tinycc/tccpp.c

/*
 *  TCC - Tiny C Compiler
 * 
 *  Copyright (c) 2001-2004 Fabrice Bellard
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */


static const char tcc_keywords[] = 
#define DEF(id, str) str "\0"
#include "tcctok.h"
#undef DEF
;

/* WARNING: the content of this string encodes token numbers */
static const unsigned char tok_two_chars[] =
    "<=\236>=\235!=\225&&\240||\241++\244--\242==\224<<\1>>\2+=\253"
    "-=\255*=\252/=\257%=\245&=\246^=\336|=\374->\313..\250##\266";

/* true if isid(c) || isnum(c) */
static unsigned char isidnum_table[256-CH_EOF];


struct macro_level {
    struct macro_level *prev;
    int *p;
};

static void next_nomacro(void);
static void next_nomacro_spc(void);
static void macro_subst(TokenString *tok_str, Sym **nested_list,
                        const int *macro_str, struct macro_level **can_read_stream);


/* allocate a new token */
static TokenSym *tok_alloc_new(TokenSym **pts, const char *str, int len)
{
    TokenSym *ts, **ptable;
    int i;

    if (tok_ident >= SYM_FIRST_ANOM) 
        error("memory full");

    /* expand token table if needed */
    i = tok_ident - TOK_IDENT;
    if ((i % TOK_ALLOC_INCR) == 0) {
        ptable = tcc_realloc(table_ident, (i + TOK_ALLOC_INCR) * sizeof(TokenSym *));
        if (!ptable)
            error("memory full");
        table_ident = ptable;
    }

    ts = tcc_malloc(sizeof(TokenSym) + len);
    table_ident[i] = ts;
    ts->tok = tok_ident++;
    ts->sym_define = NULL;
    ts->sym_label = NULL;
    ts->sym_struct = NULL;
    ts->sym_identifier = NULL;
    ts->len = len;
    ts->hash_next = NULL;
    memcpy(ts->str, str, len);
    ts->str[len] = '\0';
    *pts = ts;
    return ts;
}

#define TOK_HASH_INIT 1
#define TOK_HASH_FUNC(h, c) ((h) * 263 + (c))

/* find a token and add it if not found */
static TokenSym *tok_alloc(const char *str, int len)
{
    TokenSym *ts, **pts;
    int i;
    unsigned int h;
    
    h = TOK_HASH_INIT;
    for(i=0;i<len;i++)
        h = TOK_HASH_FUNC(h, ((unsigned char *)str)[i]);
    h &= (TOK_HASH_SIZE - 1);

    pts = &hash_ident[h];
    for(;;) {
        ts = *pts;
        if (!ts)
            break;
        if (ts->len == len && !memcmp(ts->str, str, len))
            return ts;
        pts = &(ts->hash_next);
    }
    return tok_alloc_new(pts, str, len);
}

/* XXX: buffer overflow */
/* XXX: float tokens */
char *get_tok_str(int v, CValue *cv)
{
    static char buf[STRING_MAX_SIZE + 1];
    static CString cstr_buf;
    CString *cstr;
    char *p;
    int i, len;

    /* NOTE: to go faster, we give a fixed buffer for small strings */
    cstr_reset(&cstr_buf);
    cstr_buf.data = buf;
    cstr_buf.size_allocated = sizeof(buf);
    p = buf;

    switch(v) {
    case TOK_CINT:
    case TOK_CUINT:
        /* XXX: not quite exact, but only useful for testing */
        sprintf(p, "%u", cv->ui);
        break;
    case TOK_CLLONG:
    case TOK_CULLONG:
        /* XXX: not quite exact, but only useful for testing  */
#ifdef _WIN32
        sprintf(p, "%u", (unsigned)cv->ull);
#else
        sprintf(p, "%Lu", cv->ull);
#endif
        break;
    case TOK_LCHAR:
        cstr_ccat(&cstr_buf, 'L');
    case TOK_CCHAR:
        cstr_ccat(&cstr_buf, '\'');
        add_char(&cstr_buf, cv->i);
        cstr_ccat(&cstr_buf, '\'');
        cstr_ccat(&cstr_buf, '\0');
        break;
    case TOK_PPNUM:
        cstr = cv->cstr;
        len = cstr->size - 1;
        for(i=0;i<len;i++)
            add_char(&cstr_buf, ((unsigned char *)cstr->data)[i]);
        cstr_ccat(&cstr_buf, '\0');
        break;
    case TOK_LSTR:
        cstr_ccat(&cstr_buf, 'L');
    case TOK_STR:
        cstr = cv->cstr;
        cstr_ccat(&cstr_buf, '\"');
        if (v == TOK_STR) {
            len = cstr->size - 1;
            for(i=0;i<len;i++)
                add_char(&cstr_buf, ((unsigned char *)cstr->data)[i]);
        } else {
            len = (cstr->size / sizeof(nwchar_t)) - 1;
            for(i=0;i<len;i++)
                add_char(&cstr_buf, ((nwchar_t *)cstr->data)[i]);
        }
        cstr_ccat(&cstr_buf, '\"');
        cstr_ccat(&cstr_buf, '\0');
        break;
    case TOK_LT:
        v = '<';
        goto addv;
    case TOK_GT:
        v = '>';
        goto addv;
    case TOK_DOTS:
        return strcpy(p, "...");
    case TOK_A_SHL:
        return strcpy(p, "<<=");
    case TOK_A_SAR:
        return strcpy(p, ">>=");
    default:
        if (v < TOK_IDENT) {
            /* search in two bytes table */
            const unsigned char *q = tok_two_chars;
            while (*q) {
                if (q[2] == v) {
                    *p++ = q[0];
                    *p++ = q[1];
                    *p = '\0';
                    return buf;
                }
                q += 3;
            }
        addv:
            *p++ = v;
            *p = '\0';
        } else if (v < tok_ident) {
            return table_ident[v - TOK_IDENT]->str;
        } else if (v >= SYM_FIRST_ANOM) {
            /* special name for anonymous symbol */
            sprintf(p, "L.%u", v - SYM_FIRST_ANOM);
        } else {
            /* should never happen */
            return NULL;
        }
        break;
    }
    return cstr_buf.data;
}

/* fill input buffer and peek next char */
static int tcc_peekc_slow(BufferedFile *bf)
{
    int len;
    /* only tries to read if really end of buffer */
    if (bf->buf_ptr >= bf->buf_end) {
        if (bf->fd != -1) {
#if defined(PARSE_DEBUG)
            len = 8;
#else
            len = IO_BUF_SIZE;
#endif
            len = read(bf->fd, bf->buffer, len);
            if (len < 0)
                len = 0;
        } else {
            len = 0;
        }
        total_bytes += len;
        bf->buf_ptr = bf->buffer;
        bf->buf_end = bf->buffer + len;
        *bf->buf_end = CH_EOB;
    }
    if (bf->buf_ptr < bf->buf_end) {
        return bf->buf_ptr[0];
    } else {
        bf->buf_ptr = bf->buf_end;
        return CH_EOF;
    }
}

/* return the current character, handling end of block if necessary
   (but not stray) */
static int handle_eob(void)
{
    return tcc_peekc_slow(file);
}

/* read next char from current input file and handle end of input buffer */
static inline void inp(void)
{
    ch = *(++(file->buf_ptr));
    /* end of buffer/file handling */
    if (ch == CH_EOB)
        ch = handle_eob();
}

/* handle '\[\r]\n' */
static int handle_stray_noerror(void)
{
    while (ch == '\\') {
        inp();
        if (ch == '\n') {
            file->line_num++;
            inp();
        } else if (ch == '\r') {
            inp();
            if (ch != '\n')
                goto fail;
            file->line_num++;
            inp();
        } else {
        fail:
            return 1;
        }
    }
    return 0;
}

static void handle_stray(void)
{
    if (handle_stray_noerror())
        error("stray '\\' in program");
}

/* skip the stray and handle the \\n case. Output an error if
   incorrect char after the stray */
static int handle_stray1(uint8_t *p)
{
    int c;

    if (p >= file->buf_end) {
        file->buf_ptr = p;
        c = handle_eob();
        p = file->buf_ptr;
        if (c == '\\')
            goto parse_stray;
    } else {
    parse_stray:
        file->buf_ptr = p;
        ch = *p;
        handle_stray();
        p = file->buf_ptr;
        c = *p;
    }
    return c;
}

/* handle just the EOB case, but not stray */
#define PEEKC_EOB(c, p)\
{\
    p++;\
    c = *p;\
    if (c == '\\') {\
        file->buf_ptr = p;\
        c = handle_eob();\
        p = file->buf_ptr;\
    }\
}

/* handle the complicated stray case */
#define PEEKC(c, p)\
{\
    p++;\
    c = *p;\
    if (c == '\\') {\
        c = handle_stray1(p);\
        p = file->buf_ptr;\
    }\
}

/* input with '\[\r]\n' handling. Note that this function cannot
   handle other characters after '\', so you cannot call it inside
   strings or comments */
static void minp(void)
{
    inp();
    if (ch == '\\') 
        handle_stray();
}


/* single line C++ comments */
static uint8_t *parse_line_comment(uint8_t *p)
{
    int c;

    p++;
    for(;;) {
        c = *p;
    redo:
        if (c == '\n' || c == CH_EOF) {
            break;
        } else if (c == '\\') {
            file->buf_ptr = p;
            c = handle_eob();
            p = file->buf_ptr;
            if (c == '\\') {
                PEEKC_EOB(c, p);
                if (c == '\n') {
                    file->line_num++;
                    PEEKC_EOB(c, p);
                } else if (c == '\r') {
                    PEEKC_EOB(c, p);
                    if (c == '\n') {
                        file->line_num++;
                        PEEKC_EOB(c, p);
                    }
                }
            } else {
                goto redo;
            }
        } else {
            p++;
        }
    }
    return p;
}

