自己动手写C语言编译器（4）

Statement由Expression构成，Expression由Token构成，Token由char构成。

从上到下呈现树形结构。

程序是由statement组成的，其实我们要的就是一种判断Staement开始和结束的方式：

如果能判断开始和结束，以及Expression开始和结束。也许应该这样设计语言 ,

BEGIN:IFSTATEMENT
BEGIN:EXPRESSION
i > 3
END:EXPRESION
BEGIN: GENERAL_STATEMENT
i --;
END: GENERAL_STATEMENT
END:IFSTATMENT

虽然很丑，但是意思很清楚。我们的编译器几乎不用费任何力气就能够识别这样的代码。

这让我想到了XML，如果用XML来描述我们的程序呢？

<IFSTATEMENT>
<EXPRESSION>i biggerthan 3</EXPRESSION>
<GENERAL_STATEMENT>    i -- </GENERAL_STATEMENT>
</IFSTATEMENT>

这是个很馊的注意，但是表达了这样的意思，就是说，我们需要知道，我们遇到的语句到底是什么东西！怎么开始的，又如

何结束的？

我们完全可以定义个XML形式的C语言代码结构

但是现代语言没有采用这样的方式定义的，我认为原因如下：

1. 太丑陋，阅读性不好。
2. 跟人们的生活语言相差太大，不好理解。就相当于每次讲话的时候，说，我要说什么什么了啊，我说完了什么什么了啊 。估计没有人会耐心听下去。
3.增加程序员的工作量。

简洁性是任何语言所追求的！现代的语言规则采用的叙述方式是“上下文无关的叙述”方式。

XML的规则是DTD描述的，那么现代编译语言的规则是用什么描述呢？有没有工具可以识别呢？

针对“针对上下文无关”的描述方式，不管是LL分析大法还是LR分析大法，都有相应的工具可以直接生成程序的源代码:LEX和YACC。

参照：

http://dinosaur.compilertools.net/

http://www.ibm.com/developerworks/cn/linux/sdk/lex/index.html

http://www.ibm.com/developerworks/cn/linux/l-lexyac.html

既然是C语言编译，就要弄到C语言的“上下文无关文法”，对应的标准是：

http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1256.pdf

摘抄如下：

6.1 Notation

1 In the syntax notation used in this clause, syntactic categories (nonterminals) are

indicated by italic type, and literal words and character set members (terminals) by bold

type. A colon (:) following a nonterminal introduces its definition. Alternative

definitions are listed on separate lines, except when prefaced by the words ‘‘one of’’. An

optional symbol is indicated by the subscript ‘‘opt’’, so that

{ expressionopt }

indicates an optional expression enclosed in braces.

2 When syntactic categories are referred to in the main text, they are not italicized and

words are separated by spaces instead of hyphens.

3 A summary of the language syntax is given in annex A.

主要意思是说：在本条款的语法标记中，语法类别（非终结符）用斜线表示，字母单词和字符集（终结符）用黑体表示。

跟在非终结符后面的冒号指示它的定义。“可选择的”定义放在在不同的行，除非以 one of 开始。一个可选的符号用opt表示。

{ expressionopt }表示一个可选的expression

当使用这个语言的时候，不是斜体，也没有连字号"-"。

语言语法的概要在附录 A中。

总结如下：

1.opt表示"此"可有可无。注意这个与alternative不同，这个是可以无的。0或者1个。（用编译原理的术语叫做可空）

2.换行表示可供选择的。就相当于oneof，唯一的不同是：不是以空格分割。

然后看附录A：

A.2.3 Statements

(6.8) statement:

labeled-statement

compound-statement

expression-statement

selection-statement

iteration-statement

jump-statement

语句有这么多种：

1.标签语句

2.组合语句

3.表达式语句

4.选择语句

5.遍历语句。

6.跳转语句。

上代码：

union statement
{
enum
{
E_LS,
E_CS,
E_ES,
E_SS,
E_IS,
E_JS
} _type;
labeled_statement ls;
compound_statement cs;
expression_statement es;
selection_statement ss;
iteration_statement is;
jump_statement js;
};
struct labeled_statement
{
}
struct compound_statement
{
};
struct expression_statement
{
};
struct selection_statement
{
};
struct iteration_statement
{
};
struct jump_statement
{
};

