1adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard/**
2adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard * xzlib.c: front end for the transparent suport of lzma compression
3adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard *          at the I/O layer, based on an example file from lzma project
4adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard *
5adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard * See Copyright for the status of this software.
6adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard *
7adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard * Anders F Bjorklund <afb@users.sourceforge.net>
8adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard */
9adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#define IN_LIBXML
10adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#include "libxml.h"
1118b8988511b0954272cac4d6c3e6724f9dbf6e0aDaniel Veillard#ifdef LIBXML_LZMA_ENABLED
12adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
13adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#include <string.h>
14adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#ifdef HAVE_ERRNO_H
15adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#include <errno.h>
16adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#endif
17adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
18adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
19adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#ifdef HAVE_SYS_TYPES_H
20adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#include <sys/types.h>
21adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#endif
22adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#ifdef HAVE_SYS_STAT_H
23adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#include <sys/stat.h>
24adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#endif
25adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#ifdef HAVE_FCNTL_H
26adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#include <fcntl.h>
27adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#endif
28adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#ifdef HAVE_UNISTD_H
29adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#include <unistd.h>
30adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#endif
31adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#ifdef HAVE_STDLIB_H
32adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#include <stdlib.h>
33adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#endif
34adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#ifdef HAVE_ZLIB_H
35adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#include <zlib.h>
36adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#endif
3718b8988511b0954272cac4d6c3e6724f9dbf6e0aDaniel Veillard#ifdef HAVE_LZMA_H
38adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#include <lzma.h>
3918b8988511b0954272cac4d6c3e6724f9dbf6e0aDaniel Veillard#endif
40adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
416bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#include "xzlib.h"
429f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard#include <libxml/xmlmemory.h>
436bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
446bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund/* values for xz_state how */
45adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#define LOOK 0                  /* look for a gzip/lzma header */
46adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#define COPY 1                  /* copy input directly */
47adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#define GZIP 2                  /* decompress a gzip stream */
48adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard#define LZMA 3                  /* decompress a lzma stream */
496bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
506bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund/* internal lzma file state data structure */
516bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklundtypedef struct {
52adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    int mode;                   /* see lzma modes above */
53adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    int fd;                     /* file descriptor */
54adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    char *path;                 /* path or fd for error messages */
55adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    uint64_t pos;               /* current position in uncompressed data */
5672789ef21fed132bb15bd4e858cb524f41c79ed1Daniel Veillard    unsigned int size;          /* buffer size, zero if not allocated yet */
5772789ef21fed132bb15bd4e858cb524f41c79ed1Daniel Veillard    unsigned int want;          /* requested buffer size, default is BUFSIZ */
58adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    unsigned char *in;          /* input buffer */
59adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    unsigned char *out;         /* output buffer (double-sized when reading) */
60adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    unsigned char *next;        /* next output data to deliver or write */
6172789ef21fed132bb15bd4e858cb524f41c79ed1Daniel Veillard    unsigned int have;          /* amount of output data unused at next */
62adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    int eof;                    /* true if end of input file reached */
63adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    uint64_t start;             /* where the lzma data started, for rewinding */
64adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    uint64_t raw;               /* where the raw data started, for seeking */
65adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    int how;                    /* 0: get header, 1: copy, 2: decompress */
66adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    int direct;                 /* true if last read direct, false if lzma */
67adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* seek request */
68adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    uint64_t skip;              /* amount to skip (already rewound if backwards) */
69adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    int seek;                   /* true if seek request pending */
70adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* error information */
71adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    int err;                    /* error code */
72adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    char *msg;                  /* error message */
73adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* lzma stream */
749f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard    int init;                   /* is the iniflate stream initialized */
75adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    lzma_stream strm;           /* stream structure in-place (not a pointer) */
76adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    char padding1[32];          /* padding allowing to cope with possible
77adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                                   extensions of above structure without
78adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard				   too much side effect */
796bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#ifdef HAVE_ZLIB_H
80adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* zlib inflate or deflate stream */
81adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    z_stream zstrm;             /* stream structure in-place (not a pointer) */
826bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
83adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    char padding2[32];          /* padding allowing to cope with possible
84adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                                   extensions of above structure without
85adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard				   too much side effect */
866bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund} xz_state, *xz_statep;
876bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
88adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardstatic void
89adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardxz_error(xz_statep state, int err, const char *msg)
906bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
916bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* free previously allocated message and clear */
926bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (state->msg != NULL) {
936bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (state->err != LZMA_MEM_ERROR)
949f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard            xmlFree(state->msg);
956bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->msg = NULL;
966bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
976bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
986bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* set error code, and if no message, then done */
996bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->err = err;
1006bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (msg == NULL)
1016bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return;
1026bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
1036bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* for an out of memory error, save as static string */
1046bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (err == LZMA_MEM_ERROR) {
105adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        state->msg = (char *) msg;
1066bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return;
1076bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
1086bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
1096bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* construct error message with path */
110adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    if ((state->msg =
1119f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard         xmlMalloc(strlen(state->path) + strlen(msg) + 3)) == NULL) {
1126bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->err = LZMA_MEM_ERROR;
113adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        state->msg = (char *) "out of memory";
1146bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return;
1156bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
1166bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    strcpy(state->msg, state->path);
1176bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    strcat(state->msg, ": ");
1186bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    strcat(state->msg, msg);
1196bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return;
1206bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
1216bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
122adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardstatic void
123adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardxz_reset(xz_statep state)
1246bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
125adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state->have = 0;            /* no output data available */
126adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state->eof = 0;             /* not at end of file */
127adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state->how = LOOK;          /* look for gzip header */
128adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state->direct = 1;          /* default for empty file */
129adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state->seek = 0;            /* no seek request pending */
130adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    xz_error(state, LZMA_OK, NULL);     /* clear error */
131adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state->pos = 0;             /* no uncompressed data yet */
132adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state->strm.avail_in = 0;   /* no input data yet */
1336bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#ifdef HAVE_ZLIB_H
134adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state->zstrm.avail_in = 0;  /* no input data yet */
1356bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
1366bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
1376bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
138adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardstatic xzFile
139adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardxz_open(const char *path, int fd, const char *mode ATTRIBUTE_UNUSED)
1406bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
1416bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    xz_statep state;
1426bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
1436bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* allocate xzFile structure to return */
1449f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard    state = xmlMalloc(sizeof(xz_state));
1456bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (state == NULL)
1466bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return NULL;
1476bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->size = 0;            /* no buffers allocated yet */
1486bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->want = BUFSIZ;       /* requested buffer size */
1496bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->msg = NULL;          /* no error message yet */
1509f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard    state->init = 0;            /* initialization of zlib data */
1516bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
152adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* save the path name for error messages */
1539f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard    state->path = xmlMalloc(strlen(path) + 1);
1546bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (state->path == NULL) {
1559f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard        xmlFree(state);
1566bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return NULL;
1576bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
1586bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    strcpy(state->path, path);
1596bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
1606bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* open the file with the appropriate mode (or just use fd) */
161adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state->fd = fd != -1 ? fd : open(path,
1626bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#ifdef O_LARGEFILE
163adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                                     O_LARGEFILE |
1646bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
1656bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#ifdef O_BINARY
166adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                                     O_BINARY |
1676bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
168adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                                     O_RDONLY, 0666);
1696bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (state->fd == -1) {
1709f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard        xmlFree(state->path);
1719f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard        xmlFree(state);
1726bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return NULL;
1736bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
1746bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
1756bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* save the current position for rewinding (only if reading) */
1766bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->start = lseek(state->fd, 0, SEEK_CUR);
177adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    if (state->start == (uint64_t) - 1)
178adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        state->start = 0;
1796bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
1806bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* initialize stream */
1816bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    xz_reset(state);
1826bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
1836bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* return stream */
184adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    return (xzFile) state;
