int main(void){
const char * file_to_open = "test/c89/empty-main-return-zero.c";
struct c_lexer_state c_lexer_state;
struct memory_pool_collection memory_pool_collection;
struct unsigned_char_list input_characters;
int rtn = 0;
struct parser_state parser_state;
struct unsigned_char_list generated_code;
struct unsigned_char_list lexer_output;
struct unsigned_char_list buffered_symbol_table;
struct preprocessor_state * preprocessor_state;
struct unsigned_char_list preprocssed_characters;
struct struct_c_lexer_token_ptr_list output_tokens;
struct code_gen_state code_gen_state;
struct_c_lexer_token_ptr_list_create(&output_tokens);
unsigned_char_list_create(&lexer_output);
unsigned_char_list_create(&input_characters);
unsigned_char_list_create(&buffered_symbol_table);
unsigned_char_list_create(&preprocssed_characters);
memory_pool_collection_create(&memory_pool_collection);
unsigned_char_list_create(&generated_code);
preprocessor_state = create_preprocessor_state(&memory_pool_collection);
if(!(rtn = get_preprocessed_output_from_file(preprocessor_state, (unsigned char *)file_to_open, &output_tokens))){
unsigned int i;
for(i = 0; i < struct_c_lexer_token_ptr_list_size(&output_tokens); i++){
struct c_lexer_token * current_token = struct_c_lexer_token_ptr_list_get(&output_tokens, i);
unsigned char * j;
for(j = current_token->first_byte; j < (current_token->last_byte + 1); j++){
unsigned_char_list_add_end(&preprocssed_characters, *j);
putchar(*j);
}
}
}else{
printf("Nothing to output. Preprocessing failed for %s\n", file_to_open);
}
printf("\nCompleted preprocessing\n");
printf("\n");
create_c_lexer_state(&c_lexer_state, &lexer_output, &memory_pool_collection, (unsigned char *)file_to_open, unsigned_char_list_data(&preprocssed_characters), unsigned_char_list_size(&preprocssed_characters));
rtn = lex_c(&c_lexer_state);
if(!rtn){
unsigned int i;
for(i = 0; i < struct_c_lexer_token_ptr_list_size(&c_lexer_state.tokens); i++){
struct c_lexer_token * tok = struct_c_lexer_token_ptr_list_get(&c_lexer_state.tokens, i);
printf("%s\n", get_c_token_type_names()[tok->type]);
}
printf("Lex was successful.\n");
}
printf("Begin parsing:\n");
create_parser_state(&parser_state, &memory_pool_collection, &c_lexer_state, &generated_code, unsigned_char_list_data(&preprocssed_characters));
if(parse(&parser_state)){
printf("Parsing failed.\n");
}else{
printf("Full parser true:\n");
print_parser_nodes(parser_state.top_node, 0);
}
create_code_gen_state(&code_gen_state, &parser_state, &generated_code, &buffered_symbol_table);
if(generate_code(&code_gen_state)){
printf("Parsing failed.\n");
}else{
unsigned int i;
unsigned int size;
unsigned char * data;
printf("Code generation was successful.\n");
size = unsigned_char_list_size(&generated_code);
data = unsigned_char_list_data(&generated_code);
printf("About to print\n");
printf("%c\n",'a');
printf("print %d size\n", size);
printf("print %c first \n", data[0]);
for(i = 0; i < size; i++){
printf("%c", data[i]);
}
}
struct_c_lexer_token_ptr_list_destroy(&output_tokens);
unsigned_char_list_destroy(&input_characters);
unsigned_char_list_destroy(&generated_code);
unsigned_char_list_destroy(&buffered_symbol_table);
unsigned_char_list_destroy(&lexer_output);
destroy_code_gen_state(&code_gen_state);
destroy_parser_state(&parser_state);
