lexing up and running

2025-04-15 20:02:25 +02:00 · 2025-04-15 20:02:25 +02:00 · b9266cdf96
commit b9266cdf96
parent 10e16147ba
6 changed files with 434 additions and 102 deletions
--- a/include/msph/token.h
+++ b/include/msph/token.h
@ -0,0 +1,193 @@
+#ifndef _MSPH_EXPR_H
+#define _MSPH_EXPR_H
+
+/*
+ * TYPES:
+ * Conj
+ * Disj
+ * Impl
+ * Arrow
+ * Box
+ * (Sub)
+ * Forall
+ * 
+ * True
+ * False
+ * Var
+ * Nominal
+ *
+ * Record
+ *
+ * DEFINITIONS/DIRECTIVES:
+ * Type definition		(type A[X..] = T)
+ * Nominal definition		(nominal N)
+ * Member definition		(member mname : T)
+ *
+ * {
+ * member a: A
+ * member b: B
+ * }
+ *
+ * Subtyping check		(check A <: B, checks if A <: B)
+ *
+ * EXTRA DEFINITIONS
+ * Class definition		(class C[...] { ... }, shorthand)
+ * 
+ * EXPRESSIONS
+ * Conj
+ * Disj
+ * Impl
+ * Arrow
+ * Box
+ * (Sub)
+ * Forall
+ * 
+ * True
+ * False
+ * Var
+ * Nominal
+ *
+ * Trait			({ ... }, creates scoping)
+ *
+ *
+ * TOKENS
+ * lbrace		{
+ * rbrace		}
+ * lbrak		[
+ * rbrak		]
+ * lparen		(
+ * rparen		)
+ * colon		:
+ * equals		=
+ *
+ * op_and		&
+ * op_or		|
+ * op_impl		=>
+ * op_sub		<:
+ *
+ * kw_type		type
+ * kw_nominal		nominal
+ * kw_member		member
+ * kw_check		check
+ * kw_box		box
+ * kw_forall		forall
+ *
+ * k_true		True
+ * k_false		False
+ *
+ * ident		C, anid, ...
+ *
+ */
+
+struct msph_ctx {
+	int	err;
+	int	err_info;
+	char	errbuf[SPHO_ERR_BUF_LEN];
+
+#ifdef SPHO_DEBUG
+	char	filebuf[SPHO_ERR_FILEBUF_LEN];
+	int	errline;
+#endif
+};
+
+enum msph_tok_type {
+	TOK_LBRACE,		// {
+	TOK_RBRACE,		// }
+	TOK_LBRAK,		// [
+	TOK_RBRAK,		// ]
+	TOK_LPAREN,		// (
+	TOK_RPAREN,		// )
+	TOK_COLON,		// :
+	TOK_EQUALS,		// =
+
+	TOK_AMP,		// &
+	TOK_PIPE,		// |
+	TOK_RARROW,		// =>
+	TOK_SUB,		// <:
+
+	TOK_KW_TYPE,		// type
+	TOK_KW_NOMINAL,		// nominal
+	TOK_KW_MEMBER,		// member
+	TOK_KW_CHECK,		// check
+	TOK_KW_BOX,		// box
+	TOK_KW_FORALL,		// forall
+
+	TOK_CONST_TRUE,		// True
+	TOK_CONST_FALSE,	// False
+
+	TOK_IDENT,		// identifiers
+	TOK_WSPACE,		// special: whitespace
+	TOK_END			// special: denote end of array
+};
+
+#define MSPH_TOKEN_BUF_LEN	128
+
+struct token_s {
+	char	buf[MSPH_TOKEN_BUF_LEN];
+};
+
+union token_data {
+	struct token_s	s;
+};
+
+struct msph_token {
+	int			type;
+	union token_data	d;
+};
+
+#define MSPH_FILE_NAME_LEN	1024
+#define MSPH_FILE_BUF_LEN	4096
+#define MSPH_TOKEN_SRC_FILE	1
+
+struct msph_token_src_file {
+	FILE		*f;
+
+	/* circular buffer for reading */
+	size_t		pos;
+	size_t		end;
+	char		buf[MSPH_FILE_BUF_LEN];
+
+	/* file path */
+	char		name[MSPH_FILE_NAME_LEN];
+};
+
+#define MSPH_TOKEN_SRC_STR	2
+struct msph_token_src_str {
+	const char	*s;
+	size_t		len;
+	size_t		pos;
+};
+
+union msph_token_src_data {
+	struct msph_token_src_file	file;
+	struct msph_token_src_str	str;
+};
+
+struct msph_token_src {
+	int				type;
+	union msph_token_src_data	inner;
+};
+
+struct msph_token_stream {
+	struct msph_ctx		*ctx;
+
+	struct msph_token_src	src;
+};
+
+void msph_ctx_init(struct msph_ctx *);
+
+struct msph_token_stream	*msph_token_stream_file(struct msph_ctx *,
+				    FILE *, const char *);
+struct msph_token_stream	*msph_token_stream_frombuf(struct msph_ctx *,
+				    const char *, size_t);
+
+int	msph_token_stream_close(struct msph_token_stream*);
+
+ssize_t	msph_token_stream_read_tok(struct msph_token *, size_t,
+	    struct msph_token_stream *);
+int	msph_token_stream_print(struct msph_token_stream *, FILE *);
+
+size_t	msph_token_str(char *buf, size_t len, struct msph_token *tok);
+
+
+#endif /* _MSPH_EXPR_H */