From 7a4c25d78030b3a43ed5c8dd1a456f73cb990f44 Mon Sep 17 00:00:00 2001
From: Szabolcs Nagy <nsz@port70.net>
Date: Thu, 24 Nov 2016 01:44:49 +0100
Subject: handle ^ and $ in BRE subexpression start and end as anchors

In BRE, ^ is an anchor at the beginning of an expression, optionally
it may be an anchor at the beginning of a subexpression and must be
treated as a literal otherwise.

Previously musl treated ^ in subexpressions as literal, but at least
glibc and gnu sed treats it as an anchor and that's the more useful
behaviour: it can always be escaped to get back the literal meaning.

Same for $ at the end of a subexpression.

Portable BRE should not rely on this, but there are sed commands in
build scripts which do.

This changes the meaning of the BREs:

	\(^a\)
	\(a\|^b\)
	\(a$\)
	\(a$\|b\)
---
 src/regex/regcomp.c | 21 ++++++++++++---------
 1 file changed, 12 insertions(+), 9 deletions(-)

diff --git a/src/regex/regcomp.c b/src/regex/regcomp.c
index 65f2fd0b..5a7b53a7 100644
--- a/src/regex/regcomp.c
+++ b/src/regex/regcomp.c
@@ -401,8 +401,8 @@ typedef struct {
 	tre_ast_node_t *n;
 	/* Position in the regexp pattern after a parse function returns. */
 	const char *s;
-	/* The first character of the regexp. */
-	const char *re;
+	/* The first character of the last subexpression parsed. */
+	const char *start;
 	/* Current submatch ID. */
 	int submatch_id;
 	/* Current position (number of literal). */
@@ -876,14 +876,14 @@ static reg_errcode_t parse_atom(tre_parse_ctx_t *ctx, const char *s)
 		break;
 	case '^':
 		/* '^' has a special meaning everywhere in EREs, and at beginning of BRE. */
-		if (!ere && s != ctx->re)
+		if (!ere && s != ctx->start)
 			goto parse_literal;
 		node = tre_ast_new_literal(ctx->mem, ASSERTION, ASSERT_AT_BOL, -1);
 		s++;
 		break;
 	case '$':
-		/* '$' is special everywhere in EREs, and in the end of the string in BREs. */
-		if (!ere && s[1])
+		/* '$' is special everywhere in EREs, and at the end of a BRE subexpression. */
+		if (!ere && s[1] && (s[1]!='\\'|| (s[2]!=')' && s[2]!='|')))
 			goto parse_literal;
 		node = tre_ast_new_literal(ctx->mem, ASSERTION, ASSERT_AT_EOL, -1);
 		s++;
@@ -944,7 +944,7 @@ static reg_errcode_t tre_parse(tre_parse_ctx_t *ctx)
 {
 	tre_ast_node_t *nbranch=0, *nunion=0;
 	int ere = ctx->cflags & REG_EXTENDED;
-	const char *s = ctx->re;
+	const char *s = ctx->start;
 	int subid = 0;
 	int depth = 0;
 	reg_errcode_t err;
@@ -962,6 +962,7 @@ static reg_errcode_t tre_parse(tre_parse_ctx_t *ctx)
 				s++;
 			depth++;
 			nbranch = nunion = 0;
+			ctx->start = s;
 			continue;
 		}
 		if ((!ere && *s == '\\' && s[1] == ')') ||
@@ -994,8 +995,8 @@ static reg_errcode_t tre_parse(tre_parse_ctx_t *ctx)
 			if (*s=='\\')
 				s++;
 
-			/* handle ^* at the start of a complete BRE. */
-			if (!ere && s==ctx->re+1 && s[-1]=='^')
+			/* handle ^* at the start of a BRE. */
+			if (!ere && s==ctx->start+1 && s[-1]=='^')
 				break;
 
 			/* extension: multiple consecutive *+?{,} is unspecified,
@@ -1038,8 +1039,10 @@ static reg_errcode_t tre_parse(tre_parse_ctx_t *ctx)
 
 			if (c == '\\' && s[1] == '|') {
 				s+=2;
+				ctx->start = s;
 			} else if (c == '|') {
 				s++;
+				ctx->start = s;
 			} else {
 				if (c == '\\') {
 					if (!depth) return REG_EPAREN;
@@ -2705,7 +2708,7 @@ regcomp(regex_t *restrict preg, const char *restrict regex, int cflags)
   memset(&parse_ctx, 0, sizeof(parse_ctx));
   parse_ctx.mem = mem;
   parse_ctx.stack = stack;
-  parse_ctx.re = regex;
+  parse_ctx.start = regex;
   parse_ctx.cflags = cflags;
   parse_ctx.max_backref = -1;
   errcode = tre_parse(&parse_ctx);
-- 
cgit v1.2.1