first (big) step to support wide chars

1 files changed, 118 insertions, 112 deletions

diff --git a/llex.c b/llex.c
index 386ea316..0bbc43be 100644
--- a/llex.c
+++ b/llex.c
@@ -1,5 +1,5 @@
 /*
-** $Id: llex.c,v 1.78 2001/02/22 17:15:18 roberto Exp roberto $
+** $Id: llex.c,v 1.79 2001/02/22 18:59:59 roberto Exp roberto $
 ** Lexical Analyzer
 ** See Copyright Notice in lua.h
 */
@@ -25,10 +25,13 @@
 
 
 /* ORDER RESERVED */
-static const char *const token2string [] = {
-    "and", "break", "do", "else", "elseif", "end", "for",
-    "function", "if", "local", "nil", "not", "or", "repeat", "return", "then",
-    "until", "while", "", "..", "...", "==", ">=", "<=", "~=", "", "", "<eof>"};
+static const l_char *const token2string [] = {
+    l_s("and"), l_s("break"), l_s("do"), l_s("else"), l_s("elseif"),
+    l_s("end"), l_s("for"), l_s("function"), l_s("if"), l_s("local"),
+    l_s("nil"), l_s("not"), l_s("or"), l_s("repeat"), l_s("return"),
+    l_s("then"), l_s("until"), l_s("while"), l_s(""), l_s(".."), l_s("..."),
+    l_s("=="), l_s(">="), l_s("<="), l_s("~="), l_s(""), l_s(""), l_s("<eof>")
+};
 
 
 void luaX_init (lua_State *L) {
@@ -44,37 +47,38 @@ void luaX_init (lua_State *L) {
 #define MAXSRC          80
 
 
-void luaX_checklimit (LexState *ls, int val, int limit, const char *msg) {
+void luaX_checklimit (LexState *ls, int val, int limit, const l_char *msg) {
   if (val > limit) {
-    char buff[90];
-    sprintf(buff, "too many %.40s (limit=%d)", msg, limit);
+    l_char buff[90];
+    sprintf(buff, l_s("too many %.40s (limit=%d)"), msg, limit);
     luaX_error(ls, buff, ls->t.token);
   }
 }
 
 
-void luaX_syntaxerror (LexState *ls, const char *s, const char *token) {
-  char buff[MAXSRC];
+void luaX_syntaxerror (LexState *ls, const l_char *s, const l_char *token) {
+  l_char buff[MAXSRC];
   luaO_chunkid(buff, getstr(ls->source), sizeof(buff));
-  luaO_verror(ls->L, "%.99s;\n  last token read: `%.30s' at line %d in %.80s",
+  luaO_verror(ls->L,
+              l_s("%.99s;\n  last token read: `%.30s' at line %d in %.80s"),
               s, token, ls->linenumber, buff);
 }
 
 
-void luaX_error (LexState *ls, const char *s, int token) {
-  char buff[TOKEN_LEN];
+void luaX_error (LexState *ls, const l_char *s, int token) {
+  l_char buff[TOKEN_LEN];
   luaX_token2str(token, buff);
-  if (buff[0] == '\0')
+  if (buff[0] == l_c('\0'))
     luaX_syntaxerror(ls, s, G(ls->L)->Mbuffer);
   else
     luaX_syntaxerror(ls, s, buff);
 }
 
 
-void luaX_token2str (int token, char *s) {
+void luaX_token2str (int token, l_char *s) {
   if (token < 256) {
-    s[0] = (char)token;
-    s[1] = '\0';
+    s[0] = (l_char)token;
+    s[1] = l_c('\0');
   }
   else
     strcpy(s, token2string[token-FIRST_RESERVED]);
@@ -82,16 +86,16 @@ void luaX_token2str (int token, char *s) {
 
 
 static void luaX_invalidchar (LexState *ls, int c) {
-  char buff[8];
-  sprintf(buff, "0x%02X", c);
-  luaX_syntaxerror(ls, "invalid control char", buff);
+  l_char buff[8];
+  sprintf(buff, l_s("0x%02X"), c);
+  luaX_syntaxerror(ls, l_s("invalid control l_char"), buff);
 }
 
 
 static void inclinenumber (LexState *LS) {
   next(LS);  /* skip `\n' */
   ++LS->linenumber;
-  luaX_checklimit(LS, LS->linenumber, MAX_INT, "lines in a chunk");
+  luaX_checklimit(LS, LS->linenumber, MAX_INT, l_s("lines in a chunk"));
 }
 
