tif_lzw.c

一个国人自己实现图像库的程序（有参考价值）

	 */
	if (sp->dec_restart) {
		long residue;

		codep = sp->dec_codep;
		residue = codep->length - sp->dec_restart;
		if (residue > occ) {
			/*
			 * Residue from previous decode is sufficient
			 * to satisfy decode request.  Skip to the
			 * start of the decoded string, place decoded
			 * values in the output buffer, and return.
			 */
			sp->dec_restart += occ;
			do {
				codep = codep->next;
			} while (--residue > occ);
			tp = op + occ;
			do {
				*--tp = codep->value;
				codep = codep->next;
			} while (--occ);
			return (1);
		}
		/*
		 * Residue satisfies only part of the decode request.
		 */
		op += residue, occ -= residue;
		tp = op;
		do {
			*--tp = codep->value;
			codep = codep->next;
		} while (--residue);
		sp->dec_restart = 0;
	}

	bp = (u_char *)tif->tif_rawcp;
	nbits = sp->lzw_nbits;
	nextdata = sp->lzw_nextdata;
	nextbits = sp->lzw_nextbits;
	nbitsmask = sp->dec_nbitsmask;
	oldcodep = sp->dec_oldcodep;
	free_entp = sp->dec_free_entp;
	maxcodep = sp->dec_maxcodep;

	while (occ > 0) {
		NextCode(tif, sp, bp, code, GetNextCodeCompat);
		if (code == CODE_EOI)
			break;
		if (code == CODE_CLEAR) {
			free_entp = sp->dec_codetab + CODE_FIRST;
			nbits = BITS_MIN;
			nbitsmask = MAXCODE(BITS_MIN);
			maxcodep = sp->dec_codetab + nbitsmask;
			NextCode(tif, sp, bp, code, GetNextCodeCompat);
			if (code == CODE_EOI)
				break;
			*op++ = code, occ--;
			oldcodep = sp->dec_codetab + code;
			continue;
		}
		codep = sp->dec_codetab + code;

		/*
	 	 * Add the new entry to the code table.
	 	 */
		assert(&sp->dec_codetab[0] <= free_entp && free_entp < &sp->dec_codetab[CSIZE]);
		free_entp->next = oldcodep;
		free_entp->firstchar = free_entp->next->firstchar;
		free_entp->length = free_entp->next->length+1;
		free_entp->value = (codep < free_entp) ?
		    codep->firstchar : free_entp->firstchar;
		if (++free_entp > maxcodep) {
			if (++nbits > BITS_MAX)		/* should not happen */
				nbits = BITS_MAX;
			nbitsmask = MAXCODE(nbits);
			maxcodep = sp->dec_codetab + nbitsmask;
		}
		oldcodep = codep;
		if (code >= 256) {
			/*
		 	 * Code maps to a string, copy string
			 * value to output (written in reverse).
		 	 */
			if (codep->length > occ) {
				/*
				 * String is too long for decode buffer,
				 * locate portion that will fit, copy to
				 * the decode buffer, and setup restart
				 * logic for the next decoding call.
				 */
				sp->dec_codep = codep;
				do {
					codep = codep->next;
				} while (codep->length > occ);
				sp->dec_restart = occ;
				tp = op + occ;
				do  {
					*--tp = codep->value;
					codep = codep->next;
				}  while (--occ);
				break;
			}
			op += codep->length, occ -= codep->length;
			tp = op;
			do {
				*--tp = codep->value;
			} while( (codep = codep->next) != NULL);
		} else
			*op++ = code, occ--;
	}

	tif->tif_rawcp = (tidata_t) bp;
	sp->lzw_nbits = nbits;
	sp->lzw_nextdata = nextdata;
	sp->lzw_nextbits = nextbits;
	sp->dec_nbitsmask = nbitsmask;
	sp->dec_oldcodep = oldcodep;
	sp->dec_free_entp = free_entp;
	sp->dec_maxcodep = maxcodep;

	if (occ > 0) {
		TIFFError(tif->tif_name,
	    "LZWDecodeCompat: Not enough data at scanline %d (short %d bytes)",
		    tif->tif_row, occ);
		return (0);
	}
	return (1);
}
#endif /* LZW_COMPAT */