destroy_preprocessor_state(preprocessor_state);
unsigned_char_list_destroy(&preprocssed_characters);
destroy_c_lexer_state(&c_lexer_state);
memory_pool_collection_destroy(&memory_pool_collection);
return 0;
}
int lex_c(struct c_lexer_state * c_lexer_state){
c_lexer_state->c.position = 0;
c_lexer_state->c.current_line = 0;
struct_c_lexer_token_ptr_list_create(&c_lexer_state->tokens);
while(c_lexer_state->c.position < c_lexer_state->c.buffer_size){
unsigned int rtn = 0;
unsigned char * first_byte = &c_lexer_state->c.buf[c_lexer_state->c.position];
enum c_token_type type;
struct c_lexer_token * new_token;
if((rtn = t_space(&c_lexer_state->c, c_lexer_state->c.position))){
type = SPACE;
}else if((rtn = t_newline(&c_lexer_state->c, c_lexer_state->c.position))){
type = NEWLINE;
c_lexer_state->c.current_line = c_lexer_state->c.current_line + rtn;
}else if((rtn = t_comment(&c_lexer_state->c, c_lexer_state->c.position, &c_lexer_state->c.current_line))){
type = COMMENT;
}else if((rtn = t_keyword((const unsigned char *)"auto", &c_lexer_state->c, c_lexer_state->c.position))){
type = AUTO;
}else if((rtn = t_keyword((const unsigned char *)"break", &c_lexer_state->c, c_lexer_state->c.position))){
type = BREAK;
}else if((rtn = t_keyword((const unsigned char *)"case", &c_lexer_state->c, c_lexer_state->c.position))){
type = CASE;
}else if((rtn = t_keyword((const unsigned char *)"char", &c_lexer_state->c, c_lexer_state->c.position))){
type = CHAR;
}else if((rtn = t_keyword((const unsigned char *)"const", &c_lexer_state->c, c_lexer_state->c.position))){
type = CONST;
}else if((rtn = t_keyword((const unsigned char *)"continue", &c_lexer_state->c, c_lexer_state->c.position))){
type = CONTINUE;
}else if((rtn = t_keyword((const unsigned char *)"default", &c_lexer_state->c, c_lexer_state->c.position))){
type = DEFAULT;
}else if((rtn = t_keyword((const unsigned char *)"do", &c_lexer_state->c, c_lexer_state->c.position))){
type = DO;
}else if((rtn = t_keyword((const unsigned char *)"double", &c_lexer_state->c, c_lexer_state->c.position))){
type = DOUBLE;
}else if((rtn = t_keyword((const unsigned char *)"else", &c_lexer_state->c, c_lexer_state->c.position))){
type = ELSE;
}else if((rtn = t_keyword((const unsigned char *)"enum", &c_lexer_state->c, c_lexer_state->c.position))){
type = ENUM;
}else if((rtn = t_keyword((const unsigned char *)"extern", &c_lexer_state->c, c_lexer_state->c.position))){
type = EXTERN;
}else if((rtn = t_keyword((const unsigned char *)"float", &c_lexer_state->c, c_lexer_state->c.position))){
type = FLOAT;
}else if((rtn = t_keyword((const unsigned char *)"for", &c_lexer_state->c, c_lexer_state->c.position))){
type = FOR;
}else if((rtn = t_keyword((const unsigned char *)"goto", &c_lexer_state->c, c_lexer_state->c.position))){
type = GOTO;
}else if((rtn = t_keyword((const unsigned char *)"if", &c_lexer_state->c, c_lexer_state->c.position))){
type = IF;
}else if((rtn = t_keyword((const unsigned char *)"int", &c_lexer_state->c, c_lexer_state->c.position))){
type = INT;
}else if((rtn = t_keyword((const unsigned char *)"long", &c_lexer_state->c, c_lexer_state->c.position))){
type = LONG;
}else if((rtn = t_keyword((const unsigned char *)"register", &c_lexer_state->c, c_lexer_state->c.position))){
type = REGISTER;
