apprentice.c

来自「sleuthit-2.09 一个磁盘的工具集」· C语言 代码 · 共 1,854 行 · 第 1/3 页

			case 'd':
				return 0;
			default:
				return -1;
			}

		case 'i':
		case 'c':
		case 'd':
		case 'u':
		case 'x':
		case 'X':
			return 0;
			
		default:
			return -1;
		}
		
	case FILE_FMT_STR:
		if (*ptr == '-')
			ptr++;
		while (isdigit((unsigned char )*ptr))
			ptr++;
		if (*ptr == '.') {
			ptr++;
			while (isdigit((unsigned char )*ptr))
				ptr++;
		}
		
		switch (*ptr++) {
		case 's':
			return 0;
		default:
			return -1;
		}
		
	default:
		/* internal error */
		abort();
	}
	/*NOTREACHED*/
	return -1;
}
	
/*
 * Check that the optional printf format in description matches
 * the type of the magic.
 */
private int
check_format(struct magic_set *ms, struct magic *m)
{
	char *ptr;

	for (ptr = m->desc; *ptr; ptr++)
		if (*ptr == '%')
			break;
	if (*ptr == '\0') {
		/* No format string; ok */
		return 1;
	}

	assert(file_nformats == file_nnames);

	if (m->type >= file_nformats) {
		file_error(ms, 0, "Internal error inconsistency between "
		    "m->type and format strings");		
		return -1;
	}
	if (file_formats[m->type] == FILE_FMT_NONE) {
		file_error(ms, 0, "No format string for `%s' with description "
		    "`%s'", m->desc, file_names[m->type]);
		return -1;
	}

	ptr++;
	if (check_format_type(ptr, file_formats[m->type]) == -1) {
		/*
		 * TODO: this error message is unhelpful if the format
		 * string is not one character long
		 */
		file_error(ms, 0, "Printf format `%c' is not valid for type "
		    " `%s' in description `%s'", *ptr,
		    file_names[m->type], m->desc);
		return -1;
	}
	
	for (; *ptr; ptr++) {
		if (*ptr == '%') {
			file_error(ms, 0,
			    "Too many format strings (should have at most one) "
			    "for `%s' with description `%s'",
			    file_names[m->type], m->desc);
			return -1;
		}
	}
	return 0;
}

/* 
 * Read a numeric value from a pointer, into the value union of a magic 
 * pointer, according to the magic type.  Update the string pointer to point 
 * just after the number read.  Return 0 for success, non-zero for failure.
 */
private int
getvalue(struct magic_set *ms, struct magic *m, const char **p, int action)
{
	int slen;

	switch (m->type) {
	case FILE_BESTRING16:
	case FILE_LESTRING16:
	case FILE_STRING:
	case FILE_PSTRING:
	case FILE_REGEX:
	case FILE_SEARCH:
		*p = getstr(ms, *p, m->value.s, sizeof(m->value.s), &slen, action);
		if (*p == NULL) {
			if (ms->flags & MAGIC_CHECK)
				file_magwarn(ms, "cannot get string from `%s'",
				    m->value.s);
			return -1;
		}
		m->vallen = slen;
		return 0;
	default:
		if (m->reln != 'x') {
			char *ep;
			m->value.q = file_signextend(ms, m,
			    (uint64_t)strtoull(*p, &ep, 0));
			*p = ep;
			eatsize(p);
		}
		return 0;
	}
}

/*
 * Convert a string containing C character escapes.  Stop at an unescaped
 * space or tab.
 * Copy the converted version to "p", returning its length in *slen.
 * Return updated scan pointer as function result.
 */
private const char *
getstr(struct magic_set *ms, const char *s, char *p, int plen, int *slen, int action)
{
	const char *origs = s;
	char 	*origp = p;
	char	*pmax = p + plen - 1;
	int	c;
	int	val;

	while ((c = *s++) != '\0') {
		if (isspace((unsigned char) c))
			break;
		if (p >= pmax) {
			file_error(ms, 0, "string too long: `%s'", origs);
			return NULL;
		}
		if (c == '\\') {
			switch(c = *s++) {

			case '\0':
				if (action == FILE_COMPILE)
					file_magwarn(ms, "incomplete escape");
				goto out;

			case '\t':
				if (action == FILE_COMPILE) {
					file_magwarn(ms,
					    "escaped tab found, use \\t instead");
					action++;
				}
				/*FALLTHROUGH*/
			default:
				if (action == FILE_COMPILE) {
					if (isprint((unsigned char)c))
					    file_magwarn(ms,
						"no need to escape `%c'", c);
					else
					    file_magwarn(ms,
						"unknown escape sequence: \\%03o", c);
				}
				/*FALLTHROUGH*/
			/* space, perhaps force people to use \040? */
			case ' ':
#if 0
			/*
			 * Other things people escape, but shouldn't need to,
			 * so we disallow them
			 */
			case '\'':
			case '"':
			case '?':
#endif
			/* Relations */
			case '>':
			case '<':
			case '&':
			case '^':
			case '=':
			case '!':
			/* and baskslash itself */
			case '\\':
				*p++ = (char) c;
				break;

			case 'a':
				*p++ = '\a';
				break;

			case 'b':
				*p++ = '\b';
				break;

			case 'f':
				*p++ = '\f';
				break;

			case 'n':
				*p++ = '\n';
				break;

			case 'r':
				*p++ = '\r';
				break;

			case 't':
				*p++ = '\t';
				break;

			case 'v':
				*p++ = '\v';
				break;