1856bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
1866bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
18763588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillardstatic int
18863588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillardxz_compressed(xzFile f) {
18963588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard    xz_statep state;
19063588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard
19163588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard    if (f == NULL)
19263588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard        return(-1);
19363588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard    state = (xz_statep) f;
19463588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard    if (state->init <= 0)
19563588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard        return(-1);
19663588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard
19763588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard    switch (state->how) {
19863588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard        case COPY:
19963588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard	    return(0);
20063588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard	case GZIP:
20163588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard	case LZMA:
20263588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard	    return(1);
20363588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard    }
20463588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard    return(-1);
20563588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard}
20663588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard
207adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel VeillardxzFile
208adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard__libxml2_xzopen(const char *path, const char *mode)
2096bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
2106bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return xz_open(path, -1, mode);
2116bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
2126bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
21363588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillardint
21463588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard__libxml2_xzcompressed(xzFile f) {
21563588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard    return xz_compressed(f);
21663588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard}
21763588f476f2dc89d4c6ef70a474d7230fbf4d45eDaniel Veillard
218adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel VeillardxzFile
219adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard__libxml2_xzdopen(int fd, const char *mode)
2206bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
221adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    char *path;                 /* identifier for error messages */
2226bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    xzFile xz;
2236bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
2249f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard    if (fd == -1 || (path = xmlMalloc(7 + 3 * sizeof(int))) == NULL)
2256bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return NULL;
226adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    sprintf(path, "<fd:%d>", fd);       /* for debugging */
2276bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    xz = xz_open(path, fd, mode);
2289f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard    xmlFree(path);
2296bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return xz;
2306bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
2316bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
232adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardstatic int
23372789ef21fed132bb15bd4e858cb524f41c79ed1Daniel Veillardxz_load(xz_statep state, unsigned char *buf, unsigned int len,
23472789ef21fed132bb15bd4e858cb524f41c79ed1Daniel Veillard        unsigned int *have)
2356bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
2366bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    int ret;
2376bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
2386bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    *have = 0;
2396bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    do {
2406bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        ret = read(state->fd, buf + *have, len - *have);
2416bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (ret <= 0)
2426bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            break;
2436bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        *have += ret;
2446bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    } while (*have < len);
2456bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (ret < 0) {
2466bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        xz_error(state, -1, strerror(errno));
2476bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return -1;
2486bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
2496bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (ret == 0)
2506bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->eof = 1;
2516bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return 0;
2526bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
2536bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
254adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardstatic int
255adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardxz_avail(xz_statep state)
2566bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
2576bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    lzma_stream *strm = &(state->strm);
2586bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
2596bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (state->err != LZMA_OK)
2606bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return -1;
2616bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (state->eof == 0) {
262996449273fd4a8fc656d42088779e236d456e47aMarcus Meissner        /* avail_in is size_t, which is not necessary sizeof(unsigned) */
263996449273fd4a8fc656d42088779e236d456e47aMarcus Meissner        unsigned tmp = strm->avail_in;
264996449273fd4a8fc656d42088779e236d456e47aMarcus Meissner
265996449273fd4a8fc656d42088779e236d456e47aMarcus Meissner        if (xz_load(state, state->in, state->size, &tmp) == -1) {
266996449273fd4a8fc656d42088779e236d456e47aMarcus Meissner            strm->avail_in = tmp;
2676bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
268996449273fd4a8fc656d42088779e236d456e47aMarcus Meissner        }
269996449273fd4a8fc656d42088779e236d456e47aMarcus Meissner        strm->avail_in = tmp;
2706bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        strm->next_in = state->in;
2716bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
2726bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return 0;
2736bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
2746bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
275a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander#ifdef HAVE_ZLIB_H
276a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexanderstatic int
277a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexanderxz_avail_zstrm(xz_statep state)
278a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander{
279a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    int ret;
280a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    state->strm.avail_in = state->zstrm.avail_in;
281a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    state->strm.next_in = state->zstrm.next_in;
282a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    ret = xz_avail(state);
283a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    state->zstrm.avail_in = (uInt) state->strm.avail_in;
284a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    state->zstrm.next_in = (Bytef *) state->strm.next_in;
285a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    return ret;
286a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander}
287a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander#endif