}else if((rtn = t_keyword((const unsigned char *)"return", &c_lexer_state->c, c_lexer_state->c.position))){
type = RETURN;
}else if((rtn = t_keyword((const unsigned char *)"short", &c_lexer_state->c, c_lexer_state->c.position))){
type = SHORT;
}else if((rtn = t_keyword((const unsigned char *)"signed", &c_lexer_state->c, c_lexer_state->c.position))){
type = SIGNED;
}else if((rtn = t_keyword((const unsigned char *)"sizeof", &c_lexer_state->c, c_lexer_state->c.position))){
type = SIZEOF;
}else if((rtn = t_keyword((const unsigned char *)"static", &c_lexer_state->c, c_lexer_state->c.position))){
type = STATIC;
}else if((rtn = t_keyword((const unsigned char *)"struct", &c_lexer_state->c, c_lexer_state->c.position))){
type = STRUCT;
}else if((rtn = t_keyword((const unsigned char *)"switch", &c_lexer_state->c, c_lexer_state->c.position))){
type = SWITCH;
}else if((rtn = t_keyword((const unsigned char *)"typedef", &c_lexer_state->c, c_lexer_state->c.position))){
type = TYPEDEF;
}else if((rtn = t_keyword((const unsigned char *)"union", &c_lexer_state->c, c_lexer_state->c.position))){
type = UNION;
}else if((rtn = t_keyword((const unsigned char *)"unsigned", &c_lexer_state->c, c_lexer_state->c.position))){
type = UNSIGNED;
}else if((rtn = t_keyword((const unsigned char *)"void", &c_lexer_state->c, c_lexer_state->c.position))){
type = VOID;
}else if((rtn = t_keyword((const unsigned char *)"volatile", &c_lexer_state->c, c_lexer_state->c.position))){
type = VOLATILE;
}else if((rtn = t_keyword((const unsigned char *)"while", &c_lexer_state->c, c_lexer_state->c.position))){
type = WHILE;
}else if((rtn = t_identifier(&c_lexer_state->c, c_lexer_state->c.position))){
type = IDENTIFIER;
}else if((rtn = t_constant_hex(&c_lexer_state->c, c_lexer_state->c.position))){
type = CONSTANT_HEX;
}else if((rtn = t_constant_exponent(&c_lexer_state->c, c_lexer_state->c.position))){
type = CONSTANT_EXPONENT;
}else if((rtn = t_constant_float_small(&c_lexer_state->c, c_lexer_state->c.position))){
type = CONSTANT_FLOAT_SMALL;
}else if((rtn = t_constant_float_large(&c_lexer_state->c, c_lexer_state->c.position))){
type = CONSTANT_FLOAT_LARGE;
}else if((rtn = t_constant_decimal(&c_lexer_state->c, c_lexer_state->c.position))){
type = CONSTANT_DECIMAL;
}else if((rtn = t_constant_character(&c_lexer_state->c, c_lexer_state->c.position))){
type = CONSTANT_CHARACTER;
}else if((rtn = t_constant_string(&c_lexer_state->c, c_lexer_state->c.position))){
type = STRING_LITERAL;
}else if((rtn = t_symbol((const unsigned char *)"...", &c_lexer_state->c, c_lexer_state->c.position))){
type = ELLIPSIS;
}else if((rtn = t_symbol((const unsigned char *)">>=", &c_lexer_state->c, c_lexer_state->c.position))){
type = RIGHT_ASSIGN;
}else if((rtn = t_symbol((const unsigned char *)"<<=", &c_lexer_state->c, c_lexer_state->c.position))){
type = LEFT_ASSIGN;
}else if((rtn = t_symbol((const unsigned char *)"+=", &c_lexer_state->c, c_lexer_state->c.position))){
type = ADD_ASSIGN;
}else if((rtn = t_symbol((const unsigned char *)"-=", &c_lexer_state->c, c_lexer_state->c.position))){
type = SUB_ASSIGN;
}else if((rtn = t_symbol((const unsigned char *)"*=", &c_lexer_state->c, c_lexer_state->c.position))){
type = MUL_ASSIGN;
}else if((rtn = t_symbol((const unsigned char *)"/=", &c_lexer_state->c, c_lexer_state->c.position))){
type = DIV_ASSIGN;
