- Rework the way literals are handled, these are now kept as strings until

[cparser] / token.c

diff --git a/token.c b/token.c
index 52ff730..1fbaafd 100644 (file)
--- a/token.c
+++ b/token.c
@@ -1,6 +1,6 @@
 /*
  * This file is part of cparser.
- * Copyright (C) 2007-2008 Matthias Braun <matze@braunis.de>
+ * Copyright (C) 2007-2009 Matthias Braun <matze@braunis.de>
  *
  * This program is free software; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License
@@ -32,7 +32,7 @@
 static symbol_t *token_symbols[T_LAST_TOKEN];
 static symbol_t *pp_token_symbols[TP_LAST_TOKEN];
 
-source_position_t builtin_source_position = { "<built-in>", 0 };
+const source_position_t builtin_source_position = { "<built-in>", 0 };
 
 void init_tokens(void)
 {
@@ -51,13 +51,15 @@ void init_tokens(void)
                assert(T_##x >= 0 && T_##x < T_LAST_TOKEN);                \
                symbol               = symbol_table_insert(str);           \
                symbol->ID           = T_##x;                              \
-               token_symbols[T_##x] = symbol;                             \
+               if (token_symbols[T_##x] == NULL)                          \
+                       token_symbols[T_##x] = symbol;                         \
        }
 
 #define TS(x,str,val)                                              \
        assert(T_##x >= 0 && T_##x < T_LAST_TOKEN);                    \
        symbol               = symbol_table_insert(str);               \
-       token_symbols[T_##x] = symbol;
+       if (token_symbols[T_##x] == NULL)                              \
+               token_symbols[T_##x] = symbol;                             \
 
 #include "tokens.inc"
 
@@ -65,15 +67,19 @@ void init_tokens(void)
 #undef T
 
 #define T(mode,x,str,val)                                          \
-       assert(TP_##x >= 0 && TP_##x < TP_LAST_TOKEN);                 \
-       symbol                   = symbol_table_insert(str);           \
-       symbol->pp_ID            = TP_##x;                             \
-       pp_token_symbols[TP_##x] = symbol;
+       if (c_mode & (mode)) {                                         \
+               assert(TP_##x >= 0 && TP_##x < TP_LAST_TOKEN);             \
+               symbol                   = symbol_table_insert(str);       \
+               symbol->pp_ID            = TP_##x;                         \
+               if (pp_token_symbols[TP_##x] == NULL)                      \
+                       pp_token_symbols[TP_##x] = symbol;                     \
+       }
 
 #define TS(x,str,val)                                              \
        assert(TP_##x >= 0 && TP_##x < T_LAST_TOKEN);                  \
        symbol                   = symbol_table_insert(str);           \
-       pp_token_symbols[TP_##x] = symbol;
+       if (pp_token_symbols[TP_##x] == NULL)                          \
+               pp_token_symbols[TP_##x] = symbol;
 
 #include "tokens_preprocessor.inc"
 
@@ -103,60 +109,89 @@ void print_token_type(FILE *f, token_type_t token_type)
 
        const symbol_t *symbol = token_symbols[token_type];
        if(symbol != NULL) {
-               fprintf(f, "'%s'", symbol->string);
+               fputs(symbol->string, f);
        } else {
                if(token_type >= 0 && token_type < 256) {
-                       fprintf(f, "'%c'", token_type);
+                       fputc(token_type, f);
                        return;
                }
                fputs("unknown token", f);
        }
 }
 
+symbol_t *get_token_symbol(const token_t *token)
+{
+       return token_symbols[token->type];
+}
+
+static void print_stringrep(const string_t *string, FILE *f)
+{
+       for (size_t i = 0; i < string->size; ++i) {
+               fputc(string->begin[i], f);
+       }
+}
+
 void print_token(FILE *f, const token_t *token)
 {
        switch(token->type) {
        case T_IDENTIFIER:
-               fprintf(f, "symbol '%s'", token->v.symbol->string);
+               fprintf(f, "identifier '%s'", token->symbol->string);
                break;
        case T_INTEGER:
-               fprintf(f, "integer number %lld", token->v.intvalue);
-               break;
+       case T_INTEGER_OCTAL:
+       case T_INTEGER_HEXADECIMAL:
        case T_FLOATINGPOINT:
-               fprintf(f, "floatingpointer number %LF", token->v.floatvalue);
+       case T_FLOATINGPOINT_HEXADECIMAL:
+               print_token_type(f, (token_type_t)token->type);
+               fputs(" '", f);
+               print_stringrep(&token->literal, f);
+               if (token->symbol != NULL)
+                       fputs(token->symbol->string, f);
+               fputc('\'', f);
                break;
+       case T_WIDE_STRING_LITERAL:
        case T_STRING_LITERAL:
-               fprintf(f, "string '%s'", token->v.string.begin); /* TODO suboptimal */
+               print_token_type(f, (token_type_t)token->type);
+               fprintf(f, " \"%s\"", token->literal.begin);
+               break;
+       case T_CHARACTER_CONSTANT:
+       case T_WIDE_CHARACTER_CONSTANT:
+               print_token_type(f, (token_type_t)token->type);
+               fputs(" \'", f);
+               print_stringrep(&token->literal, f);
+               fputs("'", f);
                break;
        default:
+               fputc('\'', f);
                print_token_type(f, (token_type_t)token->type);
+               fputc('\'', f);
                break;
        }
 }
 
-void print_pp_token_type(FILE *f, preprocessor_token_type_t token_type)
+void print_pp_token_type(FILE *f, int token_type)
 {
-       if(token_type == TP_EOF) {
+       if (token_type == TP_EOF) {
                fputs("end of file", f);
                return;
        }
-       if(token_type == TP_ERROR) {
+       if (token_type == TP_ERROR) {
                fputs("error", f);
                return;
        }
 
        int token_symbols_len = TP_LAST_TOKEN;
-       if(token_type < 0 || token_type >= token_symbols_len) {
+       if (token_type < 0 || token_type >= token_symbols_len) {
                fputs("invalid token", f);
                return;
        }
 
        const symbol_t *symbol = pp_token_symbols[token_type];
-       if(symbol != NULL) {
-               fprintf(f, "'%s'", symbol->string);
+       if (symbol != NULL) {
+               fputs(symbol->string, f);
        } else {
                if(token_type >= 0 && token_type < 256) {
-                       fprintf(f, "'%c'", token_type);
+                       fputc(token_type, f);
                        return;
                }
                fputs("unknown token", f);
@@ -167,13 +202,13 @@ void print_pp_token(FILE *f, const token_t *token)
 {
        switch((preprocessor_token_type_t) token->type) {
        case TP_IDENTIFIER:
-               fprintf(f, "symbol '%s'", token->v.symbol->string);
+               fprintf(f, "identifier '%s'", token->symbol->string);
                break;
        case TP_NUMBER:
-               fprintf(f, "number %s", token->v.string.begin);
+               fprintf(f, "number '%s'", token->literal.begin);
                break;
        case TP_STRING_LITERAL:
-               fprintf(f, "string '%s'", token->v.string.begin);
+               fprintf(f, "string \"%s\"", token->literal.begin);
                break;
        default:
                print_pp_token_type(f, (preprocessor_token_type_t) token->type);