lutf8lib.c File Reference

#include "lprefix.h"
#include <assert.h>
#include <limits.h>
#include <stdlib.h>
#include <string.h>
#include "lua.h"
#include "lauxlib.h"
#include "lualib.h"

Go to the source code of this file.

Macros
#define	lutf8lib_c
#define	LUA_LIB
#define	MAXUNICODE   0x10FFFFu
#define	MAXUTF   0x7FFFFFFFu
#define	iscont(p)   ((*(p) & 0xC0) == 0x80)
#define	UTF8PATT   "[\0-\x7F\xC2-\xFD][\x80-\xBF]*"

Typedefs
typedef unsigned long	utfint

Functions
static lua_Integer	u_posrelat (lua_Integer pos, size_t len)
static const char *	utf8_decode (const char *s, utfint *val, int strict)
static int	utflen (lua_State *L)
static int	codepoint (lua_State *L)
static void	pushutfchar (lua_State *L, int arg)
static int	utfchar (lua_State *L)
static int	byteoffset (lua_State *L)
static int	iter_aux (lua_State *L, int strict)
static int	iter_auxstrict (lua_State *L)
static int	iter_auxlax (lua_State *L)
static int	iter_codes (lua_State *L)
LUAMOD_API int	luaopen_utf8 (lua_State *L)

Variables
static const luaL_Reg	funcs []

Macro Definition Documentation

iscont

#define iscont

(

p

)

((*(p) & 0xC0) == 0x80)

Definition at line 38 of file lua-5.4.3/src/lutf8lib.c.

Referenced by byteoffset(), and iter_aux().

LUA_LIB

#define LUA_LIB

Definition at line 8 of file lua-5.4.3/src/lutf8lib.c.

lutf8lib_c

#define lutf8lib_c

Definition at line 7 of file lua-5.4.3/src/lutf8lib.c.

MAXUNICODE

#define MAXUNICODE   0x10FFFFu

Definition at line 24 of file lua-5.4.3/src/lutf8lib.c.

Referenced by utf8_decode().

MAXUTF

#define MAXUTF   0x7FFFFFFFu

Definition at line 26 of file lua-5.4.3/src/lutf8lib.c.

Referenced by pushutfchar(), and utf8_decode().

UTF8PATT

#define UTF8PATT   "[\0-\x7F\xC2-\xFD][\x80-\xBF]*"

Definition at line 268 of file lua-5.4.3/src/lutf8lib.c.

Referenced by luaopen_utf8().

Typedef Documentation

utfint

typedef unsigned long utfint

Definition at line 34 of file lua-5.4.3/src/lutf8lib.c.

Function Documentation

byteoffset()

static int byteoffset ( lua_State * L )

static

Definition at line 183 of file lua-5.4.3/src/lutf8lib.c.

                                     {
  size_t len;
  const char *s = luaL_checklstring(L, 1, &len);
  lua_Integer n  = luaL_checkinteger(L, 2);
  lua_Integer posi = (n >= 0) ? 1 : len + 1;
  posi = u_posrelat(luaL_optinteger(L, 3, posi), len);
  luaL_argcheck(L, 1 <= posi && --posi <= (lua_Integer)len, 3,
                   "position out of bounds");
  if (n == 0) {
    /* find beginning of current byte sequence */
    while (posi > 0 && iscont(s + posi)) posi--;
  }
  else {
    if (iscont(s + posi))
      return luaL_error(L, "initial position is a continuation byte");
    if (n < 0) {
       while (n < 0 && posi > 0) {  /* move back */
         do {  /* find beginning of previous character */
           posi--;
         } while (posi > 0 && iscont(s + posi));
         n++;
       }
     }
     else {
       n--;  /* do not move for 1st character */
       while (n > 0 && posi < (lua_Integer)len) {
         do {  /* find beginning of next character */
           posi++;
         } while (iscont(s + posi));  /* (cannot pass final '\0') */
         n--;
       }
     }
  }
  if (n == 0)  /* did it find given character? */
    lua_pushinteger(L, posi + 1);
  else  /* no such character */
    luaL_pushfail(L);
  return 1;
}

References iscont, lua_pushinteger(), luaL_argcheck, luaL_checkinteger(), luaL_checklstring(), luaL_error(), luaL_optinteger(), luaL_pushfail, s, and u_posrelat().

codepoint()

static int codepoint ( lua_State * L )

static

Definition at line 122 of file lua-5.4.3/src/lutf8lib.c.