/* C comments */
static uint8_t *parse_comment(uint8_t *p)
{
    int c;
    
    p++;
    for(;;) {
        /* fast skip loop */
        for(;;) {
            c = *p;
            if (c == '\n' || c == '*' || c == '\\')
                break;
            p++;
            c = *p;
            if (c == '\n' || c == '*' || c == '\\')
                break;
            p++;
        }
        /* now we can handle all the cases */
        if (c == '\n') {
            file->line_num++;
            p++;
        } else if (c == '*') {
            p++;
            for(;;) {
                c = *p;
                if (c == '*') {
                    p++;
                } else if (c == '/') {
                    goto end_of_comment;
                } else if (c == '\\') {
                    file->buf_ptr = p;
                    c = handle_eob();
                    p = file->buf_ptr;
                    if (c == '\\') {
                        /* skip '\[\r]\n', otherwise just skip the stray */
                        while (c == '\\') {
                            PEEKC_EOB(c, p);
                            if (c == '\n') {
                                file->line_num++;
                                PEEKC_EOB(c, p);
                            } else if (c == '\r') {
                                PEEKC_EOB(c, p);
                                if (c == '\n') {
                                    file->line_num++;
                                    PEEKC_EOB(c, p);
                                }
                            } else {
                                goto after_star;
                            }
                        }
                    }
                } else {
                    break;
                }
            }
        after_star: ;
        } else {
            /* stray, eob or eof */
            file->buf_ptr = p;
            c = handle_eob();
            p = file->buf_ptr;
            if (c == CH_EOF) {
                error("unexpected end of file in comment");
            } else if (c == '\\') {
                p++;
            }
        }
    }
 end_of_comment:
    p++;
    return p;
}

#define cinp minp

static inline void skip_spaces(void)
{
    while (is_space(ch))
        cinp();
}

static inline int check_space(int t, int *spc) 
{
    if (is_space(t)) {
        if (*spc) 
            return 1;
        *spc = 1;
    } else 
        *spc = 0;
    return 0;
}

/* parse a string without interpreting escapes */
static uint8_t *parse_pp_string(uint8_t *p,
                                int sep, CString *str)
{
    int c;
    p++;
    for(;;) {
        c = *p;
        if (c == sep) {
            break;
        } else if (c == '\\') {
            file->buf_ptr = p;
            c = handle_eob();
            p = file->buf_ptr;
            if (c == CH_EOF) {
            unterminated_string:
                /* XXX: indicate line number of start of string */
                error("missing terminating %c character", sep);
            } else if (c == '\\') {
                /* escape : just skip \[\r]\n */
                PEEKC_EOB(c, p);
                if (c == '\n') {
                    file->line_num++;
                    p++;
                } else if (c == '\r') {
                    PEEKC_EOB(c, p);
                    if (c != '\n')
                        expect("'\n' after '\r'");
                    file->line_num++;
                    p++;
                } else if (c == CH_EOF) {
                    goto unterminated_string;
                } else {
                    if (str) {
                        cstr_ccat(str, '\\');
                        cstr_ccat(str, c);
                    }
                    p++;
                }
            }
        } else if (c == '\n') {
            file->line_num++;
            goto add_char;
        } else if (c == '\r') {
            PEEKC_EOB(c, p);
            if (c != '\n') {
                if (str)
                    cstr_ccat(str, '\r');
            } else {
                file->line_num++;
                goto add_char;
            }
        } else {
        add_char:
            if (str)
                cstr_ccat(str, c);
            p++;
        }
    }
    p++;
    return p;
}

/* skip block of text until #else, #elif or #endif. skip also pairs of
   #if/#endif */
void preprocess_skip(void)
{
    int a, start_of_line, c, in_warn_or_error;
    uint8_t *p;

    p = file->buf_ptr;
    a = 0;
redo_start:
    start_of_line = 1;
    in_warn_or_error = 0;
    for(;;) {
    redo_no_start:
        c = *p;
        switch(c) {
        case ' ':
        case '\t':
        case '\f':
        case '\v':
        case '\r':
            p++;
            goto redo_no_start;
        case '\n':
            file->line_num++;
            p++;
            goto redo_start;
        case '\\':
            file->buf_ptr = p;
            c = handle_eob();
            if (c == CH_EOF) {
                expect("#endif");
            } else if (c == '\\') {
                ch = file->buf_ptr[0];
                handle_stray_noerror();
            }
            p = file->buf_ptr;
            goto redo_no_start;
        /* skip strings */
        case '\"':
        case '\'':
            if (in_warn_or_error)
                goto _default;
            p = parse_pp_string(p, c, NULL);
            break;
        /* skip comments */
        case '/':
            if (in_warn_or_error)
                goto _default;
            file->buf_ptr = p;
            ch = *p;
            minp();
            p = file->buf_ptr;
            if (ch == '*') {
                p = parse_comment(p);
            } else if (ch == '/') {
                p = parse_line_comment(p);
            }
            break;
        case '#':
            p++;
            if (start_of_line) {
                file->buf_ptr = p;
                next_nomacro();
                p = file->buf_ptr;
                if (a == 0 && 
                    (tok == TOK_ELSE || tok == TOK_ELIF || tok == TOK_ENDIF))
                    goto the_end;
                if (tok == TOK_IF || tok == TOK_IFDEF || tok == TOK_IFNDEF)
                    a++;
                else if (tok == TOK_ENDIF)
                    a--;
                else if( tok == TOK_ERROR || tok == TOK_WARNING)
                    in_warn_or_error = 1;
            }
            break;
_default:
        default:
            p++;
            break;
        }
        start_of_line = 0;
    }
 the_end: ;
    file->buf_ptr = p;
}

/* ParseState handling */

/* XXX: currently, no include file info is stored. Thus, we cannot display
   accurate messages if the function or data definition spans multiple
   files */

/* save current parse state in 's' */
void save_parse_state(ParseState *s)
{
    s->line_num = file->line_num;
    s->macro_ptr = macro_ptr;
    s->tok = tok;
    s->tokc = tokc;
}

/* restore parse state from 's' */
void restore_parse_state(ParseState *s)
{
    file->line_num = s->line_num;
    macro_ptr = s->macro_ptr;
    tok = s->tok;
    tokc = s->tokc;
}

/* return the number of additional 'ints' necessary to store the
   token */
static inline int tok_ext_size(int t)
{
    switch(t) {
        /* 4 bytes */
    case TOK_CINT:
    case TOK_CUINT:
    case TOK_CCHAR:
    case TOK_LCHAR:
    case TOK_CFLOAT:
    case TOK_LINENUM:
        return 1;
    case TOK_STR:
    case TOK_LSTR:
    case TOK_PPNUM:
        error("unsupported token");
        return 1;
    case TOK_CDOUBLE:
    case TOK_CLLONG:
    case TOK_CULLONG:
        return 2;
    case TOK_CLDOUBLE:
        return LDOUBLE_SIZE / 4;
    default:
        return 0;
    }
}

/* token string handling */

static inline void tok_str_new(TokenString *s)
{
    s->str = NULL;
    s->len = 0;
    s->allocated_len = 0;
    s->last_line_num = -1;
}

static void tok_str_free(int *str)
{
    tcc_free(str);
}

static int *tok_str_realloc(TokenString *s)
{
    int *str, len;

    if (s->allocated_len == 0) {
        len = 8;
    } else {
        len = s->allocated_len * 2;
    }
    str = tcc_realloc(s->str, len * sizeof(int));
    if (!str)
        error("memory full");
    s->allocated_len = len;
    s->str = str;
    return str;
}

static void tok_str_add(TokenString *s, int t)
{
    int len, *str;

    len = s->len;
    str = s->str;
    if (len >= s->allocated_len)
        str = tok_str_realloc(s);
    str[len++] = t;
    s->len = len;
}

static void tok_str_add2(TokenString *s, int t, CValue *cv)
{
    int len, *str;

    len = s->len;
    str = s->str;

    /* allocate space for worst case */
    if (len + TOK_MAX_SIZE > s->allocated_len)
        str = tok_str_realloc(s);
    str[len++] = t;
    switch(t) {
    case TOK_CINT:
    case TOK_CUINT:
    case TOK_CCHAR:
    case TOK_LCHAR:
    case TOK_CFLOAT:
    case TOK_LINENUM:
        str[len++] = cv->tab[0];
        break;
    case TOK_PPNUM:
    case TOK_STR:
    case TOK_LSTR:
        {
            int nb_words;
            CString *cstr;

            nb_words = (sizeof(CString) + cv->cstr->size + 3) >> 2;
            while ((len + nb_words) > s->allocated_len)
                str = tok_str_realloc(s);
            cstr = (CString *)(str + len);
            cstr->data = NULL;
            cstr->size = cv->cstr->size;
            cstr->data_allocated = NULL;
            cstr->size_allocated = cstr->size;
            memcpy((char *)cstr + sizeof(CString), 
                   cv->cstr->data, cstr->size);
            len += nb_words;
        }
        break;
    case TOK_CDOUBLE:
    case TOK_CLLONG:
    case TOK_CULLONG:
#if LDOUBLE_SIZE == 8
    case TOK_CLDOUBLE:
#endif
        str[len++] = cv->tab[0];
        str[len++] = cv->tab[1];
        break;
#if LDOUBLE_SIZE == 12
    case TOK_CLDOUBLE:
        str[len++] = cv->tab[0];
        str[len++] = cv->tab[1];
        str[len++] = cv->tab[2];
#elif LDOUBLE_SIZE == 16
    case TOK_CLDOUBLE:
        str[len++] = cv->tab[0];
        str[len++] = cv->tab[1];
        str[len++] = cv->tab[2];
        str[len++] = cv->tab[3];
#elif LDOUBLE_SIZE != 8
#error add long double size support
#endif
        break;
    default:
        break;
    }
    s->len = len;
}

/* add the current parse token in token string 's' */
static void tok_str_add_tok(TokenString *s)
{
    CValue cval;

    /* save line number info */
    if (file->line_num != s->last_line_num) {
        s->last_line_num = file->line_num;
        cval.i = s->last_line_num;
        tok_str_add2(s, TOK_LINENUM, &cval);
    }
    tok_str_add2(s, tok, &tokc);
}