}else if((rtn = t_symbol((const unsigned char *)"%=", &c_lexer_state->c, c_lexer_state->c.position))){
type = MOD_ASSIGN;
}else if((rtn = t_symbol((const unsigned char *)"&=", &c_lexer_state->c, c_lexer_state->c.position))){
type = AND_ASSIGN;
}else if((rtn = t_symbol((const unsigned char *)"^=", &c_lexer_state->c, c_lexer_state->c.position))){
type = XOR_ASSIGN;
}else if((rtn = t_symbol((const unsigned char *)"|=", &c_lexer_state->c, c_lexer_state->c.position))){
type = OR_ASSIGN;
}else if((rtn = t_symbol((const unsigned char *)">>", &c_lexer_state->c, c_lexer_state->c.position))){
type = RIGHT_OP;
}else if((rtn = t_symbol((const unsigned char *)"<<", &c_lexer_state->c, c_lexer_state->c.position))){
type = LEFT_OP;
}else if((rtn = t_symbol((const unsigned char *)"++", &c_lexer_state->c, c_lexer_state->c.position))){
type = INC_OP;
}else if((rtn = t_symbol((const unsigned char *)"->", &c_lexer_state->c, c_lexer_state->c.position))){
type = PTR_OP;
}else if((rtn = t_symbol((const unsigned char *)"--", &c_lexer_state->c, c_lexer_state->c.position))){
type = DEC_OP;
}else if((rtn = t_symbol((const unsigned char *)"&&", &c_lexer_state->c, c_lexer_state->c.position))){
type = AND_OP;
}else if((rtn = t_symbol((const unsigned char *)"||", &c_lexer_state->c, c_lexer_state->c.position))){
type = OR_OP;
}else if((rtn = t_symbol((const unsigned char *)"<=", &c_lexer_state->c, c_lexer_state->c.position))){
type = LE_OP;
}else if((rtn = t_symbol((const unsigned char *)">=", &c_lexer_state->c, c_lexer_state->c.position))){
type = GE_OP;
}else if((rtn = t_symbol((const unsigned char *)"==", &c_lexer_state->c, c_lexer_state->c.position))){
type = EQ_OP;
}else if((rtn = t_symbol((const unsigned char *)"!=", &c_lexer_state->c, c_lexer_state->c.position))){
type = NE_OP;
}else if((rtn = t_symbol((const unsigned char *)";", &c_lexer_state->c, c_lexer_state->c.position))){
type = SEMICOLON_CHAR;
}else if(((rtn = t_symbol((const unsigned char *)"{", &c_lexer_state->c, c_lexer_state->c.position)) || (rtn = t_symbol((const unsigned char *)"<%", &c_lexer_state->c, c_lexer_state->c.position)))){
type = OPEN_BRACE_CHAR;
}else if(((rtn = t_symbol((const unsigned char *)"}", &c_lexer_state->c, c_lexer_state->c.position)) || (rtn = t_symbol((const unsigned char *)"%>", &c_lexer_state->c, c_lexer_state->c.position)))){
type = CLOSE_BRACE_CHAR;
}else if(((rtn = t_symbol((const unsigned char *)"[", &c_lexer_state->c, c_lexer_state->c.position)) || (rtn = t_symbol((const unsigned char *)"<:", &c_lexer_state->c, c_lexer_state->c.position)))){
type = OPEN_SQUARE_BRACKET_CHAR;
}else if(((rtn = t_symbol((const unsigned char *)"]", &c_lexer_state->c, c_lexer_state->c.position)) || (rtn = t_symbol((const unsigned char *)":>", &c_lexer_state->c, c_lexer_state->c.position)))){
type = CLOSE_SQUARE_BRACKET_CHAR;
}else if((rtn = t_symbol((const unsigned char *)",", &c_lexer_state->c, c_lexer_state->c.position))){
type = COMMA_CHAR;
}else if((rtn = t_symbol((const unsigned char *)":", &c_lexer_state->c, c_lexer_state->c.position))){
type = COLON_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"=", &c_lexer_state->c, c_lexer_state->c.position))){
type = EQUALS_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"(", &c_lexer_state->c, c_lexer_state->c.position))){
type = OPEN_PAREN_CHAR;
