new library: utf8

4 files changed, 244 insertions, 6 deletions

diff --git a/linit.c b/linit.c
index 8fd67fcd..1c0474e6 100644
--- a/linit.c
+++ b/linit.c
@@ -1,5 +1,5 @@
 /*
-** $Id: linit.c,v 1.31 2011/01/26 16:30:02 roberto Exp roberto $
+** $Id: linit.c,v 1.32 2011/04/08 19:17:36 roberto Exp roberto $
 ** Initialization of libraries for lua.c and other clients
 ** See Copyright Notice in lua.h
 */
@@ -34,6 +34,7 @@ static const luaL_Reg loadedlibs[] = {
   {LUA_IOLIBNAME, luaopen_io},
   {LUA_OSLIBNAME, luaopen_os},
   {LUA_STRLIBNAME, luaopen_string},
+  {LUA_UTF8LIBNAME, luaopen_utf8},
   {LUA_BITLIBNAME, luaopen_bit32},
   {LUA_MATHLIBNAME, luaopen_math},
   {LUA_DBLIBNAME, luaopen_debug},
diff --git a/lualib.h b/lualib.h
index 15adcdb4..ea073444 100644
--- a/lualib.h
+++ b/lualib.h
@@ -1,5 +1,5 @@
 /*
-** $Id: lualib.h,v 1.42 2011/05/25 14:12:28 roberto Exp roberto $
+** $Id: lualib.h,v 1.43 2011/12/08 12:11:37 roberto Exp roberto $
 ** Lua standard libraries
 ** See Copyright Notice in lua.h
 */
@@ -29,6 +29,9 @@ LUAMOD_API int (luaopen_os) (lua_State *L);
 #define LUA_STRLIBNAME  "string"
 LUAMOD_API int (luaopen_string) (lua_State *L);
 
+#define LUA_UTF8LIBNAME "utf8"
+LUAMOD_API int (luaopen_utf8) (lua_State *L);
+
 #define LUA_BITLIBNAME  "bit32"
 LUAMOD_API int (luaopen_bit32) (lua_State *L);
 