#if LDOUBLE_SIZE == 16
#define LDOUBLE_GET(p, cv)                      \
        cv.tab[0] = p[0];                       \
        cv.tab[1] = p[1];                       \
        cv.tab[2] = p[2];                       \
        cv.tab[3] = p[3];
#elif LDOUBLE_SIZE == 12
#define LDOUBLE_GET(p, cv)                      \
        cv.tab[0] = p[0];                       \
        cv.tab[1] = p[1];                       \
        cv.tab[2] = p[2];
#elif LDOUBLE_SIZE == 8
#define LDOUBLE_GET(p, cv)                      \
        cv.tab[0] = p[0];                       \
        cv.tab[1] = p[1];
#else
#error add long double size support
#endif


/* get a token from an integer array and increment pointer
   accordingly. we code it as a macro to avoid pointer aliasing. */
#define TOK_GET(t, p, cv)                       \
{                                               \
    t = *p++;                                   \
    switch(t) {                                 \
    case TOK_CINT:                              \
    case TOK_CUINT:                             \
    case TOK_CCHAR:                             \
    case TOK_LCHAR:                             \
    case TOK_CFLOAT:                            \
    case TOK_LINENUM:                           \
        cv.tab[0] = *p++;                       \
        break;                                  \
    case TOK_STR:                               \
    case TOK_LSTR:                              \
    case TOK_PPNUM:                             \
        cv.cstr = (CString *)p;                 \
        cv.cstr->data = (char *)p + sizeof(CString);\
        p += (sizeof(CString) + cv.cstr->size + 3) >> 2;\
        break;                                  \
    case TOK_CDOUBLE:                           \
    case TOK_CLLONG:                            \
    case TOK_CULLONG:                           \
        cv.tab[0] = p[0];                       \
        cv.tab[1] = p[1];                       \
        p += 2;                                 \
        break;                                  \
    case TOK_CLDOUBLE:                          \
        LDOUBLE_GET(p, cv);                     \
        p += LDOUBLE_SIZE / 4;                  \
        break;                                  \
    default:                                    \
        break;                                  \
    }                                           \
}

/* defines handling */
static inline void define_push(int v, int macro_type, int *str, Sym *first_arg)
{
    Sym *s;

    s = sym_push2(&define_stack, v, macro_type, 0);
    s->d = str;
    s->next = first_arg;
    table_ident[v - TOK_IDENT]->sym_define = s;
}

/* undefined a define symbol. Its name is just set to zero */
static void define_undef(Sym *s)
{
    int v;
    v = s->v;
    if (v >= TOK_IDENT && v < tok_ident)
        table_ident[v - TOK_IDENT]->sym_define = NULL;
    s->v = 0;
}

static inline Sym *define_find(int v)
{
    v -= TOK_IDENT;
    if ((unsigned)v >= (unsigned)(tok_ident - TOK_IDENT))
        return NULL;
    return table_ident[v]->sym_define;
}

/* free define stack until top reaches 'b' */
static void free_defines(Sym *b)
{
    Sym *top, *top1;
    int v;

    top = define_stack;
    while (top != b) {
        top1 = top->prev;
        /* do not free args or predefined defines */
        if (top->d)
            tok_str_free(top->d);
        v = top->v;
        if (v >= TOK_IDENT && v < tok_ident)
            table_ident[v - TOK_IDENT]->sym_define = NULL;
        sym_free(top);
        top = top1;
    }
    define_stack = b;
}

/* label lookup */
static Sym *label_find(int v)
{
    v -= TOK_IDENT;
    if ((unsigned)v >= (unsigned)(tok_ident - TOK_IDENT))
        return NULL;
    return table_ident[v]->sym_label;
}

static Sym *label_push(Sym **ptop, int v, int flags)
{
    Sym *s, **ps;
    s = sym_push2(ptop, v, 0, 0);
    s->r = flags;
    ps = &table_ident[v - TOK_IDENT]->sym_label;
    if (ptop == &global_label_stack) {
        /* modify the top most local identifier, so that
           sym_identifier will point to 's' when popped */
        while (*ps != NULL)
            ps = &(*ps)->prev_tok;
    }
    s->prev_tok = *ps;
    *ps = s;
    return s;
}

/* pop labels until element last is reached. Look if any labels are
   undefined. Define symbols if '&&label' was used. */
static void label_pop(Sym **ptop, Sym *slast)
{
    Sym *s, *s1;
    for(s = *ptop; s != slast; s = s1) {
        s1 = s->prev;
        if (s->r == LABEL_DECLARED) {
            warning("label '%s' declared but not used", get_tok_str(s->v, NULL));
        } else if (s->r == LABEL_FORWARD) {
                error("label '%s' used but not defined",
                      get_tok_str(s->v, NULL));
        } else {
            if (s->c) {
                /* define corresponding symbol. A size of
                   1 is put. */
                put_extern_sym(s, cur_text_section, s->jnext, 1);
            }
        }
        /* remove label */
        table_ident[s->v - TOK_IDENT]->sym_label = s->prev_tok;
        sym_free(s);
    }
    *ptop = slast;
}

/* eval an expression for #if/#elif */
static int expr_preprocess(void)
{
    int c, t;
    TokenString str;
    
    tok_str_new(&str);
    while (tok != TOK_LINEFEED && tok != TOK_EOF) {
        next(); /* do macro subst */
        if (tok == TOK_DEFINED) {
            next_nomacro();
            t = tok;
            if (t == '(') 
                next_nomacro();
            c = define_find(tok) != 0;
            if (t == '(')
                next_nomacro();
            tok = TOK_CINT;
            tokc.i = c;
        } else if (tok >= TOK_IDENT) {
            /* if undefined macro */
            tok = TOK_CINT;
            tokc.i = 0;
        }
        tok_str_add_tok(&str);
    }
    tok_str_add(&str, -1); /* simulate end of file */
    tok_str_add(&str, 0);
    /* now evaluate C constant expression */
    macro_ptr = str.str;
    next();
    c = expr_const();
    macro_ptr = NULL;
    tok_str_free(str.str);
    return c != 0;
}

#if defined(PARSE_DEBUG) || defined(PP_DEBUG)
static void tok_print(int *str)
{
    int t;
    CValue cval;

    printf("<");
    while (1) {
        TOK_GET(t, str, cval);
        if (!t)
            break;
        printf("%s", get_tok_str(t, &cval));
    }
    printf(">\n");
}
#endif

/* parse after #define */
static void parse_define(void)
{
    Sym *s, *first, **ps;
    int v, t, varg, is_vaargs, spc;
    TokenString str;
    
    v = tok;
    if (v < TOK_IDENT)
        error("invalid macro name '%s'", get_tok_str(tok, &tokc));
    /* XXX: should check if same macro (ANSI) */
    first = NULL;
    t = MACRO_OBJ;
    /* '(' must be just after macro definition for MACRO_FUNC */
    next_nomacro_spc();
    if (tok == '(') {
        next_nomacro();
        ps = &first;
        while (tok != ')') {
            varg = tok;
            next_nomacro();
            is_vaargs = 0;
            if (varg == TOK_DOTS) {
                varg = TOK___VA_ARGS__;
                is_vaargs = 1;
            } else if (tok == TOK_DOTS && gnu_ext) {
                is_vaargs = 1;
                next_nomacro();
            }
            if (varg < TOK_IDENT)
                error("badly punctuated parameter list");
            s = sym_push2(&define_stack, varg | SYM_FIELD, is_vaargs, 0);
            *ps = s;
            ps = &s->next;
            if (tok != ',')
                break;
            next_nomacro();
        }
        if (tok == ')')
            next_nomacro_spc();
        t = MACRO_FUNC;
    }
    tok_str_new(&str);
    spc = 2;
    /* EOF testing necessary for '-D' handling */
    while (tok != TOK_LINEFEED && tok != TOK_EOF) {
        /* remove spaces around ## and after '#' */        
        if (TOK_TWOSHARPS == tok) {
            if (1 == spc)
                --str.len;
            spc = 2;
        } else if ('#' == tok) {
            spc = 2;
        } else if (check_space(tok, &spc)) {
            goto skip;
        }
        tok_str_add2(&str, tok, &tokc);
    skip:
        next_nomacro_spc();
    }
    if (spc == 1)
        --str.len; /* remove trailing space */
    tok_str_add(&str, 0);
#ifdef PP_DEBUG
    printf("define %s %d: ", get_tok_str(v, NULL), t);
    tok_print(str.str);
#endif
    define_push(v, t, str.str, first);
}

static inline int hash_cached_include(int type, const char *filename)
{
    const unsigned char *s;
    unsigned int h;

    h = TOK_HASH_INIT;
    h = TOK_HASH_FUNC(h, type);
    s = filename;
    while (*s) {
        h = TOK_HASH_FUNC(h, *s);
        s++;
    }
    h &= (CACHED_INCLUDES_HASH_SIZE - 1);
    return h;
}

