32 *freq,
- Int32 alphaSize,
- Int32 maxLen )
-{
- /*--
- Nodes and heap entries run from 1. Entry 0
- for both the heap and nodes is a sentinel.
- --*/
- Int32 nNodes, nHeap, n1, n2, i, j, k;
- Bool tooLong;
-
- Int32 heap [ BZ_MAX_ALPHA_SIZE + 2 ];
- Int32 weight [ BZ_MAX_ALPHA_SIZE * 2 ];
- Int32 parent [ BZ_MAX_ALPHA_SIZE * 2 ];
-
- for (i = 0; i < alphaSize; i++)
- weight[i+1] = (freq[i] == 0 ? 1 : freq[i]) << 8;
-
- while (True) {
-
- nNodes = alphaSize;
- nHeap = 0;
-
- heap[0] = 0;
- weight[0] = 0;
- parent[0] = -2;
-
- for (i = 1; i <= alphaSize; i++) {
- parent[i] = -1;
- nHeap++;
- heap[nHeap] = i;
- UPHEAP(nHeap);
- }
-
- AssertH( nHeap < (BZ_MAX_ALPHA_SIZE+2), 2001 );
-
- while (nHeap > 1) {
- n1 = heap[1]; heap[1] = heap[nHeap]; nHeap--; DOWNHEAP(1);
- n2 = heap[1]; heap[1] = heap[nHeap]; nHeap--; DOWNHEAP(1);
- nNodes++;
- parent[n1] = parent[n2] = nNodes;
- weight[nNodes] = ADDWEIGHTS(weight[n1], weight[n2]);
- parent[nNodes] = -1;
- nHeap++;
- heap[nHeap] = nNodes;
- UPHEAP(nHeap);
- }
-
- AssertH( nNodes < (BZ_MAX_ALPHA_SIZE * 2), 2002 );
-
- tooLong = False;
- for (i = 1; i <= alphaSize; i++) {
- j = 0;
- k = i;
- while (parent[k] >= 0) { k = parent[k]; j++; }
- len[i-1] = j;
- if (j > maxLen) tooLong = True;
- }
-
- if (! tooLong) break;
-
- /* 17 Oct 04: keep-going condition for the following loop used
- to be 'i < alphaSize', which missed the last element,
- theoretically leading to the possibility of the compressor
- looping. However, this count-scaling step is only needed if
- one of the generated Huffman code words is longer than
- maxLen, which up to and including version 1.0.2 was 20 bits,
- which is extremely unlikely. In version 1.0.3 maxLen was
- changed to 17 bits, which has minimal effect on compression
- ratio, but does mean this scaling step is used from time to
- time, enough to verify that it works.
-
- This means that bzip2-1.0.3 and later will only produce
- Huffman codes with a maximum length of 17 bits. However, in
- order to preserve backwards compatibility with bitstreams
- produced by versions pre-1.0.3, the decompressor must still
- handle lengths of up to 20. */
-
- for (i = 1; i <= alphaSize; i++) {
- j = weight[i] >> 8;
- j = 1 + (j / 2);
- weight[i] = j << 8;
- }
- }
-}
-
-
-/*---------------------------------------------------*/
-void BZ2_hbAssignCodes ( Int32 *code,
- UChar *length,
- Int32 minLen,
- Int32 maxLen,
- Int32 alphaSize )
-{
- Int32 n, vec, i;
-
- vec = 0;
- for (n = minLen; n <= maxLen; n++) {
- for (i = 0; i < alphaSize; i++)
- if (length[i] == n) { code[i] = vec; vec++; };
- vec <<= 1;
- }
-}
-
-
-/*---------------------------------------------------*/
-void BZ2_hbCreateDecodeTables ( Int32 *limit,
- Int32 *base,
- Int32 *perm,
- UChar *length,
- Int32 minLen,
- Int32 maxLen,
- Int32 alphaSize )
-{
- Int32 pp, i, j, vec;
-
- pp = 0;
- for (i = minLen; i <= maxLen; i++)
- for (j = 0; j < alphaSize; j++)
- if (length[j] == i) { perm[pp] = j; pp++; };
-
- for (i = 0; i < BZ_MAX_CODE_LEN; i++) base[i] = 0;
- for (i = 0; i < alphaSize; i++) base[length[i]+1]++;
-
- for (i = 1; i < BZ_MAX_CODE_LEN; i++) base[i] += base[i-1];
-
- for (i = 0; i < BZ_MAX_CODE_LEN; i++) limit[i] = 0;
- vec = 0;
-
- for (i = minLen; i <= maxLen; i++) {
- vec += (base[i+1] - base[i]);
- limit[i] = vec-1;
- vec <<= 1;
- }
- for (i = minLen + 1; i <= maxLen; i++)
- base[i] = ((limit[i-1] + 1) << 1) - base[i];
-}
-
-
-/*-------------------------------------------------------------*/
-/*--- end huffman.c ---*/
-/*-------------------------------------------------------------*/
projects / tools / pristine-tar.git / commitdiff