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

1 files changed, 185 insertions, 0 deletions

diff --git a/src/regex/tbnf_regex.c b/src/regex/tbnf_regex.c
new file mode 100644
index 0000000..2cdcb48
--- /dev/null
+++ b/src/regex/tbnf_regex.c
@@ -0,0 +1,185 @@
+/**************************************************************/
+/*  treebnf: a tree oriented bnf library                      */
+/*  Copyright (C) 2024  SysDeer Technologies, LLC             */
+/*  Released under GPLv2 and GPLv3; see COPYING.TREEBNF.      */
+/**************************************************************/
+
+#include <treebnf/treebnf.h>
+
+#include "treebnf_regex_impl.h"
+#include "treebnf_visibility_impl.h"
+
+#include "tbnf_regex_defs.h"
+#include "tbnf_regex_scanfns.h"
+
+#define TBNF_STATE_STACK_SIZE   (512)
+
+/* init state scan table*/
+static struct tbnf_scan_tbl tbnf_regex_scan_tbl__init[TBNF_REGEX_TOK_CAP] = {
+	/* --> brace */
+	[TBNF_REGEX_TOK_LBRACE] = {
+		.tok_scan_fn    = tbnf_regex_scan_lbrace,
+		.tok_state_op   = TBNF_STATE_PUSH,
+		.tok_state_next = TBNF_REGEX_STATE_BRACE,
+	},
+
+
+	/* --> bracket */
+	[TBNF_REGEX_TOK_LBRACKET_CIRCUMFLEX_RBRACKET] = {
+		.tok_scan_fn    = tbnf_regex_scan_lbracket_circumflex_rbracket,
+		.tok_state_op   = TBNF_STATE_PUSH,
+		.tok_state_next = TBNF_REGEX_STATE_BRACKET,
+	},
+
+	[TBNF_REGEX_TOK_LBRACKET_CIRCUMFLEX] = {
+		.tok_scan_fn    = tbnf_regex_scan_lbracket_circumflex,
+		.tok_state_op   = TBNF_STATE_PUSH,
+		.tok_state_next = TBNF_REGEX_STATE_BRACKET,
+	},
+
+	[TBNF_REGEX_TOK_LBRACKET_RBRACKET] = {
+		.tok_scan_fn    = tbnf_regex_scan_lbracket_rbracket,
+		.tok_state_op   = TBNF_STATE_PUSH,
+		.tok_state_next = TBNF_REGEX_STATE_BRACKET,
+	},
+
+	[TBNF_REGEX_TOK_LBRACKET] = {
+		.tok_scan_fn    = tbnf_regex_scan_lbracket,
+		.tok_state_op   = TBNF_STATE_PUSH,
+		.tok_state_next = TBNF_REGEX_STATE_BRACKET,
+	},
+
+
+	/* (expression) */
+	[TBNF_REGEX_TOK_ESCAPED_CHAR] = {
+		.tok_scan_fn    = tbnf_regex_scan_escaped_char,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_CIRCUMFLEX_ASTERISK] = {
+		.tok_scan_fn    = tbnf_regex_scan_circumflex_asterisk,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_CIRCUMFLEX] = {
+		.tok_scan_fn    = tbnf_regex_scan_circumflex,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_LPAREN] = {
+		.tok_scan_fn    = tbnf_regex_scan_lparen,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_RPAREN] = {
+		.tok_scan_fn    = tbnf_regex_scan_rparen,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_ASTERISK] = {
+		.tok_scan_fn    = tbnf_regex_scan_asterisk,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_PERIOD] = {
+		.tok_scan_fn    = tbnf_regex_scan_period,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_DOLLAR] = {
+		.tok_scan_fn    = tbnf_regex_scan_dollar,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_VLINE] = {
+		.tok_scan_fn    = tbnf_regex_scan_vline,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_QMARK] = {
+		.tok_scan_fn    = tbnf_regex_scan_qmark,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_PLUS] = {
+		.tok_scan_fn    = tbnf_regex_scan_plus,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_CHAR] = {
+		.tok_scan_fn    = tbnf_regex_scan_char,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+};
+
+
+/* brace state scan table*/
+static struct tbnf_scan_tbl tbnf_regex_scan_tbl__brace[TBNF_REGEX_TOK_CAP] = {
+	[TBNF_REGEX_TOK_BRACE_RBRACE] = {
+		.tok_scan_fn    = tbnf_regex_scan_brace_rbrace,
+		.tok_state_op   = TBNF_STATE_POP,
+	},
+
+	[TBNF_REGEX_TOK_BRACE_DIGIT] = {
+		.tok_scan_fn    = tbnf_regex_scan_brace_digit,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_BRACE_COMMA] = {
+		.tok_scan_fn    = tbnf_regex_scan_brace_comma,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+};
+
+
+/* bracket state scan table*/
+static struct tbnf_scan_tbl tbnf_regex_scan_tbl__bracket[TBNF_REGEX_TOK_CAP] = {
+	[TBNF_REGEX_TOK_BRACKET_RBRACKET] = {
+		.tok_scan_fn    = tbnf_regex_scan_bracket_rbracket,
+		.tok_state_op   = TBNF_STATE_POP,
+	},
+
+	[TBNF_REGEX_TOK_BRACKET_ESCAPED_CHAR] = {
+		.tok_scan_fn    = tbnf_regex_scan_bracket_escaped_char,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_BRACKET_CHARACTER_CLASS] = {
+		.tok_scan_fn    = tbnf_regex_scan_bracket_character_class,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_BRACKET_COLLATION_SYMBOL] = {
+		.tok_scan_fn    = tbnf_regex_scan_bracket_collation_symbol,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_BRACKET_EQUIVALENCE_CLASS] = {
+		.tok_scan_fn    = tbnf_regex_scan_bracket_equivalence_class,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_BRACKET_HYPHEN] = {
+		.tok_scan_fn    = tbnf_regex_scan_bracket_hyphen,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_BRACKET_ERROR] = {
+		.tok_scan_fn    = tbnf_regex_scan_bracket_error,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+
+	[TBNF_REGEX_TOK_BRACKET_CHAR] = {
+		.tok_scan_fn    = tbnf_regex_scan_char,
+		.tok_state_op   = TBNF_STATE_KEEP,
+	},
+};
+
+
+/* extended regex expression token scan table */
+tbnf_hidden struct tbnf_scan_tbl * tbnf_regex_scan_tbl[] = {
+	[TBNF_REGEX_STATE_INIT]    = tbnf_regex_scan_tbl__init,
+	[TBNF_REGEX_STATE_BRACE]   = tbnf_regex_scan_tbl__brace,
+	[TBNF_REGEX_STATE_BRACKET] = tbnf_regex_scan_tbl__bracket,
+	[TBNF_REGEX_STATE_CAP]     = 0,
+};