diff --git a/lutf8lib.c b/lutf8lib.c
new file mode 100644
index 00000000..dfd52832
--- /dev/null
+++ b/lutf8lib.c
@@ -0,0 +1,233 @@
+/* gcc -shared -o utf8.so -fpic -O2 -Wall -I.. utf8.c */
+
+#include <assert.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "lua.h"
+#include "lauxlib.h"
+
+#define MAXUNICODE      0x10FFFF
+
+#define iscont(p)       ((*(p) & 0xC0) == 0x80)
+
+
+/* from strlib */
+/* translate a relative string position: negative means back from end */
+static lua_Integer posrelat (lua_Integer pos, size_t len) {
+  if (pos >= 0) return pos;
+  else if (0u - (size_t)pos > len) return 0;
+  else return (lua_Integer)len + pos + 1;
+}
+
+
+/*
+** Decode an UTF-8 sequence, returning NULL if byte sequence is invalid.
+*/
+static const char *utf8_decode (const char *o, int *val) {
+  static unsigned int limits[] = {0xFF, 0x7F, 0x7FF, 0xFFFF};
+  const unsigned char *s = (const unsigned char *)o;
+  unsigned int c = s[0];
+  unsigned int res = 0;  /* final result */
+  if (c < 0x80)  /* ascii? */
+    res = c;
+  else {
+    int count = 0;  /* to count number of continuation bytes */
+    while (c & 0x40) {  /* still have continuation bytes? */
+      int cc = 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 */
+      c <<= 1;  /* to test next bit */
+    }
+    res |= ((c & 0x7F) << (count * 5));  /* add first byte */
+    if (count > 3 || res > MAXUNICODE || res <= limits[count])
+      return NULL;  /* invalid byte sequence */
+    s += count;  /* skip continuation bytes read */
+  }
+  if (val) *val = res;
+  return (const char *)s + 1;  /* +1 to include first byte */
+}
+
+
+/*
+** utf8len(s, [i])   --> number of codepoints in 's' after 'i';
+** nil if 's' not well formed
+*/
+static int utflen (lua_State *L) {
+  int n = 0;
+  const char *ends;
+  size_t len;
+  const char *s = luaL_checklstring(L, 1, &len);
+  lua_Integer posi = posrelat(luaL_optinteger(L, 2, 1), 1);
+  luaL_argcheck(L, 1 <= posi && posi <= (lua_Integer)len, 1,
+                   "initial position out of string");
+  ends = s + len;
+  s += posi - 1;
+  while (s < ends && (s = utf8_decode(s, NULL)) != NULL)
+    n++;
+  if (s == ends)
+    lua_pushinteger(L, n);
+  else
+    lua_pushnil(L);
+  return 1;
+}
+
+
+/*
+** codepoint(s, [i, [j]])  -> returns codepoints for all characters
+** between i and j
+*/
+static int codepoint (lua_State *L) {
+  size_t len;
+  const char *s = luaL_checklstring(L, 1, &len);
+  lua_Integer posi = posrelat(luaL_optinteger(L, 2, 1), len);
+  lua_Integer pose = posrelat(luaL_optinteger(L, 3, posi), len);
+  int n;
+  const char *se;
+  luaL_argcheck(L, posi >= 1, 2, "out of range");
+  luaL_argcheck(L, pose <= (lua_Integer)len, 3, "out of range");
+  if (posi > pose) return 0;  /* empty interval; return no values */
+  n = (int)(pose -  posi + 1);
+  if (posi + n <= pose)  /* (lua_Integer -> int) overflow? */
+    return luaL_error(L, "string slice too long");
+  luaL_checkstack(L, n, "string slice too long");
+  n = 0;
+  se = s + pose;
+  for (s += posi - 1; s < se;) {
+    int code;
+    s = utf8_decode(s, &code);
+    if (s == NULL)
+      luaL_error(L, "invalid UTF-8 code");
+    lua_pushinteger(L, code);
+    n++;
+  }
+  return n;
+}
+
+
+static void pushutfchar (lua_State *L, int arg) {
+  int code = luaL_checkint(L, arg);
+  luaL_argcheck(L, 0 <= code && code <= MAXUNICODE, arg, "value out of range");
+  lua_pushfstring(L, "%U", code);
+}
+
+
+/*
+** utfchar(n1, n2, ...)  -> char(n1)..char(n2)...
+*/
+static int utfchar (lua_State *L) {
+  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;
+}
+
+
+/*
+** offset(s, n, [i])  -> index where n-th character *after*
+**   position 'i' starts; 0 means character at 'i'.
+*/
+static int byteoffset (lua_State *L) {
+  size_t len;
+  const char *s = luaL_checklstring(L, 1, &len);
+  int n  = luaL_checkint(L, 2);
+  lua_Integer posi = posrelat(luaL_optinteger(L, 3, 1), len) - 1;
+  luaL_argcheck(L, 0 <= posi && posi <= (lua_Integer)len, 3,
+                   "position out of range");
+  if (n == 0) {
+    /* find beginning of current byte sequence */
+    while (posi > 0 && iscont(s + posi)) posi--;
+  }
+  else 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));  /* ('\0' is not continuation) */
+      n--;
+    }
+  }
+  if (n == 0)
+    lua_pushinteger(L, posi + 1);
+  else
+    lua_pushnil(L);  /* no such position */
+  return 1;  
+}
+
+
+static int iter_aux (lua_State *L) {
+  size_t len;
+  const char *s = luaL_checklstring(L, 1, &len);
+  int 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 {
+    int code;
+    const char *next = utf8_decode(s + n, &code);
+    if (next == NULL || iscont(next))
+      luaL_error(L, "invalid UTF-8 code");
+    lua_pushinteger(L, n + 1);
+    lua_pushinteger(L, code);
+    return 2;
+  }
+}
+
+
+static int iter_codes (lua_State *L) {
+  luaL_checkstring(L, 1);
+  lua_pushcfunction(L, iter_aux);
+  lua_pushvalue(L, 1);
+  lua_pushinteger(L, 0);
+  return 3;
+}
+
+
+/* pattern to match a single UTF-8 character */
+#define UTF8PATT        "[\0-\x7F\xC2-\xF4][\x80-\xBF]*"
+
+
+static struct luaL_Reg funcs[] = {
+  {"offset", byteoffset},
+  {"codepoint", codepoint},
+  {"char", utfchar},
+  {"len", utflen},
+  {"codes", iter_codes},
+  {NULL, NULL}
+};
+
+
+
+int luaopen_utf8 (lua_State *L);
+
+int luaopen_utf8 (lua_State *L) {
+  luaL_newlib(L, funcs);
+  lua_pushliteral(L, UTF8PATT);
+  lua_setfield(L, -2, "charpatt");
+  return 1;
+}
+
diff --git a/makefile b/makefile
index 55aacf9a..7869fbac 100644
--- a/makefile
+++ b/makefile
@@ -75,7 +75,7 @@ CORE_O=	lapi.o lcode.o lctype.o ldebug.o ldo.o ldump.o lfunc.o lgc.o llex.o \
         ltm.o lundump.o lvm.o lzio.o ltests.o
 AUX_O=  lauxlib.o
 LIB_O=  lbaselib.o ldblib.o liolib.o lmathlib.o loslib.o ltablib.o lstrlib.o \
-        lbitlib.o loadlib.o lcorolib.o linit.o
+        lutf8lib.o lbitlib.o loadlib.o lcorolib.o linit.o
 
