Browse Source

'dumpint' and related functions replaced by 'string.pack'/'string.unpack'

pull/9/head
Roberto Ierusalimschy 10 years ago
parent
commit
c172a4f7c2
  1. 497
      lstrlib.c

497
lstrlib.c

@ -1,5 +1,5 @@
/*
** $Id: lstrlib.c,v 1.201 2014/08/20 22:06:41 roberto Exp roberto $
** $Id: lstrlib.c,v 1.202 2014/10/01 11:54:56 roberto Exp roberto $
** Standard library for string operations and pattern-matching
** See Copyright Notice in lua.h
*/
@ -950,214 +950,399 @@ static int str_format (lua_State *L) {
*/
/* maximum size for the binary representation of an integer */
#define MAXINTSIZE 16
/* number of bits in a character */
#define NB CHAR_BIT
/* mask for one character (NB 1's) */
#define MC ((1 << NB) - 1)
/* mask for one character without sign bit ((NB - 1) 1's) */
#define SM (MC >> 1)
/* size of a lua_Integer */
#define SZINT ((int)sizeof(lua_Integer))
/* maximum size for the binary representation of an integer */
#define MAXINTSIZE 12
/* mask for all ones in last byte in a lua Integer */
#define HIGHERBYTE ((lua_Unsigned)MC << (NB * (SZINT - 1)))
static union {
/* dummy union to get native endianness */
static const union {
int dummy;
char little; /* true iff machine is little endian */
} const nativeendian = {1};
} nativeendian = {1};
static int getendian (lua_State *L, int arg) {
const char *endian = luaL_optstring(L, arg,
(nativeendian.little ? "l" : "b"));
if (*endian == 'n') /* native? */
return nativeendian.little;
luaL_argcheck(L, *endian == 'l' || *endian == 'b', arg,
"endianness must be 'l'/'b'/'n'");
return (*endian == 'l');
}
/* dummy structure to get native alignment requirements */
struct cD {
char c;
union { double d; void *p; lua_Integer i; lua_Number n; } u;
};
#define MAXALIGN (offsetof(struct cD, u))
static int getintsize (lua_State *L, int arg) {
lua_Integer size = luaL_optinteger(L, arg, 0);
if (size == 0) size = SZINT;
luaL_argcheck(L, 1 <= size && size <= MAXINTSIZE, arg,
"integer size out of valid range");
return (int)size;
}
/*
** Union for serializing floats
*/
typedef union Ftypes {
float f;
double d;
lua_Number n;
char buff[5 * sizeof(lua_Number)]; /* enough for any float type */
} Ftypes;
/* mask for all ones in last byte in a lua Integer */
#define HIGHERBYTE ((lua_Unsigned)MC << (NB * (SZINT - 1)))
/*
** information to pack/unpack stuff
*/
typedef struct Header {
lua_State *L;
int islittle;
int maxalign;
} Header;
static int dumpint (char *buff, lua_Integer m, int littleendian, int size) {
int i;
lua_Unsigned n = (lua_Unsigned)m;
lua_Unsigned mask = (m >= 0) ? 0 : HIGHERBYTE; /* sign extension */
if (littleendian) {
for (i = 0; i < size - 1; i++) {
buff[i] = (n & MC);
n = (n >> NB) | mask;
}
}
typedef enum KOption {Kint, Kuint, Kfloat, Kchar, Kstring, Kstring0,
Kspace, Kpadding, Kpaddalig} KOption;
/*
** Read an integer numeral from string 'fmt' or return 'df' if
** there is no numeral
*/
static int digit (int c) { return '0' <= c && c <= '9'; }
static int getnum (const char **fmt, int df) {
if (!digit(**fmt)) /* no number? */
return df; /* return default value */
else {
for (i = size - 1; i > 0; i--) {
buff[i] = (n & MC);
n = (n >> NB) | mask;
}
}
buff[i] = (n & MC); /* last byte */
if (size < SZINT) { /* need test for overflow? */
/* OK if there are only zeros left in higher bytes,
or only ones left (excluding non-signal bits in last byte) */
return ((n & ~(lua_Unsigned)MC) == 0 ||
(n | SM) == ~(lua_Unsigned)0);
int a = 0;
do {
a = a*10 + *((*fmt)++) - '0';
} while (digit(**fmt) && a < (INT_MAX/10 - 10));
return a;
}
else return 1; /* no overflow can occur with full size */
}
static int dumpint_l (lua_State *L) {
char buff[MAXINTSIZE];
lua_Integer n = luaL_checkinteger(L, 1);
int size = getintsize(L, 2);
int endian = getendian(L, 3);
if (dumpint(buff, n, endian, size))
lua_pushlstring(L, buff, size);
else
luaL_error(L, "integer does not fit into given size (%d)", size);
return 1;
