1 /* unpack.c -- decompress files in pack format.
2 * Copyright (C) 1992-1993 Jean-loup Gailly
3 * This is free software; you can redistribute it and/or modify it under the
4 * terms of the GNU General Public License, see the file COPYING.
8 static char rcsid[] = "$Id: unpack.c,v 1.4 1993/06/11 19:25:36 jloup Exp $";
16 #define MIN(a,b) ((a) <= (b) ? (a) : (b))
17 /* The arguments must not have side effects. */
20 /* Maximum length of Huffman codes. (Minor modifications to the code
21 * would be needed to support 32 bits codes, but pack never generates
22 * more than 24 bits anyway.)
26 /* Number of literals, excluding the End of Block (EOB) code */
29 /* Maximum number of 'peek' bits used to optimize traversal of the
33 local ulg orig_len; /* original uncompressed length */
34 local int max_len; /* maximum bit length of Huffman codes */
36 local uch literal[LITERALS];
37 /* The literal bytes present in the Huffman tree. The EOB code is not
41 local int lit_base[MAX_BITLEN+1];
42 /* All literals of a given bit length are contiguous in literal[] and
43 * have contiguous codes. literal[code+lit_base[len]] is the literal
44 * for a code of len bits.
47 local int leaves [MAX_BITLEN+1]; /* Number of leaves for each bit length */
48 local int parents[MAX_BITLEN+1]; /* Number of parents for each bit length */
50 local int peek_bits; /* Number of peek bits currently used */
52 /* local uch prefix_len[1 << MAX_PEEK]; */
53 #define prefix_len outbuf
54 /* For each bit pattern b of peek_bits bits, prefix_len[b] is the length
55 * of the Huffman code starting with a prefix of b (upper bits), or 0
56 * if all codes of prefix b have more than peek_bits bits. It is not
57 * necessary to have a huge table (large MAX_PEEK) because most of the
58 * codes encountered in the input stream are short codes (by construction).
59 * So for most codes a single lookup will be necessary.
61 #if (1<<MAX_PEEK) > OUTBUFSIZ
62 error cannot overlay prefix_len and outbuf
66 /* Bits are added on the low part of bitbuf and read from the high part. */
68 local int valid; /* number of valid bits in bitbuf */
69 /* all bits above the last valid bit are always zero */
71 /* Set code to the next 'bits' input bits without skipping them. code
72 * must be the name of a simple variable and bits must not have side effects.
73 * IN assertions: bits <= 25 (so that we still have room for an extra byte
74 * when valid is only 24), and mask = (1<<bits)-1.
76 #define look_bits(code,bits,mask) \
78 while (valid < (bits)) bitbuf = (bitbuf<<8) | (ulg)get_byte(), valid += 8; \
79 code = (bitbuf >> (valid-(bits))) & (mask); \
82 /* Skip the given number of bits (after having peeked at them): */
83 #define skip_bits(bits) (valid -= (bits))
85 #define clear_bitbuf() (valid = 0, bitbuf = 0)
89 local void read_tree OF((void));
90 local void build_tree OF((void));
92 /* ===========================================================================
93 * Read the Huffman tree.
95 local void read_tree()
97 int len; /* bit length */
98 int base; /* base offset for a sequence of leaves */
101 /* Read the original input size, MSB first */
103 for (n = 1; n <= 4; n++) orig_len = (orig_len << 8) | (ulg)get_byte();
105 max_len = (int)get_byte(); /* maximum bit length of Huffman codes */
106 if (max_len > MAX_BITLEN) {
107 error("invalid compressed data -- Huffman code > 32 bits");
110 /* Get the number of leaves at each bit length */
112 for (len = 1; len <= max_len; len++) {
113 leaves[len] = (int)get_byte();
117 error("too many leaves in Huffman tree");
119 Trace((stderr, "orig_len %lu, max_len %d, leaves %d\n",
120 orig_len, max_len, n));
121 /* There are at least 2 and at most 256 leaves of length max_len.