 LUA_T=  lua
 LUA_O=  lua.o
@@ -153,7 +153,7 @@ lgc.o: lgc.c lua.h luaconf.h ldebug.h lstate.h lobject.h llimits.h ltm.h \
 linit.o: linit.c lua.h luaconf.h lualib.h lauxlib.h
 liolib.o: liolib.c lua.h luaconf.h lauxlib.h lualib.h
 llex.o: llex.c lua.h luaconf.h lctype.h llimits.h ldo.h lobject.h \
- lstate.h ltm.h lzio.h lmem.h llex.h lparser.h lstring.h lgc.h ltable.h
+ lstate.h ltm.h lzio.h lmem.h lgc.h llex.h lparser.h lstring.h ltable.h
 lmathlib.o: lmathlib.c lua.h luaconf.h lauxlib.h lualib.h
 lmem.o: lmem.c lua.h luaconf.h ldebug.h lstate.h lobject.h llimits.h \
  ltm.h lzio.h lmem.h ldo.h lgc.h
@@ -168,8 +168,8 @@ lparser.o: lparser.c lua.h luaconf.h lcode.h llex.h lobject.h llimits.h \
 lstate.o: lstate.c lua.h luaconf.h lapi.h llimits.h lstate.h lobject.h \
  ltm.h lzio.h lmem.h ldebug.h ldo.h lfunc.h lgc.h llex.h lstring.h \
  ltable.h
-lstring.o: lstring.c lua.h luaconf.h lmem.h llimits.h lobject.h lstate.h \
- ltm.h lzio.h lstring.h lgc.h
+lstring.o: lstring.c lua.h luaconf.h ldebug.h lstate.h lobject.h \
+ llimits.h ltm.h lzio.h lmem.h ldo.h lstring.h lgc.h
 lstrlib.o: lstrlib.c lua.h luaconf.h lauxlib.h lualib.h
 ltable.o: ltable.c lua.h luaconf.h ldebug.h lstate.h lobject.h llimits.h \
  ltm.h lzio.h lmem.h ldo.h lgc.h lstring.h ltable.h lvm.h
@@ -182,6 +182,7 @@ ltm.o: ltm.c lua.h luaconf.h ldebug.h lstate.h lobject.h llimits.h ltm.h \
 lua.o: lua.c lua.h luaconf.h lauxlib.h lualib.h
 lundump.o: lundump.c lua.h luaconf.h ldebug.h lstate.h lobject.h \
  llimits.h ltm.h lzio.h lmem.h ldo.h lfunc.h lstring.h lgc.h lundump.h
+lutf8lib.o: lutf8lib.c lua.h luaconf.h lauxlib.h
 lvm.o: lvm.c lua.h luaconf.h ldebug.h lstate.h lobject.h llimits.h ltm.h \
  lzio.h lmem.h ldo.h lfunc.h lgc.h lopcodes.h lstring.h ltable.h lvm.h
 lzio.o: lzio.c lua.h luaconf.h llimits.h lmem.h lstate.h lobject.h ltm.h \