Home | History | Annotate | Line # | Download | only in gzip
unpack.c revision 1.3
      1  1.1  mrg /*	$FreeBSD: head/usr.bin/gzip/unpack.c 194579 2009-06-21 09:39:43Z delphij $	*/
      2  1.3  mrg /*	$NetBSD: unpack.c,v 1.3 2017/08/04 07:27:08 mrg Exp $	*/
      3  1.1  mrg 
      4  1.1  mrg /*-
      5  1.1  mrg  * Copyright (c) 2009 Xin LI <delphij (at) FreeBSD.org>
      6  1.1  mrg  * All rights reserved.
      7  1.1  mrg  *
      8  1.1  mrg  * Redistribution and use in source and binary forms, with or without
      9  1.1  mrg  * modification, are permitted provided that the following conditions
     10  1.1  mrg  * are met:
     11  1.1  mrg  * 1. Redistributions of source code must retain the above copyright
     12  1.1  mrg  *    notice, this list of conditions and the following disclaimer.
     13  1.1  mrg  * 2. Redistributions in binary form must reproduce the above copyright
     14  1.1  mrg  *    notice, this list of conditions and the following disclaimer in the
     15  1.1  mrg  *    documentation and/or other materials provided with the distribution.
     16  1.1  mrg  *
     17  1.1  mrg  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
     18  1.1  mrg  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     19  1.1  mrg  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     20  1.1  mrg  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
     21  1.1  mrg  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     22  1.1  mrg  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     23  1.1  mrg  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     24  1.1  mrg  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     25  1.1  mrg  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     26  1.1  mrg  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     27  1.1  mrg  * SUCH DAMAGE.
     28  1.1  mrg  */
     29  1.1  mrg 
     30  1.1  mrg /* This file is #included by gzip.c */
     31  1.1  mrg 
     32  1.1  mrg /*
     33  1.1  mrg  * pack(1) file format:
     34  1.1  mrg  *
     35  1.1  mrg  * The first 7 bytes is the header:
     36  1.1  mrg  *	00, 01 - Signature (US, RS), we already validated it earlier.
     37  1.1  mrg  *	02..05 - Uncompressed size
     38  1.1  mrg  *	    06 - Level for the huffman tree (<=24)
     39  1.1  mrg  *
     40  1.1  mrg  * pack(1) will then store symbols (leaf) nodes count in each huffman
     41  1.1  mrg  * tree levels, each level would consume 1 byte (See [1]).
     42  1.1  mrg  *
     43  1.1  mrg  * After the symbol count table, there is the symbol table, storing
     44  1.2  mrg  * symbols represented by corresponding leaf node.  EOB is not being
     45  1.1  mrg  * explicitly transmitted (not necessary anyway) in the symbol table.
     46  1.1  mrg  *
     47  1.1  mrg  * Compressed data goes after the symbol table.
     48  1.1  mrg  *
     49  1.1  mrg  * NOTES
     50  1.1  mrg  *
     51  1.1  mrg  * [1] If we count EOB into the symbols, that would mean that we will
     52  1.1  mrg  * have at most 256 symbols in the huffman tree.  pack(1) rejects empty
     53  1.1  mrg  * file and files that just repeats one character, which means that we
     54  1.1  mrg  * will have at least 2 symbols.  Therefore, pack(1) would reduce the
     55  1.1  mrg  * last level symbol count by 2 which makes it a number in
     56  1.1  mrg  * range [0..254], so all levels' symbol count would fit into 1 byte.
     57  1.1  mrg  */
     58  1.1  mrg 
     59  1.1  mrg #define	PACK_HEADER_LENGTH	7
     60  1.1  mrg #define	HTREE_MAXLEVEL		24
     61  1.1  mrg 
     62  1.1  mrg /*
     63  1.1  mrg  * unpack descriptor
     64  1.1  mrg  *
     65  1.2  mrg  * Represent the huffman tree in a similar way that pack(1) would
     66  1.1  mrg  * store in a packed file.  We store all symbols in a linear table,
     67  1.1  mrg  * and store pointers to each level's first symbol.  In addition to
     68  1.1  mrg  * that, maintain two counts for each level: inner nodes count and
     69  1.1  mrg  * leaf nodes count.