                                    {
  size_t len;
  const char *s = luaL_checklstring(L, 1, &len);
  lua_Integer posi = u_posrelat(luaL_optinteger(L, 2, 1), len);
  lua_Integer pose = u_posrelat(luaL_optinteger(L, 3, posi), len);
  int lax = lua_toboolean(L, 4);
  int n;
  const char *se;
  luaL_argcheck(L, posi >= 1, 2, "out of bounds");
  luaL_argcheck(L, pose <= (lua_Integer)len, 3, "out of bounds");
  if (posi > pose) return 0;  /* empty interval; return no values */
  if (pose - posi >= INT_MAX)  /* (lua_Integer -> int) overflow? */
    return luaL_error(L, "string slice too long");
  n = (int)(pose -  posi) + 1;  /* upper bound for number of returns */
  luaL_checkstack(L, n, "string slice too long");
  n = 0;  /* count the number of returns */
  se = s + pose;  /* string end */
  for (s += posi - 1; s < se;) {
    utfint code;
    s = utf8_decode(s, &code, !lax);
    if (s == NULL)
      return luaL_error(L, "invalid UTF-8 code");
    lua_pushinteger(L, code);
    n++;
  }
  return n;
}

References lua_pushinteger(), lua_toboolean(), luaL_argcheck, luaL_checklstring(), luaL_checkstack(), luaL_error(), luaL_optinteger(), NULL, s, u_posrelat(), and utf8_decode().

iter_aux()

static int iter_aux ( lua_State * L, int strict )

static

Definition at line 224 of file lua-5.4.3/src/lutf8lib.c.

                                               {
  size_t len;
  const char *s = luaL_checklstring(L, 1, &len);
  lua_Integer n = lua_tointeger(L, 2) - 1;
  if (n < 0)  /* first iteration? */
    n = 0;  /* start from here */
  else if (n < (lua_Integer)len) {
    n++;  /* skip current byte */
    while (iscont(s + n)) n++;  /* and its continuations */
  }
  if (n >= (lua_Integer)len)
    return 0;  /* no more codepoints */
  else {
    utfint code;
    const char *next = utf8_decode(s + n, &code, strict);
    if (next == NULL)
      return luaL_error(L, "invalid UTF-8 code");
    lua_pushinteger(L, n + 1);
    lua_pushinteger(L, code);
    return 2;
  }
}

References iscont, lua_pushinteger(), lua_tointeger, luaL_checklstring(), luaL_error(), next, NULL, s, and utf8_decode().

Referenced by iter_auxlax(), and iter_auxstrict().

iter_auxlax()

static int iter_auxlax ( lua_State * L )

static

Definition at line 252 of file lua-5.4.3/src/lutf8lib.c.

                                      {
  return iter_aux(L, 0);
}

References iter_aux().

Referenced by iter_codes().

iter_auxstrict()

static int iter_auxstrict ( lua_State * L )

static

Definition at line 248 of file lua-5.4.3/src/lutf8lib.c.

                                         {
  return iter_aux(L, 1);
}

References iter_aux().

Referenced by iter_codes().

iter_codes()

static int iter_codes ( lua_State * L )

static

Definition at line 257 of file lua-5.4.3/src/lutf8lib.c.

                                     {
  int lax = lua_toboolean(L, 2);
  luaL_checkstring(L, 1);
  lua_pushcfunction(L, lax ? iter_auxlax : iter_auxstrict);
  lua_pushvalue(L, 1);
  lua_pushinteger(L, 0);
  return 3;
}

References iter_auxlax(), iter_auxstrict(), lua_pushcfunction, lua_pushinteger(), lua_pushvalue(), lua_toboolean(), and luaL_checkstring.

luaopen_utf8()

LUAMOD_API int luaopen_utf8

(

lua_State *

L

)

Definition at line 283 of file lua-5.4.3/src/lutf8lib.c.

                                           {
  luaL_newlib(L, funcs);
  lua_pushlstring(L, UTF8PATT, sizeof(UTF8PATT)/sizeof(char) - 1);
  lua_setfield(L, -2, "charpattern");
  return 1;
}

References funcs, funcs, lua_pushlstring(), lua_setfield(), luaL_newlib, and UTF8PATT.

pushutfchar()

static void pushutfchar ( lua_State * L, int arg )

static

Definition at line 151 of file lua-5.4.3/src/lutf8lib.c.

                                                {
  lua_Unsigned code = (lua_Unsigned)luaL_checkinteger(L, arg);
  luaL_argcheck(L, code <= MAXUTF, arg, "value out of range");
  lua_pushfstring(L, "%U", (long)code);
}

References arg, lua_pushfstring(), luaL_argcheck, luaL_checkinteger(), and MAXUTF.

Referenced by utfchar().

u_posrelat()

static lua_Integer u_posrelat ( lua_Integer pos, size_t len )

static

Definition at line 43 of file lua-5.4.3/src/lutf8lib.c.