/*
 * LZW Encoding.
 */

static int
LZWSetupEncode(TIFF* tif)
{
	LZWEncodeState* sp = EncoderState(tif);
	static const char module[] = "LZWSetupEncode";

	assert(sp != NULL);
	sp->enc_hashtab = (hash_t*) _TIFFmalloc(HSIZE*sizeof (hash_t));
	if (sp->enc_hashtab == NULL) {
		TIFFError(module, "No space for LZW hash table");
		return (0);
	}
	return (1);
}

/*
 * Reset encoding state at the start of a strip.
 */
static int
LZWPreEncode(TIFF* tif, tsample_t s)
{
	LZWEncodeState *sp = EncoderState(tif);

	(void) s;
	assert(sp != NULL);
	sp->lzw_nbits = BITS_MIN;
	sp->lzw_maxcode = MAXCODE(BITS_MIN);
	sp->lzw_free_ent = CODE_FIRST;
	sp->lzw_nextbits = 0;
	sp->lzw_nextdata = 0;
	sp->enc_checkpoint = CHECK_GAP;
	sp->enc_ratio = 0;
	sp->enc_incount = 0;
	sp->enc_outcount = 0;
	/*
	 * The 4 here insures there is space for 2 max-sized
	 * codes in LZWEncode and LZWPostDecode.
	 */
	sp->enc_rawlimit = tif->tif_rawdata + tif->tif_rawdatasize-1 - 4;
	cl_hash(sp);		/* clear hash table */
	sp->enc_oldcode = (hcode_t) -1;	/* generates CODE_CLEAR in LZWEncode */
	return (1);
}

#define	CALCRATIO(sp, rat) {					\
	if (incount > 0x007fffff) { /* NB: shift will overflow */\
		rat = outcount >> 8;				\
		rat = (rat == 0 ? 0x7fffffff : incount/rat);	\
	} else							\
		rat = (incount<<8) / outcount;			\
}
#define	PutNextCode(op, c) {					\
	nextdata = (nextdata << nbits) | c;			\
	nextbits += nbits;					\
	*op++ = (u_char)(nextdata >> (nextbits-8));		\
	nextbits -= 8;						\
	if (nextbits >= 8) {					\
		*op++ = (u_char)(nextdata >> (nextbits-8));	\
		nextbits -= 8;					\
	}							\
	outcount += nbits;					\
}

/*
 * Encode a chunk of pixels.
 *
 * Uses an open addressing double hashing (no chaining) on the 
 * prefix code/next character combination.  We do a variant of
 * Knuth's algorithm D (vol. 3, sec. 6.4) along with G. Knott's
 * relatively-prime secondary probe.  Here, the modular division
 * first probe is gives way to a faster exclusive-or manipulation. 
 * Also do block compression with an adaptive reset, whereby the
 * code table is cleared when the compression ratio decreases,
 * but after the table fills.  The variable-length output codes
 * are re-sized at this point, and a CODE_CLEAR is generated
 * for the decoder. 
 */
static int
LZWEncode(TIFF* tif, tidata_t bp, tsize_t cc, tsample_t s)
{
	register LZWEncodeState *sp = EncoderState(tif);
	register long fcode;
	register hash_t *hp;
	register int h, c;
	hcode_t ent;
	long disp;
	long incount, outcount, checkpoint;
	long nextdata, nextbits;
	int free_ent, maxcode, nbits;
	tidata_t op, limit;

	(void) s;
	if (sp == NULL)
		return (0);
	/*
	 * Load local state.
	 */
	incount = sp->enc_incount;
	outcount = sp->enc_outcount;
	checkpoint = sp->enc_checkpoint;
	nextdata = sp->lzw_nextdata;
	nextbits = sp->lzw_nextbits;
	free_ent = sp->lzw_free_ent;
	maxcode = sp->lzw_maxcode;
	nbits = sp->lzw_nbits;
	op = tif->tif_rawcp;
	limit = sp->enc_rawlimit;
	ent = sp->enc_oldcode;