/* XXX: use a token or a hash table to accelerate matching ? */
static CachedInclude *search_cached_include(TCCState *s1,
                                            int type, const char *filename)
{
    CachedInclude *e;
    int i, h;
    h = hash_cached_include(type, filename);
    i = s1->cached_includes_hash[h];
    for(;;) {
        if (i == 0)
            break;
        e = s1->cached_includes[i - 1];
        if (e->type == type && !PATHCMP(e->filename, filename))
            return e;
        i = e->hash_next;
    }
    return NULL;
}

static inline void add_cached_include(TCCState *s1, int type, 
                                      const char *filename, int ifndef_macro)
{
    CachedInclude *e;
    int h;

    if (search_cached_include(s1, type, filename))
        return;
#ifdef INC_DEBUG
    printf("adding cached '%s' %s\n", filename, get_tok_str(ifndef_macro, NULL));
#endif
    e = tcc_malloc(sizeof(CachedInclude) + strlen(filename));
    if (!e)
        return;
    e->type = type;
    strcpy(e->filename, filename);
    e->ifndef_macro = ifndef_macro;
    dynarray_add((void ***)&s1->cached_includes, &s1->nb_cached_includes, e);
    /* add in hash table */
    h = hash_cached_include(type, filename);
    e->hash_next = s1->cached_includes_hash[h];
    s1->cached_includes_hash[h] = s1->nb_cached_includes;
}

static void pragma_parse(TCCState *s1)
{
    int val;

    next();
    if (tok == TOK_pack) {
        /*
          This may be:
          #pragma pack(1) // set
          #pragma pack() // reset to default
          #pragma pack(push,1) // push & set
          #pragma pack(pop) // restore previous
        */
        next();
        skip('(');
        if (tok == TOK_ASM_pop) {
            next();
            if (s1->pack_stack_ptr <= s1->pack_stack) {
            stk_error:
                error("out of pack stack");
            }
            s1->pack_stack_ptr--;
        } else {
            val = 0;
            if (tok != ')') {
                if (tok == TOK_ASM_push) {
                    next();
                    if (s1->pack_stack_ptr >= s1->pack_stack + PACK_STACK_SIZE - 1)
                        goto stk_error;
                    s1->pack_stack_ptr++;
                    skip(',');
                }
                if (tok != TOK_CINT) {
                pack_error:
                    error("invalid pack pragma");
                }
                val = tokc.i;
                if (val < 1 || val > 16 || (val & (val - 1)) != 0)
                    goto pack_error;
                next();
            }
            *s1->pack_stack_ptr = val;
            skip(')');
        }
    }
}

/* is_bof is true if first non space token at beginning of file */
static void preprocess(int is_bof)
{
    TCCState *s1 = tcc_state;
    int i, c, n, saved_parse_flags;
    char buf[1024], *q;
    Sym *s;

    saved_parse_flags = parse_flags;
    parse_flags = PARSE_FLAG_PREPROCESS | PARSE_FLAG_TOK_NUM | 
        PARSE_FLAG_LINEFEED;
    next_nomacro();
 redo:
    switch(tok) {
    case TOK_DEFINE:
        next_nomacro();
        parse_define();
        break;
    case TOK_UNDEF:
        next_nomacro();
        s = define_find(tok);
        /* undefine symbol by putting an invalid name */
        if (s)
            define_undef(s);
        break;
    case TOK_INCLUDE:
    case TOK_INCLUDE_NEXT:
        ch = file->buf_ptr[0];
        /* XXX: incorrect if comments : use next_nomacro with a special mode */
        skip_spaces();
        if (ch == '<') {
            c = '>';
            goto read_name;
        } else if (ch == '\"') {
            c = ch;
        read_name:
            inp();
            q = buf;
            while (ch != c && ch != '\n' && ch != CH_EOF) {
                if ((q - buf) < sizeof(buf) - 1)
                    *q++ = ch;
                if (ch == '\\') {
                    if (handle_stray_noerror() == 0)
                        --q;
                } else
                    inp();
            }
            *q = '\0';
            minp();
#if 0
            /* eat all spaces and comments after include */
            /* XXX: slightly incorrect */
            while (ch1 != '\n' && ch1 != CH_EOF)
                inp();
#endif
        } else {
            /* computed #include : either we have only strings or
               we have anything enclosed in '<>' */
            next();
            buf[0] = '\0';
            if (tok == TOK_STR) {
                while (tok != TOK_LINEFEED) {
                    if (tok != TOK_STR) {
                    include_syntax:
                        error("'#include' expects \"FILENAME\" or <FILENAME>");
                    }
                    pstrcat(buf, sizeof(buf), (char *)tokc.cstr->data);
                    next();
                }
                c = '\"';
            } else {
                int len;
                while (tok != TOK_LINEFEED) {
                    pstrcat(buf, sizeof(buf), get_tok_str(tok, &tokc));
                    next();
                }
                len = strlen(buf);
                /* check syntax and remove '<>' */
                if (len < 2 || buf[0] != '<' || buf[len - 1] != '>')
                    goto include_syntax;
                memmove(buf, buf + 1, len - 2);
                buf[len - 2] = '\0';
                c = '>';
            }
        }

        if (s1->include_stack_ptr >= s1->include_stack + INCLUDE_STACK_SIZE)
            error("#include recursion too deep");

        n = s1->nb_include_paths + s1->nb_sysinclude_paths;
        for (i = -2; i < n; ++i) {
            char buf1[sizeof file->filename];
            BufferedFile *f;
            CachedInclude *e;
            const char *path;
            int size;

            if (i == -2) {
                /* check absolute include path */
                if (!IS_ABSPATH(buf))
                    continue;
                buf1[0] = 0;

            } else if (i == -1) {
                /* search in current dir if "header.h" */
                if (c != '\"')
                    continue;
                size = tcc_basename(file->filename) - file->filename;
                memcpy(buf1, file->filename, size);
                buf1[size] = '\0';

            } else {
                /* search in all the include paths */
                if (i < s1->nb_include_paths)
                    path = s1->include_paths[i];
                else
                    path = s1->sysinclude_paths[i - s1->nb_include_paths];
                pstrcpy(buf1, sizeof(buf1), path);
                pstrcat(buf1, sizeof(buf1), "/");
            }

            pstrcat(buf1, sizeof(buf1), buf);

            e = search_cached_include(s1, c, buf1);
            if (e && define_find(e->ifndef_macro)) {
                /* no need to parse the include because the 'ifndef macro'
                   is defined */
#ifdef INC_DEBUG
                printf("%s: skipping %s\n", file->filename, buf);
#endif
                f = NULL;
            }  else {
                f = tcc_open(s1, buf1);
                if (!f)
                    continue;
            }

            if (tok == TOK_INCLUDE_NEXT) {
                tok = TOK_INCLUDE;
                if (f)
                    tcc_close(f);
                continue;
            }

            if (!f)
                goto include_done;

#ifdef INC_DEBUG
            printf("%s: including %s\n", file->filename, buf1);
#endif

           /* XXX: fix current line init */
           /* push current file in stack */
            *s1->include_stack_ptr++ = file;
            f->inc_type = c;
            pstrcpy(f->inc_filename, sizeof(f->inc_filename), buf1);
            file = f;
            /* add include file debug info */
            if (tcc_state->do_debug) {
                put_stabs(file->filename, N_BINCL, 0, 0, 0);
            }
            tok_flags |= TOK_FLAG_BOF | TOK_FLAG_BOL;
            ch = file->buf_ptr[0];
            goto the_end;
        }
        error("include file '%s' not found", buf);
include_done:
        break;
    case TOK_IFNDEF:
        c = 1;
        goto do_ifdef;
    case TOK_IF:
        c = expr_preprocess();
        goto do_if;
    case TOK_IFDEF:
        c = 0;
    do_ifdef:
        next_nomacro();
        if (tok < TOK_IDENT)
            error("invalid argument for '#if%sdef'", c ? "n" : "");
        if (is_bof) {
            if (c) {
#ifdef INC_DEBUG
                printf("#ifndef %s\n", get_tok_str(tok, NULL));
#endif
                file->ifndef_macro = tok;
            }
        }
        c = (define_find(tok) != 0) ^ c;
    do_if:
        if (s1->ifdef_stack_ptr >= s1->ifdef_stack + IFDEF_STACK_SIZE)
            error("memory full");
        *s1->ifdef_stack_ptr++ = c;
        goto test_skip;
    case TOK_ELSE:
        if (s1->ifdef_stack_ptr == s1->ifdef_stack)
            error("#else without matching #if");
        if (s1->ifdef_stack_ptr[-1] & 2)
            error("#else after #else");
        c = (s1->ifdef_stack_ptr[-1] ^= 3);
        goto test_else;
    case TOK_ELIF:
        if (s1->ifdef_stack_ptr == s1->ifdef_stack)
            error("#elif without matching #if");
        c = s1->ifdef_stack_ptr[-1];
        if (c > 1)
            error("#elif after #else");
        /* last #if/#elif expression was true: we skip */
        if (c == 1)
            goto skip;
        c = expr_preprocess();
        s1->ifdef_stack_ptr[-1] = c;
    test_else:
        if (s1->ifdef_stack_ptr == file->ifdef_stack_ptr + 1)
            file->ifndef_macro = 0;
    test_skip:
        if (!(c & 1)) {
        skip:
            preprocess_skip();
            is_bof = 0;
            goto redo;
        }
        break;
    case TOK_ENDIF:
        if (s1->ifdef_stack_ptr <= file->ifdef_stack_ptr)
            error("#endif without matching #if");
        s1->ifdef_stack_ptr--;
        /* '#ifndef macro' was at the start of file. Now we check if
           an '#endif' is exactly at the end of file */
        if (file->ifndef_macro &&
            s1->ifdef_stack_ptr == file->ifdef_stack_ptr) {
            file->ifndef_macro_saved = file->ifndef_macro;
            /* need to set to zero to avoid false matches if another
               #ifndef at middle of file */
            file->ifndef_macro = 0;
            while (tok != TOK_LINEFEED)
                next_nomacro();
            tok_flags |= TOK_FLAG_ENDIF;
            goto the_end;
        }
        break;
    case TOK_LINE:
        next();
        if (tok != TOK_CINT)
            error("#line");
        file->line_num = tokc.i - 1; /* the line number will be incremented after */
        next();
        if (tok != TOK_LINEFEED) {
            if (tok != TOK_STR)
                error("#line");
            pstrcpy(file->filename, sizeof(file->filename), 
                    (char *)tokc.cstr->data);
        }
        break;
    case TOK_ERROR:
    case TOK_WARNING:
        c = tok;
        ch = file->buf_ptr[0];
        skip_spaces();
        q = buf;
        while (ch != '\n' && ch != CH_EOF) {
            if ((q - buf) < sizeof(buf) - 1)
                *q++ = ch;
            if (ch == '\\') {
                if (handle_stray_noerror() == 0)
                    --q;
            } else
                inp();
        }
        *q = '\0';
        if (c == TOK_ERROR)
            error("#error %s", buf);
        else
            warning("#warning %s", buf);
        break;
    case TOK_PRAGMA:
        pragma_parse(s1);
        break;
    default:
        if (tok == TOK_LINEFEED || tok == '!' || tok == TOK_CINT) {
            /* '!' is ignored to allow C scripts. numbers are ignored
               to emulate cpp behaviour */
        } else {
            if (!(saved_parse_flags & PARSE_FLAG_ASM_COMMENTS))
                warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok, &tokc));
        }
        break;
    }
    /* ignore other preprocess commands or #! for C scripts */
    while (tok != TOK_LINEFEED)
        next_nomacro();
 the_end:
    parse_flags = saved_parse_flags;
}