 
@@ -104,10 +108,10 @@ void luaX_setinput (lua_State *L, LexState *LS, ZIO *z, TString *source) {
   LS->lastline = 1;
   LS->source = source;
   next(LS);  /* read first char */
-  if (LS->current == '#') {
+  if (LS->current == l_c('#')) {
     do {  /* skip first line */
       next(LS);
-    } while (LS->current != '\n' && LS->current != EOZ);
+    } while (LS->current != l_c('\n') && LS->current != EOZ);
   }
 }
 
@@ -126,7 +130,7 @@ void luaX_setinput (lua_State *L, LexState *LS, ZIO *z, TString *source) {
 #define checkbuffer(L, n, len)  if ((len)+(n) > G(L)->Mbuffsize) \
                                   luaO_openspace(L, (len)+(n)+EXTRABUFF)
 
-#define save(L, c, l)   (G(L)->Mbuffer[l++] = (char)c)
+#define save(L, c, l)   (G(L)->Mbuffer[l++] = (l_char)c)
 #define save_and_next(L, LS, l)  (save(L, LS->current, l), next(LS))
 
 
@@ -137,8 +141,8 @@ static size_t readname (LexState *LS) {
   do {
     checkbuffer(L, 10, l);
     save_and_next(L, LS, l);
-  } while (isalnum(LS->current) || LS->current == '_');
-  save(L, '\0', l);
+  } while (isalnum(LS->current) || LS->current == l_c('_'));
+  save(L, l_c('\0'), l);
   return l-1;
 }
 
@@ -148,36 +152,37 @@ static void read_number (LexState *LS, int comma, SemInfo *seminfo) {
   lua_State *L = LS->L;
   size_t l = 0;
   checkbuffer(L, 10, l);
-  if (comma) save(L, '.', l);
+  if (comma) save(L, l_c('.'), l);
   while (isdigit(LS->current)) {
     checkbuffer(L, 10, l);
     save_and_next(L, LS, l);
   }
-  if (LS->current == '.') {
+  if (LS->current == l_c('.')) {
     save_and_next(L, LS, l);
-    if (LS->current == '.') {
+    if (LS->current == l_c('.')) {
       save_and_next(L, LS, l);
-      save(L, '\0', l);
-      luaX_error(LS, "ambiguous syntax"
-           " (decimal point x string concatenation)", TK_NUMBER);
+      save(L, l_c('\0'), l);
+      luaX_error(LS,
+                 l_s("ambiguous syntax (decimal point x string concatenation)"),
+                 TK_NUMBER);
     }
   }
   while (isdigit(LS->current)) {
     checkbuffer(L, 10, l);
     save_and_next(L, LS, l);
   }
-  if (LS->current == 'e' || LS->current == 'E') {
+  if (LS->current == l_c('e') || LS->current == l_c('E')) {
     save_and_next(L, LS, l);  /* read `E' */
-    if (LS->current == '+' || LS->current == '-')
+    if (LS->current == l_c('+') || LS->current == l_c('-'))
       save_and_next(L, LS, l);  /* optional exponent sign */
     while (isdigit(LS->current)) {
       checkbuffer(L, 10, l);
       save_and_next(L, LS, l);
     }
   }
-  save(L, '\0', l);
+  save(L, l_c('\0'), l);
   if (!luaO_str2d(G(L)->Mbuffer, &seminfo->r))
-    luaX_error(LS, "malformed number", TK_NUMBER);
+    luaX_error(LS, l_s("malformed number"), TK_NUMBER);
 }
 