/*
** Read an integer numeral and raises an error if it is larger
** than the maximum size for integers.
*/
static int getnumlimit (Header *h, const char **fmt, int df) {
int sz = getnum(fmt, df);
if (sz > MAXINTSIZE || sz <= 0)
luaL_error(h->L, "integral size (%d) out of limits [1,%d]", sz, MAXINTSIZE);
return sz;
}
/* mask to check higher-order byte + signal bit of next (lower) byte */
#define HIGHERBYTE1 (HIGHERBYTE | (HIGHERBYTE >> 1))
/*
** Reads an option endianness indication ('<'/'>') and
** returns true if operation should use little endian.
*/
static int getendian (const char **fmt) {
if (**fmt == '<' || **fmt == '>') /* explicit endianness? */
return (*((*fmt)++) == '<'); /* true iff little endian */
else /* no endian indication */
return nativeendian.little; /* use native */
}
static int undumpint (const char *buff, lua_Integer *res,
int littleendian, int size) {
lua_Unsigned n = 0;
int i;
for (i = 0; i < size; i++) {
if (i >= SZINT) { /* will throw away a byte? */
/* check for overflow: it is OK to throw away leading zeros for a
positive number, leading ones for a negative number, and a
leading zero byte to allow unsigned integers with a 1 in
its "signal bit" */
if (!((n & HIGHERBYTE1) == 0 || /* zeros for positive number */
(n & HIGHERBYTE1) == HIGHERBYTE1 || /* ones for negative number */
(i == size - 1 && (n & HIGHERBYTE) == 0))) /* leading zero */
return 0; /* overflow */
}
n <<= NB;
n |= (lua_Unsigned)(unsigned char)buff[littleendian ? size - 1 - i : i];
}
if (size < SZINT) { /* need sign extension? */
lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1);
*res = (lua_Integer)((n ^ mask) - mask); /* do sign extension */
/*
** Read and return maximum alignment to be used
*/
static int getalignment (Header *h, const char **fmt) {
if (**fmt == '!') { /* explicit alignment? */
(*fmt)++; /* skip '!' */
return getnumlimit(h, fmt, MAXALIGN);
}
else
*res = (lua_Integer)n;
return 1;
return 1; /* default is no alignment */
}
static int undumpint_l (lua_State *L) {
lua_Integer res;
size_t len;
const char *s = luaL_checklstring(L, 1, &len);
lua_Integer pos = posrelat(luaL_optinteger(L, 2, 1), len);
int size = getintsize(L, 3);
int endian = getendian(L, 4);
luaL_argcheck(L, 1 <= pos && (size_t)pos + size - 1 <= len, 1,
"string too short");
if(undumpint(s + pos - 1, &res, endian, size))
lua_pushinteger(L, res);
else
luaL_error(L, "result does not fit into a Lua integer");
return 1;
/*
** Read optional endianness and alignment indications
*/
static void getheader (lua_State *L, Header *h, const char **fmt) {
h->L = L;
h->islittle = getendian(fmt);
h->maxalign = getalignment(h, fmt);
}
static void correctendianness (lua_State *L, char *b, int size, int endianarg) {
int endian = getendian(L, endianarg);
if (endian != nativeendian.little) { /* not native endianness? */
int i = 0;
while (i < --size) {
char temp = b[i];
b[i++] = b[size];
b[size] = temp;