上面这个例子用union表达了这个oneof的思想。

当然也可以用多态，用多态也许更符合现代语言的编程思想，但是工具可能不支持多态，因此就算了。

/* Parse C expressions for CCCP.
Copyright (C) 1987, 1992, 94 - 97, 1998 Free Software Foundation.
This program is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2, or (at your option) any
later version.
This program 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 General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, 59 Temple Place - Suite 330,
Boston, MA 02111-1307, USA.
In other words, you are welcome to use, share and improve this program.
You are forbidden to forbid anyone else to use, share and improve
what you give them.   Help stamp out software-hoarding!
Adapted from expread.y of GDB by Paul Rubin, July 1986.  */
/* Parse a C expression from text in a string  */
%{
#include "config.h"
#ifdef __STDC__
# include <stdarg.h>
#else
# include <varargs.h>
#endif
#define PRINTF_PROTO(ARGS, m, n) PVPROTO (ARGS) ATTRIBUTE_PRINTF(m, n)
#define PRINTF_PROTO_1(ARGS) PRINTF_PROTO(ARGS, 1, 2)
#include "system.h"
#include <setjmp.h>
/* #define YYDEBUG 1 */
#ifdef MULTIBYTE_CHARS
#include <locale.h>
#endif
#include "gansidecl.h"
typedef unsigned char U_CHAR;
/* This is used for communicating lists of keywords with cccp.c.  */
struct arglist {
struct arglist *next;
U_CHAR *name;
int length;
int argno;
};
/* Find the largest host integer type and set its size and type.
Watch out: on some crazy hosts `long' is shorter than `int'.  */
#ifndef HOST_WIDE_INT
# if HAVE_INTTYPES_H
#  include <inttypes.h>
#  define HOST_WIDE_INT intmax_t
#  define unsigned_HOST_WIDE_INT uintmax_t
# else
#  if (HOST_BITS_PER_LONG <= HOST_BITS_PER_INT && HOST_BITS_PER_LONGLONG <= HOST_BITS_PER_INT)
#   define HOST_WIDE_INT int
#  else
#  if (HOST_BITS_PER_LONGLONG <= HOST_BITS_PER_LONG || ! (defined LONG_LONG_MAX || defined LLONG_MAX))
#   define HOST_WIDE_INT long
#  else
#   define HOST_WIDE_INT long long
#  endif
#  endif
# endif
#endif
#ifndef unsigned_HOST_WIDE_INT
#define unsigned_HOST_WIDE_INT unsigned HOST_WIDE_INT
#endif
#ifndef CHAR_BIT
#define CHAR_BIT 8
#endif
#ifndef HOST_BITS_PER_WIDE_INT
#define HOST_BITS_PER_WIDE_INT (CHAR_BIT * sizeof (HOST_WIDE_INT))
#endif
HOST_WIDE_INT parse_c_expression PROTO((char *, int));
static int yylex PROTO((void));
static void yyerror PROTO((char *)) __attribute__ ((noreturn));
static HOST_WIDE_INT expression_value;
#ifdef TEST_EXP_READER
static int expression_signedp;
#endif
static jmp_buf parse_return_error;
/* Nonzero means count most punctuation as part of a name.  */
static int keyword_parsing = 0;
/* Nonzero means do not evaluate this expression.
This is a count, since unevaluated expressions can nest.  */
static int skip_evaluation;
/* Nonzero means warn if undefined identifiers are evaluated.  */
static int warn_undef;
/* some external tables of character types */
extern unsigned char is_idstart[], is_idchar[], is_space[];
/* Flag for -pedantic.  */
extern int pedantic;
/* Flag for -traditional.  */
extern int traditional;
/* Flag for -lang-c89.  */
extern int c89;
#ifndef CHAR_TYPE_SIZE
#define CHAR_TYPE_SIZE BITS_PER_UNIT
#endif
#ifndef INT_TYPE_SIZE
#define INT_TYPE_SIZE BITS_PER_WORD
#endif
#ifndef LONG_TYPE_SIZE
#define LONG_TYPE_SIZE BITS_PER_WORD
#endif
#ifndef WCHAR_TYPE_SIZE
#define WCHAR_TYPE_SIZE INT_TYPE_SIZE
#endif
#ifndef MAX_CHAR_TYPE_SIZE
#define MAX_CHAR_TYPE_SIZE CHAR_TYPE_SIZE
#endif
#ifndef MAX_INT_TYPE_SIZE
#define MAX_INT_TYPE_SIZE INT_TYPE_SIZE
#endif
#ifndef MAX_LONG_TYPE_SIZE
#define MAX_LONG_TYPE_SIZE LONG_TYPE_SIZE
#endif
#ifndef MAX_WCHAR_TYPE_SIZE
#define MAX_WCHAR_TYPE_SIZE WCHAR_TYPE_SIZE
#endif
#define MAX_CHAR_TYPE_MASK (MAX_CHAR_TYPE_SIZE < HOST_BITS_PER_WIDE_INT \
? (~ (~ (HOST_WIDE_INT) 0 << MAX_CHAR_TYPE_SIZE)) \
: ~ (HOST_WIDE_INT) 0)
#define MAX_WCHAR_TYPE_MASK (MAX_WCHAR_TYPE_SIZE < HOST_BITS_PER_WIDE_INT \
? ~ (~ (HOST_WIDE_INT) 0 << MAX_WCHAR_TYPE_SIZE) \
: ~ (HOST_WIDE_INT) 0)
/* Suppose A1 + B1 = SUM1, using 2's complement arithmetic ignoring overflow.
Suppose A, B and SUM have the same respective signs as A1, B1, and SUM1.
Suppose SIGNEDP is negative if the result is signed, zero if unsigned.
Then this yields nonzero if overflow occurred during the addition.
Overflow occurs if A and B have the same sign, but A and SUM differ in sign,
