#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <ctype.h>


/******************************************************************************/
/*                                                                            */
/*   TOKENISER                                                                */
/*                                                                            */
/******************************************************************************/

enum token_type {
	TOK_NONE,
	TOK_PAREN_OPEN, TOK_PAREN_CLOSE, TOK_COMMA, TOK_COLON, 
	TOK_NEWLINE, TOK_EOF,
	TOK_FUN, TOK_END,
	TOK_IDENT, TOK_LITERAL_INT
};

struct token {
	enum token_type type;
	char* str;
};

int ch;
char* buf;
int buf_len;

struct token* tokens;
int token_len;

int isidentch(int c) {
	return isalnum(c) || c == '_';
}

void next_ch() {
	ch = getchar();
}

void accept_ch() {
	buf = realloc(buf, ++buf_len);
	buf[buf_len-1] = (char)ch;	
}

void skip_ws() {
	//printf("attempting to skip whitespace, next (%d)\n", ch);
	while(ch == ' ' || ch == '\t') {
		//printf("skipping ws: (%d)\n", ch);
		next_ch();
	}
}

void finalise_buf() {
	buf = realloc(buf, ++buf_len);
	buf[buf_len-1] = 0;
}

void reset_buf() { /* warning, memory leaking by design */
	buf = 0;
	buf_len = 0;
}

void add_token(enum token_type type, char* str) {
	tokens = realloc(tokens, sizeof(struct token) * (++token_len));
	tokens[token_len - 1].type = type;
	tokens[token_len - 1].str = str;
}

void tokenise() {
	enum token_type mode = TOK_NONE;
	reset_buf();
	tokens = 0;
	token_len = 0;
	next_ch();
	do {
		printf(">> %c <<\n", ch);
		switch(mode) {
			case TOK_NONE:
				if(ch == '(') {
					add_token(TOK_PAREN_OPEN, 0);
					next_ch();
					skip_ws();
				}
				else if(ch == ')') {
					add_token(TOK_PAREN_CLOSE, 0);
					next_ch();
					skip_ws();
				}
				else if(ch == ',') {
					add_token(TOK_COMMA, 0);
					next_ch();
					skip_ws();
				}
				else if(ch == ':') {
					add_token(TOK_COLON, 0);
					next_ch();
					skip_ws();
				}
				else if(ch == '\n' || ch == '\r') {
					add_token(TOK_NEWLINE, 0);
					next_ch();
					/* collapse all versions, and multiples to a single token */
					while(ch == '\n' || ch == '\r') next_ch();
					skip_ws();
				}
				else if(isalpha(ch)) {
					mode = TOK_IDENT;
					accept_ch();
					next_ch();
				}
				else if(isdigit(ch)) {
					mode = TOK_LITERAL_INT;
					accept_ch();
					next_ch();
				}
				else {
					printf("ERROR: unexpected character '%c' (%d)\n", ch, ch);
					return;
				}
				break;
			case TOK_IDENT:
				if(isidentch(ch)) {
					accept_ch();
					next_ch();
				}
				else {
					finalise_buf();
					if(strcmp(buf, "fun") == 0)
						add_token(TOK_FUN, buf);
					else if(strcmp(buf, "end") == 0)
						add_token(TOK_END, buf);
					else
						add_token(TOK_IDENT, buf);
					printf("ident: >>%s<<\n", buf);
					reset_buf();
					mode = TOK_NONE;
					skip_ws();
				}
				break;
			case TOK_LITERAL_INT:
				if(isdigit(ch)) {
					accept_ch();
					next_ch();
				}
				else {
					finalise_buf();
					add_token(TOK_LITERAL_INT, buf);
					printf("int_literal: >>%s<<\n", buf);
					reset_buf();
					mode = TOK_NONE;
					skip_ws();
				}
				break;
			default:
				printf("ERROR: unknown mode\n");
				return;
		}
	} while(ch != EOF);
	add_token(TOK_EOF, 0);
}


/******************************************************************************/
/*                                                                            */
/*   PARSER                                                                   */
/*                                                                            */
/******************************************************************************/

int parse_pos;

enum node_type { NODE_ROOT, NODE_FUNDEF, NODE_FUNCALL, NODE_LITERAL_INT };

struct data_type {
	char* name;
};

struct node {
	struct node** childs;
	int child_count;
	char* name;
	enum node_type type;
	char* val;
	struct data_type data_type;
};

struct node* node_new(enum node_type type) {
	struct node* node = malloc(sizeof(struct node));
	node->childs = 0;
	node->child_count = 0;
	node->name = 0;
	node->type = type;
	node->val = 0;
	node->data_type.name = 0;
	return node;
}