	if (ent == (hcode_t) -1 && cc > 0) {
		/*
		 * NB: This is safe because it can only happen
		 *     at the start of a strip where we know there
		 *     is space in the data buffer.
		 */
		PutNextCode(op, CODE_CLEAR);
		ent = *bp++; cc--; incount++;
	}
	while (cc > 0) {
		c = *bp++; cc--; incount++;
		fcode = ((long)c << BITS_MAX) + ent;
		h = (c << HSHIFT) ^ ent;	/* xor hashing */
#ifdef _WINDOWS
		/*
		 * Check hash index for an overflow.
		 */
		if (h >= HSIZE)
			h -= HSIZE;
#endif
		hp = &sp->enc_hashtab[h];
		if (hp->hash == fcode) {
			ent = hp->code;
			continue;
		}
		if (hp->hash >= 0) {
			/*
			 * Primary hash failed, check secondary hash.
			 */
			disp = HSIZE - h;
			if (h == 0)
				disp = 1;
			do {
				/*
				 * Avoid pointer arithmetic 'cuz of
				 * wraparound problems with segments.
				 */
				if ((h -= disp) < 0)
					h += HSIZE;
				hp = &sp->enc_hashtab[h];
				if (hp->hash == fcode) {
					ent = hp->code;
					goto hit;
				}
			} while (hp->hash >= 0);
		}
		/*
		 * New entry, emit code and add to table.
		 */
		/*
		 * Verify there is space in the buffer for the code
		 * and any potential Clear code that might be emitted
		 * below.  The value of limit is setup so that there
		 * are at least 4 bytes free--room for 2 codes.
		 */
		if (op > limit) {
			tif->tif_rawcc = (tsize_t)(op - tif->tif_rawdata);
			TIFFFlushData1(tif);
			op = tif->tif_rawdata;
		}
		PutNextCode(op, ent);
		ent = c;
		hp->code = free_ent++;
		hp->hash = fcode;
		if (free_ent == CODE_MAX-1) {
			/* table is full, emit clear code and reset */
			cl_hash(sp);
			sp->enc_ratio = 0;
			incount = 0;
			outcount = 0;
			free_ent = CODE_FIRST;
			PutNextCode(op, CODE_CLEAR);
			nbits = BITS_MIN;
			maxcode = MAXCODE(BITS_MIN);
		} else {
			/*
			 * If the next entry is going to be too big for
			 * the code size, then increase it, if possible.
			 */
			if (free_ent > maxcode) {
				nbits++;
				assert(nbits <= BITS_MAX);
				maxcode = (int) MAXCODE(nbits);
			} else if (incount >= checkpoint) {
				long rat;
				/*
				 * Check compression ratio and, if things seem
				 * to be slipping, clear the hash table and
				 * reset state.  The compression ratio is a
				 * 24+8-bit fractional number.
				 */
				checkpoint = incount+CHECK_GAP;
				CALCRATIO(sp, rat);
				if (rat <= sp->enc_ratio) {
					cl_hash(sp);
					sp->enc_ratio = 0;
					incount = 0;
					outcount = 0;
					free_ent = CODE_FIRST;
					PutNextCode(op, CODE_CLEAR);
					nbits = BITS_MIN;
					maxcode = MAXCODE(BITS_MIN);
				} else
					sp->enc_ratio = rat;
			}
		}
	hit:
		;
	}

	/*
	 * Restore global state.
	 */
	sp->enc_incount = incount;
	sp->enc_outcount = outcount;
	sp->enc_checkpoint = checkpoint;
	sp->enc_oldcode = ent;
	sp->lzw_nextdata = nextdata;
	sp->lzw_nextbits = nextbits;
	sp->lzw_free_ent = free_ent;
	sp->lzw_maxcode = maxcode;
	sp->lzw_nbits = nbits;
	tif->tif_rawcp = op;
	return (1);
}

/*
 * Finish off an encoded strip by flushing the last
 * string and tacking on an End Of Information code.
 */
static int
LZWPostEncode(TIFF* tif)
{
	register LZWEncodeState *sp = EncoderState(tif);
	tidata_t op = tif->tif_rawcp;
	long nextbits = sp->lzw_nextbits;
	long nextdata = sp->lzw_nextdata;
	long outcount = sp->enc_outcount;
	int nbits = sp->lzw_nbits;

	if (op > sp->enc_rawlimit) {
		tif->tif_rawcc = (tsize_t)(op - tif->tif_rawdata);
		TIFFFlushData1(tif);
		op = tif->tif_rawdata;
	}
	if (sp->enc_oldcode != (hcode_t) -1) {
		PutNextCode(op, sp->enc_oldcode);
		sp->enc_oldcode = (hcode_t) -1;
	}
	PutNextCode(op, CODE_EOI);
	if (nextbits > 0) 
		*op++ = (u_char)(nextdata << (8-nextbits));
	tif->tif_rawcc = (tsize_t)(op - tif->tif_rawdata);
	return (1);
}