 
@@ -186,32 +191,32 @@ static void read_long_string (LexState *LS, SemInfo *seminfo) {
   int cont = 0;
   size_t l = 0;
   checkbuffer(L, 10, l);
-  save(L, '[', l);  /* save first `[' */
+  save(L, l_c('['), l);  /* save first `[' */
   save_and_next(L, LS, l);  /* pass the second `[' */
   for (;;) {
     checkbuffer(L, 10, l);
     switch (LS->current) {
       case EOZ:
-        save(L, '\0', l);
-        luaX_error(LS, "unfinished long string", TK_STRING);
+        save(L, l_c('\0'), l);
+        luaX_error(LS, l_s("unfinished long string"), TK_STRING);
         break;  /* to avoid warnings */
-      case '[':
+      case l_c('['):
         save_and_next(L, LS, l);
-        if (LS->current == '[') {
+        if (LS->current == l_c('[')) {
           cont++;
           save_and_next(L, LS, l);
         }
         continue;
-      case ']':
+      case l_c(']'):
         save_and_next(L, LS, l);
-        if (LS->current == ']') {
+        if (LS->current == l_c(']')) {
           if (cont == 0) goto endloop;
           cont--;
           save_and_next(L, LS, l);
         }
         continue;
-      case '\n':
-        save(L, '\n', l);
+      case l_c('\n'):
+        save(L, l_c('\n'), l);
         inclinenumber(LS);
         continue;
       default:
@@ -219,7 +224,7 @@ static void read_long_string (LexState *LS, SemInfo *seminfo) {
     }
   } endloop:
   save_and_next(L, LS, l);  /* skip the second `]' */
-  save(L, '\0', l);
+  save(L, l_c('\0'), l);
   seminfo->ts = luaS_newlstr(L, G(L)->Mbuffer+2, l-5);
 }
 
@@ -232,38 +237,38 @@ static void read_string (LexState *LS, int del, SemInfo *seminfo) {
   while (LS->current != del) {
     checkbuffer(L, 10, l);
     switch (LS->current) {
-      case EOZ:  case '\n':
-        save(L, '\0', l);
-        luaX_error(LS, "unfinished string", TK_STRING);
+      case EOZ:  case l_c('\n'):
+        save(L, l_c('\0'), l);
+        luaX_error(LS, l_s("unfinished string"), TK_STRING);
         break;  /* to avoid warnings */
-      case '\\':
+      case l_c('\\'):
         next(LS);  /* do not save the `\' */
         switch (LS->current) {
-          case 'a': save(L, '\a', l); next(LS); break;
-          case 'b': save(L, '\b', l); next(LS); break;
-          case 'f': save(L, '\f', l); next(LS); break;
-          case 'n': save(L, '\n', l); next(LS); break;
-          case 'r': save(L, '\r', l); next(LS); break;
-          case 't': save(L, '\t', l); next(LS); break;
-          case 'v': save(L, '\v', l); next(LS); break;
-          case '\n': save(L, '\n', l); inclinenumber(LS); break;
-          case '0': case '1': case '2': case '3': case '4':
-          case '5': case '6': case '7': case '8': case '9': {
-            int c = 0;
-            int i = 0;
-            do {
-              c = 10*c + (LS->current-'0');
-              next(LS);
-            } while (++i<3 && isdigit(LS->current));
-            if (c > UCHAR_MAX) {
-              save(L, '\0', l);
-              luaX_error(LS, "escape sequence too large", TK_STRING);
+          case l_c('a'): save(L, l_c('\a'), l); next(LS); break;
+          case l_c('b'): save(L, l_c('\b'), l); next(LS); break;
+          case l_c('f'): save(L, l_c('\f'), l); next(LS); break;
+          case l_c('n'): save(L, l_c('\n'), l); next(LS); break;
+          case l_c('r'): save(L, l_c('\r'), l); next(LS); break;
+          case l_c('t'): save(L, l_c('\t'), l); next(LS); break;
+          case l_c('v'): save(L, l_c('\v'), l); next(LS); break;
+          case l_c('\n'): save(L, l_c('\n'), l); inclinenumber(LS); break;
+          default: {
+            if (!isdigit(LS->current))
+              save_and_next(L, LS, l);  /* handles \\, \", \', and \? */
+            else {  /* \xxx */
+              int c = 0;
+              int i = 0;
+              do {
+                c = 10*c + (LS->current-l_c('0'));
+                next(LS);
+              } while (++i<3 && isdigit(LS->current));
+              if (c > UCHAR_MAX) {
+                save(L, l_c('\0'), l);
+                luaX_error(LS, l_s("escape sequence too large"), TK_STRING);
+              }
+              save(L, c, l);
             }
-            save(L, c, l);
-            break;
           }
-          default:  /* handles \\, \", \', and \? */
-            save_and_next(L, LS, l);
         }
         break;
       default:
@@ -271,7 +276,7 @@ static void read_string (LexState *LS, int del, SemInfo *seminfo) {
     }
   }
   save_and_next(L, LS, l);  /* skip delimiter */
-  save(L, '\0', l);
+  save(L, l_c('\0'), l);
   seminfo->ts = luaS_newlstr(L, G(L)->Mbuffer+1, l-3);
 }
 