}else if((rtn = t_symbol((const unsigned char *)")", &c_lexer_state->c, c_lexer_state->c.position))){
type = CLOSE_PAREN_CHAR;
}else if((rtn = t_symbol((const unsigned char *)".", &c_lexer_state->c, c_lexer_state->c.position))){
type = DOT_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"&", &c_lexer_state->c, c_lexer_state->c.position))){
type = AMPERSAND_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"!", &c_lexer_state->c, c_lexer_state->c.position))){
type = EXCLAMATION_MARK_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"~", &c_lexer_state->c, c_lexer_state->c.position))){
type = TILDE_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"-", &c_lexer_state->c, c_lexer_state->c.position))){
type = MINUS_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"+", &c_lexer_state->c, c_lexer_state->c.position))){
type = PLUS_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"*", &c_lexer_state->c, c_lexer_state->c.position))){
type = MULTIPLY_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"/", &c_lexer_state->c, c_lexer_state->c.position))){
type = DIVIDE_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"%", &c_lexer_state->c, c_lexer_state->c.position))){
type = PERCENT_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"<", &c_lexer_state->c, c_lexer_state->c.position))){
type = OPEN_ANGLE_BRACKET_CHAR;
}else if((rtn = t_symbol((const unsigned char *)">", &c_lexer_state->c, c_lexer_state->c.position))){
type = CLOSE_ANGLE_BRACKET_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"^", &c_lexer_state->c, c_lexer_state->c.position))){
type = CARET_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"|", &c_lexer_state->c, c_lexer_state->c.position))){
type = PIPE_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"?", &c_lexer_state->c, c_lexer_state->c.position))){
type = QUESTION_MARK_CHAR;
}else if((rtn = t_symbol((const unsigned char *)"#", &c_lexer_state->c, c_lexer_state->c.position))){
type = NUMBER_SIGN_CHAR;
}else{
unsigned int i = 0;
printf("Lexer stopping on character 0x%02x. This was character %u of %u\nFollowing chars are:", c_lexer_state->c.buf[c_lexer_state->c.position], (c_lexer_state->c.position + 1), c_lexer_state->c.buffer_size);
for(i = c_lexer_state->c.position; i < (c_lexer_state->c.position + 100) && i < c_lexer_state->c.buffer_size; i++){
if(c_lexer_state->c.buf[i]){
printf("%c", c_lexer_state->c.buf[i]);
}
}
buffered_printf(c_lexer_state->c.buffered_output, "\n");
assert(0 && "Invalid character in lexer.");
return 1;
}
new_token = struct_c_lexer_token_memory_pool_malloc(c_lexer_state->c.memory_pool_collection->struct_c_lexer_token_pool);
new_token->type = type;
new_token->first_byte = first_byte;
new_token->last_byte = (unsigned char *)((first_byte + rtn) - 1);
{
/*
unsigned char *g = new_token->first_byte;
printf("Got token: %s ", get_c_token_type_names()[new_token->type]);
do{
if(*g == '\n'){
printf("\\n");
}else if(*g == '\r'){
printf("\\r");
}else{
printf("%c", *g);
}
}while(g++ != new_token->last_byte);
printf("\n");
*/
}
struct_c_lexer_token_ptr_list_add_end(&c_lexer_state->tokens, new_token);
show_lexer_token(c_lexer_state->c.buffered_output, get_c_token_type_names()[new_token->type], new_token->first_byte, new_token->last_byte, SHOW_LEXER_TOKENS);
c_lexer_state->c.position += rtn;
}
return 0;
}
