add utf8 library from Lua 5.3; use struct for string packing

3 files changed, 335 insertions, 0 deletions

diff --git a/compat53.lua b/compat53.lua
new file mode 100644
index 0000000..98e1a3d
--- /dev/null
+++ b/compat53.lua
@@ -0,0 +1,24 @@
+local lua_version = _VERSION:sub(-3)
+
+if lua_version ~= "5.3" then
+
+   -- load utf8 library
+   utf8 = require("compat53.utf8")
+   package.loaded["utf8"] = utf8
+   if lua_version == "5.1" then
+      utf8.charpattern = "[%z\1-\127\194-\244][\128-\191]*"
+   end
+
+   -- use Roberto's struct module for string packing/unpacking for now
+   -- maybe we'll later extract the functions from the 5.3 string
+   -- library for greater compatiblity, but it uses the 5.3 buffer API
+   -- which cannot easily be backported to Lua 5.1.
+   local ok, struct = pcall(require, "struct")
+   if ok then
+      string.pack = struct.pack
+      string.packsize = struct.size
+      string.unpack = struct.unpack
+   end
+
+end
+
diff --git a/lprefix.h b/lprefix.h
new file mode 100644
index 0000000..788da4e
--- /dev/null
+++ b/lprefix.h
@@ -0,0 +1,56 @@
+/*
+** $Id: lprefix.h,v 1.2 2014/12/29 16:54:13 roberto Exp $
+** Definitions for Lua code that must come before any other header file
+** See Copyright Notice in lua.h
+*/
+
+#ifndef lprefix_h
+#define lprefix_h
+
+
+/*
+** Allows POSIX/XSI stuff
+*/
+#if !defined(LUA_USE_C89)       /* { */
+
+#if !defined(_XOPEN_SOURCE)
+#define _XOPEN_SOURCE           600
+#elif _XOPEN_SOURCE == 0
+#undef _XOPEN_SOURCE  /* use -D_XOPEN_SOURCE=0 to undefine it */
+#endif
+
+/*
+** Allows manipulation of large files in gcc and some other compilers
+*/
+#if !defined(LUA_32BITS) && !defined(_FILE_OFFSET_BITS)
+#define _LARGEFILE_SOURCE       1
+#define _FILE_OFFSET_BITS       64
+#endif
+
+#endif                          /* } */
+
+
+/*
+** Windows stuff
+*/
+#if defined(_WIN32)     /* { */
+
+#if !defined(_CRT_SECURE_NO_WARNINGS)
+#define _CRT_SECURE_NO_WARNINGS  /* avoid warnings about ISO C functions */
+#endif
+
+#endif                  /* } */
+
+
+/* COMPAT53 adaptation */
+#include "c-api/compat-5.3.h"
+
+#undef LUAMOD_API
+#define LUAMOD_API extern
+
+#ifndef luaopen_utf8
+#  define luaopen_utf8 luaopen_compat53_utf8
+#endif
+
+#endif
+
diff --git a/lutf8lib.c b/lutf8lib.c
new file mode 100644
index 0000000..be4f087
--- /dev/null
+++ b/lutf8lib.c
@@ -0,0 +1,255 @@
+/*
+** $Id: lutf8lib.c,v 1.13 2014/11/02 19:19:04 roberto Exp $
+** Standard library for UTF-8 manipulation
+** See Copyright Notice in lua.h
+*/
+
+#define lutf8lib_c
+#define LUA_LIB
+
+#include "lprefix.h"
+
+
+#include <assert.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "lua.h"
+
+#include "lauxlib.h"
+#include "lualib.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 u_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 one 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 [, j]]) --> number of characters that start in the
+** range [i,j], or nil + current position if 's' is not well formed in
+** that interval
+*/
+static int utflen (lua_State *L) {
+  int n = 0;
+  size_t len;
+  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);
+  luaL_argcheck(L, 1 <= posi && --posi <= (lua_Integer)len, 2,
+                   "initial position out of string");
+  luaL_argcheck(L, --posj < (lua_Integer)len, 3,
+                   "final position out of string");
+  while (posi <= posj) {
+    const char *s1 = utf8_decode(s + posi, NULL);
+    if (s1 == NULL) {  /* conversion error? */
+      lua_pushnil(L);  /* return nil ... */
+      lua_pushinteger(L, posi + 1);  /* ... and current position */
+      return 2;
+    }
+    posi = s1 - s;
+    n++;
+  }
+  lua_pushinteger(L, n);
+  return 1;
+}
+
+
+/*
+** codepoint(s, [i, [j]])  -> returns codepoints for all characters
+** that start in the range [i,j]
+*/
+static int codepoint (lua_State *L) {
+  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 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)
+      return luaL_error(L, "invalid UTF-8 code");
+    lua_pushinteger(L, code);
+    n++;
+  }
+  return n;
+}
+
+
+static void pushutfchar (lua_State *L, int arg) {
+  lua_Integer code = luaL_checkinteger(L, arg);
+  luaL_argcheck(L, 0 <= code && code <= MAXUNICODE, arg, "value out of range");
+  lua_pushfstring(L, "%U", (long)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 counting from
+**   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);
+  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 range");
+  if (n == 0) {
+    /* find beginning of current byte sequence */
+    while (posi > 0 && iscont(s + posi)) posi--;
+  }
+  else {
+    if (iscont(s + posi))
+      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 */
+    lua_pushnil(L);
+  return 1;  
+}
+
+
+static int iter_aux (lua_State *L) {
+  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 {
+    int code;
+    const char *next = utf8_decode(s + n, &code);
+    if (next == NULL || iscont(next))
+      return 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},
+  /* placeholders */
+  {"charpattern", NULL},
+  {NULL, NULL}
+};
+
+
+LUAMOD_API int luaopen_utf8 (lua_State *L) {
+  luaL_newlib(L, funcs);
+  lua_pushliteral(L, UTF8PATT);
+  lua_setfield(L, -2, "charpattern");
+  return 1;
+}
+