     70  1.1  mrg  */
     71  1.1  mrg typedef struct {
     72  1.1  mrg 	int		symbol_size;	/* Size of the symbol table */
     73  1.1  mrg 	int		treelevels;	/* Levels for the huffman tree */
     74  1.1  mrg 
     75  1.1  mrg 	int		*symbolsin;	/* Table of leaf symbols count in
     76  1.1  mrg 					   each level */
     77  1.1  mrg 	int		*inodesin;	/* Table of internal nodes count in
     78  1.1  mrg 					   each level */
     79  1.1  mrg 
     80  1.1  mrg 	char		*symbol;	/* The symbol table */
     81  1.1  mrg 	char		*symbol_eob;	/* Pointer to the EOB symbol */
     82  1.1  mrg 	char		**tree;		/* Decoding huffman tree (pointers to
     83  1.1  mrg 					   first symbol of each tree level */
     84  1.1  mrg 
     85  1.1  mrg 	off_t		uncompressed_size; /* Uncompressed size */
     86  1.1  mrg 	FILE		*fpIn;		/* Input stream */
     87  1.1  mrg 	FILE		*fpOut;		/* Output stream */
     88  1.1  mrg } unpack_descriptor_t;
     89  1.1  mrg 
     90  1.1  mrg /*
     91  1.1  mrg  * Release resource allocated to an unpack descriptor.
     92  1.1  mrg  *
     93  1.1  mrg  * Caller is responsible to make sure that all of these pointers are
     94  1.1  mrg  * initialized (in our case, they all point to valid memory block).
     95  1.1  mrg  * We don't zero out pointers here because nobody else would ever
     96  1.2  mrg  * reference the memory block without scrubbing them.
     97  1.1  mrg  */
     98  1.1  mrg static void
     99  1.1  mrg unpack_descriptor_fini(unpack_descriptor_t *unpackd)
    100  1.1  mrg {
    101  1.1  mrg 
    102  1.1  mrg 	free(unpackd->symbolsin);
    103  1.1  mrg 	free(unpackd->inodesin);
    104  1.1  mrg 	free(unpackd->symbol);
    105  1.1  mrg 	free(unpackd->tree);
    106  1.1  mrg 
    107  1.1  mrg 	fclose(unpackd->fpIn);
    108  1.1  mrg 	fclose(unpackd->fpOut);
    109  1.1  mrg }
    110  1.1  mrg 
    111  1.1  mrg /*
    112  1.1  mrg  * Recursively fill the internal node count table
    113  1.1  mrg  */
    114  1.1  mrg static void
    115  1.1  mrg unpackd_fill_inodesin(const unpack_descriptor_t *unpackd, int level)
    116  1.1  mrg {
    117  1.1  mrg 
    118  1.1  mrg 	/*
    119  1.1  mrg 	 * The internal nodes would be 1/2 of total internal nodes and
    120  1.1  mrg 	 * leaf nodes in the next level.  For the last level there
    121  1.2  mrg 	 * would be no internal node by definition.
    122  1.1  mrg 	 */
    123  1.1  mrg 	if (level < unpackd->treelevels) {
    124  1.1  mrg 		unpackd_fill_inodesin(unpackd, level + 1);
    125  1.1  mrg 		unpackd->inodesin[level] = (unpackd->inodesin[level + 1] +
    126  1.1  mrg 					  unpackd->symbolsin[level + 1]) / 2;
    127  1.1  mrg 	} else
    128  1.1  mrg 		unpackd->inodesin[level] = 0;
    129  1.1  mrg }
    130  1.1  mrg 
    131  1.1  mrg /*
    132  1.1  mrg  * Update counter for accepted bytes
    133  1.1  mrg  */
    134  1.1  mrg static void
    135  1.1  mrg accepted_bytes(off_t *bytes_in, off_t newbytes)
    136  1.1  mrg {
    137  1.1  mrg 
    138  1.1  mrg 	if (bytes_in != NULL)
    139  1.1  mrg 		(*bytes_in) += newbytes;
    140  1.1  mrg }
    141  1.1  mrg 
    142  1.1  mrg /*
    143  1.1  mrg  * Read file header and construct the tree.  Also, prepare the buffered I/O
    144  1.2  mrg  * for decode routine.