and SIGNEDP is negative.
Use `^' to test whether signs differ, and `< 0' to isolate the sign.  */
#define overflow_sum_sign(a, b, sum, signedp) \
((~((a) ^ (b)) & ((a) ^ (sum)) & (signedp)) < 0)
struct constant;
GENERIC_PTR xmalloc PROTO((size_t));
HOST_WIDE_INT parse_escape PROTO((char **, HOST_WIDE_INT));
int check_assertion PROTO((U_CHAR *, int, int, struct arglist *));
struct hashnode *lookup PROTO((U_CHAR *, int, int));
void error PRINTF_PROTO_1((char *, ...));
void pedwarn PRINTF_PROTO_1((char *, ...));
void warning PRINTF_PROTO_1((char *, ...));
static int parse_number PROTO((int));
static HOST_WIDE_INT left_shift PROTO((struct constant *, unsigned_HOST_WIDE_INT));
static HOST_WIDE_INT right_shift PROTO((struct constant *, unsigned_HOST_WIDE_INT));
static void integer_overflow PROTO((void));
/* `signedp' values */
#define SIGNED (~0)
#define UNSIGNED 0
%}
%union {
struct constant {HOST_WIDE_INT value; int signedp;} integer;
struct name {U_CHAR *address; int length;} name;
struct arglist *keywords;
}
%type <integer> exp exp1 start
%type <keywords> keywords
%token <integer> INT CHAR
%token <name> NAME
%token <integer> ERROR
%right '?' ':'
%left ','
%left OR
%left AND
%left '|'
%left '^'
%left '&'
%left EQUAL NOTEQUAL
%left '<' '>' LEQ GEQ
%left LSH RSH
%left '+' '-'
%left '*' '/' '%'
%right UNARY
/* %expect 40 */
%%
start   :   exp1
{
expression_value = $1.value;
#ifdef TEST_EXP_READER
expression_signedp = $1.signedp;
#endif
}
;
/* Expressions, including the comma operator.  */
exp1    :   exp
|   exp1 ',' exp
{ if (pedantic)
pedwarn ("comma operator in operand of `#if'");
$$ = $3; }
;
/* Expressions, not including the comma operator.  */
exp :   '-' exp    %prec UNARY
{ $$.value = - $2.value;
$$.signedp = $2.signedp;
if (($$.value & $2.value & $$.signedp) < 0)
integer_overflow (); }
|   '!' exp    %prec UNARY
{ $$.value = ! $2.value;
$$.signedp = SIGNED; }
|   '+' exp    %prec UNARY
{ $$ = $2; }
|   '~' exp    %prec UNARY
{ $$.value = ~ $2.value;
$$.signedp = $2.signedp; }
|   '#' NAME
{ $$.value = check_assertion ($2.address, $2.length,
0, NULL_PTR);
$$.signedp = SIGNED; }
|   '#' NAME
{ keyword_parsing = 1; }
'(' keywords ')'
{ $$.value = check_assertion ($2.address, $2.length,
1, $5);
keyword_parsing = 0;
$$.signedp = SIGNED; }
|   '(' exp1 ')'
{ $$ = $2; }
;
/* Binary operators in order of decreasing precedence.  */
exp :   exp '*' exp
{ $$.signedp = $1.signedp & $3.signedp;
if ($$.signedp)
{
$$.value = $1.value * $3.value;
if ($1.value
&& ($$.value / $1.value != $3.value
|| ($$.value & $1.value & $3.value) < 0))
integer_overflow ();
}
else
$$.value = ((unsigned_HOST_WIDE_INT) $1.value
* $3.value); }
|   exp '/' exp
{ if ($3.value == 0)
{
if (!skip_evaluation)
error ("division by zero in #if");
$3.value = 1;
}
$$.signedp = $1.signedp & $3.signedp;
if ($$.signedp)
{
$$.value = $1.value / $3.value;
if (($$.value & $1.value & $3.value) < 0)
integer_overflow ();
}
else
$$.value = ((unsigned_HOST_WIDE_INT) $1.value
/ $3.value); }
|   exp '%' exp
{ if ($3.value == 0)
{
if (!skip_evaluation)
error ("division by zero in #if");
$3.value = 1;
}
$$.signedp = $1.signedp & $3.signedp;
if ($$.signedp)
$$.value = $1.value % $3.value;
else
$$.value = ((unsigned_HOST_WIDE_INT) $1.value
% $3.value); }
|   exp '+' exp
{ $$.value = $1.value + $3.value;
$$.signedp = $1.signedp & $3.signedp;
if (overflow_sum_sign ($1.value, $3.value,
$$.value, $$.signedp))
integer_overflow (); }
|   exp '-' exp
{ $$.value = $1.value - $3.value;
$$.signedp = $1.signedp & $3.signedp;
if (overflow_sum_sign ($$.value, $3.value,
$1.value, $$.signedp))
integer_overflow (); }
|   exp LSH exp
{ $$.signedp = $1.signedp;
if (($3.value & $3.signedp) < 0)
$$.value = right_shift (&$1, -$3.value);
else
$$.value = left_shift (&$1, $3.value); }
|   exp RSH exp
{ $$.signedp = $1.signedp;
if (($3.value & $3.signedp) < 0)
$$.value = left_shift (&$1, -$3.value);
else
$$.value = right_shift (&$1, $3.value); }
|   exp EQUAL exp
{ $$.value = ($1.value == $3.value);
$$.signedp = SIGNED; }
|   exp NOTEQUAL exp
{ $$.value = ($1.value != $3.value);
$$.signedp = SIGNED; }
|   exp LEQ exp
{ $$.signedp = SIGNED;
if ($1.signedp & $3.signedp)
$$.value = $1.value <= $3.value;
else
$$.value = ((unsigned_HOST_WIDE_INT) $1.value
