コード例 #1
0
ファイル: ppmd_compress.c プロジェクト: muyu/pcompress
int
ppmd_state_init(void **data, int *level, int alloc)
{
	CPpmd8 *_ppmd;

	pthread_mutex_lock(&mem_init_lock);
	if (!mem_inited) {
		slab_cache_add(sizeof (CPpmd8));
		slab_cache_add(ppmd8_mem_sz[*level]);
	}
	pthread_mutex_unlock(&mem_init_lock);
	_ppmd = (CPpmd8 *)slab_alloc(NULL, sizeof (CPpmd8));
	if (!_ppmd)
		return (-1);

	/* Levels 0 - 14 correspond to PPMd model orders 0 - 14. */
	if (*level > 14) *level = 14;
	_ppmd->Order = *level;

	_ppmd->Base = 0;
	_ppmd->Size = 0;
	*data = _ppmd;
	if (*level > 9) *level = 9;

	if (alloc)
		return (ppmd_alloc(*data));
	return (0);
}
コード例 #2
0
ファイル: adaptive_compress.c プロジェクト: sehe/pcompress
int
adapt_compress(void *src, uint64_t srclen, void *dst,
	uint64_t *dstlen, int level, uchar_t chdr, int btype, void *data)
{
	struct adapt_data *adat = (struct adapt_data *)(data);
	uchar_t *src1 = (uchar_t *)src;
	int rv = 0, bsc_type = 0;

	if (btype == TYPE_UNKNOWN) {
		uint64_t i, tot8b, tag1, tag2, tag3;
		double tagcnt, pct_tag;
		uchar_t cur_byte, prev_byte;
		/*
		* Count number of 8-bit binary bytes and XML tags in source.
		*/
		tot8b = 0;
		tag1 = 0;
		tag2 = 0;
		tag3 = 0;
		prev_byte = cur_byte = 0;
		for (i = 0; i < srclen; i++) {
			cur_byte = src1[i];
			tot8b += (cur_byte & 0x80); // This way for possible auto-vectorization
			tag1 += (cur_byte == '<');
			tag2 += (cur_byte == '>');
			tag3 += ((prev_byte == '<') & (cur_byte == '/'));
			tag3 += ((prev_byte == '/') & (cur_byte == '>'));
			if (cur_byte != ' ')
				prev_byte = cur_byte;
		}

		tot8b /= 0x80;
		tagcnt = tag1 + tag2 + tag3;
		pct_tag = tagcnt / (double)srclen;
		if (adat->adapt_mode == 2 && tot8b > FORTY_PCT(srclen)) {
			btype = TYPE_BINARY;
		} else if (adat->adapt_mode == 1 && tot8b > FIFTY_PCT(srclen)) {
			btype = TYPE_BINARY;
		} else {
			btype = TYPE_TEXT;
			if (tag1 > tag2 - 4 && tag1 < tag2 + 4 && tag3 > (double)tag1 * 0.40 &&
			    tagcnt > (double)srclen * 0.001)
				btype |= TYPE_MARKUP;
		}
	}

	/*
	 * Use PPMd if some percentage of source is 7-bit textual bytes, otherwise
	 * use Bzip2 or LZMA. For totally incompressible data we always use LZ4. There
	 * is no point trying to compress such data, like Jpegs. However some archive headers
	 * and zero paddings can exist which LZ4 can easily take care of very fast.
	 */
#ifdef ENABLE_PC_LIBBSC
	bsc_type = is_bsc_type(btype);
#endif
	if (is_incompressible(btype)) {
		rv = lz4_compress(src, srclen, dst, dstlen, level, chdr, btype, adat->lz4_data);
		if (rv < 0)
			return (rv);
		rv = ADAPT_COMPRESS_LZ4;
		lz4_count++;

	} else if (adat->adapt_mode == 2 && PC_TYPE(btype) == TYPE_BINARY && !bsc_type) {
		rv = lzma_compress(src, srclen, dst, dstlen, level, chdr, btype, adat->lzma_data);
		if (rv < 0)
			return (rv);
		rv = ADAPT_COMPRESS_LZMA;
		lzma_count++;

	} else if (adat->adapt_mode == 1 && PC_TYPE(btype) == TYPE_BINARY && !bsc_type) {
		rv = bzip2_compress(src, srclen, dst, dstlen, level, chdr, btype, NULL);
		if (rv < 0)
			return (rv);
		rv = ADAPT_COMPRESS_BZIP2;
		bzip2_count++;

	} else {
#ifdef ENABLE_PC_LIBBSC
		if (adat->bsc_data && bsc_type) {
			rv = libbsc_compress(src, srclen, dst, dstlen, level, chdr, btype, adat->bsc_data);
			if (rv < 0)
				return (rv);
			rv = ADAPT_COMPRESS_BSC;
			bsc_count++;
		} else {
#endif
			rv = ppmd_alloc(adat->ppmd_data);
			if (rv < 0)
				return (rv);
			rv = ppmd_compress(src, srclen, dst, dstlen, level, chdr, btype, adat->ppmd_data);
			ppmd_free(adat->ppmd_data);
			if (rv < 0)
				return (rv);
			rv = ADAPT_COMPRESS_PPMD;
			ppmd_count++;
#ifdef ENABLE_PC_LIBBSC
		}
#endif
	}

	return (rv);
}