			/* \ and up to 3 octal digits */
			case '0':
			case '1':
			case '2':
			case '3':
			case '4':
			case '5':
			case '6':
			case '7':
				val = c - '0';
				c = *s++;  /* try for 2 */
				if (c >= '0' && c <= '7') {
					val = (val << 3) | (c - '0');
					c = *s++;  /* try for 3 */
					if (c >= '0' && c <= '7')
						val = (val << 3) | (c-'0');
					else
						--s;
				}
				else
					--s;
				*p++ = (char)val;
				break;

			/* \x and up to 2 hex digits */
			case 'x':
				val = 'x';	/* Default if no digits */
				c = hextoint(*s++);	/* Get next char */
				if (c >= 0) {
					val = c;
					c = hextoint(*s++);
					if (c >= 0)
						val = (val << 4) + c;
					else
						--s;
				} else
					--s;
				*p++ = (char)val;
				break;
			}
		} else
			*p++ = (char)c;
	}
out:
	*p = '\0';
	*slen = p - origp;
	return s;
}


/* Single hex char to int; -1 if not a hex char. */
private int
hextoint(int c)
{
	if (!isascii((unsigned char) c))
		return -1;
	if (isdigit((unsigned char) c))
		return c - '0';
	if ((c >= 'a') && (c <= 'f'))
		return c + 10 - 'a';
	if (( c>= 'A') && (c <= 'F'))
		return c + 10 - 'A';
	return -1;
}


/*
 * Print a string containing C character escapes.
 */
protected void
file_showstr(FILE *fp, const char *s, size_t len)
{
	char	c;

	for (;;) {
		c = *s++;
		if (len == ~0U) {
			if (c == '\0')
				break;
		}
		else  {
			if (len-- == 0)
				break;
		}
		if (c >= 040 && c <= 0176)	/* TODO isprint && !iscntrl */
			(void) fputc(c, fp);
		else {
			(void) fputc('\\', fp);
			switch (c) {
			case '\a':
				(void) fputc('a', fp);
				break;

			case '\b':
				(void) fputc('b', fp);
				break;

			case '\f':
				(void) fputc('f', fp);
				break;

			case '\n':
				(void) fputc('n', fp);
				break;

			case '\r':
				(void) fputc('r', fp);
				break;

			case '\t':
				(void) fputc('t', fp);
				break;

			case '\v':
				(void) fputc('v', fp);
				break;

			default:
				(void) fprintf(fp, "%.3o", c & 0377);
				break;
			}
		}
	}
}

/*
 * eatsize(): Eat the size spec from a number [eg. 10UL]
 */
private void
eatsize(const char **p)
{
	const char *l = *p;

	if (LOWCASE(*l) == 'u') 
		l++;

	switch (LOWCASE(*l)) {
	case 'l':    /* long */
	case 's':    /* short */
	case 'h':    /* short */
	case 'b':    /* char/byte */
	case 'c':    /* char/byte */
		l++;
		/*FALLTHROUGH*/
	default:
		break;
	}

	*p = l;
}

/*
 * handle a compiled file.
 */
private int
apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
    const char *fn)
{
	int fd;
	struct stat st;
	uint32_t *ptr;
	uint32_t version;
	int needsbyteswap;
	char buf[MAXPATHLEN];
	char *dbname = mkdbname(fn, buf, sizeof(buf), 0);
	void *mm = NULL;

	if (dbname == NULL)
		return -1;

	if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1)
		return -1;

	if (fstat(fd, &st) == -1) {
		file_error(ms, errno, "cannot stat `%s'", dbname);
		goto error;
	}
	if (st.st_size < 16) {
		file_error(ms, 0, "file `%s' is too small", dbname);
		goto error;
	}