288a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander
2896bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklundstatic int
2906bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklundis_format_xz(xz_statep state)
2916bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
2926bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    lzma_stream *strm = &(state->strm);
2936bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
294adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    return strm->avail_in >= 6 && memcmp(state->in, "\3757zXZ", 6) == 0;
2956bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
2966bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
2976bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklundstatic int
2986bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklundis_format_lzma(xz_statep state)
2996bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
3006bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    lzma_stream *strm = &(state->strm);
3016bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
302adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    lzma_filter filter;
303adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    lzma_options_lzma *opt;
304adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    uint32_t dict_size;
305adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    uint64_t uncompressed_size;
306adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    size_t i;
307adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
308adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    if (strm->avail_in < 13)
309adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        return 0;
310adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
311adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    filter.id = LZMA_FILTER_LZMA1;
312adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    if (lzma_properties_decode(&filter, NULL, state->in, 5) != LZMA_OK)
313adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        return 0;
314adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
315adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    opt = filter.options;
316adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    dict_size = opt->dict_size;
31794431ecba6c458c56cb0f5b2a919ed4cf27107baDaniel Veillard    free(opt); /* we can't use xmlFree on a string returned by zlib */
318adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
319adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* A hack to ditch tons of false positives: We allow only dictionary
320adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard     * sizes that are 2^n or 2^n + 2^(n-1) or UINT32_MAX. LZMA_Alone
321adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard     * created only files with 2^n, but accepts any dictionary size.
322adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard     * If someone complains, this will be reconsidered.
323adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard     */
324adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    if (dict_size != UINT32_MAX) {
325adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        uint32_t d = dict_size - 1;
326adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
327adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        d |= d >> 2;
328adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        d |= d >> 3;
329adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        d |= d >> 4;
330adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        d |= d >> 8;
331adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        d |= d >> 16;
332adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        ++d;
333adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        if (d != dict_size || dict_size == 0)
334adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            return 0;
335adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    }
336adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
337adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* Another hack to ditch false positives: Assume that if the
338adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard     * uncompressed size is known, it must be less than 256 GiB.
339adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard     * Again, if someone complains, this will be reconsidered.
340adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard     */
341adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    uncompressed_size = 0;
342adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    for (i = 0; i < 8; ++i)
343adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        uncompressed_size |= (uint64_t) (state->in[5 + i]) << (i * 8);
344adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
345adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    if (uncompressed_size != UINT64_MAX
346adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        && uncompressed_size > (UINT64_C(1) << 38))
347adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        return 0;
348adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
349adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    return 1;
3506bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
3516bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
3526bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#ifdef HAVE_ZLIB_H
353adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
3546bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund/* Get next byte from input, or -1 if end or error. */
3556bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#define NEXT() ((strm->avail_in == 0 && xz_avail(state) == -1) ? -1 : \
3566bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                (strm->avail_in == 0 ? -1 : \
3576bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                 (strm->avail_in--, *(strm->next_in)++)))
358a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander/* Same thing, but from zstrm */
359a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander#define NEXTZ() ((strm->avail_in == 0 && xz_avail_zstrm(state) == -1) ? -1 : \
360a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander                (strm->avail_in == 0 ? -1 : \
361a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander                 (strm->avail_in--, *(strm->next_in)++)))
3626bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
3636bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund/* Get a four-byte little-endian integer and return 0 on success and the value
3646bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund   in *ret.  Otherwise -1 is returned and *ret is not modified. */
365adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardstatic int
366adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardgz_next4(xz_statep state, unsigned long *ret)
3676bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
3686bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    int ch;
3696bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    unsigned long val;
3706bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    z_streamp strm = &(state->zstrm);
3716bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
372a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    val = NEXTZ();
373a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    val += (unsigned) NEXTZ() << 8;
374a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    val += (unsigned long) NEXTZ() << 16;
375a1313a6f8c48cc27b02f65aca6b04e6cd59d939aMike Alexander    ch = NEXTZ();
3766bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (ch == -1)
3776bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return -1;
378adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    val += (unsigned long) ch << 24;
3796bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    *ret = val;
3806bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return 0;
3816bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
3826bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