    145  1.1  mrg  *
    146  1.1  mrg  * Return value is uncompressed size.
    147  1.1  mrg  */
    148  1.1  mrg static void
    149  1.1  mrg unpack_parse_header(int in, int out, char *pre, size_t prelen, off_t *bytes_in,
    150  1.1  mrg     unpack_descriptor_t *unpackd)
    151  1.1  mrg {
    152  1.1  mrg 	unsigned char hdr[PACK_HEADER_LENGTH];	/* buffer for header */
    153  1.1  mrg 	ssize_t bytesread;		/* Bytes read from the file */
    154  1.1  mrg 	int i, j, thisbyte;
    155  1.1  mrg 
    156  1.3  mrg 	if (prelen > sizeof hdr)
    157  1.3  mrg 		maybe_err("prelen too long");
    158  1.3  mrg 
    159  1.1  mrg 	/* Prepend the header buffer if we already read some data */
    160  1.1  mrg 	if (prelen != 0)
    161  1.1  mrg 		memcpy(hdr, pre, prelen);
    162  1.1  mrg 
    163  1.1  mrg 	/* Read in and fill the rest bytes of header */
    164  1.1  mrg 	bytesread = read(in, hdr + prelen, PACK_HEADER_LENGTH - prelen);
    165  1.1  mrg 	if (bytesread < 0)
    166  1.1  mrg 		maybe_err("Error reading pack header");
    167  1.3  mrg 	infile_newdata(bytesread);
    168  1.1  mrg 
    169  1.1  mrg 	accepted_bytes(bytes_in, PACK_HEADER_LENGTH);
    170  1.1  mrg 
    171  1.1  mrg 	/* Obtain uncompressed length (bytes 2,3,4,5)*/
    172  1.1  mrg 	unpackd->uncompressed_size = 0;
    173  1.1  mrg 	for (i = 2; i <= 5; i++) {
    174  1.1  mrg 		unpackd->uncompressed_size <<= 8;
    175  1.1  mrg 		unpackd->uncompressed_size |= hdr[i];
    176  1.1  mrg 	}
    177  1.1  mrg 
    178  1.1  mrg 	/* Get the levels of the tree */
    179  1.1  mrg 	unpackd->treelevels = hdr[6];
    180  1.1  mrg 	if (unpackd->treelevels > HTREE_MAXLEVEL || unpackd->treelevels < 1)
    181  1.1  mrg 		maybe_errx("Huffman tree has insane levels");
    182  1.1  mrg 
    183  1.1  mrg 	/* Let libc take care for buffering from now on */
    184  1.1  mrg 	if ((unpackd->fpIn = fdopen(in, "r")) == NULL)
    185  1.1  mrg 		maybe_err("Can not fdopen() input stream");
    186  1.1  mrg 	if ((unpackd->fpOut = fdopen(out, "w")) == NULL)
    187  1.1  mrg 		maybe_err("Can not fdopen() output stream");
    188  1.1  mrg 
    189  1.1  mrg 	/* Allocate for the tables of bounds and the tree itself */
    190  1.1  mrg 	unpackd->inodesin =
    191  1.1  mrg 	    calloc(unpackd->treelevels, sizeof(*(unpackd->inodesin)));
    192  1.1  mrg 	unpackd->symbolsin =
    193  1.1  mrg 	    calloc(unpackd->treelevels, sizeof(*(unpackd->symbolsin)));
    194  1.1  mrg 	unpackd->tree =
    195  1.1  mrg 	    calloc(unpackd->treelevels, (sizeof (*(unpackd->tree))));
    196  1.1  mrg 	if (unpackd->inodesin == NULL || unpackd->symbolsin == NULL ||
    197  1.1  mrg 	    unpackd->tree == NULL)
    198  1.1  mrg 		maybe_err("calloc");
    199  1.1  mrg 
    200  1.1  mrg 	/* We count from 0 so adjust to match array upper bound */
    201  1.1  mrg 	unpackd->treelevels--;
    202  1.1  mrg 
    203  1.2  mrg 	/* Read the levels symbol count table and calculate total */
    204  1.1  mrg 	unpackd->symbol_size = 1;		/* EOB */
    205  1.1  mrg 	for (i = 0; i <= unpackd->treelevels; i++) {
