mirror of
https://github.com/DarkPlacesEngine/gmqcc.git
synced 2024-11-27 14:12:36 +00:00
Copying my old lexer
This commit is contained in:
parent
3943e27ebb
commit
9b6598f049
6 changed files with 855 additions and 100 deletions
3
Makefile
3
Makefile
|
@ -17,14 +17,13 @@ ifeq ($(CC), clang)
|
|||
|
||||
endif
|
||||
OBJ = \
|
||||
error.o \
|
||||
util.o \
|
||||
code.o \
|
||||
ast.o \
|
||||
ir.o
|
||||
OBJ_A = test/ast-test.o
|
||||
OBJ_I = test/ir-test.o
|
||||
OBJ_C = main.o
|
||||
OBJ_C = main.o lexer.o parser.o
|
||||
|
||||
#default is compiler only
|
||||
default: gmqcc
|
||||
|
|
91
gmqcc.h
91
gmqcc.h
|
@ -179,95 +179,6 @@ typedef char int64_size_is_correct [sizeof(int64_t) == 8?1:-1];
|
|||
typedef char uintptr_size_is_correct[sizeof(intptr_t) == sizeof(int*)?1:-1];
|
||||
typedef char intptr_size_is_correct [sizeof(uintptr_t)== sizeof(int*)?1:-1];
|
||||
|
||||
/*===================================================================*/
|
||||
/*============================ lex.c ================================*/
|
||||
/*===================================================================*/
|
||||
typedef struct lex_file_t {
|
||||
FILE *file; /* file handler */
|
||||
char *name; /* name of file */
|
||||
char peek [5];
|
||||
char lastok[8192];
|
||||
|
||||
int last; /* last token */
|
||||
int current; /* current token */
|
||||
int length; /* bytes left to parse */
|
||||
int size; /* never changes (size of file) */
|
||||
int line; /* what line are we on? */
|
||||
} lex_file;
|
||||
|
||||
/*
|
||||
* It's important that this table never exceed 32 keywords, the ascii
|
||||
* table starts at 33 (and we don't want conflicts)
|
||||
*/
|
||||
enum {
|
||||
TOKEN_DO ,
|
||||
TOKEN_ELSE ,
|
||||
TOKEN_IF ,
|
||||
TOKEN_WHILE ,
|
||||
TOKEN_BREAK ,
|
||||
TOKEN_CONTINUE ,
|
||||
TOKEN_RETURN ,
|
||||
TOKEN_GOTO ,
|
||||
TOKEN_FOR , /* extension */
|
||||
TOKEN_TYPEDEF , /* extension */
|
||||
|
||||
/* ensure the token types are out of the */
|
||||
/* bounds of anyothers that may conflict. */
|
||||
TOKEN_FLOAT = 110,
|
||||
TOKEN_VECTOR ,
|
||||
TOKEN_STRING ,
|
||||
TOKEN_ENTITY ,
|
||||
TOKEN_VOID
|
||||
};
|
||||
|
||||
/*
|
||||
* Lexer state constants, these are numbers for where exactly in
|
||||
* the lexing the lexer is at. Or where it decided to stop if a lexer
|
||||
* error occurs. These numbers must be > where the ascii-table ends
|
||||
* and > the last type token which is TOKEN_VOID
|
||||
*/
|
||||
enum {
|
||||
LEX_COMMENT = 1128,
|
||||
LEX_CHRLIT ,
|
||||
LEX_STRLIT ,
|
||||
LEX_IDENT
|
||||
};
|
||||
|
||||
int lex_token (lex_file *);
|
||||
void lex_reset (lex_file *);
|
||||
void lex_close (lex_file *);
|
||||
void lex_parse (lex_file *);
|
||||
lex_file *lex_include(lex_file *, const char *);
|
||||
void lex_init (const char *, lex_file **);
|
||||
|
||||
/*===================================================================*/
|
||||
/*========================== error.c ================================*/
|
||||
/*===================================================================*/
|
||||
#define ERROR_LEX (SHRT_MAX+0)
|
||||
#define ERROR_PARSE (SHRT_MAX+1)
|
||||
#define ERROR_INTERNAL (SHRT_MAX+2)
|
||||
#define ERROR_COMPILER (SHRT_MAX+3)
|
||||
#define ERROR_PREPRO (SHRT_MAX+4)
|
||||
int error(lex_file *, int, const char *, ...);
|
||||
|
||||
/*===================================================================*/
|
||||
/*========================== parse.c ================================*/
|
||||
/*===================================================================*/
|
||||
int parse_gen(lex_file *);
|
||||
|
||||
/*===================================================================*/
|
||||
/*========================== typedef.c ==============================*/
|
||||
/*===================================================================*/
|
||||
typedef struct typedef_node_t {
|
||||
char *name;
|
||||
} typedef_node;
|
||||
|
||||
void typedef_init();
|
||||
