From 678295f25ee1915d93d80c3650f76e55b7eefbf0 Mon Sep 17 00:00:00 2001
From: Kyryl Melekhin <k.melekhin@gmail.com>
Date: Wed, 21 Jul 2021 15:39:11 +0000
Subject: [PATCH] add word boundary assert

---
 pike.c  | 31 ++++++++++++++++++++++++++++++-
 test.sh |  9 +++++++++
 2 files changed, 39 insertions(+), 1 deletion(-)

diff --git a/pike.c b/pike.c
index e2bb77f..c371745 100644
--- a/pike.c
+++ b/pike.c
@@ -47,6 +47,19 @@ else if (~dst & 0x08) \
 else \
 	dst = 0; \
 
+static int isword(const char *s)
+{
+	int c = (unsigned char) s[0];
+	return isalnum(c) || c == '_' || c > 127;
+}
+
+static char *uc_beg(char *beg, char *s)
+{
+	while (s > beg && (((unsigned char) *s) & 0xc0) == 0x80)
+		s--;
+	return s;
+}
+
 typedef struct rinst rinst;
 struct rinst
 {
@@ -84,6 +97,7 @@ enum	/* rinst.opcode */
 	ASSERT,
 	BOL,
 	EOL,
+	WBND,
 	// Instructions which take relative offset as arg
 	JMP,
 	SPLIT,
@@ -196,7 +210,13 @@ void re_dumpcode(rcode *prog)
 			printf("save %d\n", code[pc++]);
 			break;
 		case ASSERT:
-			printf("assert %s\n", code[pc++] == BOL ? "bol" : "eol");
+			if (code[pc] == BOL)
+				printf("assert bol\n");
+			else if (code[pc] == EOL)
+				printf("assert eol\n");
+			else if (code[pc] == WBND)
+				printf("assert WBND\n");
+			pc++;
 			break;
 		}
 	}
@@ -215,6 +235,13 @@ static int _compilecode(const char **re_loc, rcode *prog, int sizecode)
 		case '\\':
 			re++;
 			if (!*re) goto syntax_error; // Trailing backslash
+			if (*re == 'b') {
+				EMIT(PC++, ASSERT);
+				EMIT(PC++, WBND);
+				prog->len++;
+				term = PC;
+				break;
+			} 
 		default:
 			term = PC;
 			EMIT(PC++, CHAR);
@@ -529,6 +556,8 @@ int re_comp(rcode *prog, const char *re, int anchored)
 			goto rec_check##nn; \
 		if(*pc == EOL && *_sp) \
 			goto rec_check##nn; \
+		if(*pc == WBND && isword(sp)) \
+			goto rec_check##nn; \
 		pc++; goto rec##nn; \
 	} \
 } \
diff --git a/test.sh b/test.sh
index 0e4b33b..c2c1b98 100755
--- a/test.sh
+++ b/test.sh
@@ -49,6 +49,9 @@ abc+h+d+f
 [A-Fa-f0-9]{64}
 <tag>[^<]*</tag>
 ^([a-z0-9_.-]+)@([0-9a-z.-]+)\\\\.([a-z.]{2,5})$
+\\\\babc
+ab\\\\bd
+\\\\b(as|js)
 "
 input="\
 abcdef
@@ -99,6 +102,9 @@ abcccccccccccchdf
 bf33d4a0dbbee85061531c9d47e5aae692c0729e5c9c1fa21c46d9bcab5f52c5
 ajdas <tag> sidufisudf hsdfhshdfh sdf </tag> asjdfjs
 veloval596@godpeed.com
+   	   abc
+ab   d
+     js hashasd
 "
 expect="\
 (0,3)
@@ -149,6 +155,9 @@ expect="\
 (0,64)
 (6,44)
 (0,22)(0,10)(11,18)(19,22)
+(7,10)
+-nomatch-
+(5,7)(5,7)
 (0,0)
 "
 c=1