<= $3.value); }
|   exp GEQ exp
{ $$.signedp = SIGNED;
if ($1.signedp & $3.signedp)
$$.value = $1.value >= $3.value;
else
$$.value = ((unsigned_HOST_WIDE_INT) $1.value
>= $3.value); }
|   exp '<' exp
{ $$.signedp = SIGNED;
if ($1.signedp & $3.signedp)
$$.value = $1.value < $3.value;
else
$$.value = ((unsigned_HOST_WIDE_INT) $1.value
< $3.value); }
|   exp '>' exp
{ $$.signedp = SIGNED;
if ($1.signedp & $3.signedp)
$$.value = $1.value > $3.value;
else
$$.value = ((unsigned_HOST_WIDE_INT) $1.value
> $3.value); }
|   exp '&' exp
{ $$.value = $1.value & $3.value;
$$.signedp = $1.signedp & $3.signedp; }
|   exp '^' exp
{ $$.value = $1.value ^ $3.value;
$$.signedp = $1.signedp & $3.signedp; }
|   exp '|' exp
{ $$.value = $1.value | $3.value;
$$.signedp = $1.signedp & $3.signedp; }
|   exp AND
{ skip_evaluation += !$1.value; }
exp
{ skip_evaluation -= !$1.value;
$$.value = ($1.value && $4.value);
$$.signedp = SIGNED; }
|   exp OR
{ skip_evaluation += !!$1.value; }
exp
{ skip_evaluation -= !!$1.value;
$$.value = ($1.value || $4.value);
$$.signedp = SIGNED; }
|   exp '?'
{ skip_evaluation += !$1.value; }
exp ':'
{ skip_evaluation += !!$1.value - !$1.value; }
exp
{ skip_evaluation -= !!$1.value;
$$.value = $1.value ? $4.value : $7.value;
$$.signedp = $4.signedp & $7.signedp; }
|   INT
{ $$ = yylval.integer; }
|   CHAR
{ $$ = yylval.integer; }
|   NAME
{ if (warn_undef && !skip_evaluation)
warning ("`%.*s' is not defined",
$1.length, $1.address);
$$.value = 0;
$$.signedp = SIGNED; }
;
keywords :
{ $$ = 0; }
|   '(' keywords ')' keywords
{ struct arglist *temp;
$$ = (struct arglist *) xmalloc (sizeof (struct arglist));
$$->next = $2;
$$->name = (U_CHAR *) "(";
$$->length = 1;
temp = $$;
while (temp != 0 && temp->next != 0)
temp = temp->next;
temp->next = (struct arglist *) xmalloc (sizeof (struct arglist));
temp->next->next = $4;
temp->next->name = (U_CHAR *) ")";
temp->next->length = 1; }
|   NAME keywords
{ $$ = (struct arglist *) xmalloc (sizeof (struct arglist));
$$->name = $1.address;
$$->length = $1.length;
$$->next = $2; }
;
%%
/* During parsing of a C expression, the pointer to the next character
is in this variable.  */
static char *lexptr;
/* Take care of parsing a number (anything that starts with a digit).
Set yylval and return the token type; update lexptr.
LEN is the number of characters in it.  */
/* maybe needs to actually deal with floating point numbers */
static int
parse_number (olen)
int olen;
{
register char *p = lexptr;
register int c;
register unsigned_HOST_WIDE_INT n = 0, nd, max_over_base;
register int base = 10;
register int len = olen;
register int overflow = 0;
register int digit, largest_digit = 0;
int spec_long = 0;
yylval.integer.signedp = SIGNED;
if (*p == '0') {
base = 8;
if (len >= 3 && (p[1] == 'x' || p[1] == 'X')) {
p += 2;
base = 16;
len -= 2;
}
}
max_over_base = (unsigned_HOST_WIDE_INT) -1 / base;
for (; len > 0; len--) {
c = *p++;
if (c >= '0' && c <= '9')
digit = c - '0';
else if (base == 16 && c >= 'a' && c <= 'f')
digit = c - 'a' + 10;
else if (base == 16 && c >= 'A' && c <= 'F')
digit = c - 'A' + 10;
else {
/* `l' means long, and `u' means unsigned.  */
while (1) {
if (c == 'l' || c == 'L')
{
if (!pedantic < spec_long)
yyerror ("too many `l's in integer constant");
spec_long++;
}
else if (c == 'u' || c == 'U')
{
if (! yylval.integer.signedp)
yyerror ("two `u's in integer constant");
yylval.integer.signedp = UNSIGNED;
}
else {
if (c == '.' || c == 'e' || c == 'E' || c == 'p' || c == 'P')
yyerror ("Floating point numbers not allowed in #if expressions");
else {
char *buf = (char *) alloca (p - lexptr + 40);
sprintf (buf, "missing white space after number `%.*s'",
(int) (p - lexptr - 1), lexptr);
yyerror (buf);
}
}
if (--len == 0)
break;
c = *p++;
}
/* Don't look for any more digits after the suffixes.  */
break;
}
if (largest_digit < digit)
largest_digit = digit;
nd = n * base + digit;
overflow |= (max_over_base < n) | (nd < n);
n = nd;
}
if (base <= largest_digit)
pedwarn ("integer constant contains digits beyond the radix");
if (overflow)
pedwarn ("integer constant out of range");
/* If too big to be signed, consider it unsigned.  */
if (((HOST_WIDE_INT) n & yylval.integer.signedp) < 0)
{
if (base == 10)
