vm/src/lexer.h

#ifndef LEXER_H
#define LEXER_H

#define LEXER_REGEX_WHITESPACE "\\s+"
#define LEXER_REGEX_IDENTIFIER "[a-zA-Z][a-zA-Z0-9]*"
#define LEXER_REGEX_KEYWORD "if|ret|func|str|arr|err|null"
#define LEXER_REGEX_OPERATOR "=|=="

#define LEXER_REGEX_INTEGER "[0-9]+"
#define LEXER_REGEX_STRING "\\\"[^\\\"]*\\\""

#define LEXER_REGEX_COMMA "\\,"
#define LEXER_REGEX_OPENBRACE "\\{"
#define LEXER_REGEX_CLOSEBRACE "\\}"
#define LEXER_REGEX_OPENPAREN "\\("
#define LEXER_REGEX_CLOSEPAREN "\\)"

typedef enum lexer_token_class
{
	LEXER_TOKEN_WHITESPACE,
	LEXER_TOKEN_IDENTIFIER,
	LEXER_TOKEN_KEYWORD,
	LEXER_TOKEN_OPERATOR,

	LEXER_TOKEN_INTEGER,
	LEXER_TOKEN_STRING,

	LEXER_TOKEN_COMMA,
	LEXER_TOKEN_OPENBRACE,
	LEXER_TOKEN_CLOSEBRACE,
	LEXER_TOKEN_OPENPAREN,
	LEXER_TOKEN_CLOSEPAREN,

	LEXER_TOKEN_NONE
} lexer_token_class;

typedef struct lexer_token
{
	lexer_token_class tokenClass;
	char* str;
	size_t len;
} lexer_token;

typedef struct lexer_tokens
{
	lexer_token* pairs;
	size_t length;
	size_t allocd;
} lexer_tokens;

lexer_tokens* lexer_analyze(char* str);

#endif