void typedef_clear();
|
||||
typedef_node *typedef_find(const char *);
|
||||
int typedef_add (lex_file *file, const char *, const char *);
|
||||
|
||||
|
||||
/*===================================================================*/
|
||||
/*=========================== util.c ================================*/
|
||||
/*===================================================================*/
|
||||
|
@ -364,7 +275,7 @@ enum {
|
|||
TYPE_FIELD ,
|
||||
TYPE_FUNCTION ,
|
||||
TYPE_POINTER ,
|
||||
/* TYPE_INTEGER , */
|
||||
TYPE_INTEGER ,
|
||||
TYPE_QUATERNION ,
|
||||
TYPE_MATRIX ,
|
||||
TYPE_VARIANT ,
|
||||
|
|
632
lexer.c
Normal file
632
lexer.c
Normal file
|
@ -0,0 +1,632 @@
|
|||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
#include <string.h>
|
||||
#include <stdarg.h>
|
||||
|
||||
#include "gmqcc.h"
|
||||
#include "lexer.h"
|
||||
|
||||
MEM_VEC_FUNCTIONS(token, char, value)
|
||||
|
||||
void lexerror(lex_file *lex, const char *fmt, ...)
|
||||
{
|
||||
va_list ap;
|
||||
|
||||
if (lex)
|
||||
printf("error %s:%lu: ", lex->name, (unsigned long)lex->sline);
|
||||
else
|
||||
printf("error: ");
|
||||
|
||||
va_start(ap, fmt);
|
||||
vprintf(fmt, ap);
|
||||
va_end(ap);
|
||||
|
||||
printf("\n");
|
||||
}
|
||||
|
||||
token* token_new()
|
||||
{
|
||||
token *tok = (token*)mem_a(sizeof(token));
|
||||
if (!tok)
|
||||
return NULL;
|
||||
memset(tok, 0, sizeof(*tok));
|
||||
return tok;
|
||||
}
|
||||
|
||||
void token_delete(token *self)
|
||||
{
|
||||
if (self->next && self->next->prev == self)
|
||||
self->next->prev = self->prev;
|
||||
if (self->prev && self->prev->next == self)
|
||||
self->prev->next = self->next;
|
||||
MEM_VECTOR_CLEAR(self, value);
|
||||
mem_d(self);
|
||||
}
|
||||
|
||||
token* token_copy(const token *cp)
|
||||
{
|
||||
token* self = token_new();
|
||||
if (!self)
|
||||
return NULL;
|
||||
/* copy the value */
|
||||
self->value_alloc = cp->value_count + 1;
|
||||
self->value_count = cp->value_count;
|
||||
self->value = (char*)mem_a(self->value_alloc);
|
||||
if (!self->value) {
|
||||
mem_d(self);
|
||||
return NULL;
|
||||
}
|
||||
memcpy(self->value, cp->value, cp->value_count);
|
||||
self->value[self->value_alloc-1] = 0;
|
||||
|
||||
/* rest */
|
||||
self->ctx = cp->ctx;
|
||||
self->ttype = cp->ttype;
|
||||
memcpy(&self->constval, &cp->constval, sizeof(self->constval));
|
||||
return self;
|
||||
}
|
||||
|
||||
void token_delete_all(token *t)
|
||||
{
|
||||
token *n;
|
||||
|
||||
do {
|
||||
n = t->next;
|
||||
token_delete(t);
|
||||
t = n;
|
||||
} while(t);
|
||||
}
|
||||
|
||||
token* token_copy_all(const token *cp)
|
||||
{
|
||||
token *cur;
|
||||
token *out;
|
||||
|
||||
out = cur = token_copy(cp);
|
||||
if (!out)
|
||||
return NULL;
|
||||
|
||||
while (cp->next) {
|
||||
cp = cp->next;
|
||||
cur->next = token_copy(cp);
|
||||
if (!cur->next) {
|
||||
token_delete_all(out);
|
||||
return NULL;
|
||||
}
|
||||
cur->next->prev = cur;
|
||||
cur = cur->next;
|
||||
}
|
||||
|
||||
return out;
|
||||
}
|
||||
|
||||
lex_file* lex_open(const char *file)
|
||||
{
|
||||
lex_file *lex;
|
||||
FILE *in = fopen(file, "rb");
|
||||
|
||||
if (!in) {
|
||||
lexerror(NULL, "open failed: '%s'\n", file);
|
||||
return NULL;
|
||||
}
|
||||
|
||||
lex = (lex_file*)mem_a(sizeof(*lex));
|
||||
if (!lex) {
|
||||
fclose(in);
|
||||
lexerror(NULL, "out of memory\n");
|
||||
return NULL;
|
||||
}
|
||||
|
||||
memset(lex, 0, sizeof(*lex));
|
||||
|
||||
lex->file = in;
|
||||
lex->name = util_strdup(file);
|
||||
lex->line = 1; /* we start counting at 1 */
|
||||
|
||||
lex->peekpos = 0;
|
||||
|
||||
return lex;
|
||||
}
|
||||
|
||||
void lex_close(lex_file *lex)
|
||||
{
|
||||
if (lex->file)
|
||||
fclose(lex->file);
|
||||
if (lex->tok)
|
||||
token_delete(lex->tok);
|
||||
mem_d(lex->name);
|
||||
mem_d(lex);
|
||||
}
|
||||
|
||||
/* Get or put-back data
|
||||
* The following to functions do NOT understand what kind of data they
|
||||
* are working on.