warning ("integer constant is so large that it is unsigned");
yylval.integer.signedp = UNSIGNED;
}
lexptr = p;
yylval.integer.value = n;
return INT;
}
struct token {
char *operator;
int token;
};
static struct token tokentab2[] = {
{"&&", AND},
{"||", OR},
{"<<", LSH},
{">>", RSH},
{"==", EQUAL},
{"!=", NOTEQUAL},
{"<=", LEQ},
{">=", GEQ},
{"++", ERROR},
{"--", ERROR},
{NULL, ERROR}
};
/* Read one token, getting characters through lexptr.  */
static int
yylex ()
{
register int c;
register int namelen;
register unsigned char *tokstart;
register struct token *toktab;
int wide_flag;
HOST_WIDE_INT mask;
retry:
tokstart = (unsigned char *) lexptr;
c = *tokstart;
/* See if it is a special token of length 2.  */
if (! keyword_parsing)
for (toktab = tokentab2; toktab->operator != NULL; toktab++)
if (c == *toktab->operator && tokstart[1] == toktab->operator[1]) {
lexptr += 2;
if (toktab->token == ERROR)
{
char *buf = (char *) alloca (40);
sprintf (buf, "`%s' not allowed in operand of `#if'", toktab->operator);
yyerror (buf);
}
return toktab->token;
}
switch (c) {
case '\n':
return 0;
case ' ':
case '\t':
case '\r':
lexptr++;
goto retry;
case 'L':
/* Capital L may start a wide-string or wide-character constant.  */
if (lexptr[1] == '\'')
{
lexptr++;
wide_flag = 1;
mask = MAX_WCHAR_TYPE_MASK;
goto char_constant;
}
if (lexptr[1] == '"')
{
lexptr++;
wide_flag = 1;
mask = MAX_WCHAR_TYPE_MASK;
goto string_constant;
}
break;
case '\'':
wide_flag = 0;
mask = MAX_CHAR_TYPE_MASK;
char_constant:
lexptr++;
if (keyword_parsing) {
char *start_ptr = lexptr - 1;
while (1) {
c = *lexptr++;
if (c == '\\')
c = parse_escape (&lexptr, mask);
else if (c == '\'')
break;
}
yylval.name.address = tokstart;
yylval.name.length = lexptr - start_ptr;
return NAME;
}
/* This code for reading a character constant
handles multicharacter constants and wide characters.
It is mostly copied from c-lex.c.  */
{
register HOST_WIDE_INT result = 0;
register int num_chars = 0;
unsigned width = MAX_CHAR_TYPE_SIZE;
int max_chars;
char *token_buffer;
if (wide_flag)
{
width = MAX_WCHAR_TYPE_SIZE;
#ifdef MULTIBYTE_CHARS
max_chars = MB_CUR_MAX;
#else
max_chars = 1;
#endif
}
else
max_chars = MAX_LONG_TYPE_SIZE / width;
token_buffer = (char *) alloca (max_chars + 1);
while (1)
{
c = *lexptr++;
if (c == '\'' || c == EOF)
break;
if (c == '\\')
{
c = parse_escape (&lexptr, mask);
}
num_chars++;
/* Merge character into result; ignore excess chars.  */
if (num_chars <= max_chars)
{
if (width < HOST_BITS_PER_WIDE_INT)
result = (result << width) | c;
else
result = c;
token_buffer[num_chars - 1] = c;
}
}
token_buffer[num_chars] = 0;
if (c != '\'')
error ("malformatted character constant");
else if (num_chars == 0)
error ("empty character constant");
else if (num_chars > max_chars)
{
num_chars = max_chars;
error ("character constant too long");
}
else if (num_chars != 1 && ! traditional)
warning ("multi-character character constant");
/* If char type is signed, sign-extend the constant.  */
if (! wide_flag)
{
int num_bits = num_chars * width;
if (lookup ((U_CHAR *) "__CHAR_UNSIGNED__",
sizeof ("__CHAR_UNSIGNED__") - 1, -1)
|| ((result >> (num_bits - 1)) & 1) == 0)
yylval.integer.value
= result & (~ (unsigned_HOST_WIDE_INT) 0
>> (HOST_BITS_PER_WIDE_INT - num_bits));
else
yylval.integer.value
= result | ~(~ (unsigned_HOST_WIDE_INT) 0
>> (HOST_BITS_PER_WIDE_INT - num_bits));
}
else
{
#ifdef MULTIBYTE_CHARS
/* Set the initial shift state and convert the next sequence.  */
result = 0;
/* In all locales L'\0' is zero and mbtowc will return zero,
so don't use it.  */
if (num_chars > 1
|| (num_chars == 1 && token_buffer[0] != '\0'))
{
wchar_t wc;
(void) mbtowc (NULL_PTR, NULL_PTR, 0);
if (mbtowc (& wc, token_buffer, num_chars) == num_chars)
result = wc;
else
pedwarn ("Ignoring invalid multibyte character");
}
#endif
yylval.integer.value = result;
}
}
/* This is always a signed type.  */
yylval.integer.signedp = SIGNED;
return CHAR;
/* some of these chars are invalid in constant expressions;
maybe do something about them later */
case '/':
case '+':
case '-':
case '*':
case '%':
case '|':
case '&':
case '^':
case '~':
case '!':
case '@':
case '<':
case '>':
case '[':
case ']':
case '.':
case '?':
case ':':
case '=':
case '{':
case '}':
case ',':
case '#':