    206  1.1  mrg 		if ((thisbyte = fgetc(unpackd->fpIn)) == EOF)
    207  1.1  mrg 			maybe_err("File appears to be truncated");
    208  1.1  mrg 		unpackd->symbolsin[i] = (unsigned char)thisbyte;
    209  1.1  mrg 		unpackd->symbol_size += unpackd->symbolsin[i];
    210  1.1  mrg 	}
    211  1.1  mrg 	accepted_bytes(bytes_in, unpackd->treelevels);
    212  1.1  mrg 	if (unpackd->symbol_size > 256)
    213  1.1  mrg 		maybe_errx("Bad symbol table");
    214  1.3  mrg 	infile_newdata(unpackd->treelevels);
    215  1.1  mrg 
    216  1.1  mrg 	/* Allocate for the symbol table, point symbol_eob at the beginning */
    217  1.1  mrg 	unpackd->symbol_eob = unpackd->symbol = calloc(1, unpackd->symbol_size);
    218  1.1  mrg 	if (unpackd->symbol == NULL)
    219  1.1  mrg 		maybe_err("calloc");
    220  1.1  mrg 
    221  1.1  mrg 	/*
    222  1.1  mrg 	 * Read in the symbol table, which contain [2, 256] symbols.
    223  1.1  mrg 	 * In order to fit the count in one byte, pack(1) would offset
    224  1.1  mrg 	 * it by reducing 2 from the actual number from the last level.
    225  1.1  mrg 	 *
    226  1.1  mrg 	 * We adjust the last level's symbol count by 1 here, because
    227  1.1  mrg 	 * the EOB symbol is not being transmitted explicitly.  Another
    228  1.1  mrg 	 * adjustment would be done later afterward.
    229  1.1  mrg 	 */
    230  1.1  mrg 	unpackd->symbolsin[unpackd->treelevels]++;
    231  1.1  mrg 	for (i = 0; i <= unpackd->treelevels; i++) {
    232  1.1  mrg 		unpackd->tree[i] = unpackd->symbol_eob;
    233  1.1  mrg 		for (j = 0; j < unpackd->symbolsin[i]; j++) {
    234  1.1  mrg 			if ((thisbyte = fgetc(unpackd->fpIn)) == EOF)
    235  1.1  mrg 				maybe_errx("Symbol table truncated");
    236  1.1  mrg 			*unpackd->symbol_eob++ = (char)thisbyte;
    237  1.1  mrg 		}
    238  1.3  mrg 		infile_newdata(unpackd->symbolsin[i]);
    239  1.1  mrg 		accepted_bytes(bytes_in, unpackd->symbolsin[i]);
    240  1.1  mrg 	}
    241  1.1  mrg 
    242  1.1  mrg 	/* Now, take account for the EOB symbol as well */
    243  1.1  mrg 	unpackd->symbolsin[unpackd->treelevels]++;
    244  1.1  mrg 
    245  1.1  mrg 	/*
    246  1.1  mrg 	 * The symbolsin table has been constructed now.
    247  1.2  mrg 	 * Calculate the internal nodes count table based on it.
    248  1.1  mrg 	 */
    249  1.1  mrg 	unpackd_fill_inodesin(unpackd, 0);
    250  1.1  mrg }
    251  1.1  mrg 
    252  1.1  mrg /*
    253  1.1  mrg  * Decode huffman stream, based on the huffman tree.
    254  1.1  mrg  */
    255  1.1  mrg static void
    256  1.1  mrg unpack_decode(const unpack_descriptor_t *unpackd, off_t *bytes_in)
    257  1.1  mrg {
    258  1.1  mrg 	int thislevel, thiscode, thisbyte, inlevelindex;
    259  1.1  mrg 	int i;
    260  1.1  mrg 	off_t bytes_out = 0;
    261  1.1  mrg 	const char *thissymbol;	/* The symbol pointer decoded from stream */
    262  1.1  mrg 
    263  1.1  mrg 	/*
    264  1.1  mrg 	 * Decode huffman.  Fetch every bytes from the file, get it
    265  1.1  mrg 	 * into 'thiscode' bit-by-bit, then output the symbol we got
    266  1.1  mrg 	 * when one has been found.
