gzip.c

一些关于HTTP协议处理的文档,还包括了gzip包(它是用来解压缩HTTP传送的压缩数据)

	    return ERROR;
	}
	/* Create the output file */
	remove_ofname = 1;
	ofd = OPEN(ofname, flags, RW_USER);
	if (ofd == -1) {
	    perror(ofname);
	    close(ifd);
	    exit_code = ERROR;
	    return ERROR;
	}

	/* Check for name truncation on new file (1234567890123.gz) */
#ifdef NO_FSTAT
	if (stat(ofname, &ostat) != 0) {
#else
	if (fstat(ofd, &ostat) != 0) {
#endif
	    fprintf(stderr, "%s: ", progname);
	    perror(ofname);
	    close(ifd); close(ofd);
	    unlink(ofname);
	    exit_code = ERROR;
	    return ERROR;
	}
	if (!name_too_long(ofname, &ostat)) return OK;

	if (decompress) {
	    /* name might be too long if an original name was saved */
	    WARN((stderr, "%s: %s: warning, name truncated\n",
		  progname, ofname));
	    return OK;
	}
	close(ofd);
	unlink(ofname);
#ifdef NO_MULTIPLE_DOTS
	/* Should never happen, see check_ofname() */
	fprintf(stderr, "%s: %s: name too long\n", progname, ofname);
	do_exit(ERROR);
#endif
	shorten_name(ofname);
    }
}

/* ========================================================================
 * Use lstat if available, except for -c or -f. Use stat otherwise.
 * This allows links when not removing the original file.
 */
local int do_stat(name, sbuf)
    char *name;
    struct stat *sbuf;
{
    errno = 0;
#if (defined(S_IFLNK) || defined (S_ISLNK)) && !defined(NO_SYMLINK)
    if (!to_stdout && !force) {
	return lstat(name, sbuf);
    }
#endif
    return stat(name, sbuf);
}

/* ========================================================================
 * Return a pointer to the 'z' suffix of a file name, or NULL. For all
 * systems, ".gz", ".z", ".Z", ".taz", ".tgz", "-gz", "-z" and "_z" are
 * accepted suffixes, in addition to the value of the --suffix option.
 * ".tgz" is a useful convention for tar.z files on systems limited
 * to 3 characters extensions. On such systems, ".?z" and ".??z" are
 * also accepted suffixes. For Unix, we do not want to accept any
 * .??z suffix as indicating a compressed file; some people use .xyz
 * to denote volume data.
 *   On systems allowing multiple versions of the same file (such as VMS),
 * this function removes any version suffix in the given name.
 */
local char *get_suffix(name)
    char *name;
{
    int nlen, slen;
    char suffix[MAX_SUFFIX+3]; /* last chars of name, forced to lower case */
    static char *known_suffixes[] =
       {z_suffix, ".gz", ".z", ".taz", ".tgz", "-gz", "-z", "_z",
#ifdef MAX_EXT_CHARS
          "z",
#endif
          NULL};
    char **suf = known_suffixes;

    if (strequ(z_suffix, "z")) suf++; /* check long suffixes first */

#ifdef SUFFIX_SEP
    /* strip a version number from the file name */
    {
	char *v = strrchr(name, SUFFIX_SEP);
 	if (v != NULL) *v = '\0';
    }
#endif
    nlen = strlen(name);
    if (nlen <= MAX_SUFFIX+2) {
        strcpy(suffix, name);
    } else {
        strcpy(suffix, name+nlen-MAX_SUFFIX-2);
    }
    strlwr(suffix);
    slen = strlen(suffix);
    do {
       int s = strlen(*suf);
       if (slen > s && suffix[slen-s-1] != PATH_SEP
           && strequ(suffix + slen - s, *suf)) {
           return name+nlen-s;
       }
    } while (*++suf != NULL);

    return NULL;
}


/* ========================================================================
 * Set ifname to the input file name (with a suffix appended if necessary)
 * and istat to its stats. For decompression, if no file exists with the
 * original name, try adding successively z_suffix, .gz, .z, -z and .Z.
 * For MSDOS, we try only z_suffix and z.
 * Return OK or ERROR.
 */
local int get_istat(iname, sbuf)
    char *iname;
    struct stat *sbuf;
{
    int ilen;  /* strlen(ifname) */
    static char *suffixes[] = {z_suffix, ".gz", ".z", "-z", ".Z", NULL};
    char **suf = suffixes;
    char *s;
#ifdef NO_MULTIPLE_DOTS
    char *dot; /* pointer to ifname extension, or NULL */
#endif

    strcpy(ifname, iname);

    /* If input file exists, return OK. */
    if (do_stat(ifname, sbuf) == 0) return OK;

    if (!decompress || errno != ENOENT) {
	perror(ifname);
	exit_code = ERROR;
	return ERROR;
    }
    /* file.ext doesn't exist, try adding a suffix (after removing any
     * version number for VMS).
     */
    s = get_suffix(ifname);
    if (s != NULL) {
	perror(ifname); /* ifname already has z suffix and does not exist */
	exit_code = ERROR;
	return ERROR;
    }
#ifdef NO_MULTIPLE_DOTS
    dot = strrchr(ifname, '.');
    if (dot == NULL) {
        strcat(ifname, ".");
        dot = strrchr(ifname, '.');
    }
#endif
    ilen = strlen(ifname);
    if (strequ(z_suffix, ".gz")) suf++;