|
||||
* The are merely wrapping get/put in order to count line numbers.
|
||||
*/
|
||||
static int lex_getch(lex_file *lex)
|
||||
{
|
||||
int ch;
|
||||
|
||||
if (lex->peekpos) {
|
||||
lex->peekpos--;
|
||||
if (lex->peek[lex->peekpos] == '\n')
|
||||
lex->line++;
|
||||
return lex->peek[lex->peekpos];
|
||||
}
|
||||
|
||||
ch = fgetc(lex->file);
|
||||
if (ch == '\n')
|
||||
lex->line++;
|
||||
return ch;
|
||||
}
|
||||
|
||||
static void lex_ungetch(lex_file *lex, int ch)
|
||||
{
|
||||
lex->peek[lex->peekpos++] = ch;
|
||||
if (ch == '\n')
|
||||
lex->line--;
|
||||
}
|
||||
|
||||
/* classify characters
|
||||
* some additions to the is*() functions of ctype.h
|
||||
*/
|
||||
|
||||
/* Idents are alphanumberic, but they start with alpha or _ */
|
||||
static bool isident_start(int ch)
|
||||
{
|
||||
return isalpha(ch) || ch == '_';
|
||||
}
|
||||
|
||||
static bool isident(int ch)
|
||||
{
|
||||
return isident_start(ch) || isdigit(ch);
|
||||
}
|
||||
|
||||
/* isxdigit_only is used when we already know it's not a digit
|
||||
* and want to see if it's a hex digit anyway.
|
||||
*/
|
||||
static bool isxdigit_only(int ch)
|
||||
{
|
||||
return (ch >= 'a' && ch <= 'f') || (ch >= 'A' && ch <= 'F');
|
||||
}
|
||||
|
||||
/* Skip whitespace and comments and return the first
|
||||
* non-white character.
|
||||
* As this makes use of the above getch() ungetch() functions,
|
||||
* we don't need to care at all about line numbering anymore.
|
||||
*
|
||||
* In theory, this function should only be used at the beginning
|
||||
* of lexing, or when we *know* the next character is part of the token.
|
||||
* Otherwise, if the parser throws an error, the linenumber may not be
|
||||
* the line of the error, but the line of the next token AFTER the error.
|
||||
*
|
||||
* This is currently only problematic when using c-like string-continuation,
|
||||
* since comments and whitespaces are allowed between 2 such strings.
|
||||
* Example:
|
||||
printf( "line one\n"
|
||||
// A comment
|
||||
"A continuation of the previous string"
|
||||
// This line is skipped
|
||||
, foo);
|
||||
|
||||
* In this case, if the parse decides it didn't actually want a string,
|
||||
* and uses lex->line to print an error, it will show the ', foo);' line's
|
||||
* linenumber.
|
||||
*
|
||||
* On the other hand, the parser is supposed to remember the line of the next
|
||||
* token's beginning. In this case we would want skipwhite() to be called
|
||||
* AFTER reading a token, so that the parser, before reading the NEXT token,
|
||||
* doesn't store teh *comment's* linenumber, but the actual token's linenumber.
|
||||
*
|
||||
* THIS SOLUTION
|
||||
* here is to store the line of the first character after skipping
|
||||
* the initial whitespace in lex->sline, this happens in lex_do.