122 * (Pack arbitrarily rejects empty files and files consisting of
123 * a single byte even repeated.) To fit the last leaf count in a
124 * byte, it is offset by 2. However, the last literal is the EOB
125 * code, and is not transmitted explicitly in the tree, so we must
126 * adjust here by one only.
130 /* Now read the leaves themselves */
132 for (len = 1; len <= max_len; len++) {
133 /* Remember where the literals of this length start in literal[] : */
134 lit_base[len] = base;
135 /* And read the literals: */
136 for (n = leaves[len]; n > 0; n--) {
137 literal[base++] = (uch)get_byte();
140 leaves[max_len]++; /* Now include the EOB code in the Huffman tree */
143 /* ===========================================================================
144 * Build the Huffman tree and the prefix table.
146 local void build_tree()
148 int nodes = 0; /* number of nodes (parents+leaves) at current bit length */
149 int len; /* current bit length */
150 uch *prefixp; /* pointer in prefix_len */
152 for (len = max_len; len >= 1; len--) {
153 /* The number of parent nodes at this level is half the total
154 * number of nodes at parent level:
157 parents[len] = nodes;
158 /* Update lit_base by the appropriate bias to skip the parent nodes
159 * (which are not represented in the literal array):
161 lit_base[len] -= nodes;
162 /* Restore nodes to be parents+leaves: */
163 nodes += leaves[len];
165 /* Construct the prefix table, from shortest leaves to longest ones.
166 * The shortest code is all ones, so we start at the end of the table.
168 peek_bits = MIN(max_len, MAX_PEEK);
169 prefixp = &prefix_len[1<<peek_bits];
170 for (len = 1; len <= peek_bits; len++) {
171 int prefixes = leaves[len] << (peek_bits-len); /* may be 0 */
172 while (prefixes--) *--prefixp = (uch)len;
174 /* The length of all other codes is unknown: */
175 while (prefixp > prefix_len) *--prefixp = 0;
178 /* ===========================================================================
179 * Unpack in to out. This routine does not support the old pack format
180 * with magic header \037\037.
182 * IN assertions: the buffer inbuf contains already the beginning of
183 * the compressed data, from offsets inptr to insize-1 included.
184 * The magic header has already been checked. The output buffer is cleared.
187 int in, out; /* input and output file descriptors */
189 int len; /* Bit length of current code */
190 unsigned eob; /* End Of Block code */
191 register unsigned peek; /* lookahead bits */
192 unsigned peek_mask; /* Mask for peek_bits bits */
197 read_tree(); /* Read the Huffman tree */
198 build_tree(); /* Build the prefix table */
199 clear_bitbuf(); /* Initialize bit input */
200 peek_mask = (1<<peek_bits)-1;
202 /* The eob code is the largest code among all leaves of maximal length: */
203 eob = leaves[max_len]-1;
204 Trace((stderr, "eob %d %x\n", max_len, eob));
206 /* Decode the input data: */
208 /* Since eob is the longest code and not shorter than max_len,
209 * we can peek at max_len bits without having the risk of reading
210 * beyond the end of file.
212 look_bits(peek, peek_bits, peek_mask);
213 len = prefix_len[peek];
215 peek >>= peek_bits - len; /* discard the extra bits */
217 /* Code of more than peek_bits bits, we must traverse the tree */
218 ulg mask = peek_mask;
221 len++, mask = (mask<<1)+1;
222 look_bits(peek, len, mask);
223 } while (peek < (unsigned)parents[len]);
224 /* loop as long as peek is a parent node */
226 /* At this point, peek is the next complete code, of len bits */
227 if (peek == eob && len == max_len) break; /* end of file? */
228 put_ubyte(literal[peek+lit_base[len]]);
229 Tracev((stderr,"%02d %04x %c\n", len, peek,
230 literal[peek+lit_base[len]]));
235 if (orig_len != (ulg)(bytes_out & 0xffffffff)) {
236 error("invalid compressed data--length error");