tcc.c

一个GNU的C语言编译器源代码,源代码过万行.

        sprintf(p, "%Lu", cv->ull);
        break;
    case TOK_CCHAR:
    case TOK_LCHAR:
        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_STR:
    case TOK_LSTR:
        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(int)) - 1;
            for(i=0;i<len;i++)
                add_char(&cstr_buf, ((int *)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_A_SHL:
        return strcpy(p, "<<=");
    case TOK_A_SAR:
        return strcpy(p, ">>=");
    default:
        if (v < TOK_IDENT) {
            /* search in two bytes table */
            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;
}

/* push, without hashing */
static Sym *sym_push2(Sym **ps, int v, int t, int c)
{
    Sym *s;
    s = tcc_malloc(sizeof(Sym));
    s->v = v;
    s->type.t = t;
    s->c = c;
    s->next = NULL;
    /* add in stack */
    s->prev = *ps;
    *ps = s;
    return s;
}

/* find a symbol and return its associated structure. 's' is the top
   of the symbol stack */
static Sym *sym_find2(Sym *s, int v)
{
    while (s) {
        if (s->v == v)
            return s;
        s = s->prev;
    }
    return NULL;
}

/* structure lookup */
static inline Sym *struct_find(int v)
{
    v -= TOK_IDENT;
    if ((unsigned)v >= (unsigned)(tok_ident - TOK_IDENT))
        return NULL;
    return table_ident[v]->sym_struct;
}

/* find an identifier */
static inline Sym *sym_find(int v)
{
    v -= TOK_IDENT;
    if ((unsigned)v >= (unsigned)(tok_ident - TOK_IDENT))
        return NULL;
    return table_ident[v]->sym_identifier;
}

/* push a given symbol on the symbol stack */
static Sym *sym_push(int v, CType *type, int r, int c)
{
    Sym *s, **ps;
    TokenSym *ts;

    if (local_stack)
        ps = &local_stack;
    else
        ps = &global_stack;
    s = sym_push2(ps, v, type->t, c);
    s->type.ref = type->ref;
    s->r = r;
    /* don't record fields or anonymous symbols */
    /* XXX: simplify */
    if (!(v & SYM_FIELD) && (v & ~SYM_STRUCT) < SYM_FIRST_ANOM) {
        /* record symbol in token array */
        ts = table_ident[(v & ~SYM_STRUCT) - TOK_IDENT];
        if (v & SYM_STRUCT)
            ps = &ts->sym_struct;
        else
            ps = &ts->sym_identifier;
        s->prev_tok = *ps;
        *ps = s;
    }
    return s;
}

/* push a global identifier */
static Sym *global_identifier_push(int v, int t, int c)
{
    Sym *s, **ps;
    s = sym_push2(&global_stack, v, t, c);
    /* don't record anonymous symbol */
    if (v < SYM_FIRST_ANOM) {
        ps = &table_ident[v - TOK_IDENT]->sym_identifier;
        /* 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 = NULL;
        *ps = s;
    }
    return s;
}

/* pop symbols until top reaches 'b' */
static void sym_pop(Sym **ptop, Sym *b)
{
    Sym *s, *ss, **ps;
    TokenSym *ts;
    int v;

    s = *ptop;
    while(s != b) {
        ss = s->prev;
        v = s->v;
        /* remove symbol in token array */
        /* XXX: simplify */
        if (!(v & SYM_FIELD) && (v & ~SYM_STRUCT) < SYM_FIRST_ANOM) {
            ts = table_ident[(v & ~SYM_STRUCT) - TOK_IDENT];
            if (v & SYM_STRUCT)
                ps = &ts->sym_struct;
            else
                ps = &ts->sym_identifier;
            *ps = s->prev_tok;
        }
        tcc_free(s);
        s = ss;
    }
    *ptop = b;
}

/* I/O layer */

BufferedFile *tcc_open(TCCState *s1, const char *filename)
{
    int fd;
    BufferedFile *bf;

    fd = open(filename, O_RDONLY);
    if (fd < 0)
        return NULL;
    bf = tcc_malloc(sizeof(BufferedFile));
    if (!bf) {
        close(fd);
        return NULL;
    }
    bf->fd = fd;
    bf->buf_ptr = bf->buffer;
    bf->buf_end = bf->buffer;
    bf->buffer[0] = CH_EOB; /* put eob symbol */
    pstrcpy(bf->filename, sizeof(bf->filename), filename);
    bf->line_num = 1;
    bf->ifndef_macro = 0;
    bf->ifdef_stack_ptr = s1->ifdef_stack_ptr;
    //    printf("opening '%s'\n", filename);
    return bf;
}

void tcc_close(BufferedFile *bf)
{
    total_lines += bf->line_num;
    close(bf->fd);
    tcc_free(bf);
}

/* 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 void handle_stray(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:
            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;
        if (c == '\n' || c == CH_EOF) {
            break;
        } else 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 {
            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

/* space exlcuding newline */
static inline int is_space(int ch)
{
    return ch == ' ' || ch == '\t' || ch == '\v' || ch == '\f' || ch == '\r';
}

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

/* 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);