|
||||
*/
|
||||
static int lex_skipwhite(lex_file *lex)
|
||||
{
|
||||
int ch = 0;
|
||||
|
||||
do
|
||||
{
|
||||
ch = lex_getch(lex);
|
||||
while (ch != EOF && isspace(ch)) ch = lex_getch(lex);
|
||||
|
||||
if (ch == '/') {
|
||||
ch = lex_getch(lex);
|
||||
if (ch == '/')
|
||||
{
|
||||
/* one line comment */
|
||||
ch = lex_getch(lex);
|
||||
|
||||
/* check for special: '/', '/', '*', '/' */
|
||||
if (ch == '*') {
|
||||
ch = lex_getch(lex);
|
||||
if (ch == '/') {
|
||||
ch = ' ';
|
||||
continue;
|
||||
}
|
||||
}
|
||||
|
||||
while (ch != EOF && ch != '\n') {
|
||||
ch = lex_getch(lex);
|
||||
}
|
||||
continue;
|
||||
}
|
||||
if (ch == '*')
|
||||
{
|
||||
/* multiline comment */
|
||||
while (ch != EOF)
|
||||
{
|
||||
ch = lex_getch(lex);
|
||||
if (ch == '*') {
|
||||
ch = lex_getch(lex);
|
||||
if (ch == '/') {
|
||||
ch = lex_getch(lex);
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
if (ch == '/') /* allow *//* direct following comment */
|
||||
{
|
||||
lex_ungetch(lex, ch);
|
||||
ch = ' '; /* cause TRUE in the isspace check */
|
||||
}
|
||||
continue;
|
||||
}
|
||||
/* Otherwise roll back to the slash and break out of the loop */
|
||||
lex_ungetch(lex, ch);
|
||||
ch = '/';
|
||||
break;
|
||||
}
|
||||
} while (ch != EOF && isspace(ch));
|
||||
|
||||
return ch;
|
||||
}
|
||||
|
||||
/* Append a character to the token buffer */
|
||||
static bool GMQCC_WARN lex_tokench(lex_file *lex, int ch)
|
||||
{
|
||||
if (!token_value_add(lex->tok, ch)) {
|
||||
lexerror(lex, "out of memory");
|
||||
return false;
|
||||
}
|
||||
return true;
|
||||
}
|
||||
|
||||
/* Append a trailing null-byte */
|
||||
static bool GMQCC_WARN lex_endtoken(lex_file *lex)
|
||||
{
|
||||
if (!token_value_add(lex->tok, 0)) {
|
||||
lexerror(lex, "out of memory");
|
||||
return false;
|
||||
}
|
||||
lex->tok->value_count--;
|
||||
return true;
|
||||
}
|
||||
|
||||
/* Get a token */
|
||||
static bool GMQCC_WARN lex_finish_ident(lex_file *lex)
|
||||
{
|
||||
int ch;
|
||||
|
||||
ch = lex_getch(lex);
|
||||
while (ch != EOF && isident(ch))
|
||||
{
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
ch = lex_getch(lex);
|
||||
}
|
||||
|
||||
/* last ch was not an ident ch: */
|
||||
lex_ungetch(lex, ch);
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
static int GMQCC_WARN lex_finish_string(lex_file *lex, int quote)
|
||||
{
|
||||
int ch = 0;
|
||||
|
||||
while (ch != EOF)
|
||||
{
|
||||
ch = lex_getch(lex);
|
||||
if (ch == quote)
|
||||
return TOKEN_STRINGCONST;
|
||||
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
|
||||
/* as lexer we only care about \" to not terminate the string prematurely */
|
||||
if (ch == '\\') {
|
||||
ch = lex_getch(lex);
|
||||
if (ch == EOF) {
|
||||
lexerror(lex, "unexpected end of file");
|
||||
lex_ungetch(lex, EOF); /* next token to be TOKEN_EOF */
|
||||
return (lex->tok->ttype = TOKEN_ERROR);
|
||||
}
|
||||
/* so we just add the next character no matter what it actually is */
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
}
|
||||
}
|
||||
lexerror(lex, "unexpected end of file within string constant");
|
||||
lex_ungetch(lex, EOF); /* next token to be TOKEN_EOF */
|
||||
return (lex->tok->ttype = TOKEN_ERROR);
|
||||
}
|
||||
|
||||
static int GMQCC_WARN lex_finish_digit(lex_file *lex, int lastch)
|
||||
{
|
||||
bool ishex = false;
|
||||
|
||||
int ch = lastch;
|
||||
|
||||
/* parse a number... */
|
||||
lex->tok->ttype = TOKEN_INTCONST;
|
||||
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
|
||||
ch = lex_getch(lex);
|
||||
if (ch != '.' && !isdigit(ch))
|
||||
{
|
||||
if (lastch != '0' || ch != 'x')
|
||||
{
|
||||
/* end of the number or EOF */
|
||||
lex_ungetch(lex, ch);
|
||||
if (!lex_endtoken(lex))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
|
||||
lex->tok->constval.i = lastch - '0';
|
||||
return lex->tok->ttype;
|
||||
}
|
||||
|
||||
ishex = true;
|
||||
}
|
||||
|
||||
/* EOF would have been caught above */
|
||||
|
||||
if (ch != '.')