if (keyword_parsing)
break;
case '(':
case ')':
lexptr++;
return c;
case '"':
mask = MAX_CHAR_TYPE_MASK;
string_constant:
if (keyword_parsing) {
char *start_ptr = lexptr;
lexptr++;
while (1) {
c = *lexptr++;
if (c == '\\')
c = parse_escape (&lexptr, mask);
else if (c == '"')
break;
}
yylval.name.address = tokstart;
yylval.name.length = lexptr - start_ptr;
return NAME;
}
yyerror ("string constants not allowed in #if expressions");
return ERROR;
}
if (c >= '0' && c <= '9' && !keyword_parsing) {
/* It's a number */
for (namelen = 1; ; namelen++) {
int d = tokstart[namelen];
if (! ((is_idchar[d] || d == '.')
|| ((d == '-' || d == '+')
&& (c == 'e' || c == 'E'
|| ((c == 'p' || c == 'P') && ! c89))
&& ! traditional)))
break;
c = d;
}
return parse_number (namelen);
}
/* It is a name.  See how long it is.  */
if (keyword_parsing) {
for (namelen = 0;; namelen++) {
if (is_space[tokstart[namelen]])
break;
if (tokstart[namelen] == '(' || tokstart[namelen] == ')')
break;
if (tokstart[namelen] == '"' || tokstart[namelen] == '\'')
break;
}
} else {
if (!is_idstart[c]) {
yyerror ("Invalid token in expression");
return ERROR;
}
for (namelen = 0; is_idchar[tokstart[namelen]]; namelen++)
;
}
lexptr += namelen;
yylval.name.address = tokstart;
yylval.name.length = namelen;
return NAME;
}
/* Parse a C escape sequence.  STRING_PTR points to a variable
containing a pointer to the string to parse.  That pointer
is updated past the characters we use.  The value of the
escape sequence is returned.
RESULT_MASK is used to mask out the result;
an error is reported if bits are lost thereby.
A negative value means the sequence \ newline was seen,
which is supposed to be equivalent to nothing at all.
If \ is followed by a null character, we return a negative
value and leave the string pointer pointing at the null character.
If \ is followed by 000, we return 0 and leave the string pointer
after the zeros.  A value of 0 does not mean end of string.  */
HOST_WIDE_INT
parse_escape (string_ptr, result_mask)
char **string_ptr;
HOST_WIDE_INT result_mask;
{
register int c = *(*string_ptr)++;
switch (c)
{
case 'a':
return TARGET_BELL;
case 'b':
return TARGET_BS;
case 'e':
case 'E':
if (pedantic)
pedwarn ("non-ANSI-standard escape sequence, `\\%c'", c);
return 033;
case 'f':
return TARGET_FF;
case 'n':
return TARGET_NEWLINE;
case 'r':
return TARGET_CR;
case 't':
return TARGET_TAB;
case 'v':
return TARGET_VT;
case '\n':
return -2;
case 0:
(*string_ptr)--;
return 0;
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
{
register HOST_WIDE_INT i = c - '0';
register int count = 0;
while (++count < 3)
{
c = *(*string_ptr)++;
if (c >= '0' && c <= '7')
i = (i << 3) + c - '0';
else
{
(*string_ptr)--;
break;
}
}
if (i != (i & result_mask))
{
i &= result_mask;
pedwarn ("octal escape sequence out of range");
}
return i;
}
case 'x':
{
register unsigned_HOST_WIDE_INT i = 0, overflow = 0;
register int digits_found = 0, digit;
for (;;)
{
c = *(*string_ptr)++;
if (c >= '0' && c <= '9')
digit = c - '0';
else if (c >= 'a' && c <= 'f')
digit = c - 'a' + 10;
else if (c >= 'A' && c <= 'F')
digit = c - 'A' + 10;
else
{
(*string_ptr)--;
break;
}
overflow |= i ^ (i << 4 >> 4);
i = (i << 4) + digit;
digits_found = 1;
}
if (!digits_found)
yyerror ("\\x used with no following hex digits");
if (overflow | (i != (i & result_mask)))
{
i &= result_mask;
pedwarn ("hex escape sequence out of range");
}
return i;
}
default:
return c;
}
}
static void
yyerror (s)
char *s;
{
error ("%s", s);
skip_evaluation = 0;
longjmp (parse_return_error, 1);
}
static void
integer_overflow ()
{
if (!skip_evaluation && pedantic)
pedwarn ("integer overflow in preprocessor expression");
}
static HOST_WIDE_INT
left_shift (a, b)
struct constant *a;
unsigned_HOST_WIDE_INT b;
{
/* It's unclear from the C standard whether shifts can overflow.
The following code ignores overflow; perhaps a C standard
interpretation ruling is needed.  */
if (b >= HOST_BITS_PER_WIDE_INT)
return 0;
else
return (unsigned_HOST_WIDE_INT) a->value << b;
}
static HOST_WIDE_INT
right_shift (a, b)
struct constant *a;
unsigned_HOST_WIDE_INT b;
{
if (b >= HOST_BITS_PER_WIDE_INT)
return a->signedp ? a->value >> (HOST_BITS_PER_WIDE_INT - 1) : 0;
else if (a->signedp)
return a->value >> b;
else
