#ifndef CERC_LEX_H
#define CERC_LEX_H
#include <stdbool.h>
#include <stdio.h>
#include "utf8.h"

#define C_EOF UTF8_INVALID

enum lexical_token {
	// Keywords
	T_ALIGN,
	T_AS,
	T_BREAK,
	T_CONST,
	T_CONTINUE,
	T_DEFER,
	T_ELSE,
	T_ENUM,
	T_EXTERN,
	T_FALSE,
	T_FOR,
	T_FUNC,
	T_IF,
	T_INCLUDE,
	T_LET,
	T_PUB,
	T_RETURN,
	T_STRUCT,
	T_TAGOF,
	T_TRUE,
	T_UNION,
	T_VAR,
	T_LAST_KEYWORD,

	// Builtin types
	T_BOOL,
	T_FLOAT32,
	T_FLOAT64,
	T_INT8,
	T_INT16,
	T_INT32,
	T_INT64,
	T_INT,
	T_UINT8,
	T_UINT16,
	T_UINT32,
	T_UINT64,
	T_UINT,
	T_LAST_BUILTIN_TYPE,

	// Operators
	T_ADD,
	T_ADDASSIGN,
	T_AND,
	T_ASSIGN,
	T_BAND,
	T_BANDASSIGN,
	T_BANG,
	T_BNOT,
	T_BOR,
	T_BORASSIGN,
	T_BSHL,
	T_BSHR,
	T_BXOR,
	T_BXORASSIGN,
	T_COLON,
	T_COMMA,
	T_DECR,
	T_DIV,
	T_DIVASSIGN,
	T_DOT,
	T_EQ,
	T_GE,
	T_GT,
	T_INCR,
	T_LBRACE,
	T_LBRACKET,
	T_LE,
	T_LPAREN,
	T_LT,
	T_MINUS,
	T_MODASSIGN,
	T_MODDELIM,
	T_MODULO,
	T_MULASSIGN,
	T_NEQ,
	T_OR,
	T_QUESTION,
	T_RBRACE,
	T_RBRACKET,
	T_RPAREN,
	T_SEMICOLON,
	T_STAR,
	T_SUBASSIGN,
	T_UNDERSCORE,
	T_LAST_OPERATOR,

	// Tokens with additional information
	T_COMMENT,
	T_NAME,
	T_NUMBER,
	T_CHAR,
	T_STRING,

	// Magic values
	T_EOF,
	T_NONE,
};

extern const char *tokens[];

struct number {
	bool isfloat;
	union {
		uint64_t integer;
		double floatingpt;
	} value;
};

struct location {
	const char *file;
	int line, column;
};

struct token {
	enum lexical_token token;
	struct location loc;
	union {
		struct number num;
		unsigned int chr;
		const char *str;
	} info;
};

struct lexer {
	FILE *in;
	uint32_t c[2];
	char *buf;
	size_t bufsz, buflen;
	struct token un;
	struct location loc;
};

void lex_init(struct lexer *lexer, FILE *f, const char *filename);
void lex_finish(struct lexer *lexer);

struct location lex_loc(struct lexer *lexer);

enum lexical_token lex(struct lexer *lexer, struct token *out);
enum lexical_token lex_any(struct lexer *lexer, struct token *out);
void unlex(struct lexer *lexer, const struct token *in);
bool match(struct lexer *lexer, enum lexical_token token);

#endif