@@ -280,92 +285,85 @@ int luaX_lex (LexState *LS, SemInfo *seminfo) {
   for (;;) {
     switch (LS->current) {
 
-      case ' ': case '\t': case '\r':  /* `\r' to avoid problems with DOS */
+      case l_c(' '): case l_c('\t'): case l_c('\r'):  /* `\r' to avoid problems with DOS */
         next(LS);
         continue;
 
-      case '\n':
+      case l_c('\n'):
         inclinenumber(LS);
         continue;
 
-      case '$':
-        luaX_error(LS, "unexpected `$' (pragmas are no longer supported)", '$');
+      case l_c('$'):
+        luaX_error(LS,
+                   l_s("unexpected `$' (pragmas are no longer supported)"),
+                   LS->current);
         break;
 
-      case '-':
+      case l_c('-'):
         next(LS);
-        if (LS->current != '-') return '-';
-        do { next(LS); } while (LS->current != '\n' && LS->current != EOZ);
+        if (LS->current != l_c('-')) return l_c('-');
+        do { next(LS); } while (LS->current != l_c('\n') && LS->current != EOZ);
         continue;
 
-      case '[':
+      case l_c('['):
         next(LS);
-        if (LS->current != '[') return '[';
+        if (LS->current != l_c('[')) return l_c('[');
         else {
           read_long_string(LS, seminfo);
           return TK_STRING;
         }
 
-      case '=':
+      case l_c('='):
         next(LS);
-        if (LS->current != '=') return '=';
+        if (LS->current != l_c('=')) return l_c('=');
         else { next(LS); return TK_EQ; }
 
-      case '<':
+      case l_c('<'):
         next(LS);
-        if (LS->current != '=') return '<';
+        if (LS->current != l_c('=')) return l_c('<');
         else { next(LS); return TK_LE; }
 
-      case '>':
+      case l_c('>'):
         next(LS);
-        if (LS->current != '=') return '>';
+        if (LS->current != l_c('=')) return l_c('>');
         else { next(LS); return TK_GE; }
 
-      case '~':
+      case l_c('~'):
         next(LS);
-        if (LS->current != '=') return '~';
+        if (LS->current != l_c('=')) return l_c('~');
         else { next(LS); return TK_NE; }
 
-      case '"':
-      case '\'':
+      case l_c('"'):
+      case l_c('\''):
         read_string(LS, LS->current, seminfo);
         return TK_STRING;
 
-      case '.':
+      case l_c('.'):
         next(LS);
-        if (LS->current == '.') {
+        if (LS->current == l_c('.')) {
           next(LS);
-          if (LS->current == '.') {
+          if (LS->current == l_c('.')) {
             next(LS);
             return TK_DOTS;   /* ... */
           }
           else return TK_CONCAT;   /* .. */
         }
-        else if (!isdigit(LS->current)) return '.';
+        else if (!isdigit(LS->current)) return l_c('.');
         else {
           read_number(LS, 1, seminfo);
           return TK_NUMBER;
         }
 
-      case '0': case '1': case '2': case '3': case '4':
-      case '5': case '6': case '7': case '8': case '9':
-        read_number(LS, 0, seminfo);
-        return TK_NUMBER;
-
       case EOZ:
         return TK_EOS;
 
-      case '_': goto tname;
-
-      default:
-        if (!isalpha(LS->current)) {
-          int c = LS->current;
-          if (iscntrl(c))
-            luaX_invalidchar(LS, c);
-          next(LS);
-          return c;
+      default: {
+        if (isdigit(LS->current)) {
+          read_number(LS, 0, seminfo);
+          return TK_NUMBER;
         }
-        tname: {  /* identifier or reserved word */
+        else if (isalpha(LS->current) || LS->current == l_c('_')) {
+          /* identifier or reserved word */
           size_t l = readname(LS);
           TString *ts = luaS_newlstr(LS->L, G(LS->L)->Mbuffer, l);
           if (ts->marked >= RESERVEDMARK)  /* reserved word? */
@@ -373,6 +371,14 @@ int luaX_lex (LexState *LS, SemInfo *seminfo) {
           seminfo->ts = ts;
           return TK_NAME;
         }
+        else {
+          int c = LS->current;
+          if (iscntrl(c))
+            luaX_invalidchar(LS, c);
+          next(LS);
+          return c;  /* single-char tokens (+ - / ...) */
+        }
+      }
     }
   }
 }