/*
** Read and classify next option. 'size' is filled with option's size.
*/
static KOption getoption (Header *h, const char **fmt, int *size) {
int opt = *((*fmt)++);
switch (opt) {
case 'b': *size = sizeof(char); return Kint;
case 'B': *size = sizeof(char); return Kuint;
case 'h': *size = sizeof(short); return Kint;
case 'H': *size = sizeof(short); return Kuint;
case 'l': *size = sizeof(long); return Kint;
case 'L': *size = sizeof(long); return Kuint;
case 'j': *size = sizeof(lua_Integer); return Kint;
case 'J': *size = sizeof(lua_Integer); return Kuint;
case 'T': *size = sizeof(size_t); return Kuint;
case 'f': *size = sizeof(float); return Kfloat;
case 'd': *size = sizeof(double); return Kfloat;
case 'n': *size = sizeof(lua_Number); return Kfloat;
case 'i': *size = getnumlimit(h, fmt, sizeof(int)); return Kint;
case 'I': *size = getnumlimit(h, fmt, sizeof(int)); return Kuint;
case 's': *size = getnumlimit(h, fmt, sizeof(size_t)); return Kstring;
case 'c': *size = getnum(fmt, 1); return Kchar;
case 'z': *size = 0; return Kstring0;
case 'x': *size = 1; return Kpadding;
case 'X': *size = 0; return Kpaddalig;
case ' ': *size = 0; return Kspace;
default: {
*size = 0; /* to avoid warnings */
luaL_error(h->L, "invalid format option '%c'", opt);
return (KOption)0;
}
}
}
static int getfloatsize (lua_State *L, int arg) {
const char *size = luaL_optstring(L, arg, "n");
if (*size == 'n') return sizeof(lua_Number);
luaL_argcheck(L, *size == 'd' || *size == 'f', arg,
"size must be 'f'/'d'/'n'");
return (*size == 'd' ? sizeof(double) : sizeof(float));
/*
** Read, classify, and fill other details about the next option.
** 'psize' is filled with option's size, 'notoalign' with its
** alignment requirements.
** Local variable 'size' gets the size to be aligned. (Kpadal option
** always gets its full alignment, other options are limited by
** the maximum alignment ('maxalign). Kchar option needs no aligment
** despite its size.
*/
static KOption getdetails (Header *h, size_t totalsize,
const char **fmt, int *psize, int *ntoalign) {
KOption opt = getoption(h, fmt, psize);
int align = *psize; /* usually, alignment follows size */
if (opt == Kpaddalig) {
if (**fmt == '\0' || strchr("Xc ", **fmt) != NULL)
luaL_argerror(h->L, 1, "invalid next option for option 'X'");
getoption(h, fmt, &align); /* get next element's size for alignment */
}
if (align <= 1 || opt == Kchar) /* need no alignment? */
*ntoalign = 0;
else {
if (align > h->maxalign)
align = h->maxalign;
if ((align & (align - 1)) != 0) /* is 'align' not a power of 2? */
luaL_argerror(h->L, 1, "format asks for alignment not power of 2");
*ntoalign = (align - (int)(totalsize & (align - 1))) & (align - 1);
}
return opt;
}
static int dumpfloat_l (lua_State *L) {
float f; double d;
char *pn; /* pointer to number */
lua_Number n = luaL_checknumber(L, 1);
int size = getfloatsize(L, 2);
if (size == sizeof(lua_Number))
pn = (char*)&n;
else if (size == sizeof(float)) {
f = (float)n;
pn = (char*)&f;
}
else { /* native lua_Number may be neither float nor double */
lua_assert(size == sizeof(double));
d = (double)n;
pn = (char*)&d;
static void packint (luaL_Buffer *b, lua_Unsigned n,
int islittle, int size, lua_Unsigned mask) {
char *buff = luaL_prepbuffsize(b, size);
int i;
for (i = 0; i < size - 1; i++) {
buff[islittle ? i : size - 1 - i] = (n & MC);
n = (n >> NB) | mask;
}
correctendianness(L, pn, size, 3);
lua_pushlstring(L, pn, size);
return 1;
buff[islittle ? i : size - 1 - i] = (n & MC);
luaL_addsize(b, size); /* add result to buffer */
}
static int undumpfloat_l (lua_State *L) {
lua_Number res;
size_t len;
const char *s = luaL_checklstring(L, 1, &len);
lua_Integer pos = posrelat(luaL_optinteger(L, 2, 1), len);
int size = getfloatsize(L, 3);
luaL_argcheck(L, 1 <= pos && (size_t)pos + size - 1 <= len, 1,
"string too short");
if (size == sizeof(lua_Number)) {
memcpy(&res, s + pos - 1, size);
correctendianness(L, (char*)&res, size, 4);
