Fix pattern matching on anchorsHEADmaster

The previous code would find all matches by searching for a match,
chopping it off of the string view, and then matching again in a loop.
This caused bugs with anchors such as ^, because x/^a/ would
match *every* instance of ‘a’ instead of just ‘a’s at the start of the
matching context.

This PR switches the matching engine to use the offset parameter of
pcre2_jit_match() to correctly handle anchors.

1 files changed, 43 insertions, 25 deletions

diff --git a/src/work.c b/src/work.c
index 7013726..3a138fb 100644
--- a/src/work.c
+++ b/src/work.c
@@ -228,10 +228,9 @@ DEFINE_OPERATOR(h)
 
 	pcre2_match_data *md =
 		pcre2_match_data_create_from_pattern(ops[opi].re, nullptr);
-	u8view_t sv_save = sv;
 	ptrdiff_t origlen = array_len(*hl);
-	for (;;) {
-		int n = pcre2_jit_match(ops[opi].re, sv.p, sv.len, 0,
+	for (ptrdiff_t off = 0;;) {
+		int n = pcre2_jit_match(ops[opi].re, sv.p, sv.len, off,
 		                        PCRE2_NOTEMPTY, md, nullptr);
 		if (n == PCRE2_ERROR_NOMATCH)
 			break;
@@ -240,10 +239,10 @@ DEFINE_OPERATOR(h)
 
 		size_t *ov = pcre2_get_ovector_pointer(md);
 		array_push(hl, ((u8view_t){sv.p + ov[0], ov[1] - ov[0]}));
-		VSHFT(&sv, ov[1]);
+		off = ov[1];
 	}
 	pcre2_match_data_free(md);
-	operator_dispatch(opi + 1, sv_save, hl);
+	operator_dispatch(opi + 1, sv, hl);
 	array_hdr(*hl)->len = origlen;
 }
 
@@ -256,22 +255,29 @@ DEFINE_OPERATOR(H)
 
 	pcre2_match_data *md =
 		pcre2_match_data_create_from_pattern(ops[opi].re, nullptr);
-	u8view_t sv_save = sv;
-	ptrdiff_t origlen = array_len(*hl);
-	for (;;) {
-		int n = pcre2_jit_match(ops[opi].re, sv.p, sv.len, 0, PCRE2_NOTEMPTY,
-		                        md, nullptr);
+	ptrdiff_t prvend = 0, origlen = array_len(*hl);
+
+	for (ptrdiff_t off = 0;;) {
+		int n = pcre2_jit_match(ops[opi].re, sv.p, sv.len, off, 0, md, nullptr);
 		if (n == PCRE2_ERROR_NOMATCH)
 			break;
 		if (n < 0)
 			pcre2_bitch_and_die(n, "failed to match regex");
 
 		size_t *ov = pcre2_get_ovector_pointer(md);
-		array_push(hl, ((u8view_t){sv.p, ov[0]}));
-		VSHFT(&sv, ov[1]);
+		if (ov[0] - prvend != 0)
+			array_push(hl, ((u8view_t){sv.p + prvend, ov[0] - prvend}));
+
+		prvend = off = ov[1];
+		if (ov[0] == ov[1])
+			off++;
 	}
+
+	if (prvend < sv.len)
+		array_push(hl, ((u8view_t){sv.p + prvend, sv.len - prvend}));
+
 	pcre2_match_data_free(md);
-	operator_dispatch(opi + 1, sv_save, hl);
+	operator_dispatch(opi + 1, sv, hl);
 	array_hdr(*hl)->len = origlen;
 }
 
@@ -279,9 +285,9 @@ DEFINE_OPERATOR(x)
 {
 	pcre2_match_data *md =
 		pcre2_match_data_create_from_pattern(ops[opi].re, nullptr);
-	for (;;) {
-		int n = pcre2_jit_match(ops[opi].re, sv.p, sv.len, 0, PCRE2_NOTEMPTY,
-		                        md, nullptr);
+
+	for (ptrdiff_t off = 0;;) {
+		int n = pcre2_jit_match(ops[opi].re, sv.p, sv.len, off, 0, md, nullptr);
 		if (n == PCRE2_ERROR_NOMATCH)
 			break;
 		if (n < 0)
@@ -289,7 +295,9 @@ DEFINE_OPERATOR(x)
 
 		size_t *ov = pcre2_get_ovector_pointer(md);
 		operator_dispatch(opi + 1, (u8view_t){sv.p + ov[0], ov[1] - ov[0]}, hl);
-		VSHFT(&sv, ov[1]);
+		off = ov[1];
+		if (ov[0] == ov[1])
+			off++;
 	}
 	pcre2_match_data_free(md);
 }
@@ -298,21 +306,31 @@ DEFINE_OPERATOR(X)
 {
 	pcre2_match_data *md =
 		pcre2_match_data_create_from_pattern(ops[opi].re, nullptr);
-	for (;;) {
-		int n = pcre2_jit_match(ops[opi].re, sv.p, sv.len, 0, PCRE2_NOTEMPTY,
-		                        md, nullptr);
+
+	ptrdiff_t prvend = 0;
+
+	for (ptrdiff_t off = 0;;) {
+		int n = pcre2_jit_match(ops[opi].re, sv.p, sv.len, off, 0, md, nullptr);
 		if (n == PCRE2_ERROR_NOMATCH)
 			break;
 		if (n < 0)
 			pcre2_bitch_and_die(n, "failed to match regex");
 
 		size_t *ov = pcre2_get_ovector_pointer(md);
-		if (ov[0] != 0)
-			operator_dispatch(opi + 1, (u8view_t){sv.p, ov[0]}, hl);
-		VSHFT(&sv, ov[1]);
+		if (ov[0] > (size_t)prvend) {
+			u8view_t sub = {sv.p + prvend, ov[0] - prvend};
+			operator_dispatch(opi + 1, sub, hl);
+		}
+
+		prvend = off = ov[1];
+		if (ov[0] == ov[1])
+			off++;
+	}
+
+	if (prvend < sv.len) {
+		u8view_t sub = {sv.p + prvend, sv.len - prvend};
+		operator_dispatch(opi + 1, sub, hl);
 	}
-	if (sv.len != 0)
-		operator_dispatch(opi + 1, sv, hl);
 	pcre2_match_data_free(md);
 }