/*
 * Reset encoding hash table.
 */
static void
cl_hash(LZWEncodeState* sp)
{
	register hash_t *hp = &sp->enc_hashtab[HSIZE-1];
	register long i = HSIZE-8;

 	do {
		i -= 8;
		hp[-7].hash = -1;
		hp[-6].hash = -1;
		hp[-5].hash = -1;
		hp[-4].hash = -1;
		hp[-3].hash = -1;
		hp[-2].hash = -1;
		hp[-1].hash = -1;
		hp[ 0].hash = -1;
		hp -= 8;
	} while (i >= 0);
    	for (i += 8; i > 0; i--, hp--)
		hp->hash = -1;
}

static void
LZWCleanup(TIFF* tif)
{
	if (tif->tif_data) {
		if (LZWState(tif)->rw_mode == O_RDONLY) {
			if (DecoderState(tif)->dec_codetab)
				_TIFFfree(DecoderState(tif)->dec_codetab);
		} else {
			if (EncoderState(tif)->enc_hashtab)
				_TIFFfree(EncoderState(tif)->enc_hashtab);
		}
		_TIFFfree(tif->tif_data);
		tif->tif_data = NULL;
	}
}

int
TIFFInitLZW(TIFF* tif, int scheme)
{
	assert(scheme == COMPRESSION_LZW);
	/*
	 * Allocate state block so tag methods have storage to record values.
	 */
	if (tif->tif_mode == O_RDONLY) {
		tif->tif_data = (tidata_t) _TIFFmalloc(sizeof (LZWDecodeState));
		if (tif->tif_data == NULL)
			goto bad;
		DecoderState(tif)->dec_codetab = NULL;
		DecoderState(tif)->dec_decode = NULL;
	} else {
		tif->tif_data = (tidata_t) _TIFFmalloc(sizeof (LZWEncodeState));
		if (tif->tif_data == NULL)
			goto bad;
		EncoderState(tif)->enc_hashtab = NULL;
	}

        LZWState(tif)->rw_mode = tif->tif_mode;

	/*
	 * Install codec methods.
	 */
	tif->tif_setupdecode = LZWSetupDecode;
	tif->tif_predecode = LZWPreDecode;
	tif->tif_decoderow = LZWDecode;
	tif->tif_decodestrip = LZWDecode;
	tif->tif_decodetile = LZWDecode;
	tif->tif_setupencode = LZWSetupEncode;
	tif->tif_preencode = LZWPreEncode;
	tif->tif_postencode = LZWPostEncode;
	tif->tif_encoderow = LZWEncode;
	tif->tif_encodestrip = LZWEncode;
	tif->tif_encodetile = LZWEncode;
	tif->tif_cleanup = LZWCleanup;
	/*
	 * Setup predictor setup.
	 */
	(void) TIFFPredictorInit(tif);
	return (1);
bad:
	TIFFError("TIFFInitLZW", "No space for LZW state block");
	return (0);
}

/*
 * Copyright (c) 1985, 1986 The Regents of the University of California.
 * All rights reserved.
 *
 * This code is derived from software contributed to Berkeley by
 * James A. Woods, derived from original work by Spencer Thomas
 * and Joseph Orost.
 *
 * Redistribution and use in source and binary forms are permitted
 * provided that the above copyright notice and this paragraph are
 * duplicated in all such forms and that any documentation,
 * advertising materials, and other materials related to such
 * distribution and use acknowledge that the software was developed
 * by the University of California, Berkeley.  The name of the
 * University may not be used to endorse or promote products derived
 * from this software without specific prior written permission.
 * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
 * IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED
 * WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE.
 */
#endif /* LZW_SUPPORT */