3836bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
384adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardstatic int
385adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardxz_head(xz_statep state)
3866bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
3876bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    lzma_stream *strm = &(state->strm);
3886bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    lzma_stream init = LZMA_STREAM_INIT;
3896bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    int flags;
3906bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    unsigned len;
3916bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
3926bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* allocate read buffers and inflate memory */
3936bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (state->size == 0) {
3946bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* allocate buffers */
3959f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard        state->in = xmlMalloc(state->want);
3969f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard        state->out = xmlMalloc(state->want << 1);
3976bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (state->in == NULL || state->out == NULL) {
3986bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            if (state->out != NULL)
3999f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard                xmlFree(state->out);
4006bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            if (state->in != NULL)
4019f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard                xmlFree(state->in);
4026bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            xz_error(state, LZMA_MEM_ERROR, "out of memory");
4036bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
4046bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
4056bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->size = state->want;
4066bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
4076bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* allocate decoder memory */
4086bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->strm = init;
4096bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->strm.avail_in = 0;
4106bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->strm.next_in = NULL;
4116bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (lzma_auto_decoder(&state->strm, UINT64_MAX, 0) != LZMA_OK) {
4129f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard            xmlFree(state->out);
4139f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard            xmlFree(state->in);
4146bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->size = 0;
4156bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            xz_error(state, LZMA_MEM_ERROR, "out of memory");
4166bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
417adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        }
4186bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#ifdef HAVE_ZLIB_H
4196bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* allocate inflate memory */
4206bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->zstrm.zalloc = Z_NULL;
4216bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->zstrm.zfree = Z_NULL;
4226bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->zstrm.opaque = Z_NULL;
4236bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->zstrm.avail_in = 0;
4246bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->zstrm.next_in = Z_NULL;
4259f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard        if (state->init == 0) {
4269f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard            if (inflateInit2(&(state->zstrm), -15) != Z_OK) {/* raw inflate */
4279f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard                xmlFree(state->out);
4289f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard                xmlFree(state->in);
4299f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard                state->size = 0;
4309f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard                xz_error(state, LZMA_MEM_ERROR, "out of memory");
4319f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard                return -1;
4329f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard            }
4339f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard            state->init = 1;
4346bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
4356bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
4366bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
4376bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
4386bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* get some data in the input buffer */
4396bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (strm->avail_in == 0) {
4406bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (xz_avail(state) == -1)
4416bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
4426bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (strm->avail_in == 0)
4436bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return 0;
4446bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
4456bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
446adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* look for the xz magic header bytes */
4476bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (is_format_xz(state) || is_format_lzma(state)) {
448adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        state->how = LZMA;
449adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        state->direct = 0;
450adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        return 0;
4516bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
4526bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#ifdef HAVE_ZLIB_H
453adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* look for the gzip magic header bytes 31 and 139 */
4546bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (strm->next_in[0] == 31) {
4556bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        strm->avail_in--;
4566bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        strm->next_in++;
4576bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (strm->avail_in == 0 && xz_avail(state) == -1)
4586bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
4596bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (strm->avail_in && strm->next_in[0] == 139) {
4606bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            /* we have a gzip header, woo hoo! */
4616bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            strm->avail_in--;
4626bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            strm->next_in++;
4636bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
4646bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            /* skip rest of header */
465adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (NEXT() != 8) {  /* compression method */
466adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                xz_error(state, LZMA_DATA_ERROR,
467adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                         "unknown compression method");
4686bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                return -1;
4696bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            }
4706bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            flags = NEXT();
471adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (flags & 0xe0) { /* reserved flag bits */
472adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                xz_error(state, LZMA_DATA_ERROR,
473adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                         "unknown header flags set");
4746bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                return -1;
4756bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            }
476adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            NEXT();             /* modification time */
4776bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            NEXT();
4786bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            NEXT();
