bshell/bshell.runtime/include/bshell/parse/token.h

#ifndef BSHELL_PARSE_TOKEN_H_
#define BSHELL_PARSE_TOKEN_H_

#include <fx/queue.h>
#include <fx/value.h>
#include <stdbool.h>

struct bshell_char_cell {
	unsigned long c_row, c_col;
};

enum bshell_lex_token_type {
	BSHELL_TOK_NONE = 0,
	__BSHELL_TOK_INDEX_BASE = 100,
	BSHELL_TOK_KEYWORD,
	BSHELL_TOK_SYMBOL,
	BSHELL_TOK_INT,
	BSHELL_TOK_DOUBLE,
	BSHELL_TOK_WORD,
	BSHELL_TOK_WORD_START,
	BSHELL_TOK_WORD_END,
	BSHELL_TOK_FLAG,
	BSHELL_TOK_OPERATOR,
	BSHELL_TOK_VAR,
	BSHELL_TOK_VAR_SPLAT,
	BSHELL_TOK_STRING,
	BSHELL_TOK_STR_START,
	BSHELL_TOK_STR_END,
	BSHELL_TOK_LINEFEED,
	__BSHELL_TOK_INDEX_LIMIT,
};

enum bshell_lex_keyword {
	BSHELL_KW_NONE = 0,
	__BSHELL_KW_INDEX_BASE = 200,
	BSHELL_KW_FUNC,
	BSHELL_KW_IF,
	BSHELL_KW_ELSEIF,
	BSHELL_KW_ELSE,
	__BSHELL_KW_INDEX_LIMIT,
};

enum bshell_lex_operator {
	BSHELL_TKOP_NONE = 0,
	__BSHELL_TKOP_INDEX_BASE = 300,
	BSHELL_TKOP_F,
	BSHELL_TKOP_BAND,
	BSHELL_TKOP_BOR,
	BSHELL_TKOP_BXOR,
	BSHELL_TKOP_BNOT,
	BSHELL_TKOP_SHL,
	BSHELL_TKOP_SHR,
	BSHELL_TKOP_EQ,
	BSHELL_TKOP_NE,
	BSHELL_TKOP_GT,
	BSHELL_TKOP_LT,
	BSHELL_TKOP_GE,
	BSHELL_TKOP_LE,
	BSHELL_TKOP_MATCH,
	BSHELL_TKOP_NOTMATCH,
	BSHELL_TKOP_REPLACE,
	BSHELL_TKOP_LIKE,
	BSHELL_TKOP_NOTLIKE,
	BSHELL_TKOP_IN,
	BSHELL_TKOP_NOTIN,
	BSHELL_TKOP_CONTAINS,
	BSHELL_TKOP_NOTCONTAINS,
	BSHELL_TKOP_AND,
	BSHELL_TKOP_OR,
	BSHELL_TKOP_XOR,
	BSHELL_TKOP_NOT,
	BSHELL_TKOP_SPLIT,
	BSHELL_TKOP_JOIN,
	BSHELL_TKOP_IS,
	BSHELL_TKOP_ISNOT,
	BSHELL_TKOP_AS,
	__BSHELL_TKOP_INDEX_LIMIT,
};