return (unsigned_HOST_WIDE_INT) a->value >> b;
}
/* This page contains the entry point to this file.  */
/* Parse STRING as an expression, and complain if this fails
to use up all of the contents of STRING.
STRING may contain '\0' bytes; it is terminated by the first '\n'
outside a string constant, so that we can diagnose '\0' properly.
If WARN_UNDEFINED is nonzero, warn if undefined identifiers are evaluated.
We do not support C comments.  They should be removed before
this function is called.  */
HOST_WIDE_INT
parse_c_expression (string, warn_undefined)
char *string;
int warn_undefined;
{
lexptr = string;
warn_undef = warn_undefined;
/* if there is some sort of scanning error, just return 0 and assume
the parsing routine has printed an error message somewhere.
there is surely a better thing to do than this.     */
if (setjmp (parse_return_error))
return 0;
if (yyparse () != 0)
abort ();
if (*lexptr != '\n')
error ("Junk after end of expression.");
return expression_value;    /* set by yyparse () */
}
#ifdef TEST_EXP_READER
#if YYDEBUG
extern int yydebug;
#endif
int pedantic;
int traditional;
int main PROTO((int, char **));
static void initialize_random_junk PROTO((void));
static void print_unsigned_host_wide_int PROTO((unsigned_HOST_WIDE_INT));
/* Main program for testing purposes.  */
int
main (argc, argv)
int argc;
char **argv;
{
int n, c;
char buf[1024];
unsigned_HOST_WIDE_INT u;
pedantic = 1 < argc;
traditional = 2 < argc;
#if YYDEBUG
yydebug = 3 < argc;
#endif
initialize_random_junk ();
for (;;) {
printf ("enter expression: ");
n = 0;
while ((buf[n] = c = getchar ()) != '\n' && c != EOF)
n++;
if (c == EOF)
break;
parse_c_expression (buf, 1);
printf ("parser returned ");
u = (unsigned_HOST_WIDE_INT) expression_value;
if (expression_value < 0 && expression_signedp) {
u = -u;
printf ("-");
}
if (u == 0)
printf ("0");
else
print_unsigned_host_wide_int (u);
if (! expression_signedp)
printf("u");
printf ("\n");
}
return 0;
}
static void
print_unsigned_host_wide_int (u)
unsigned_HOST_WIDE_INT u;
{
if (u) {
print_unsigned_host_wide_int (u / 10);
putchar ('0' + (int) (u % 10));
}
}
/* table to tell if char can be part of a C identifier. */
unsigned char is_idchar[256];
/* table to tell if char can be first char of a c identifier. */
unsigned char is_idstart[256];
/* table to tell if c is horizontal or vertical space.  */
unsigned char is_space[256];
/*
* initialize random junk in the hash table and maybe other places
*/
static void
initialize_random_junk ()
{
register int i;
/*
* Set up is_idchar and is_idstart tables.  These should be
* faster than saying (is_alpha (c) || c == '_'), etc.
* Must do set up these things before calling any routines tthat
* refer to them.
*/
for (i = 'a'; i <= 'z'; i++) {
++is_idchar[i - 'a' + 'A'];
++is_idchar[i];
++is_idstart[i - 'a' + 'A'];
++is_idstart[i];
}
for (i = '0'; i <= '9'; i++)
++is_idchar[i];
++is_idchar['_'];
++is_idstart['_'];
++is_idchar['$'];
++is_idstart['$'];
++is_space[' '];
++is_space['\t'];
++is_space['\v'];
++is_space['\f'];
++is_space['\n'];
++is_space['\r'];
}
void
error VPROTO ((char * msg, ...))
{
#ifndef __STDC__
char * msg;
#endif
va_list args;
VA_START (args, msg);
#ifndef __STDC__
msg = va_arg (args, char *);
#endif
fprintf (stderr, "error: ");
vfprintf (stderr, msg, args);
fprintf (stderr, "\n");
va_end (args);
}
void
pedwarn VPROTO ((char * msg, ...))
{
#ifndef __STDC__
char * msg;
#endif
va_list args;
VA_START (args, msg);
#ifndef __STDC__
msg = va_arg (args, char *);
#endif
fprintf (stderr, "pedwarn: ");
vfprintf (stderr, msg, args);
fprintf (stderr, "\n");
va_end (args);
}
void
warning VPROTO ((char * msg, ...))
{
#ifndef __STDC__
char * msg;
#endif
va_list args;
VA_START (args, msg);
#ifndef __STDC__
msg = va_arg (args, char *);
#endif
fprintf (stderr, "warning: ");
vfprintf (stderr, msg, args);
fprintf (stderr, "\n");
va_end (args);
}
int
check_assertion (name, sym_length, tokens_specified, tokens)
U_CHAR *name;
int sym_length;
int tokens_specified;
struct arglist *tokens;
{
return 0;
}
struct hashnode *
lookup (name, len, hash)
U_CHAR *name;
int len;
int hash;
{
return (DEFAULT_SIGNED_CHAR) ? 0 : ((struct hashnode *) -1);
}
GENERIC_PTR
xmalloc (size)
size_t size;
{
return (GENERIC_PTR) malloc (size);
}
#endif

