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

1 files changed, 287 insertions, 0 deletions

diff --git a/src/regex/tbnf_regex_scanfns.h b/src/regex/tbnf_regex_scanfns.h
new file mode 100644
index 0000000..c57a606
--- /dev/null
+++ b/src/regex/tbnf_regex_scanfns.h
@@ -0,0 +1,287 @@
+/**************************************************************/
+/*  treebnf: a tree oriented bnf library                      */
+/*  Copyright (C) 2024  SysDeer Technologies, LLC             */
+/*  Released under GPLv2 and GPLv3; see COPYING.TREEBNF.      */
+/**************************************************************/
+
+#ifndef TBNF_REGEX_SCANFNS_H
+#define TBNF_REGEX_SCANFNS_H
+
+#include <treebnf/treebnf.h>
+
+/* in-brace scanners */
+static inline int tbnf_regex_scan_brace_rbrace(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '}')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_brace_digit(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] >= '0')
+		if (sctx->tok_scan_mark[0] <= '9')
+			return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_brace_comma(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == ',')
+		return 1;
+
+	return -1;
+}
+
+
+/* in-bracket scanners */
+static inline int tbnf_regex_scan_bracket_escaped_char(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '\\')
+		if (&sctx->tok_scan_mark[1] < sctx->tok_scan_cap)
+			if (sctx->tok_scan_mark[1])
+				return 2;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_bracket_character_class(const struct tbnf_scan_ctx * sctx)
+{
+	const char * ch = 0;
+
+	if (&sctx->tok_scan_mark[4] >= sctx->tok_scan_cap)
+		return -1;
+
+	if (sctx->tok_scan_mark[0] == '[')
+		if (sctx->tok_scan_mark[1] == ':')
+			ch = &sctx->tok_scan_mark[2];
+
+	if (!ch)
+		return -1;
+
+	for (; (*ch >= 'a') && (*ch <= 'z') && (&ch[2] < sctx->tok_scan_cap); )
+		ch++;
+
+	if ((*ch++ == ':') && (*ch++ == ']'))
+			return (ch - sctx->tok_scan_mark);
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_bracket_collation_symbol(const struct tbnf_scan_ctx * sctx)
+{
+	if (&sctx->tok_scan_mark[4] >= sctx->tok_scan_cap)
+		return -1;
+
+	if (sctx->tok_scan_mark[0] == '[')
+		if (sctx->tok_scan_mark[1] == '.')
+			if (sctx->tok_scan_mark[3] == '.')
+				if (sctx->tok_scan_mark[4] == ']')
+					if (sctx->tok_scan_mark[2])
+						return 5;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_bracket_equivalence_class(const struct tbnf_scan_ctx * sctx)
+{
+	const char * ch = 0;
+
+	if (&sctx->tok_scan_mark[4] >= sctx->tok_scan_cap)
+		return -1;
+
+	if (sctx->tok_scan_mark[0] == '[')
+		if (sctx->tok_scan_mark[1] == '=')
+			ch = &sctx->tok_scan_mark[2];
+
+	if (!ch)
+		return -1;
+
+	for (; (*ch >= 'a') && (*ch <= 'z') && (&ch[2] < sctx->tok_scan_cap); )
+		ch++;
+
+	if ((*ch++ == '=') && (*ch++ == ']'))
+			return (ch - sctx->tok_scan_mark);
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_bracket_rbracket(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == ']')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_bracket_hyphen(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '-')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_bracket_error(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '[')
+		return 1;
+
+	return -1;
+}
+
+
+
+/* bracket state entry scanners */
+static inline int tbnf_regex_scan_lbracket_circumflex_rbracket(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '[')
+		if (&sctx->tok_scan_mark[2] < sctx->tok_scan_cap)
+			if (sctx->tok_scan_mark[1] == '^')
+				if (sctx->tok_scan_mark[2] == ']')
+					return 3;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_lbracket_circumflex(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '[')
+		if (&sctx->tok_scan_mark[1] < sctx->tok_scan_cap)
+			if (sctx->tok_scan_mark[1] == '^')
+				return 2;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_lbracket_rbracket(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '[')
+		if (&sctx->tok_scan_mark[1] < sctx->tok_scan_cap)
+			if (sctx->tok_scan_mark[1] == ']')
+				return 2;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_lbracket(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '[')
+		return 1;
+
+	return -1;
+}
+
+
+/* common scanners */
+static inline int tbnf_regex_scan_escaped_char(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '\\')
+		if (&sctx->tok_scan_mark[1] < sctx->tok_scan_cap)
+			if (sctx->tok_scan_mark[1])
+				return 2;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_circumflex_asterisk(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '^')
+		if (&sctx->tok_scan_mark[1] < sctx->tok_scan_cap)
+			if (sctx->tok_scan_mark[1] == '*')
+				return 2;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_circumflex(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '^')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_lbrace(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '{')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_lparen(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '(')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_rparen(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == ')')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_asterisk(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '*')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_period(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '.')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_dollar(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '$')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_vline(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '|')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_qmark(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '?')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_plus(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0] == '+')
+		return 1;
+
+	return -1;
+}
+
+static inline int tbnf_regex_scan_char(const struct tbnf_scan_ctx * sctx)
+{
+	if (sctx->tok_scan_mark[0])
+		return 1;
+
+	return -1;
+}
+
+#endif