/* evaluate escape codes in a string. */
static void parse_escape_string(CString *outstr, const uint8_t *buf, int is_long)
{
    int c, n;
    const uint8_t *p;

    p = buf;
    for(;;) {
        c = *p;
        if (c == '\0')
            break;
        if (c == '\\') {
            p++;
            /* escape */
            c = *p;
            switch(c) {
            case '0': case '1': case '2': case '3':
            case '4': case '5': case '6': case '7':
                /* at most three octal digits */
                n = c - '0';
                p++;
                c = *p;
                if (isoct(c)) {
                    n = n * 8 + c - '0';
                    p++;
                    c = *p;
                    if (isoct(c)) {
                        n = n * 8 + c - '0';
                        p++;
                    }
                }
                c = n;
                goto add_char_nonext;
            case 'x':
            case 'u':
            case 'U':
                p++;
                n = 0;
                for(;;) {
                    c = *p;
                    if (c >= 'a' && c <= 'f')
                        c = c - 'a' + 10;
                    else if (c >= 'A' && c <= 'F')
                        c = c - 'A' + 10;
                    else if (isnum(c))
                        c = c - '0';
                    else
                        break;
                    n = n * 16 + c;
                    p++;
                }
                c = n;
                goto add_char_nonext;
            case 'a':
                c = '\a';
                break;
            case 'b':
                c = '\b';
                break;
            case 'f':
                c = '\f';
                break;
            case 'n':
                c = '\n';
                break;
            case 'r':
                c = '\r';
                break;
            case 't':
                c = '\t';
                break;
            case 'v':
                c = '\v';
                break;
            case 'e':
                if (!gnu_ext)
                    goto invalid_escape;
                c = 27;
                break;
            case '\'':
            case '\"':
            case '\\': 
            case '?':
                break;
            default:
            invalid_escape:
                if (c >= '!' && c <= '~')
                    warning("unknown escape sequence: \'\\%c\'", c);
                else
                    warning("unknown escape sequence: \'\\x%x\'", c);
                break;
            }
        }
        p++;
    add_char_nonext:
        if (!is_long)
            cstr_ccat(outstr, c);
        else
            cstr_wccat(outstr, c);
    }
    /* add a trailing '\0' */
    if (!is_long)
        cstr_ccat(outstr, '\0');
    else
        cstr_wccat(outstr, '\0');
}

/* we use 64 bit numbers */
#define BN_SIZE 2

/* bn = (bn << shift) | or_val */
void bn_lshift(unsigned int *bn, int shift, int or_val)
{
    int i;
    unsigned int v;
    for(i=0;i<BN_SIZE;i++) {
        v = bn[i];
        bn[i] = (v << shift) | or_val;
        or_val = v >> (32 - shift);
    }
}

void bn_zero(unsigned int *bn)
{
    int i;
    for(i=0;i<BN_SIZE;i++) {
        bn[i] = 0;
    }
}

/* parse number in null terminated string 'p' and return it in the
   current token */
void parse_number(const char *p)
{
    int b, t, shift, frac_bits, s, exp_val, ch;
    char *q;
    unsigned int bn[BN_SIZE];
    double d;

    /* number */
    q = token_buf;
    ch = *p++;
    t = ch;
    ch = *p++;
    *q++ = t;
    b = 10;
    if (t == '.') {
        goto float_frac_parse;
    } else if (t == '0') {
        if (ch == 'x' || ch == 'X') {
            q--;
            ch = *p++;
            b = 16;
        } else if (tcc_ext && (ch == 'b' || ch == 'B')) {
            q--;
            ch = *p++;
            b = 2;
        }
    }
    /* parse all digits. cannot check octal numbers at this stage
       because of floating point constants */
    while (1) {
        if (ch >= 'a' && ch <= 'f')
            t = ch - 'a' + 10;
        else if (ch >= 'A' && ch <= 'F')
            t = ch - 'A' + 10;
        else if (isnum(ch))
            t = ch - '0';
        else
            break;
        if (t >= b)
            break;
        if (q >= token_buf + STRING_MAX_SIZE) {
        num_too_long:
            error("number too long");
        }
        *q++ = ch;
        ch = *p++;
    }
    if (ch == '.' ||
        ((ch == 'e' || ch == 'E') && b == 10) ||
        ((ch == 'p' || ch == 'P') && (b == 16 || b == 2))) {
        if (b != 10) {
            /* NOTE: strtox should support that for hexa numbers, but
               non ISOC99 libcs do not support it, so we prefer to do
               it by hand */
            /* hexadecimal or binary floats */
            /* XXX: handle overflows */
            *q = '\0';
            if (b == 16)
                shift = 4;
            else 
                shift = 2;
            bn_zero(bn);
            q = token_buf;
            while (1) {
                t = *q++;
                if (t == '\0') {
                    break;
                } else if (t >= 'a') {
                    t = t - 'a' + 10;
                } else if (t >= 'A') {
                    t = t - 'A' + 10;
                } else {
                    t = t - '0';
                }
                bn_lshift(bn, shift, t);
            }
            frac_bits = 0;
            if (ch == '.') {
                ch = *p++;
                while (1) {
                    t = ch;
                    if (t >= 'a' && t <= 'f') {
                        t = t - 'a' + 10;
                    } else if (t >= 'A' && t <= 'F') {
                        t = t - 'A' + 10;
                    } else if (t >= '0' && t <= '9') {
                        t = t - '0';
                    } else {
                        break;
                    }
                    if (t >= b)
                        error("invalid digit");
                    bn_lshift(bn, shift, t);
                    frac_bits += shift;
                    ch = *p++;
                }
            }
            if (ch != 'p' && ch != 'P')
                expect("exponent");
            ch = *p++;
            s = 1;
            exp_val = 0;
            if (ch == '+') {
                ch = *p++;
            } else if (ch == '-') {
                s = -1;
                ch = *p++;
            }
            if (ch < '0' || ch > '9')
                expect("exponent digits");
            while (ch >= '0' && ch <= '9') {
                exp_val = exp_val * 10 + ch - '0';
                ch = *p++;
            }
            exp_val = exp_val * s;
            
            /* now we can generate the number */
            /* XXX: should patch directly float number */
            d = (double)bn[1] * 4294967296.0 + (double)bn[0];
            d = ldexp(d, exp_val - frac_bits);
            t = toup(ch);
            if (t == 'F') {
                ch = *p++;
                tok = TOK_CFLOAT;
                /* float : should handle overflow */
                tokc.f = (float)d;
            } else if (t == 'L') {
                ch = *p++;
#ifdef TCC_TARGET_PE
                tok = TOK_CDOUBLE;
                tokc.d = d;
#else
                tok = TOK_CLDOUBLE;
                /* XXX: not large enough */
                tokc.ld = (long double)d;
#endif
            } else {
                tok = TOK_CDOUBLE;
                tokc.d = d;
            }
        } else {
            /* decimal floats */
            if (ch == '.') {
                if (q >= token_buf + STRING_MAX_SIZE)
                    goto num_too_long;
                *q++ = ch;
                ch = *p++;
            float_frac_parse:
                while (ch >= '0' && ch <= '9') {
                    if (q >= token_buf + STRING_MAX_SIZE)
                        goto num_too_long;
                    *q++ = ch;
                    ch = *p++;
                }
            }
            if (ch == 'e' || ch == 'E') {
                if (q >= token_buf + STRING_MAX_SIZE)
                    goto num_too_long;
                *q++ = ch;
                ch = *p++;
                if (ch == '-' || ch == '+') {
                    if (q >= token_buf + STRING_MAX_SIZE)
                        goto num_too_long;
                    *q++ = ch;
                    ch = *p++;
                }
                if (ch < '0' || ch > '9')
                    expect("exponent digits");
                while (ch >= '0' && ch <= '9') {
                    if (q >= token_buf + STRING_MAX_SIZE)
                        goto num_too_long;
                    *q++ = ch;
                    ch = *p++;
                }
            }
            *q = '\0';
            t = toup(ch);
            errno = 0;
            if (t == 'F') {
                ch = *p++;
                tok = TOK_CFLOAT;
                tokc.f = strtof(token_buf, NULL);
            } else if (t == 'L') {
                ch = *p++;
#ifdef TCC_TARGET_PE
                tok = TOK_CDOUBLE;
                tokc.d = strtod(token_buf, NULL);
#else
                tok = TOK_CLDOUBLE;
                tokc.ld = strtold(token_buf, NULL);
#endif
            } else {
                tok = TOK_CDOUBLE;
                tokc.d = strtod(token_buf, NULL);
            }
        }
    } else {
        unsigned long long n, n1;
        int lcount, ucount;

        /* integer number */
        *q = '\0';
        q = token_buf;
        if (b == 10 && *q == '0') {
            b = 8;
            q++;
        }
        n = 0;
        while(1) {
            t = *q++;
            /* no need for checks except for base 10 / 8 errors */
            if (t == '\0') {
                break;
            } else if (t >= 'a') {
                t = t - 'a' + 10;
            } else if (t >= 'A') {
                t = t - 'A' + 10;
            } else {
                t = t - '0';
                if (t >= b)
                    error("invalid digit");
            }
            n1 = n;
            n = n * b + t;
            /* detect overflow */
            /* XXX: this test is not reliable */
            if (n < n1)
                error("integer constant overflow");
        }
        