自己动手写C语言编译器（4）相关推荐

自己动手写C语言编译器（5）
依托Bison强大的工具, 写编译器,先用编译原理搞个计算器玩玩.如下: Bison工具: http://coolshell.cn/articles/1547.html http://www.gnu. ...
自己动手写C语言编译器（暂停）
1.开源的东西其实并不好用. 2.需要更多的关注Android开发和linux的工具的使用. 3.llvm编译出错的问题一直没有解决. 4.也许需要亲自与llvm的官方进行沟通.这是一个庞大的任务.
自己动手写C语言编译器（3）
词法分析器部分完成. 支持: 1.支持单词分割 2.支持数字类型 3.支持字符串 4.支持换行 6.支持注释不支持: 1.不支持关键字 2.不支持变量. 3.不支持关键字. 4.不支操作符. 偶没有 ...
自己动手写C语言编译器（2）
直接上代码 : 支持:左右结合性 // MyCompiler.cpp : Defines the entry point for the console application. // #includ ...
自己动手写C语言编译器（1）
直接上代码 (表达式树在此初具模型 ) 由于没有做前期的"词法分析", 1.支持单个字符形式的变量 2.支持单个字符形式的操作符 3.支持优先级 . 不支持 1 ...
java c语言 for_Java能写C语言编译器吗
用java是可以写语言编译器的,用任何语言工具写都可以.各语言开发工具间只有开发效率的差异,没有可行不可行的区别. 编译器其实就是一个翻译工具,它可以把代码直接翻译成二进制文件交给CPU执行(二进制指 ...
自己动手制作C 语言编译器（8）：表达式
这是整个编译器的最后一部分,解析表达式.什么是表达式?表达式是将各种语言要素的一个组合,用来求值.例如:函数调用.变量赋值.运算符运算等等. 表达式的解析难点有二:一是运算符的优先级问题,二是如何将表 ...
自己动手制作C 语言编译器（7）：语句
整个编译器还剩下最后两个部分:语句和表达式的解析.它们的内容比较多,主要涉及如何将语句和表达式编译成汇编代码.这章讲解语句的解析,相对于表达式来说它还是较为容易的. 语句 C 语言区分"语句 ...
keil c语言编译运行,Keil的c语言编译器
我曾经通过查看反汇编代码对KEILC编译器进行了测试,大概有这么一下内容,也得出一些结论. (1)全局变量:如果程序中定义了全局变量,而且初始值不是0.此时,在程序调到main()函数执行前,除了要进 ...

自己动手写C语言编译器（4）

自己动手写C语言编译器（4）相关推荐

最新文章

热门文章