    /* Search for all suffixes */
    do {
        s = *suf;
#ifdef NO_MULTIPLE_DOTS
        if (*s == '.') s++;
#endif
#ifdef MAX_EXT_CHARS
        strcpy(ifname, iname);
        /* Needed if the suffixes are not sorted by increasing length */

        if (*dot == '\0') strcpy(dot, ".");
        dot[MAX_EXT_CHARS+1-strlen(s)] = '\0';
#endif
        strcat(ifname, s);
        if (do_stat(ifname, sbuf) == 0) return OK;
	ifname[ilen] = '\0';
    } while (*++suf != NULL);

    /* No suffix found, complain using z_suffix: */
#ifdef MAX_EXT_CHARS
    strcpy(ifname, iname);
    if (*dot == '\0') strcpy(dot, ".");
    dot[MAX_EXT_CHARS+1-z_len] = '\0';
#endif
    strcat(ifname, z_suffix);
    perror(ifname);
    exit_code = ERROR;
    return ERROR;
}

/* ========================================================================
 * Generate ofname given ifname. Return OK, or WARNING if file must be skipped.
 * Sets save_orig_name to true if the file name has been truncated.
 */
local int make_ofname()
{
    char *suff;            /* ofname z suffix */

    strcpy(ofname, ifname);
    /* strip a version number if any and get the gzip suffix if present: */
    suff = get_suffix(ofname);

    if (decompress) {
	if (suff == NULL) {
	    /* Whith -t or -l, try all files (even without .gz suffix)
	     * except with -r (behave as with just -dr).
             */
            if (!recursive && (list || test)) return OK;

	    /* Avoid annoying messages with -r */
	    if (verbose || (!recursive && !quiet)) {
		WARN((stderr,"%s: %s: unknown suffix -- ignored\n",
		      progname, ifname));
	    }
	    return WARNING;
	}
	/* Make a special case for .tgz and .taz: */
	strlwr(suff);
	if (strequ(suff, ".tgz") || strequ(suff, ".taz")) {
	    strcpy(suff, ".tar");
	} else {
	    *suff = '\0'; /* strip the z suffix */
	}
        /* ofname might be changed later if infile contains an original name */

    } else if (suff != NULL) {
	/* Avoid annoying messages with -r (see treat_dir()) */
	if (verbose || (!recursive && !quiet)) {
	    fprintf(stderr, "%s: %s already has %s suffix -- unchanged\n",
		    progname, ifname, suff);
	}
	if (exit_code == OK) exit_code = WARNING;
	return WARNING;
    } else {
        save_orig_name = 0;

#ifdef NO_MULTIPLE_DOTS
	suff = strrchr(ofname, '.');
	if (suff == NULL) {
            strcat(ofname, ".");
#  ifdef MAX_EXT_CHARS
	    if (strequ(z_suffix, "z")) {
		strcat(ofname, "gz"); /* enough room */
		return OK;
	    }
        /* On the Atari and some versions of MSDOS, name_too_long()
         * does not work correctly because of a bug in stat(). So we
         * must truncate here.
         */
        } else if (strlen(suff)-1 + z_len > MAX_SUFFIX) {
            suff[MAX_SUFFIX+1-z_len] = '\0';
            save_orig_name = 1;
#  endif
        }
#endif /* NO_MULTIPLE_DOTS */
	strcat(ofname, z_suffix);

    } /* decompress ? */
    return OK;
}


/* ========================================================================
 * Check the magic number of the input file and update ofname if an
 * original name was given and to_stdout is not set.
 * Return the compression method, -1 for error, -2 for warning.
 * Set inptr to the offset of the next byte to be processed.
 * Updates time_stamp if there is one and --no-time is not used.
 * This function may be called repeatedly for an input file consisting
 * of several contiguous gzip'ed members.
 * IN assertions: there is at least one remaining compressed member.
 *   If the member is a zip file, it must be the only one.
 */
local int get_method(in)
    int in;        /* input file descriptor */
{
    uch flags;     /* compression flags */
    char magic[2]; /* magic header */
    ulg stamp;     /* time stamp */