enum bshell_lex_symbol {
	BSHELL_SYM_NONE = 0,
	__BSHELL_SYM_INDEX_BASE = 400,
	BSHELL_SYM_BANG,
	BSHELL_SYM_PLUS,
	BSHELL_SYM_HYPHEN,
	BSHELL_SYM_FORWARD_SLASH,
	BSHELL_SYM_ASTERISK,
	BSHELL_SYM_AMPERSAND,
	BSHELL_SYM_PERCENT,
	BSHELL_SYM_SQUOTE,
	BSHELL_SYM_DQUOTE,
	BSHELL_SYM_HASH,
	BSHELL_SYM_COLON_COLON,
	BSHELL_SYM_SEMICOLON,
	BSHELL_SYM_COMMA,
	BSHELL_SYM_DOLLAR,
	BSHELL_SYM_DOLLAR_LEFT_PAREN,
	BSHELL_SYM_DOLLAR_LEFT_BRACE,
	BSHELL_SYM_DOT,
	BSHELL_SYM_DOT_DOT,
	BSHELL_SYM_PIPE,
	BSHELL_SYM_AT,
	BSHELL_SYM_AT_LEFT_PAREN,
	BSHELL_SYM_AT_LEFT_BRACE,
	BSHELL_SYM_LEFT_BRACE,
	BSHELL_SYM_RIGHT_BRACE,
	BSHELL_SYM_LEFT_BRACKET,
	BSHELL_SYM_RIGHT_BRACKET,
	BSHELL_SYM_LEFT_PAREN,
	BSHELL_SYM_RIGHT_PAREN,
	BSHELL_SYM_EQUAL,
	BSHELL_SYM_PLUS_EQUAL,
	BSHELL_SYM_HYPHEN_EQUAL,
	BSHELL_SYM_ASTERISK_EQUAL,
	BSHELL_SYM_FORWARD_SLASH_EQUAL,
	BSHELL_SYM_PERCENT_EQUAL,
	BSHELL_SYM_QUESTION_DOT,
	BSHELL_SYM_QUESTION_LEFT_BRACKET,
	__BSHELL_SYM_INDEX_LIMIT,
};

struct bshell_lex_token {
	enum bshell_lex_token_type tok_type;

	struct bshell_char_cell tok_start, tok_end;

	fx_queue_entry tok_entry;

	union {
		enum bshell_lex_keyword tok_keyword;
		enum bshell_lex_symbol tok_symbol;
		enum bshell_lex_operator tok_operator;
		fx_value tok_number;
		char *tok_str;
	};
};

extern struct bshell_lex_token *bshell_lex_token_create(
	enum bshell_lex_token_type type);
extern struct bshell_lex_token *bshell_lex_token_create_with_string(
	enum bshell_lex_token_type type,
	const char *s);
extern void bshell_lex_token_destroy(struct bshell_lex_token *tok);

extern struct bshell_lex_token *bshell_lex_token_change_type(
	struct bshell_lex_token *tok,
	enum bshell_lex_token_type new_type);
extern void bshell_lex_token_change_string(
	struct bshell_lex_token *tok,
	const char *s);

static inline bool bshell_lex_token_is_symbol(
	struct bshell_lex_token *tok,
	enum bshell_lex_symbol sym)
{
	return (tok->tok_type == BSHELL_TOK_SYMBOL && tok->tok_symbol == sym);
}
static inline bool bshell_lex_token_is_keyword(
	struct bshell_lex_token *tok,
	enum bshell_lex_keyword kw)
{
	return (tok->tok_type == BSHELL_TOK_KEYWORD && tok->tok_keyword == kw);
}
static inline bool bshell_lex_token_type_has_string_value(
	enum bshell_lex_token_type type)
{
	switch (type) {
	case BSHELL_TOK_WORD:
	case BSHELL_TOK_STRING:
	case BSHELL_TOK_FLAG:
	case BSHELL_TOK_VAR:
	case BSHELL_TOK_VAR_SPLAT:
		return true;
	default:
		return false;
	}
}
static inline bool bshell_lex_token_has_string_value(
	const struct bshell_lex_token *tok)
{
	return bshell_lex_token_type_has_string_value(tok->tok_type);
}

extern const char *bshell_token_type_to_string(enum bshell_lex_token_type type);
extern const char *bshell_lex_keyword_to_string(
	enum bshell_lex_keyword keyword);
extern const char *bshell_lex_symbol_to_string(enum bshell_lex_symbol sym);
extern const char *bshell_lex_operator_to_string(enum bshell_lex_operator op);

#endif