                                                            {
  if (pos >= 0) return pos;
  else if (0u - (size_t)pos > len) return 0;
  else return (lua_Integer)len + pos + 1;
}

Referenced by byteoffset(), codepoint(), and utflen().

utf8_decode()

static const char * utf8_decode ( const char * s, utfint * val, int strict )

static

Definition at line 57 of file lua-5.4.3/src/lutf8lib.c.

                                                                        {
  static const utfint limits[] =
        {~(utfint)0, 0x80, 0x800, 0x10000u, 0x200000u, 0x4000000u};
  unsigned int c = (unsigned char)s[0];
  utfint res = 0;  /* final result */
  if (c < 0x80)  /* ascii? */
    res = c;
  else {
    int count = 0;  /* to count number of continuation bytes */
    for (; c & 0x40; c <<= 1) {  /* while it needs continuation bytes... */
      unsigned int cc = (unsigned char)s[++count];  /* read next byte */
      if ((cc & 0xC0) != 0x80)  /* not a continuation byte? */
        return NULL;  /* invalid byte sequence */
      res = (res << 6) | (cc & 0x3F);  /* add lower 6 bits from cont. byte */
    }
    res |= ((utfint)(c & 0x7F) << (count * 5));  /* add first byte */
    if (count > 5 || res > MAXUTF || res < limits[count])
      return NULL;  /* invalid byte sequence */
    s += count;  /* skip continuation bytes read */
  }
  if (strict) {
    /* check for invalid code points; too large or surrogates */
    if (res > MAXUNICODE || (0xD800u <= res && res <= 0xDFFFu))
      return NULL;
  }
  if (val) *val = res;
  return s + 1;  /* +1 to include first byte */
}

References MAXUNICODE, MAXUTF, NULL, and s.

Referenced by codepoint(), iter_aux(), and utflen().

utfchar()

static int utfchar ( lua_State * L )

static

Definition at line 161 of file lua-5.4.3/src/lutf8lib.c.

                                  {
  int n = lua_gettop(L);  /* number of arguments */
  if (n == 1)  /* optimize common case of single char */
    pushutfchar(L, 1);
  else {
    int i;
    luaL_Buffer b;
    luaL_buffinit(L, &b);
    for (i = 1; i <= n; i++) {
      pushutfchar(L, i);
      luaL_addvalue(&b);
    }
    luaL_pushresult(&b);
  }
  return 1;
}

References lua_gettop(), luaL_addvalue(), luaL_buffinit(), luaL_pushresult(), and pushutfchar().

utflen()

static int utflen ( lua_State * L )

static

Definition at line 92 of file lua-5.4.3/src/lutf8lib.c.

                                 {
  lua_Integer n = 0;  /* counter for the number of characters */
  size_t len;  /* string length in bytes */
  const char *s = luaL_checklstring(L, 1, &len);
  lua_Integer posi = u_posrelat(luaL_optinteger(L, 2, 1), len);
  lua_Integer posj = u_posrelat(luaL_optinteger(L, 3, -1), len);
  int lax = lua_toboolean(L, 4);
  luaL_argcheck(L, 1 <= posi && --posi <= (lua_Integer)len, 2,
                   "initial position out of bounds");
  luaL_argcheck(L, --posj < (lua_Integer)len, 3,
                   "final position out of bounds");
  while (posi <= posj) {
    const char *s1 = utf8_decode(s + posi, NULL, !lax);
    if (s1 == NULL) {  /* conversion error? */
      luaL_pushfail(L);  /* return fail ... */
      lua_pushinteger(L, posi + 1);  /* ... and current position */
      return 2;
    }
    posi = s1 - s;
    n++;
  }
  lua_pushinteger(L, n);
  return 1;
}

References lua_pushinteger(), lua_toboolean(), luaL_argcheck, luaL_checklstring(), luaL_optinteger(), luaL_pushfail, NULL, s, u_posrelat(), and utf8_decode().

Variable Documentation

funcs

const luaL_Reg funcs[]

static

Initial value:

= {
  {"offset", byteoffset},
  {"codepoint", codepoint},
  {"char", utfchar},
  {"len", utflen},
  {"codes", iter_codes},
  
  {"charpattern", NULL},
  {NULL, NULL}
}

Definition at line 271 of file lua-5.4.3/src/lutf8lib.c.

                                {
  {"offset", byteoffset},
  {"codepoint", codepoint},
  {"char", utfchar},
  {"len", utflen},
  {"codes", iter_codes},
  /* placeholders */
  {"charpattern", NULL},
  {NULL, NULL}
};

Referenced by luaopen_utf8().