    267  1.1  mrg 	 *
    268  1.1  mrg 	 * Assumption: sizeof(int) > ((max tree levels + 1) / 8).
    269  1.1  mrg 	 * bad things could happen if not.
    270  1.1  mrg 	 */
    271  1.1  mrg 	thislevel = 0;
    272  1.1  mrg 	thiscode = thisbyte = 0;
    273  1.1  mrg 
    274  1.1  mrg 	while ((thisbyte = fgetc(unpackd->fpIn)) != EOF) {
    275  1.1  mrg 		accepted_bytes(bytes_in, 1);
    276  1.3  mrg 		infile_newdata(1);
    277  1.3  mrg 		check_siginfo();
    278  1.1  mrg 
    279  1.1  mrg 		/*
    280  1.1  mrg 		 * Split one bit from thisbyte, from highest to lowest,
    281  1.1  mrg 		 * feed the bit into thiscode, until we got a symbol from
    282  1.1  mrg 		 * the tree.
    283  1.1  mrg 		 */
    284  1.1  mrg 		for (i = 7; i >= 0; i--) {
    285  1.1  mrg 			thiscode = (thiscode << 1) | ((thisbyte >> i) & 1);
    286  1.1  mrg 
    287  1.1  mrg 			/* Did we got a symbol? (referencing leaf node) */
    288  1.1  mrg 			if (thiscode >= unpackd->inodesin[thislevel]) {
    289  1.1  mrg 				inlevelindex =
    290  1.1  mrg 				    thiscode - unpackd->inodesin[thislevel];
    291  1.1  mrg 				if (inlevelindex > unpackd->symbolsin[thislevel])
    292  1.1  mrg 					maybe_errx("File corrupt");
    293  1.1  mrg 
    294  1.1  mrg 				thissymbol =
    295  1.1  mrg 				    &(unpackd->tree[thislevel][inlevelindex]);
    296  1.1  mrg 				if ((thissymbol == unpackd->symbol_eob) &&
    297  1.1  mrg 				    (bytes_out == unpackd->uncompressed_size))
    298  1.1  mrg 					goto finished;
    299  1.1  mrg 
    300  1.1  mrg 				fputc((*thissymbol), unpackd->fpOut);
    301  1.1  mrg 				bytes_out++;
    302  1.1  mrg 
    303  1.1  mrg 				/* Prepare for next input */
    304  1.1  mrg 				thislevel = 0; thiscode = 0;
    305  1.1  mrg 			} else {
    306  1.1  mrg 				thislevel++;
    307  1.1  mrg 				if (thislevel > unpackd->treelevels)
    308  1.1  mrg 					maybe_errx("File corrupt");
    309  1.1  mrg 			}
    310  1.1  mrg 		}
    311  1.1  mrg 	}
    312  1.1  mrg 
    313  1.1  mrg finished:
    314  1.1  mrg 	if (bytes_out != unpackd->uncompressed_size)
    315  1.1  mrg 		maybe_errx("Premature EOF");
    316  1.1  mrg }
    317  1.1  mrg 
    318  1.1  mrg /* Handler for pack(1)'ed file */
    319  1.1  mrg static off_t
    320  1.1  mrg unpack(int in, int out, char *pre, size_t prelen, off_t *bytes_in)
    321  1.1  mrg {
    322  1.1  mrg 	unpack_descriptor_t	unpackd;
    323  1.1  mrg 
    324  1.1  mrg 	unpack_parse_header(dup(in), dup(out), pre, prelen, bytes_in, &unpackd);
    325  1.1  mrg 	unpack_decode(&unpackd, bytes_in);
    326  1.1  mrg 	unpack_descriptor_fini(&unpackd);
    327  1.1  mrg 
    328  1.1  mrg 	/* If we reached here, the unpack was successful */
    329  1.1  mrg 	return (unpackd.uncompressed_size);
    330  1.1  mrg }
    331  1.1  mrg 
    332