#ifdef QUICK
	if ((mm = mmap(0, (size_t)st.st_size, PROT_READ|PROT_WRITE,
	    MAP_PRIVATE|MAP_FILE, fd, (off_t)0)) == MAP_FAILED) {
		file_error(ms, errno, "cannot map `%s'", dbname);
		goto error;
	}
#define RET	2
#else
	if ((mm = malloc((size_t)st.st_size)) == NULL) {
		file_oomem(ms, (size_t)st.st_size);
		goto error;
	}
	if (read(fd, mm, (size_t)st.st_size) != (size_t)st.st_size) {
		file_badread(ms);
		goto error;
	}
#define RET	1
#endif
	*magicp = mm;
	(void)close(fd);
	fd = -1;
	ptr = (uint32_t *)(void *)*magicp;
	if (*ptr != MAGICNO) {
		if (swap4(*ptr) != MAGICNO) {
			file_error(ms, 0, "bad magic in `%s'");
			goto error;
		}
		needsbyteswap = 1;
	} else
		needsbyteswap = 0;
	if (needsbyteswap)
		version = swap4(ptr[1]);
	else
		version = ptr[1];
	if (version != VERSIONNO) {
		file_error(ms, 0, "version mismatch (%d != %d) in `%s'",
		    version, VERSIONNO, dbname);
		goto error;
	}
	*nmagicp = (uint32_t)(st.st_size / sizeof(struct magic)) - 1;
	(*magicp)++;
	if (needsbyteswap)
		byteswap(*magicp, *nmagicp);
	return RET;

error:
	if (fd != -1)
		(void)close(fd);
	if (mm) {
#ifdef QUICK
		(void)munmap((void *)mm, (size_t)st.st_size);
#else
		free(mm);
#endif
	} else {
		*magicp = NULL;
		*nmagicp = 0;
	}
	return -1;
}

private const uint32_t ar[] = {
    MAGICNO, VERSIONNO
};
/*
 * handle an mmaped file.
 */
private int
apprentice_compile(struct magic_set *ms, struct magic **magicp,
    uint32_t *nmagicp, const char *fn)
{
	int fd;
	char buf[MAXPATHLEN];
	char *dbname = mkdbname(fn, buf, sizeof(buf), 1);

	if (dbname == NULL) 
		return -1;

	if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) {
		file_error(ms, errno, "cannot open `%s'", dbname);
		return -1;
	}

	if (write(fd, ar, sizeof(ar)) != (ssize_t)sizeof(ar)) {
		file_error(ms, errno, "error writing `%s'", dbname);
		return -1;
	}

	if (lseek(fd, (off_t)sizeof(struct magic), SEEK_SET)
	    != sizeof(struct magic)) {
		file_error(ms, errno, "error seeking `%s'", dbname);
		return -1;
	}

	if (write(fd, *magicp, (sizeof(struct magic) * *nmagicp)) 
	    != (ssize_t)(sizeof(struct magic) * *nmagicp)) {
		file_error(ms, errno, "error writing `%s'", dbname);
		return -1;
	}

	(void)close(fd);
	return 0;
}

private const char ext[] = ".mgc";
/*
 * make a dbname
 */
private char *
mkdbname(const char *fn, char *buf, size_t bufsiz, int strip)
{
	if (strip) {
		const char *p;
		if ((p = strrchr(fn, '/')) != NULL)
			fn = ++p;
	}

	(void)snprintf(buf, bufsiz, "%s%s", fn, ext);
	return buf;
}

/*
 * Byteswap an mmap'ed file if needed
 */
private void
byteswap(struct magic *magic, uint32_t nmagic)
{
	uint32_t i;
	for (i = 0; i < nmagic; i++)
		bs1(&magic[i]);
}

/*
 * swap a short
 */
private uint16_t
swap2(uint16_t sv)
{
	uint16_t rv;
	uint8_t *s = (uint8_t *)(void *)&sv; 
	uint8_t *d = (uint8_t *)(void *)&rv; 
	d[0] = s[1];
	d[1] = s[0];
	return rv;
}

/*
 * swap an int
 */
private uint32_t
swap4(uint32_t sv)
{
	uint32_t rv;
	uint8_t *s = (uint8_t *)(void *)&sv; 
	uint8_t *d = (uint8_t *)(void *)&rv; 
	d[0] = s[3];
	d[1] = s[2];
	d[2] = s[1];
	d[3] = s[0];
	return rv;
}

/*
 * swap a quad
 */
private uint64_t
swap8(uint64_t sv)
{
	uint32_t rv;
	uint8_t *s = (uint8_t *)(void *)&sv; 
	uint8_t *d = (uint8_t *)(void *)&rv; 
	d[0] = s[3];
	d[1] = s[2];
	d[2] = s[1];
	d[3] = s[0];
	d[4] = s[7];
	d[5] = s[6];
	d[6] = s[5];
	d[7] = s[4];
	return rv;
}

/*
 * byteswap a single magic entry
 */
private void
bs1(struct magic *m)
{
	m->cont_level = swap2(m->cont_level);
	m->offset = swap4((uint32_t)m->offset);
	m->in_offset = swap4((uint32_t)m->in_offset);
	m->lineno = swap4((uint32_t)m->lineno);
	if (IS_STRING(m->type)) {
		m->str_count = swap4(m->str_count);
		m->str_flags = swap4(m->str_flags);
	}
	else {
		m->value.q = swap8(m->value.q);
		m->num_mask = swap8(m->num_mask);
	}
}