42sh/src/lexer/lexer.h

#ifndef LEXER_H
#define LEXER_H

#include <sys/types.h>

enum token_type
{
    // Special characters
    TOKEN_NULL = 0,
    TOKEN_EOF,
    TOKEN_WORD,
    TOKEN_NEWLINE,
    TOKEN_QUOTE,
    TOKEN_DOUBLE_QUOTE,
    TOKEN_GRAVE,
    TOKEN_SEMICOLON,
    TOKEN_COMMENT,
    TOKEN_PIPE,
    TOKEN_AMPERSAND,
    TOKEN_BACKSLASH,
    TOKEN_DOLLAR,
    TOKEN_LEFT_PAREN,
    TOKEN_RIGHT_PAREN,
    TOKEN_LEFT_BRACKET,
    TOKEN_RIGHT_BRACKET,
    TOKEN_LESS,
    TOKEN_GREATER,
    TOKEN_STAR,

    // Keywords
    TOKEN_IF,
    TOKEN_THEN,
    TOKEN_ELSE,
    TOKEN_FI,
    TOKEN_ELIF
};

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

/*
 * @brief: returns the next (newly allocated) token without consuming it.
 *         if end of input is reached, enters in EOF looping node,
 *         returning only the same token of type TOKEN_EOF.
 */
struct token *peek_token(void);

/*
 * @brief: returns the next (newly allocated) token and consumes it.
 *         if end of input is reached, returns a token of type TOKEN_EOF.
 *         It also frees the last token created if there was one.
 *
 * @warning: if the last returned token was a token EOF, it frees it
 *          and returns NULL. This means that after peeking a token EOF
 *          in the parser, there must be EXACTLY ONE call to pop_token().
 *
 */
struct token *pop_token(void);

/* @note: maybe usefull for subshells.
 *
 * @warning: NOT IMPLEMENTED.
 */

struct token *get_token_str(void);

/*
 * @brief: return a newly allocated token, with the corresponding type.
 *         The data contains [size] char, starting from [begin].
 *
 * @return: NULL on error, a token otherwise.
 */
struct token *new_token(char *begin, ssize_t size);

/* @brief: frees the token given in argument
 */
void free_token(struct token **tok);

/*
 * @brief: checks if the stream used for the last token creation is empty.
 *         If it is, it calls stream_read() from IO_backend,
 *         and sets [remaing_chars].
 *         If not, it starts from the end of the last token.
 *         Also trims left blanks before returning.
 *
 * @return: char* stream from which we tokenise.
 */
char *stream_init(void);

#endif /* ! LEXER_H */