void node_add_child(struct node* parent, struct node* child) {
	int new_size = sizeof(struct node*) * (++parent->child_count);
	parent->childs = realloc(parent->childs, new_size);
	parent->childs[parent->child_count-1] = child;
}

struct token* expect(enum token_type type) {
	if(type == tokens[parse_pos].type) {
		++parse_pos;
		return &tokens[parse_pos-1];
	}
	else {
		printf("expected (%d), got (%d), at (%d)\n",
		       type, tokens[parse_pos].type, parse_pos);
		exit(-1);
	}
}

int is_next_tok(enum token_type type) {
	return tokens[parse_pos].type == type;
}

struct node* p_expr();

struct node* p_func_call() {
	struct node* node = node_new(NODE_FUNCALL);
	node->name = expect(TOK_IDENT)->str;
	expect(TOK_PAREN_OPEN);
	printf("1\n");
	if(tokens[parse_pos].type != TOK_PAREN_CLOSE) {
		printf("2\n");
		node_add_child(node, p_expr());
		printf("3\n");
		while(tokens[parse_pos].type == TOK_COMMA) {
			printf("4\n");
			expect(TOK_COMMA);
			node_add_child(node, p_expr());
		}
	}
	expect(TOK_PAREN_CLOSE);
	return node;
}

struct node* p_var_use() {
	/* todo: implement */
}

struct node* p_expr() {
	switch(tokens[parse_pos].type) {
		case TOK_IDENT:
			if(tokens[parse_pos+1].type == TOK_PAREN_OPEN)
				return p_func_call();
			else
				return p_var_use();
		case TOK_LITERAL_INT:
		{
			struct node* node = node_new(NODE_LITERAL_INT);
			node->val = expect(TOK_LITERAL_INT)->str;
			return node;
		}
		default:
			printf("can't parse expression, (%d)\n", tokens[parse_pos].type);
			exit(-1);
	}
}

struct node* p_statement() {
	printf("p_statement\n");
	struct node* node;
	switch(tokens[parse_pos].type) {
		default:
			node = p_expr();
			expect(TOK_NEWLINE);
	}
	return node;
}

struct node* p_fun() {
	printf("p_fun\n");
	struct node* node = node_new(NODE_FUNDEF);
	expect(TOK_FUN);
	node->name = expect(TOK_IDENT)->str;
	expect(TOK_PAREN_OPEN);
	expect(TOK_PAREN_CLOSE);
	if(tokens[parse_pos].type == TOK_COLON) {
		expect(TOK_COLON);
		node->data_type.name = expect(TOK_IDENT)->str;
	}
	expect(TOK_NEWLINE);
	while(!is_next_tok(TOK_END)) {
		node_add_child(node, p_statement());
	}
	expect(TOK_END);
	expect(TOK_NEWLINE);
	return node;
}

struct node* parse() {
	struct node* node = node_new(NODE_ROOT);
	parse_pos = 0;
	switch(tokens[parse_pos].type) {
		case TOK_EOF:
			break;
		case TOK_FUN:
			node_add_child(node, p_fun());
	}
	return node;
}


/******************************************************************************/
/*                                                                            */
/*   MAIN                                                                     */
/*                                                                            */
/******************************************************************************/

void print_tokens() {
	for(int i = 0; i < token_len; ++i)
		printf("TOKEN %d %s\n", tokens[i].type, tokens[i].str);
}

void print_indent(int num) {
	for(int i = 0; i < num; ++i) printf("  ");
}

void print_tree(struct node* node, int indent) {
	switch(node->type) {
		case NODE_ROOT:
			for(int j = 0; j < node->child_count; ++j) {
				print_tree(node->childs[j], 0);
				printf("\n");
			}
			break;
		case NODE_FUNDEF:
			print_indent(indent);
			printf("%s %s() {\n",
			       node->data_type.name?node->data_type.name:"void",
				   node->name);
			for(int j = 0; j < node->child_count; ++j) {
				print_tree(node->childs[j], indent+1);
				printf(";\n");
			}
			printf("}\n");
			break;
		case NODE_FUNCALL:
			print_indent(indent);
			printf("%s(\n", node->name);
			for(int j = 0; j < node->child_count; ++j) {
				print_tree(node->childs[j], indent+1);
				if(j < node->child_count-1)
					printf(",\n");
				else
					printf("\n");
			}
			print_indent(indent);
			printf(")");
			break;
		case NODE_LITERAL_INT:
			print_indent(indent);
			printf("%s", node->val);
			break;
		default:
			printf("(UNKNOWN TOKEN)");
	}
}

int main() {
	tokenise();
	print_tokens();
	struct node* node = parse();
	print_tree(node, 0);
	return 0;
}