/*
** Copy 'size' bytes from 'src' to 'dest', correcting endianness if
** given 'islittle' is different from native endianness.
*/
static void copywithendian (volatile char *dest, volatile const char *src,
int size, int islittle) {
if (islittle == nativeendian.little) {
while (size-- != 0)
*(dest++) = *(src++);
}
else if (size == sizeof(float)) {
float f;
memcpy(&f, s + pos - 1, size);
correctendianness(L, (char*)&f, size, 4);
res = (lua_Number)f;
}
else { /* native lua_Number may be neither float nor double */
double d;
lua_assert(size == sizeof(double));
memcpy(&d, s + pos - 1, size);
correctendianness(L, (char*)&d, size, 4);
res = (lua_Number)d;
else {
dest += size - 1;
while (size-- != 0)
*(dest--) = *(src++);
}
}
static int str_pack (lua_State *L) {
luaL_Buffer b;
Header h;
const char *fmt = luaL_checkstring(L, 1); /* format string */
int arg = 1; /* current argument to pack */
size_t totalsize = 0; /* accumulate total size of result */
getheader(L, &h, &fmt);
lua_pushnil(L); /* mark to separate arguments from string buffer */
luaL_buffinit(L, &b);
while (*fmt != '\0') {
int size, ntoalign;
KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
totalsize += ntoalign + size;
while (ntoalign-- > 0) luaL_addchar(&b, '\0'); /* fill alignment */
arg++;
switch (opt) {
case Kint: { /* signed integers */
lua_Integer n = luaL_checkinteger(L, arg);
lua_Unsigned mask = (n < 0) ? HIGHERBYTE : 0; /* sign extension */
if (size < SZINT) { /* need overflow check? */
lua_Integer lim = (lua_Integer)1 << ((size * NB) - 1);
luaL_argcheck(L, -lim <= n && n < lim, arg, "integer overflow");
}
packint(&b, (lua_Unsigned)n, h.islittle, size, mask);
break;
}
case Kuint: { /* unsigned integers */
lua_Integer n = luaL_checkinteger(L, arg);
if (size < SZINT) /* need overflow check? */
luaL_argcheck(L, (lua_Unsigned)n < ((lua_Unsigned)1 << (size * NB)),
arg, "unsigned overflow");
packint(&b, (lua_Unsigned)n, h.islittle, size, 0);
break;
}
case Kfloat: { /* floating-point options */
volatile Ftypes u;
char *buff = luaL_prepbuffsize(&b, size);
lua_Number n = luaL_checknumber(L, arg); /* get argument */
if (size == sizeof(u.f)) u.f = (float)n; /* copy it into 'u' */
else if (size == sizeof(u.d)) u.d = (double)n;
else u.n = n;
/* move 'u' to final result, correcting endianness if needed */
copywithendian(buff, u.buff, size, h.islittle);
luaL_addsize(&b, size);
break;
}
case Kchar: { /* fixed-size string */
size_t len;
const char *s = luaL_checklstring(L, arg, &len);
luaL_argcheck(L, len == (size_t)size, arg, "wrong length");
luaL_addlstring(&b, s, size);
break;
}
case Kstring: { /* strings with length count */
size_t len;
const char *s = luaL_checklstring(L, arg, &len);
luaL_argcheck(L, size >= (int)sizeof(size_t) ||
len < ((size_t)1 << (size * NB)),
arg, "string length does not fit in given size");
packint(&b, (lua_Unsigned)len, h.islittle, size, 0); /* pack length */
luaL_addlstring(&b, s, len);
totalsize += len;
break;
}
case Kstring0: { /* zero-terminated string */
size_t len;
const char *s = luaL_checklstring(L, arg, &len);
luaL_argcheck(L, strlen(s) == len, arg, "string contains zeros");