        /* XXX: not exactly ANSI compliant */
        if ((n & 0xffffffff00000000LL) != 0) {
            if ((n >> 63) != 0)
                tok = TOK_CULLONG;
            else
                tok = TOK_CLLONG;
        } else if (n > 0x7fffffff) {
            tok = TOK_CUINT;
        } else {
            tok = TOK_CINT;
        }
        lcount = 0;
        ucount = 0;
        for(;;) {
            t = toup(ch);
            if (t == 'L') {
                if (lcount >= 2)
                    error("three 'l's in integer constant");
                lcount++;
                if (lcount == 2) {
                    if (tok == TOK_CINT)
                        tok = TOK_CLLONG;
                    else if (tok == TOK_CUINT)
                        tok = TOK_CULLONG;
                }
                ch = *p++;
            } else if (t == 'U') {
                if (ucount >= 1)
                    error("two 'u's in integer constant");
                ucount++;
                if (tok == TOK_CINT)
                    tok = TOK_CUINT;
                else if (tok == TOK_CLLONG)
                    tok = TOK_CULLONG;
                ch = *p++;
            } else {
                break;
            }
        }
        if (tok == TOK_CINT || tok == TOK_CUINT)
            tokc.ui = n;
        else
            tokc.ull = n;
    }
    if (ch)
        error("invalid number\n");
}


#define PARSE2(c1, tok1, c2, tok2)              \
    case c1:                                    \
        PEEKC(c, p);                            \
        if (c == c2) {                          \
            p++;                                \
            tok = tok2;                         \
        } else {                                \
            tok = tok1;                         \
        }                                       \
        break;

/* return next token without macro substitution */
static inline void next_nomacro1(void)
{
    int t, c, is_long;
    TokenSym *ts;
    uint8_t *p, *p1;
    unsigned int h;

    p = file->buf_ptr;
 redo_no_start:
    c = *p;
    switch(c) {
    case ' ':
    case '\t':
        tok = c;
        p++;
        goto keep_tok_flags;
    case '\f':
    case '\v':
    case '\r':
        p++;
        goto redo_no_start;
    case '\\':
        /* first look if it is in fact an end of buffer */
        if (p >= file->buf_end) {
            file->buf_ptr = p;
            handle_eob();
            p = file->buf_ptr;
            if (p >= file->buf_end)
                goto parse_eof;
            else
                goto redo_no_start;
        } else {
            file->buf_ptr = p;
            ch = *p;
            handle_stray();
            p = file->buf_ptr;
            goto redo_no_start;
        }
    parse_eof:
        {
            TCCState *s1 = tcc_state;
            if ((parse_flags & PARSE_FLAG_LINEFEED)
                && !(tok_flags & TOK_FLAG_EOF)) {
                tok_flags |= TOK_FLAG_EOF;
                tok = TOK_LINEFEED;
                goto keep_tok_flags;
            } else if (s1->include_stack_ptr == s1->include_stack ||
                       !(parse_flags & PARSE_FLAG_PREPROCESS)) {
                /* no include left : end of file. */
                tok = TOK_EOF;
            } else {
                tok_flags &= ~TOK_FLAG_EOF;
                /* pop include file */
                
                /* test if previous '#endif' was after a #ifdef at
                   start of file */
                if (tok_flags & TOK_FLAG_ENDIF) {
#ifdef INC_DEBUG
                    printf("#endif %s\n", get_tok_str(file->ifndef_macro_saved, NULL));
#endif
                    add_cached_include(s1, file->inc_type, file->inc_filename,
                                       file->ifndef_macro_saved);
                }

                /* add end of include file debug info */
                if (tcc_state->do_debug) {
                    put_stabd(N_EINCL, 0, 0);
                }
                /* pop include stack */
                tcc_close(file);
                s1->include_stack_ptr--;
                file = *s1->include_stack_ptr;
                p = file->buf_ptr;
                goto redo_no_start;
            }
        }
        break;

    case '\n':
        file->line_num++;
        tok_flags |= TOK_FLAG_BOL;
        p++;
maybe_newline:
        if (0 == (parse_flags & PARSE_FLAG_LINEFEED))
            goto redo_no_start;
        tok = TOK_LINEFEED;
        goto keep_tok_flags;

    case '#':
        /* XXX: simplify */
        PEEKC(c, p);
        if ((tok_flags & TOK_FLAG_BOL) && 
            (parse_flags & PARSE_FLAG_PREPROCESS)) {
            file->buf_ptr = p;
            preprocess(tok_flags & TOK_FLAG_BOF);
            p = file->buf_ptr;
            goto maybe_newline;
        } else {
            if (c == '#') {
                p++;
                tok = TOK_TWOSHARPS;
            } else {
                if (parse_flags & PARSE_FLAG_ASM_COMMENTS) {
                    p = parse_line_comment(p - 1);
                    goto redo_no_start;
                } else {
                    tok = '#';
                }
            }
        }
        break;

    case 'a': case 'b': case 'c': case 'd':
    case 'e': case 'f': case 'g': case 'h':
    case 'i': case 'j': case 'k': case 'l':
    case 'm': case 'n': case 'o': case 'p':
    case 'q': case 'r': case 's': case 't':
    case 'u': case 'v': case 'w': case 'x':
    case 'y': case 'z': 
    case 'A': case 'B': case 'C': case 'D':
    case 'E': case 'F': case 'G': case 'H':
    case 'I': case 'J': case 'K': 
    case 'M': case 'N': case 'O': case 'P':
    case 'Q': case 'R': case 'S': case 'T':
    case 'U': case 'V': case 'W': case 'X':
    case 'Y': case 'Z': 
    case '_':
    parse_ident_fast:
        p1 = p;
        h = TOK_HASH_INIT;
        h = TOK_HASH_FUNC(h, c);
        p++;
        for(;;) {
            c = *p;
            if (!isidnum_table[c-CH_EOF])
                break;
            h = TOK_HASH_FUNC(h, c);
            p++;
        }
        if (c != '\\') {
            TokenSym **pts;
            int len;

            /* fast case : no stray found, so we have the full token
               and we have already hashed it */
            len = p - p1;
            h &= (TOK_HASH_SIZE - 1);
            pts = &hash_ident[h];
            for(;;) {
                ts = *pts;
                if (!ts)
                    break;
                if (ts->len == len && !memcmp(ts->str, p1, len))
                    goto token_found;
                pts = &(ts->hash_next);
            }
            ts = tok_alloc_new(pts, p1, len);
        token_found: ;
        } else {
            /* slower case */
            cstr_reset(&tokcstr);

            while (p1 < p) {
                cstr_ccat(&tokcstr, *p1);
                p1++;
            }
            p--;
            PEEKC(c, p);
        parse_ident_slow:
            while (isidnum_table[c-CH_EOF]) {
                cstr_ccat(&tokcstr, c);
                PEEKC(c, p);
            }
            ts = tok_alloc(tokcstr.data, tokcstr.size);
        }
        tok = ts->tok;
        break;
    case 'L':
        t = p[1];
        if (t != '\\' && t != '\'' && t != '\"') {
            /* fast case */
            goto parse_ident_fast;
        } else {
            PEEKC(c, p);
            if (c == '\'' || c == '\"') {
                is_long = 1;
                goto str_const;
            } else {
                cstr_reset(&tokcstr);
                cstr_ccat(&tokcstr, 'L');
                goto parse_ident_slow;
            }
        }
        break;
    case '0': case '1': case '2': case '3':
    case '4': case '5': case '6': case '7':
    case '8': case '9':

        cstr_reset(&tokcstr);
        /* after the first digit, accept digits, alpha, '.' or sign if
           prefixed by 'eEpP' */
    parse_num:
        for(;;) {
            t = c;
            cstr_ccat(&tokcstr, c);
            PEEKC(c, p);
            if (!(isnum(c) || isid(c) || c == '.' ||
                  ((c == '+' || c == '-') && 
                   (t == 'e' || t == 'E' || t == 'p' || t == 'P'))))
                break;
        }
        /* We add a trailing '\0' to ease parsing */
        cstr_ccat(&tokcstr, '\0');
        tokc.cstr = &tokcstr;
        tok = TOK_PPNUM;
        break;
    case '.':
        /* special dot handling because it can also start a number */
        PEEKC(c, p);
        if (isnum(c)) {
            cstr_reset(&tokcstr);
            cstr_ccat(&tokcstr, '.');
            goto parse_num;
        } else if (c == '.') {
            PEEKC(c, p);
            if (c != '.')
                expect("'.'");
            PEEKC(c, p);
            tok = TOK_DOTS;
        } else {
            tok = '.';
        }
        break;
    case '\'':
    case '\"':
        is_long = 0;
    str_const:
        {
            CString str;
            int sep;

            sep = c;

            /* parse the string */
            cstr_new(&str);
            p = parse_pp_string(p, sep, &str);
            cstr_ccat(&str, '\0');
            