|
||||
{
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
ch = lex_getch(lex);
|
||||
while (isdigit(ch) || (ishex && isxdigit_only(ch)))
|
||||
{
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
ch = lex_getch(lex);
|
||||
}
|
||||
}
|
||||
/* NOT else, '.' can come from above as well */
|
||||
if (ch == '.' && !ishex)
|
||||
{
|
||||
/* Allow floating comma in non-hex mode */
|
||||
lex->tok->ttype = TOKEN_FLOATCONST;
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
|
||||
/* continue digits-only */
|
||||
ch = lex_getch(lex);
|
||||
while (isdigit(ch))
|
||||
{
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
ch = lex_getch(lex);
|
||||
}
|
||||
}
|
||||
/* put back the last character */
|
||||
/* but do not put back the trailing 'f' or a float */
|
||||
if (lex->tok->ttype == TOKEN_FLOATCONST && ch == 'f')
|
||||
ch = lex_getch(lex);
|
||||
|
||||
/* generally we don't want words to follow numbers: */
|
||||
if (isident(ch)) {
|
||||
lexerror(lex, "unexpected trailing characters after number");
|
||||
return (lex->tok->ttype = TOKEN_ERROR);
|
||||
}
|
||||
lex_ungetch(lex, ch);
|
||||
|
||||
if (!lex_endtoken(lex))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
if (lex->tok->ttype == TOKEN_FLOATCONST)
|
||||
lex->tok->constval.f = strtod(lex->tok->value, NULL);
|
||||
else
|
||||
lex->tok->constval.i = strtol(lex->tok->value, NULL, 0);
|
||||
return lex->tok->ttype;
|
||||
}
|
||||
|
||||
int lex_do(lex_file *lex)
|
||||
{
|
||||
int ch, nextch;
|
||||
|
||||
if (lex->tok)
|
||||
token_delete(lex->tok);
|
||||
lex->tok = token_new();
|
||||
if (!lex->tok)
|
||||
return TOKEN_FATAL;
|
||||
|
||||
ch = lex_skipwhite(lex);
|
||||
lex->sline = lex->line;
|
||||
lex->tok->ctx.line = lex->sline;
|
||||
lex->tok->ctx.file = lex->name;
|
||||
|
||||
if (ch == EOF)
|
||||
return (lex->tok->ttype = TOKEN_EOF);
|
||||
|
||||
/* single-character tokens */
|
||||
switch (ch)
|
||||
{
|
||||
case ';':
|
||||
case '(':
|
||||
case ')':
|
||||
case '{':
|
||||
case '}':
|
||||
case '[':
|
||||
case ']':
|
||||
|
||||
case ',':
|
||||
|
||||
return (lex->tok->ttype = ch);
|
||||
default:
|
||||
break;
|
||||
}
|
||||
|
||||
if (lex->flags.noops)
|
||||
{
|
||||
/* Detect characters early which are normally
|
||||
* operators OR PART of an operator.
|
||||
*/
|
||||
switch (ch)
|
||||
{
|
||||
case '+':
|
||||
case '-':
|
||||
case '*':
|
||||
case '/':
|
||||
case '<':
|
||||
case '>':
|
||||
case '=':
|
||||
case '&':
|
||||
case '|':
|
||||
case '^':
|
||||
case '~':
|
||||
return ch;
|
||||
default:
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if (ch == '+' || ch == '-' || /* ++, --, +=, -= and -> as well! */
|
||||
ch == '>' || ch == '<' || /* <<, >>, <=, >= */
|
||||
ch == '=' || /* == */
|
||||
ch == '&' || ch == '|') /* &&, ||, &=, |= */
|
||||
{
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
|
||||
nextch = lex_getch(lex);
|
||||
if (nextch == ch || nextch == '=') {
|
||||
if (!lex_tokench(lex, nextch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
} else if (ch == '-' && nextch == '>') {
|
||||
if (!lex_tokench(lex, nextch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
} else
|
||||
lex_ungetch(lex, nextch);
|
||||
|
||||
if (!lex_endtoken(lex))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
return (lex->tok->ttype = TOKEN_OPERATOR);
|
||||
}
|
||||
|
||||
if (ch == '^' || ch == '~' || ch == '!')