luaL_addlstring(&b, s, len);
luaL_addchar(&b, '\0'); /* add zero at the end */
totalsize += len + 1;
break;
}
case Kpadding: luaL_addchar(&b, '\0'); /* go through */
case Kpaddalig: case Kspace:
arg--; /* undo increment */
break;
}
}
lua_pushnumber(L, res);
luaL_pushresult(&b);
return 1;
}
static lua_Integer unpackint (lua_State *L, const char *str,
int islittle, int size, int issigned) {
lua_Unsigned res = 0;
int i;
int limit = (size <= SZINT) ? size : SZINT;
for (i = limit - 1; i >= 0; i--) {
res <<= NB;
res |= (lua_Unsigned)(unsigned char)str[islittle ? i : size - 1 - i];
}
if (size < SZINT) { /* real size smaller than lua_Integer? */
if (issigned) { /* needs sign extension? */
lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1);
res = ((res ^ mask) - mask); /* do sign extension */
}
}
else { /* must check unread bytes */
int mask = (!issigned || (lua_Integer)res >= 0) ? 0 : MC;
for (i = limit; i < size; i++) {
if ((unsigned char)str[islittle ? i : size - 1 - i] != mask)
luaL_error(L, "%d-bit integer does not fit into Lua Integer", size);
}
}
return (lua_Integer)res;
}
static int str_unpack (lua_State *L) {
Header h;
const char *fmt = luaL_checkstring(L, 1);
size_t ld;
const char *data = luaL_checklstring(L, 2, &ld);
size_t pos = (size_t)posrelat(luaL_optinteger(L, 3, 1), ld) - 1;
int n = 0; /* number of results */
luaL_argcheck(L, pos <= ld, 3, "initial position out of string");
getheader(L, &h, &fmt);
while (*fmt) {
int size, ntoalign;
KOption opt = getdetails(&h, pos, &fmt, &size, &ntoalign);
if ((size_t)ntoalign + size > ~pos || pos + ntoalign + size > ld)
luaL_argerror(L, 2, "data string too short");
pos += ntoalign;
/* stack space for item + next position */
luaL_checkstack(L, 2, "too many results");
n++;
switch (opt) {
case Kint:
case Kuint: {
lua_Integer res = unpackint(L, data + pos, h.islittle, size,
(opt == Kint));
lua_pushinteger(L, res);
break;
}
case Kfloat: {
volatile Ftypes u;
lua_Number num;
copywithendian(u.buff, data + pos, size, h.islittle);
if (size == sizeof(u.f)) num = (lua_Number)u.f;
else if (size == sizeof(u.d)) num = (lua_Number)u.d;
else num = u.n;
lua_pushnumber(L, num);
break;
}
case Kchar: {
lua_pushlstring(L, data + pos, size);
break;
}
case Kstring: {
size_t len = (size_t)unpackint(L, data + pos, h.islittle, size, 0);
luaL_argcheck(L, pos + len + size <= ld, 2, "data string too short");
lua_pushlstring(L, data + pos + size, len);
pos += len;
break;
}
case Kstring0: {
size_t len = (int)strlen(data + pos);
lua_pushlstring(L, data + pos, len);
pos += len + 1; /* skip final '\0' */
break;
}
case Kpaddalig: case Kpadding: case Kspace:
n--; /* undo increment */
break;
}
pos += size;
}
lua_pushinteger(L, pos + 1); /* next position */
return n + 1;
}
/* }====================================================== */
@ -1176,10 +1361,8 @@ static const luaL_Reg strlib[] = {
{"reverse", str_reverse},
{"sub", str_sub},
{"upper", str_upper},
{"dumpfloat", dumpfloat_l},
{"dumpinteger", dumpint_l},
{"undumpfloat", undumpfloat_l},
{"undumpinteger", undumpint_l},
{"pack", str_pack},
{"unpack", str_unpack},
{NULL, NULL}
};

Loading…
Cancel
Save