            /* eval the escape (should be done as TOK_PPNUM) */
            cstr_reset(&tokcstr);
            parse_escape_string(&tokcstr, str.data, is_long);
            cstr_free(&str);

            if (sep == '\'') {
                int char_size;
                /* XXX: make it portable */
                if (!is_long)
                    char_size = 1;
                else
                    char_size = sizeof(nwchar_t);
                if (tokcstr.size <= char_size)
                    error("empty character constant");
                if (tokcstr.size > 2 * char_size)
                    warning("multi-character character constant");
                if (!is_long) {
                    tokc.i = *(int8_t *)tokcstr.data;
                    tok = TOK_CCHAR;
                } else {
                    tokc.i = *(nwchar_t *)tokcstr.data;
                    tok = TOK_LCHAR;
                }
            } else {
                tokc.cstr = &tokcstr;
                if (!is_long)
                    tok = TOK_STR;
                else
                    tok = TOK_LSTR;
            }
        }
        break;

    case '<':
        PEEKC(c, p);
        if (c == '=') {
            p++;
            tok = TOK_LE;
        } else if (c == '<') {
            PEEKC(c, p);
            if (c == '=') {
                p++;
                tok = TOK_A_SHL;
            } else {
                tok = TOK_SHL;
            }
        } else {
            tok = TOK_LT;
        }
        break;
        
    case '>':
        PEEKC(c, p);
        if (c == '=') {
            p++;
            tok = TOK_GE;
        } else if (c == '>') {
            PEEKC(c, p);
            if (c == '=') {
                p++;
                tok = TOK_A_SAR;
            } else {
                tok = TOK_SAR;
            }
        } else {
            tok = TOK_GT;
        }
        break;
        
    case '&':
        PEEKC(c, p);
        if (c == '&') {
            p++;
            tok = TOK_LAND;
        } else if (c == '=') {
            p++;
            tok = TOK_A_AND;
        } else {
            tok = '&';
        }
        break;
        
    case '|':
        PEEKC(c, p);
        if (c == '|') {
            p++;
            tok = TOK_LOR;
        } else if (c == '=') {
            p++;
            tok = TOK_A_OR;
        } else {
            tok = '|';
        }
        break;

    case '+':
        PEEKC(c, p);
        if (c == '+') {
            p++;
            tok = TOK_INC;
        } else if (c == '=') {
            p++;
            tok = TOK_A_ADD;
        } else {
            tok = '+';
        }
        break;
        
    case '-':
        PEEKC(c, p);
        if (c == '-') {
            p++;
            tok = TOK_DEC;
        } else if (c == '=') {
            p++;
            tok = TOK_A_SUB;
        } else if (c == '>') {
            p++;
            tok = TOK_ARROW;
        } else {
            tok = '-';
        }
        break;

    PARSE2('!', '!', '=', TOK_NE)
    PARSE2('=', '=', '=', TOK_EQ)
    PARSE2('*', '*', '=', TOK_A_MUL)
    PARSE2('%', '%', '=', TOK_A_MOD)
    PARSE2('^', '^', '=', TOK_A_XOR)
        
        /* comments or operator */
    case '/':
        PEEKC(c, p);
        if (c == '*') {
            p = parse_comment(p);
            goto redo_no_start;
        } else if (c == '/') {
            p = parse_line_comment(p);
            goto redo_no_start;
        } else if (c == '=') {
            p++;
            tok = TOK_A_DIV;
        } else {
            tok = '/';
        }
        break;
        
        /* simple tokens */
    case '(':
    case ')':
    case '[':
    case ']':
    case '{':
    case '}':
    case ',':
    case ';':
    case ':':
    case '?':
    case '~':
    case '$': /* only used in assembler */
    case '@': /* dito */
        tok = c;
        p++;
        break;
    default:
        error("unrecognized character \\x%02x", c);
        break;
    }
    tok_flags = 0;
keep_tok_flags:
    file->buf_ptr = p;
#if defined(PARSE_DEBUG)
    printf("token = %s\n", get_tok_str(tok, &tokc));
#endif
}

/* return next token without macro substitution. Can read input from
   macro_ptr buffer */
static void next_nomacro_spc(void)
{
    if (macro_ptr) {
    redo:
        tok = *macro_ptr;
        if (tok) {
            TOK_GET(tok, macro_ptr, tokc);
            if (tok == TOK_LINENUM) {
                file->line_num = tokc.i;
                goto redo;
            }
        }
    } else {
        next_nomacro1();
    }
}

static void next_nomacro(void)
{
    do {
        next_nomacro_spc();
    } while (is_space(tok));
}
 
/* substitute args in macro_str and return allocated string */
static int *macro_arg_subst(Sym **nested_list, int *macro_str, Sym *args)
{
    int *st, last_tok, t, spc;
    Sym *s;
    CValue cval;
    TokenString str;
    CString cstr;

    tok_str_new(&str);
    last_tok = 0;
    while(1) {
        TOK_GET(t, macro_str, cval);
        if (!t)
            break;
        if (t == '#') {
            /* stringize */
            TOK_GET(t, macro_str, cval);
            if (!t)
                break;
            s = sym_find2(args, t);
            if (s) {
                cstr_new(&cstr);
                st = s->d;
                spc = 0;
                while (*st) {
                    TOK_GET(t, st, cval);
                    if (!check_space(t, &spc))
                        cstr_cat(&cstr, get_tok_str(t, &cval));
                }
                cstr.size -= spc;
                cstr_ccat(&cstr, '\0');
#ifdef PP_DEBUG
                printf("stringize: %s\n", (char *)cstr.data);
#endif
                /* add string */
                cval.cstr = &cstr;
                tok_str_add2(&str, TOK_STR, &cval);
                cstr_free(&cstr);
            } else {
                tok_str_add2(&str, t, &cval);
            }
        } else if (t >= TOK_IDENT) {
            s = sym_find2(args, t);
            if (s) {
                st = s->d;
                /* if '##' is present before or after, no arg substitution */
                if (*macro_str == TOK_TWOSHARPS || last_tok == TOK_TWOSHARPS) {
                    /* special case for var arg macros : ## eats the
                       ',' if empty VA_ARGS variable. */
                    /* XXX: test of the ',' is not 100%
                       reliable. should fix it to avoid security
                       problems */
                    if (gnu_ext && s->type.t &&
                        last_tok == TOK_TWOSHARPS && 
                        str.len >= 2 && str.str[str.len - 2] == ',') {
                        if (*st == 0) {
                            /* suppress ',' '##' */
                            str.len -= 2;
                        } else {
                            /* suppress '##' and add variable */
                            str.len--;
                            goto add_var;
                        }
                    } else {
                        int t1;
                    add_var:
                        for(;;) {
                            TOK_GET(t1, st, cval);
                            if (!t1)
                                break;
                            tok_str_add2(&str, t1, &cval);
                        }
                    }
                } else {
                    /* NOTE: the stream cannot be read when macro
                       substituing an argument */
                    macro_subst(&str, nested_list, st, NULL);
                }
            } else {
                tok_str_add(&str, t);
            }
        } else {
            tok_str_add2(&str, t, &cval);
        }
        last_tok = t;
    }
    tok_str_add(&str, 0);
    return str.str;
}

static char const ab_month_name[12][4] =
{
    "Jan", "Feb", "Mar", "Apr", "May", "Jun",
    "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
};

/* do macro substitution of current token with macro 's' and add
   result to (tok_str,tok_len). 'nested_list' is the list of all
   macros we got inside to avoid recursing. Return non zero if no
   substitution needs to be done */
static int macro_subst_tok(TokenString *tok_str,
                           Sym **nested_list, Sym *s, struct macro_level **can_read_stream)
{
    Sym *args, *sa, *sa1;
    int mstr_allocated, parlevel, *mstr, t, t1, *p, spc;
    TokenString str;
    char *cstrval;
    CValue cval;
    CString cstr;
    char buf[32];
    
    /* if symbol is a macro, prepare substitution */
    /* special macros */
    if (tok == TOK___LINE__) {
        snprintf(buf, sizeof(buf), "%d", file->line_num);
        cstrval = buf;
        t1 = TOK_PPNUM;
        goto add_cstr1;
    } else if (tok == TOK___FILE__) {
        cstrval = file->filename;
        goto add_cstr;
    } else if (tok == TOK___DATE__ || tok == TOK___TIME__) {
        time_t ti;
        struct tm *tm;

        time(&ti);
        tm = localtime(&ti);
        if (tok == TOK___DATE__) {
            snprintf(buf, sizeof(buf), "%s %2d %d", 
                     ab_month_name[tm->tm_mon], tm->tm_mday, tm->tm_year + 1900);
        } else {
            snprintf(buf, sizeof(buf), "%02d:%02d:%02d", 
                     tm->tm_hour, tm->tm_min, tm->tm_sec);
        }
        cstrval = buf;
    add_cstr:
        t1 = TOK_STR;
    add_cstr1:
        cstr_new(&cstr);
        cstr_cat(&cstr, cstrval);
        cstr_ccat(&cstr, '\0');
        cval.cstr = &cstr;
        tok_str_add2(tok_str, t1, &cval);
        cstr_free(&cstr);
    } else {
        mstr = s->d;
        mstr_allocated = 0;
        if (s->type.t == MACRO_FUNC) {
            /* NOTE: we do not use next_nomacro to avoid eating the
               next token. XXX: find better solution */
        redo:
            if (macro_ptr) {
                p = macro_ptr;
                while (is_space(t = *p) || TOK_LINEFEED == t) 
                    ++p;
                if (t == 0 && can_read_stream) {
                    /* end of macro stream: we must look at the token
                       after in the file */
                    struct macro_level *ml = *can_read_stream;
                    macro_ptr = NULL;
                    if (ml)
                    {
                        macro_ptr = ml->p;
                        ml->p = NULL;
                        *can_read_stream = ml -> prev;
                    }
                    goto redo;
                }
            } else {
                /* XXX: incorrect with comments */
                ch = file->buf_ptr[0];
                while (is_space(ch) || ch == '\n')
                    cinp();
                t = ch;
            }
            if (t != '(') /* no macro subst */
                return -1;
                    
            /* argument macro */
            next_nomacro();
            next_nomacro();
            args = NULL;
            sa = s->next;
            /* NOTE: empty args are allowed, except if no args */
            for(;;) {
                /* handle '()' case */
                if (!args && !sa && tok == ')')
                    break;
                if (!sa)
                    error("macro '%s' used with too many args",
                          get_tok_str(s->v, 0));
                tok_str_new(&str);
                parlevel = spc = 0;
                /* NOTE: non zero sa->t indicates VA_ARGS */
                while ((parlevel > 0 || 
                        (tok != ')' && 
                         (tok != ',' || sa->type.t))) && 
                       tok != -1) {
                    if (tok == '(')
                        parlevel++;
                    else if (tok == ')')
                        parlevel--;
                    if (tok == TOK_LINEFEED)
                        tok = ' ';
                    if (!check_space(tok, &spc))
                        tok_str_add2(&str, tok, &tokc);
                    next_nomacro_spc();
                }
                str.len -= spc;
                tok_str_add(&str, 0);
                sa1 = sym_push2(&args, sa->v & ~SYM_FIELD, sa->type.t, 0);
                sa1->d = str.str;
                sa = sa->next;
                if (tok == ')') {
                    /* special case for gcc var args: add an empty
                       var arg argument if it is omitted */
                    if (sa && sa->type.t && gnu_ext)
                        continue;
                    else
                        break;
                }
                if (tok != ',')
                    expect(",");
                next_nomacro();
            }
            if (sa) {
                error("macro '%s' used with too few args",
                      get_tok_str(s->v, 0));
            }