|
||||
{
|
||||
if (!lex_tokench(lex, ch) ||
|
||||
!lex_endtoken(lex))
|
||||
{
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
}
|
||||
return (lex->tok->ttype = TOKEN_OPERATOR);
|
||||
}
|
||||
|
||||
if (ch == '*' || ch == '/') /* *=, /= */
|
||||
{
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
|
||||
nextch = lex_getch(lex);
|
||||
if (nextch == '=') {
|
||||
if (!lex_tokench(lex, nextch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
} else
|
||||
lex_ungetch(lex, nextch);
|
||||
|
||||
if (!lex_endtoken(lex))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
return (lex->tok->ttype = TOKEN_OPERATOR);
|
||||
}
|
||||
|
||||
if (isident_start(ch))
|
||||
{
|
||||
const char *v;
|
||||
if (!lex_tokench(lex, ch))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
if (!lex_finish_ident(lex)) {
|
||||
/* error? */
|
||||
return (lex->tok->ttype = TOKEN_ERROR);
|
||||
}
|
||||
if (!lex_endtoken(lex))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
lex->tok->ttype = TOKEN_IDENT;
|
||||
|
||||
v = lex->tok->value;
|
||||
if (!strcmp(v, "void") ||
|
||||
!strcmp(v, "int") ||
|
||||
!strcmp(v, "float") ||
|
||||
!strcmp(v, "vector") )
|
||||
{
|
||||
lex->tok->ttype = TOKEN_TYPENAME;
|
||||
switch (v[1]) {
|
||||
case 'o': lex->tok->constval.t = TYPE_VOID; break;
|
||||
case 'n': lex->tok->constval.t = TYPE_INTEGER; break;
|
||||
case 'l': lex->tok->constval.t = TYPE_FLOAT; break;
|
||||
case 'e': lex->tok->constval.t = TYPE_VECTOR; break;
|
||||
}
|
||||
}
|
||||
else if (!strcmp(v, "for") ||
|
||||
!strcmp(v, "while") ||
|
||||
!strcmp(v, "do"))
|
||||
lex->tok->ttype = TOKEN_KEYWORD;
|
||||
|
||||
return lex->tok->ttype;
|
||||
}
|
||||
|
||||
if (ch == '"')
|
||||
{
|
||||
lex->tok->ttype = lex_finish_string(lex, '"');
|
||||
while (lex->tok->ttype == TOKEN_STRINGCONST)
|
||||
{
|
||||
/* Allow c style "string" "continuation" */
|
||||
ch = lex_skipwhite(lex);
|
||||
if (ch != '"') {
|
||||
lex_ungetch(lex, ch);
|
||||
break;
|
||||
}
|
||||
|
||||
lex->tok->ttype = lex_finish_string(lex, '"');
|
||||
}
|
||||
if (!lex_endtoken(lex))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
return lex->tok->ttype;
|
||||
}
|
||||
|
||||
if (ch == '\'')
|
||||
{
|
||||
/* we parse character constants like string,
|
||||
* but return TOKEN_CHARCONST, or a vector type if it fits...
|
||||
* Likewise actual unescaping has to be done by the parser.
|
||||
* The difference is we don't allow 'char' 'continuation'.
|
||||
*/
|
||||
lex->tok->ttype = lex_finish_string(lex, '\'');
|
||||
if (!lex_endtoken(lex))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
|
||||
/* It's a vector if we can successfully scan 3 floats */
|
||||
if (sscanf(lex->tok->value, " %f %f %f ", &lex->tok->constval.v.x, &lex->tok->constval.v.y, &lex->tok->constval.v.z) == 3)
|
||||
{
|
||||
lex->tok->ttype = TOKEN_VECTORCONST;
|
||||
}
|
||||
|
||||
return lex->tok->ttype;
|
||||
}
|
||||
|
||||
if (isdigit(ch))
|
||||
{
|
||||
lex->tok->ttype = lex_finish_digit(lex, ch);
|
||||
if (!lex_endtoken(lex))
|
||||
return (lex->tok->ttype = TOKEN_FATAL);
