regex module: implemented token scanners, added definitions and scan table.

1 files changed, 64 insertions, 0 deletions

diff --git a/src/regex/tbnf_regex_defs.h b/src/regex/tbnf_regex_defs.h
new file mode 100644
index 0000000..9276ae0
--- /dev/null
+++ b/src/regex/tbnf_regex_defs.h
@@ -0,0 +1,64 @@
+/**************************************************************/
+/*  treebnf: a tree oriented bnf library                      */
+/*  Copyright (C) 2024  SysDeer Technologies, LLC             */
+/*  Released under GPLv2 and GPLv3; see COPYING.TREEBNF.      */
+/**************************************************************/
+
+#ifndef TBNF_REGEX_DEFS_H
+#define TBNF_REGEX_DEFS_H
+
+enum tbnf_regex_scan_state {
+	TBNF_REGEX_STATE_INIT,
+	TBNF_REGEX_STATE_BRACE,
+	TBNF_REGEX_STATE_BRACKET,
+	TBNF_REGEX_STATE_CAP,
+};
+
+enum tbnf_regex_token_type {
+	TBNF_REGEX_TOK_NONE,
+
+	/* in-brace tokens */
+	TBNF_REGEX_TOK_BRACE_RBRACE,
+	TBNF_REGEX_TOK_BRACE_DIGIT,
+	TBNF_REGEX_TOK_BRACE_COMMA,
+
+	/* in-bracket tokens */
+	TBNF_REGEX_TOK_BRACKET_ESCAPED_CHAR,
+	TBNF_REGEX_TOK_BRACKET_CHARACTER_CLASS,
+	TBNF_REGEX_TOK_BRACKET_COLLATION_SYMBOL,
+	TBNF_REGEX_TOK_BRACKET_EQUIVALENCE_CLASS,
+	TBNF_REGEX_TOK_BRACKET_RBRACKET,
+	TBNF_REGEX_TOK_BRACKET_HYPHEN,
+	TBNF_REGEX_TOK_BRACKET_ERROR,
+	TBNF_REGEX_TOK_BRACKET_CHAR,
+
+	/* brace state initializer token */
+	TBNF_REGEX_TOK_LBRACE,
+
+	/* bracket state initializer tokens */
+	TBNF_REGEX_TOK_LBRACKET_CIRCUMFLEX_RBRACKET,
+	TBNF_REGEX_TOK_LBRACKET_CIRCUMFLEX,
+	TBNF_REGEX_TOK_LBRACKET_RBRACKET,
+	TBNF_REGEX_TOK_LBRACKET,
+
+	/* init state tokens */
+	TBNF_REGEX_TOK_ESCAPED_CHAR,
+
+	TBNF_REGEX_TOK_CIRCUMFLEX_ASTERISK,
+	TBNF_REGEX_TOK_CIRCUMFLEX,
+
+	TBNF_REGEX_TOK_LPAREN,
+	TBNF_REGEX_TOK_RPAREN,
+
+	TBNF_REGEX_TOK_ASTERISK,
+	TBNF_REGEX_TOK_PERIOD,
+	TBNF_REGEX_TOK_DOLLAR,
+	TBNF_REGEX_TOK_VLINE,
+	TBNF_REGEX_TOK_QMARK,
+	TBNF_REGEX_TOK_PLUS,
+	TBNF_REGEX_TOK_CHAR,
+
+	TBNF_REGEX_TOK_CAP,
+};
+
+#endif