            /* now subst each arg */
            mstr = macro_arg_subst(nested_list, mstr, args);
            /* free memory */
            sa = args;
            while (sa) {
                sa1 = sa->prev;
                tok_str_free(sa->d);
                sym_free(sa);
                sa = sa1;
            }
            mstr_allocated = 1;
        }
        sym_push2(nested_list, s->v, 0, 0);
        macro_subst(tok_str, nested_list, mstr, can_read_stream);
        /* pop nested defined symbol */
        sa1 = *nested_list;
        *nested_list = sa1->prev;
        sym_free(sa1);
        if (mstr_allocated)
            tok_str_free(mstr);
    }
    return 0;
}

/* handle the '##' operator. Return NULL if no '##' seen. Otherwise
   return the resulting string (which must be freed). */
static inline int *macro_twosharps(const int *macro_str)
{
    const int *ptr;
    int t;
    CValue cval;
    TokenString macro_str1;
    CString cstr;
    char *p;
    int n;

    /* we search the first '##' */
    for(ptr = macro_str;;) {
        TOK_GET(t, ptr, cval);
        if (t == TOK_TWOSHARPS)
            break;
        /* nothing more to do if end of string */
        if (t == 0)
            return NULL;
    }

    /* we saw '##', so we need more processing to handle it */
    tok_str_new(&macro_str1);
    for(ptr = macro_str;;) {
        TOK_GET(tok, ptr, tokc);
        if (tok == 0)
            break;
        if (tok == TOK_TWOSHARPS)
            continue;
        while (*ptr == TOK_TWOSHARPS) {
            t = *++ptr;
            if (t && t != TOK_TWOSHARPS) {
                TOK_GET(t, ptr, cval);

                /* We concatenate the two tokens */
                cstr_new(&cstr);
                cstr_cat(&cstr, get_tok_str(tok, &tokc));
                n = cstr.size;
                cstr_cat(&cstr, get_tok_str(t, &cval));
                cstr_ccat(&cstr, '\0');

                p = file->buf_ptr;
                file->buf_ptr = cstr.data;
                for (;;) {
                    next_nomacro1();
                    if (0 == *file->buf_ptr)
                        break;
                    tok_str_add2(&macro_str1, tok, &tokc);

                    warning("pasting \"%.*s\" and \"%s\" does not give a valid preprocessing token",
                        n, cstr.data, (char*)cstr.data + n);
                }
                file->buf_ptr = p;
                cstr_reset(&cstr);
            }
        }
        tok_str_add2(&macro_str1, tok, &tokc);
    }
    tok_str_add(&macro_str1, 0);
    return macro_str1.str;
}


/* do macro substitution of macro_str and add result to
   (tok_str,tok_len). 'nested_list' is the list of all macros we got
   inside to avoid recursing. */
static void macro_subst(TokenString *tok_str, Sym **nested_list, 
                        const int *macro_str, struct macro_level ** can_read_stream)
{
    Sym *s;
    int *macro_str1;
    const int *ptr;
    int t, ret, spc;
    CValue cval;
    struct macro_level ml;
    
    /* first scan for '##' operator handling */
    ptr = macro_str;
    macro_str1 = macro_twosharps(ptr);
    if (macro_str1) 
        ptr = macro_str1;
    spc = 0;
    while (1) {
        /* NOTE: ptr == NULL can only happen if tokens are read from
           file stream due to a macro function call */
        if (ptr == NULL)
            break;
        TOK_GET(t, ptr, cval);
        if (t == 0)
            break;
        s = define_find(t);
        if (s != NULL) {
            /* if nested substitution, do nothing */
            if (sym_find2(*nested_list, t))
                goto no_subst;
            ml.p = macro_ptr;
            if (can_read_stream)
                ml.prev = *can_read_stream, *can_read_stream = &ml;
            macro_ptr = (int *)ptr;
            tok = t;
            ret = macro_subst_tok(tok_str, nested_list, s, can_read_stream);
            ptr = (int *)macro_ptr;
            macro_ptr = ml.p;
            if (can_read_stream && *can_read_stream == &ml)
                *can_read_stream = ml.prev;
            if (ret != 0)
                goto no_subst;
        } else {
        no_subst:
            if (!check_space(t, &spc)) 
                tok_str_add2(tok_str, t, &cval);
        }
    }
    if (macro_str1)
        tok_str_free(macro_str1);
}

/* return next token with macro substitution */
static void next(void)
{
    Sym *nested_list, *s;
    TokenString str;
    struct macro_level *ml;

 redo:
    if (parse_flags & PARSE_FLAG_SPACES)
        next_nomacro_spc();
    else
        next_nomacro();
    if (!macro_ptr) {
        /* if not reading from macro substituted string, then try
           to substitute macros */
        if (tok >= TOK_IDENT &&
            (parse_flags & PARSE_FLAG_PREPROCESS)) {
            s = define_find(tok);
            if (s) {
                /* we have a macro: we try to substitute */
                tok_str_new(&str);
                nested_list = NULL;
                ml = NULL;
                if (macro_subst_tok(&str, &nested_list, s, &ml) == 0) {
                    /* substitution done, NOTE: maybe empty */
                    tok_str_add(&str, 0);
                    macro_ptr = str.str;
                    macro_ptr_allocated = str.str;
                    goto redo;
                }
            }
        }
    } else {
        if (tok == 0) {
            /* end of macro or end of unget buffer */
            if (unget_buffer_enabled) {
                macro_ptr = unget_saved_macro_ptr;
                unget_buffer_enabled = 0;
            } else {
                /* end of macro string: free it */
                tok_str_free(macro_ptr_allocated);
                macro_ptr_allocated = NULL;
                macro_ptr = NULL;
            }
            goto redo;
        }
    }
    
    /* convert preprocessor tokens into C tokens */
    if (tok == TOK_PPNUM &&
        (parse_flags & PARSE_FLAG_TOK_NUM)) {
        parse_number((char *)tokc.cstr->data);
    }
}

/* push back current token and set current token to 'last_tok'. Only
   identifier case handled for labels. */
static inline void unget_tok(int last_tok)
{
    int i, n;
    int *q;
    unget_saved_macro_ptr = macro_ptr;
    unget_buffer_enabled = 1;
    q = unget_saved_buffer;
    macro_ptr = q;
    *q++ = tok;
    n = tok_ext_size(tok) - 1;
    for(i=0;i<n;i++)
        *q++ = tokc.tab[i];
    *q = 0; /* end of token string */
    tok = last_tok;
}


/* better than nothing, but needs extension to handle '-E' option
   correctly too */
static void preprocess_init(TCCState *s1)
{
    s1->include_stack_ptr = s1->include_stack;
    /* XXX: move that before to avoid having to initialize
       file->ifdef_stack_ptr ? */
    s1->ifdef_stack_ptr = s1->ifdef_stack;
    file->ifdef_stack_ptr = s1->ifdef_stack_ptr;

    /* XXX: not ANSI compliant: bound checking says error */
    vtop = vstack - 1;
    s1->pack_stack[0] = 0;
    s1->pack_stack_ptr = s1->pack_stack;
}

void preprocess_new()
{
    int i, c;
    const char *p, *r;
    TokenSym *ts;

    /* init isid table */
    for(i=CH_EOF;i<256;i++)
        isidnum_table[i-CH_EOF] = isid(i) || isnum(i);

    /* add all tokens */
    table_ident = NULL;
    memset(hash_ident, 0, TOK_HASH_SIZE * sizeof(TokenSym *));
    
    tok_ident = TOK_IDENT;
    p = tcc_keywords;
    while (*p) {
        r = p;
        for(;;) {
            c = *r++;
            if (c == '\0')
                break;
        }
        ts = tok_alloc(p, r - p - 1);
        p = r;
    }
}

/* Preprocess the current file */
static int tcc_preprocess(TCCState *s1)
{
    Sym *define_start;
    BufferedFile *file_ref, **iptr, **iptr_new;
    int token_seen, line_ref, d;
    const char *s;

    preprocess_init(s1);
    define_start = define_stack;
    ch = file->buf_ptr[0];
    tok_flags = TOK_FLAG_BOL | TOK_FLAG_BOF;
    parse_flags = PARSE_FLAG_ASM_COMMENTS | PARSE_FLAG_PREPROCESS |
        PARSE_FLAG_LINEFEED | PARSE_FLAG_SPACES;
    token_seen = 0;
    line_ref = 0;
    file_ref = NULL;

    iptr = s1->include_stack_ptr;
    for (;;) {
        next();
        if (tok == TOK_EOF) {
            break;
        } else if (file != file_ref) {
            goto print_line;
        } else if (tok == TOK_LINEFEED) {
            if (!token_seen)
                continue;
            ++line_ref;
            token_seen = 0;
        } else if (!token_seen) {
            d = file->line_num - line_ref;
            if (file != file_ref || d < 0 || d >= 8) {
print_line:
                iptr_new = s1->include_stack_ptr;
                s = iptr_new > iptr ? " 1"
                  : iptr_new < iptr ? " 2"
                  : iptr_new > s1->include_stack ? " 3"
                  : ""
                  ;
                iptr = iptr_new;
                fprintf(s1->outfile, "# %d \"%s\"%s\n", file->line_num, file->filename, s);
            } else {
                while (d)
                    fputs("\n", s1->outfile), --d;
            }
            line_ref = (file_ref = file)->line_num;
            token_seen = tok != TOK_LINEFEED;
            if (!token_seen)
                continue;
        }
        fputs(get_tok_str(tok, &tokc), s1->outfile);
    }
    free_defines(define_start); 
    return 0;
}