|
||||
return lex->tok->ttype;
|
||||
}
|
||||
|
||||
lexerror(lex, "unknown token");
|
||||
return (lex->tok->ttype = TOKEN_ERROR);
|
||||
}
|
207
lexer.h
Normal file
207
lexer.h
Normal file
|
@ -0,0 +1,207 @@
|
|||
#ifndef GMQCC_LEXER_HDR_
|
||||
#define GMQCC_LEXER_HDR_
|
||||
|
||||
typedef struct token_s token;
|
||||
|
||||
#include "ast.h"
|
||||
|
||||
struct token_s {
|
||||
int ttype;
|
||||
|
||||
MEM_VECTOR_MAKE(char, value);
|
||||
|
||||
union {
|
||||
vector v;
|
||||
int i;
|
||||
double f;
|
||||
int t; /* type */
|
||||
} constval;
|
||||
|
||||
struct token_s *next;
|
||||
struct token_s *prev;
|
||||
|
||||
lex_ctx ctx;
|
||||
};
|
||||
|
||||
token* token_new();
|
||||
void token_delete(token*);
|
||||
token* token_copy(const token *cp);
|
||||
void token_delete_all(token *t);
|
||||
token* token_copy_all(const token *cp);
|
||||
|
||||
/* Lexer
|
||||
*
|
||||
*/
|
||||
enum {
|
||||
/* Other tokens which we can return: */
|
||||
TOKEN_NONE = 0,
|
||||
TOKEN_START = 128,
|
||||
|
||||
TOKEN_IDENT,
|
||||
|
||||
TOKEN_TYPENAME,
|
||||
|
||||
TOKEN_OPERATOR,
|
||||
|
||||
TOKEN_KEYWORD, /* loop */
|
||||
|
||||
TOKEN_STRINGCONST, /* not the typename but an actual "string" */
|
||||
TOKEN_CHARCONST,
|
||||
TOKEN_VECTORCONST,
|
||||
TOKEN_INTCONST,
|
||||
TOKEN_FLOATCONST,
|
||||
|
||||
TOKEN_EOF,
|
||||
|
||||
/* We use '< TOKEN_ERROR', so TOKEN_FATAL must come after it and any
|
||||
* other error related tokens as well
|
||||
*/
|
||||
TOKEN_ERROR,
|
||||
TOKEN_FATAL /* internal error, eg out of memory */
|
||||
};
|
||||
|
||||
static const char *_tokennames[] = {
|
||||
"TOKEN_START",
|
||||
"TOKEN_IDENT",
|
||||
"TOKEN_TYPENAME",
|
||||
"TOKEN_OPERATOR",
|
||||
"TOKEN_KEYWORD",
|
||||
"TOKEN_STRINGCONST",
|
||||
"TOKEN_CHARCONST",
|
||||
"TOKEN_VECTORCONST",
|
||||
"TOKEN_INTCONST",
|
||||
"TOKEN_FLOATCONST",
|
||||
"TOKEN_EOF",
|
||||
"TOKEN_ERROR",
|
||||
"TOKEN_FATAL",
|
||||
};
|
||||
typedef int
|
||||
_all_tokennames_added_[
|
||||
((TOKEN_FATAL - TOKEN_START + 1) ==
|
||||
(sizeof(_tokennames)/sizeof(_tokennames[0])))
|
||||
? 1 : -1];
|
||||
|
||||
typedef struct {
|
||||
FILE *file;
|
||||
char *name;
|
||||
size_t line;
|
||||
size_t sline; /* line at the start of a token */
|
||||
|
||||
char peek[256];
|
||||
size_t peekpos;
|
||||
|
||||
token *tok;
|
||||
|
||||
struct {
|
||||
bool noops;
|
||||
} flags;
|
||||
} lex_file;
|
||||
|
||||
MEM_VECTOR_PROTO(lex_file, char, token);
|
||||
|
||||
lex_file* lex_open (const char *file);
|
||||
void lex_close(lex_file *lex);
|
||||
int lex_do (lex_file *lex);
|
||||
|
||||
/* Parser
|
||||
*
|
||||
*/
|
||||
|
||||
enum {
|
||||
ASSOC_LEFT,
|
||||
ASSOC_RIGHT
|
||||
};
|
||||
|
||||
#define OP_SUFFIX 1
|
||||
#define OP_PREFIX 2
|
||||
|
||||
typedef struct {
|
||||
const char *op;
|
||||
unsigned int assoc;
|
||||
unsigned int prec;
|
||||
unsigned int flags;
|
||||
} oper_info;
|
||||
|
||||
static const oper_info operators[] = {
|
||||
{ "++", ASSOC_LEFT, 16, OP_SUFFIX},
|
||||
{ "--", ASSOC_LEFT, 16, OP_SUFFIX},
|
||||
|
||||