    /* If --force and --stdout, zcat == cat, so do not complain about
     * premature end of file: use try_byte instead of get_byte.
     */
    if (force && to_stdout) {
	magic[0] = (char)try_byte();
	magic[1] = (char)try_byte();
	/* If try_byte returned EOF, magic[1] == 0xff */
    } else {
	magic[0] = (char)get_byte();
	magic[1] = (char)get_byte();
    }
    method = -1;                 /* unknown yet */
    part_nb++;                   /* number of parts in gzip file */
    header_bytes = 0;
    last_member = RECORD_IO;
    /* assume multiple members in gzip file except for record oriented I/O */

    if (memcmp(magic, GZIP_MAGIC, 2) == 0
        || memcmp(magic, OLD_GZIP_MAGIC, 2) == 0) {

	method = (int)get_byte();
	if (method != DEFLATED) {
	    fprintf(stderr,
		    "%s: %s: unknown method %d -- get newer version of gzip\n",
		    progname, ifname, method);
	    exit_code = ERROR;
	    return -1;
	}
	work = unzip;
	flags  = (uch)get_byte();

	if ((flags & ENCRYPTED) != 0) {
	    fprintf(stderr,
		    "%s: %s is encrypted -- get newer version of gzip\n",
		    progname, ifname);
	    exit_code = ERROR;
	    return -1;
	}
	if ((flags & CONTINUATION) != 0) {
	    fprintf(stderr,
	   "%s: %s is a a multi-part gzip file -- get newer version of gzip\n",
		    progname, ifname);
	    exit_code = ERROR;
	    if (force <= 1) return -1;
	}
	if ((flags & RESERVED) != 0) {
	    fprintf(stderr,
		    "%s: %s has flags 0x%x -- get newer version of gzip\n",
		    progname, ifname, flags);
	    exit_code = ERROR;
	    if (force <= 1) return -1;
	}
	stamp  = (ulg)get_byte();
	stamp |= ((ulg)get_byte()) << 8;
	stamp |= ((ulg)get_byte()) << 16;
	stamp |= ((ulg)get_byte()) << 24;
	if (stamp != 0 && !no_time) time_stamp = stamp;

	(void)get_byte();  /* Ignore extra flags for the moment */
	(void)get_byte();  /* Ignore OS type for the moment */

	if ((flags & CONTINUATION) != 0) {
	    unsigned part = (unsigned)get_byte();
	    part |= ((unsigned)get_byte())<<8;
	    if (verbose) {
		fprintf(stderr,"%s: %s: part number %u\n",
			progname, ifname, part);
	    }
	}
	if ((flags & EXTRA_FIELD) != 0) {
	    unsigned len = (unsigned)get_byte();
	    len |= ((unsigned)get_byte())<<8;
	    if (verbose) {
		fprintf(stderr,"%s: %s: extra field of %u bytes ignored\n",
			progname, ifname, len);
	    }
	    while (len--) (void)get_byte();
	}

	/* Get original file name if it was truncated */
	if ((flags & ORIG_NAME) != 0) {
	    if (no_name || (to_stdout && !list) || part_nb > 1) {
		/* Discard the old name */
		char c; /* dummy used for NeXTstep 3.0 cc optimizer bug */
		do {c=get_byte();} while (c != 0);
	    } else {
		/* Copy the base name. Keep a directory prefix intact. */
                char *p = basename(ofname);
                char *base = p;
		for (;;) {
		    *p = (char)get_char();
		    if (*p++ == '\0') break;
		    if (p >= ofname+sizeof(ofname)) {
			error("corrupted input -- file name too large");
		    }
		}
                /* If necessary, adapt the name to local OS conventions: */
                if (!list) {
                   MAKE_LEGAL_NAME(base);
		   if (base) list=0; /* avoid warning about unused variable */
                }
	    } /* no_name || to_stdout */
	} /* ORIG_NAME */

	/* Discard file comment if any */
	if ((flags & COMMENT) != 0) {
	    while (get_char() != 0) /* null */ ;
	}
	if (part_nb == 1) {
	    header_bytes = inptr + 2*sizeof(long); /* include crc and size */
	}

    } else if (memcmp(magic, PKZIP_MAGIC, 2) == 0 && inptr == 2
	    && memcmp((char*)inbuf, PKZIP_MAGIC, 4) == 0) {
	/* To simplify the code, we support a zip file when alone only.
         * We are thus guaranteed that the entire local header fits in inbuf.
         */
        inptr = 0;
	work = unzip;
	if (check_zipfile(in) != OK) return -1;
	/* check_zipfile may get ofname from the local header */
	last_member = 1;

    } else if (memcmp(magic, PACK_MAGIC, 2) == 0) {
	work = unpack;
	method = PACKED;

    } else if (memcmp(magic, LZW_MAGIC, 2) == 0) {
	work = unlzw;
	method = COMPRESSED;
	last_member = 1;

    } else if (memcmp(magic, LZH_MAGIC, 2) == 0) {
	work = unlzh;
	method = LZHED;
	last_member = 1;

    } else if (force && to_stdout && !list) { /* pass input unchanged */
	method = STORED;
	work = copy;
        inptr = 0;
	last_member = 1;
    }
    if (method >= 0) return method;

    if (part_nb == 1) {
	fprintf(stderr, "\n%s: %s: not in gzip format\n", progname, ifname);
	exit_code = ERROR;
	return -1;
    } else {
	WARN((stderr, "\n%s: %s: decompression OK, trailing garbage ignored\n",
	      progname, ifname));
	return -2;
    }