4796bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            NEXT();
480adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            NEXT();             /* extra flags */
481adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            NEXT();             /* operating system */
482adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (flags & 4) {    /* extra field */
483adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                len = (unsigned) NEXT();
484adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                len += (unsigned) NEXT() << 8;
4856bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                while (len--)
4866bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                    if (NEXT() < 0)
4876bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                        break;
4886bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            }
489adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (flags & 8)      /* file name */
490adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                while (NEXT() > 0) ;
491adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (flags & 16)     /* comment */
492adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                while (NEXT() > 0) ;
493adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (flags & 2) {    /* header crc */
4946bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                NEXT();
4956bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                NEXT();
4966bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            }
4976bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            /* an unexpected end of file is not checked for here -- it will be
498adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard             * noticed on the first request for uncompressed data */
4996bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
5006bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            /* set up for decompression */
5016bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            inflateReset(&state->zstrm);
5026bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->zstrm.adler = crc32(0L, Z_NULL, 0);
5036bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->how = GZIP;
5046bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->direct = 0;
5056bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return 0;
506adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        } else {
5076bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            /* not a gzip file -- save first byte (31) and fall to raw i/o */
5086bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->out[0] = 31;
5096bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->have = 1;
5106bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
5116bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
5126bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
5136bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
5146bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* doing raw i/o, save start of raw data for seeking, copy any leftover
515adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard     * input to output -- this assumes that the output buffer is larger than
516adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard     * the input buffer, which also assures space for gzungetc() */
5176bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->raw = state->pos;
5186bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->next = state->out;
5196bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (strm->avail_in) {
5206bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        memcpy(state->next + state->have, strm->next_in, strm->avail_in);
5216bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->have += strm->avail_in;
5226bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        strm->avail_in = 0;
5236bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
5246bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->how = COPY;
5256bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->direct = 1;
5266bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return 0;
5276bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
5286bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
529adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardstatic int
530adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardxz_decomp(xz_statep state)
5316bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
5326bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    int ret;
5336bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    unsigned had;
5346bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    unsigned long crc, len;
5356bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    lzma_stream *strm = &(state->strm);
5366bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
5376bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    lzma_action action = LZMA_RUN;
5386bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
5396bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* fill output buffer up to end of deflate stream */
5406bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    had = strm->avail_out;
5416bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    do {
5426bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* get more input for inflate() */
5436bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (strm->avail_in == 0 && xz_avail(state) == -1)
5446bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
5456bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (strm->avail_in == 0) {
5466bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            xz_error(state, LZMA_DATA_ERROR, "unexpected end of file");
5476bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
5486bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
5496bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (state->eof)
5506bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            action = LZMA_FINISH;
551adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard
5526bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* decompress and handle errors */
5536bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#ifdef HAVE_ZLIB_H
5546bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (state->how == GZIP) {
555adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->zstrm.avail_in = (uInt) state->strm.avail_in;
556adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->zstrm.next_in = (Bytef *) state->strm.next_in;
557adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->zstrm.avail_out = (uInt) state->strm.avail_out;
558adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->zstrm.next_out = (Bytef *) state->strm.next_out;
559adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            ret = inflate(&state->zstrm, Z_NO_FLUSH);
560adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
561adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                xz_error(state, Z_STREAM_ERROR,
562adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                         "internal error: inflate stream corrupt");
563adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                return -1;
564adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            }
565adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (ret == Z_MEM_ERROR)
566adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                ret = LZMA_MEM_ERROR;
567adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (ret == Z_DATA_ERROR)
568adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                ret = LZMA_DATA_ERROR;
569adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (ret == Z_STREAM_END)
570adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                ret = LZMA_STREAM_END;
571adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->strm.avail_in = state->zstrm.avail_in;
572adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->strm.next_in = state->zstrm.next_in;
573adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->strm.avail_out = state->zstrm.avail_out;
574adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->strm.next_out = state->zstrm.next_out;
575adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        } else                  /* state->how == LZMA */
5766bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
577adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            ret = lzma_code(strm, action);
5786bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (ret == LZMA_MEM_ERROR) {
5796bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            xz_error(state, LZMA_MEM_ERROR, "out of memory");
5806bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
5816bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
5826bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (ret == LZMA_DATA_ERROR) {
5836bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            xz_error(state, LZMA_DATA_ERROR, "compressed data error");
5846bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
5856bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
586f0709e3ca8f8947f2d91ed34e92e38a4c23eae63Daniel Veillard        if (ret == LZMA_PROG_ERROR) {
587f0709e3ca8f8947f2d91ed34e92e38a4c23eae63Daniel Veillard            xz_error(state, LZMA_PROG_ERROR, "compression error");
588f0709e3ca8f8947f2d91ed34e92e38a4c23eae63Daniel Veillard            return -1;
589f0709e3ca8f8947f2d91ed34e92e38a4c23eae63Daniel Veillard        }
5906bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    } while (strm->avail_out && ret != LZMA_STREAM_END);
5916bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
5926bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* update available output and crc check value */
5936bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->have = had - strm->avail_out;
5946bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    state->next = strm->next_out - state->have;
5956bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#ifdef HAVE_ZLIB_H
596adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state->zstrm.adler =
597adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        crc32(state->zstrm.adler, state->next, state->have);
5986bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
5996bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
6006bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (ret == LZMA_STREAM_END) {
6016bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#ifdef HAVE_ZLIB_H
6026bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (state->how == GZIP) {
603adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (gz_next4(state, &crc) == -1 || gz_next4(state, &len) == -1) {
604adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                xz_error(state, LZMA_DATA_ERROR, "unexpected end of file");
605adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                return -1;
606adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            }
607adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (crc != state->zstrm.adler) {
608adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                xz_error(state, LZMA_DATA_ERROR, "incorrect data check");
609adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                return -1;
610adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            }
611adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            if (len != (state->zstrm.total_out & 0xffffffffL)) {
612adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                xz_error(state, LZMA_DATA_ERROR, "incorrect length check");
613adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                return -1;
614adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            }
615adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->strm.avail_in = 0;
616adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->strm.next_in = NULL;
617adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->strm.avail_out = 0;
618adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            state->strm.next_out = NULL;
619adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        } else
6206bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
621adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        if (strm->avail_in != 0 || !state->eof) {
6226bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            xz_error(state, LZMA_DATA_ERROR, "trailing garbage");
6236bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
6246bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
6256bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->how = LOOK;      /* ready for next stream, once have is 0 (leave
626adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                                 * state->direct unchanged to remember how) */
6276bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
6286bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
6296bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* good decompression */
6306bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return 0;
6316bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
6326bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
633adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardstatic int
634adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardxz_make(xz_statep state)
6356bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
6366bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    lzma_stream *strm = &(state->strm);
6376bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
638adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    if (state->how == LOOK) {   /* look for lzma / gzip header */
6396bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (xz_head(state) == -1)
6406bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
641adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        if (state->have)        /* got some data from xz_head() */
6426bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return 0;
6436bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
644adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    if (state->how == COPY) {   /* straight copy */
645adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        if (xz_load(state, state->out, state->size << 1, &(state->have)) ==
646adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            -1)
6476bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
6486bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->next = state->out;
649adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    } else if (state->how == LZMA || state->how == GZIP) {      /* decompress */
6506bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        strm->avail_out = state->size << 1;
6516bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        strm->next_out = state->out;
6526bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (xz_decomp(state) == -1)
6536bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
6546bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
6556bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return 0;
6566bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
6576bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
658adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardstatic int
659adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardxz_skip(xz_statep state, uint64_t len)
6606bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
6616bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    unsigned n;
6626bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
6636bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* skip over len bytes or reach end-of-file, whichever comes first */
6646bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    while (len)
6656bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* skip over whatever is in output buffer */
6666bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (state->have) {
667adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            n = (uint64_t) state->have > len ?
668adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                (unsigned) len : state->have;
6696bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->have -= n;
6706bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->next += n;
6716bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->pos += n;
6726bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            len -= n;
6736bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
6746bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
675adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* output buffer empty -- return if we're at the end of the input */
6766bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        else if (state->eof && state->strm.avail_in == 0)
6776bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            break;
6786bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
679adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    /* need more data to skip -- load up output buffer */
6806bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        else {
6816bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            /* get more output, looking for header if required */
6826bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            if (xz_make(state) == -1)
6836bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                return -1;
6846bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
6856bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return 0;
6866bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
6876bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
688adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardint
689adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard__libxml2_xzread(xzFile file, void *buf, unsigned len)
6906bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
6916bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    unsigned got, n;
6926bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    xz_statep state;
6936bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    lzma_stream *strm;
6946bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
6956bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* get internal structure */
6966bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (file == NULL)
6976bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return -1;
698adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state = (xz_statep) file;
6996bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    strm = &(state->strm);
7006bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7016bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* check that we're reading and that there's no error */
7026bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (state->err != LZMA_OK)
7036bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return -1;
7046bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7056bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* since an int is returned, make sure len fits in one, otherwise return
706adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard     * with an error (this avoids the flaw in the interface) */
707adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    if ((int) len < 0) {
708adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        xz_error(state, LZMA_BUF_ERROR,
709adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard                 "requested length does not fit in int");
7106bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return -1;
7116bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
7126bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7136bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* if len is zero, avoid unnecessary operations */
7146bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (len == 0)
7156bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return 0;
7166bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7176bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* process a skip request */
7186bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (state->seek) {
7196bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->seek = 0;
7206bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (xz_skip(state, state->skip) == -1)
7216bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            return -1;
7226bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
7236bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7246bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* get len bytes to buf, or less than len if at the end */
7256bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    got = 0;
7266bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    do {
7276bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* first just try copying data from the output buffer */
7286bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        if (state->have) {
7296bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            n = state->have > len ? len : state->have;
7306bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            memcpy(buf, state->next, n);
7316bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->next += n;
7326bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->have -= n;
7336bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
7346bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7356bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* output buffer empty -- return if we're at the end of the input */
7366bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        else if (state->eof && strm->avail_in == 0)
7376bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            break;
7386bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7396bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* need output data -- for small len or new stream load up our output
740adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard         * buffer */
7416bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        else if (state->how == LOOK || len < (state->size << 1)) {
7426bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            /* get more output, looking for header if required */
7436bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            if (xz_make(state) == -1)
7446bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                return -1;
745adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard            continue;           /* no progress yet -- go back to memcpy() above */
7466bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            /* the copy above assures that we will leave with space in the
747adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard             * output buffer, allowing at least one gzungetc() to succeed */
7486bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
7496bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7506bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* large len -- read directly into user buffer */
751adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        else if (state->how == COPY) {  /* read directly */
7526bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            if (xz_load(state, buf, len, &n) == -1)
7536bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                return -1;
7546bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
7556bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7566bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* large len -- decompress directly into user buffer */
757adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        else {                  /* state->how == LZMA */
7586bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            strm->avail_out = len;
7596bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            strm->next_out = buf;
7606bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            if (xz_decomp(state) == -1)
7616bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund                return -1;
7626bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            n = state->have;
7636bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund            state->have = 0;
7646bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        }
7656bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7666bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        /* update progress */
7676bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        len -= n;
768adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard        buf = (char *) buf + n;
7696bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        got += n;
7706bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        state->pos += n;
7716bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    } while (len);
7726bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7736bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* return number of bytes read into user buffer (will fit in int) */
774adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    return (int) got;
7756bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
7766bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
777adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillardint
778adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard__libxml2_xzclose(xzFile file)
7796bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund{
7806bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    int ret;
7816bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    xz_statep state;
7826bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7836bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* get internal structure */
7846bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (file == NULL)
7856bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        return LZMA_DATA_ERROR;
786adf5ec9496828b18e683fb23bd1e670824cfb6d0Daniel Veillard    state = (xz_statep) file;
7876bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund
7886bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    /* free memory and close file */
7896bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    if (state->size) {
7906bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund        lzma_end(&(state->strm));
7919f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard#ifdef HAVE_ZLIB_H
7929f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard        if (state->init == 1)
7939f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard            inflateEnd(&(state->zstrm));
7949f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard        state->init = 0;
7956bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund#endif
7969f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard        xmlFree(state->out);
7979f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard        xmlFree(state->in);
7986bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    }
7999f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard    xmlFree(state->path);
8006bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    ret = close(state->fd);
8019f3cdef08a5d45c82c71bf740a54e2bc5d07f3ecDaniel Veillard    xmlFree(state);
8026bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund    return ret ? ret : LZMA_OK;
8036bdc774350d11117849a53f7baf2bb7e26b00009Anders F Bjorklund}
80418b8988511b0954272cac4d6c3e6724f9dbf6e0aDaniel Veillard#endif /* LIBXML_LZMA_ENABLED */
805