{ ".", ASSOC_LEFT, 15, 0 },
|
||||
|
||||
{ "!", ASSOC_RIGHT, 14, 0 },
|
||||
{ "~", ASSOC_RIGHT, 14, 0 },
|
||||
{ "+", ASSOC_RIGHT, 14, OP_PREFIX },
|
||||
{ "-", ASSOC_RIGHT, 14, OP_PREFIX },
|
||||
{ "++", ASSOC_RIGHT, 14, OP_PREFIX },
|
||||
{ "--", ASSOC_RIGHT, 14, OP_PREFIX },
|
||||
/* { "&", ASSOC_RIGHT, 14, OP_PREFIX }, */
|
||||
|
||||
{ "*", ASSOC_LEFT, 13, 0 },
|
||||
{ "/", ASSOC_LEFT, 13, 0 },
|
||||
{ "%", ASSOC_LEFT, 13, 0 },
|
||||
|
||||
{ "+", ASSOC_LEFT, 12, 0 },
|
||||
{ "-", ASSOC_LEFT, 12, 0 },
|
||||
|
||||
{ "<<", ASSOC_LEFT, 11, 0 },
|
||||
{ ">>", ASSOC_LEFT, 11, 0 },
|
||||
|
||||
{ "<", ASSOC_LEFT, 10, 0 },
|
||||
{ ">", ASSOC_LEFT, 10, 0 },
|
||||
{ "<=", ASSOC_LEFT, 10, 0 },
|
||||
{ ">=", ASSOC_LEFT, 10, 0 },
|
||||
|
||||
{ "==", ASSOC_LEFT, 9, 0 },
|
||||
{ "!=", ASSOC_LEFT, 9, 0 },
|
||||
|
||||
{ "&", ASSOC_LEFT, 8, 0 },
|
||||
|
||||
{ "^", ASSOC_LEFT, 7, 0 },
|
||||
|
||||
{ "|", ASSOC_LEFT, 6, 0 },
|
||||
|
||||
{ "&&", ASSOC_LEFT, 5, 0 },
|
||||
|
||||
{ "||", ASSOC_LEFT, 4, 0 },
|
||||
|
||||
{ "?", ASSOC_RIGHT, 3, 0 },
|
||||
|
||||
{ "=", ASSOC_RIGHT, 2, 0 },
|
||||
{ "+=", ASSOC_RIGHT, 2, 0 },
|
||||
{ "-=", ASSOC_RIGHT, 2, 0 },
|
||||
{ "*=", ASSOC_RIGHT, 2, 0 },
|
||||
{ "/=", ASSOC_RIGHT, 2, 0 },
|
||||
{ "%=", ASSOC_RIGHT, 2, 0 },
|
||||
{ ">>=", ASSOC_RIGHT, 2, 0 },
|
||||
{ "<<=", ASSOC_RIGHT, 2, 0 },
|
||||
{ "&=", ASSOC_RIGHT, 2, 0 },
|
||||
{ "^=", ASSOC_RIGHT, 2, 0 },
|
||||
{ "|=", ASSOC_RIGHT, 2, 0 },
|
||||
};
|
||||
|
||||
typedef struct
|
||||
{
|
||||
lex_file *lex;
|
||||
int error;
|
||||
lex_ctx ctx;
|
||||
|
||||
token *tokens;
|
||||
token *lastok;
|
||||
|
||||
token *tok; /* current token */
|
||||
|
||||
MEM_VECTOR_MAKE(ast_value*, globals);
|
||||
} parse_file;
|
||||
|
||||
MEM_VECTOR_PROTO(parse_file, ast_value*, globals);
|
||||
|
||||
parse_file* parse_open(const char *file);
|
||||
void parse_file_close(parse_file*);
|
||||
|
||||
bool parse(parse_file*);
|
||||
|
||||
bool parse_iskey(parse_file *self, const char *ident);
|
||||
|
||||
void lexerror(lex_file*, const char *fmt, ...);
|
||||
|
||||
#endif
|
15
main.c
15
main.c
|
@ -21,19 +21,18 @@
|
|||
* SOFTWARE.
|
||||
*/
|
||||
#include "gmqcc.h"
|
||||
typedef struct { char *name, type; } argitem;
|
||||
VECTOR_MAKE(argitem, items);
|
||||
|
||||
bool parser_compile(const char *filename);
|
||||
int main(int argc, char **argv) {
|
||||
size_t itr;
|
||||
|
||||
util_debug("COM", "starting ...\n");
|
||||
|
||||
if (argc == 2) {
|
||||
if (!parser_compile(argv[1])) {
|
||||
printf("There were compile errors\n");
|
||||
}
|
||||
}
|
||||
|
||||
util_debug("COM", "cleaning ...\n");
|
||||
/* clean list */
|
||||
for (itr = 0; itr < items_elements; itr++)
|
||||
mem_d(items_data[itr].name);
|
||||
mem_d(items_data);
|
||||
|
||||
util_meminfo();
|
||||
return 0;
|
||||
|
|
7
parser.c
Normal file
7
parser.c
Normal file
|
@ -0,0 +1,7 @@
|
|||
#include "gmqcc.h"
|
||||
#include "lexer.h"
|
||||
|
||||
bool parser_compile(const char *filename)
|
||||
{
|
||||
return false;
|
||||
}
|
Loading…
Reference in a new issue