make regex line ends more consistent across chunks

3 files changed, 27 insertions, 5 deletions

diff --git a/main.c b/main.c
index 71c7938..5df9e37 100644
--- a/main.c
+++ b/main.c
@@ -681,9 +681,10 @@ int search_next_regex(TxtLoc l, Str src, TxtLoc *out) {
 		return 0;
 	}
 	int match_found = 0;
-search_from_start:
+search_from_start:;
 	TxtLoc t = l;
 	re_search_start(&s, &re, &e.scratch);
+	if (at_start(t) || txt_chr(cprev(t)) == '\n') re_search_first_chunk(&s);
 	while (!at_end(t)) {
 		Str chnk = txt_next_chunk(&t);
 		re_search_chunk(&s, chnk.s, chnk.n);
@@ -720,6 +721,7 @@ search_from_end:
 		Arena a = e.scratch;
 		re_search_start(&s, &re, &e.scratch);
 		TxtLoc t = l;
+		if (at_start(t) || txt_chr(cprev(t)) == '\n') re_search_first_chunk(&s);
 		while (!at_end(t)) {
 			TxtLoc p = t;
 			Str chnk = txt_next_chunk(&t);
diff --git a/regex.c b/regex.c
index 6fe4809..1c42126 100644
--- a/regex.c
+++ b/regex.c
@@ -824,7 +824,7 @@ static inline isize re_search_try_match(ReSearch *s, size_t i, size_t n) {
 	}
 	isize found_i = -1;
 	ReThreadStepFlags f = 0;
-	if (i + start == 0) f = RE_THREAD_AT_START;
+	if (i == 0 && (s->flags & (RE_SEARCH_FIRST_CHUNK | RE_SEARCH_WAS_NEWLINE))) f = RE_THREAD_AT_START;
 
 	for (; i < n && s->tcur.n > 0; i++) {
 		s->c = (unsigned char)s->buf[i];
@@ -854,7 +854,7 @@ void re_search_start(ReSearch *s, RegEx *re, Arena *a) {
 		.a = a,
 		.re = re,
 		.grp = new_arr(a, ReSpan, re->groups),
-		.flags = RE_SEARCH_FIRST_CHUNK
+		.flags = 0
 	};
 	re_threadlist_alloc(&s->tcur, re->op.n, a);
 	re_threadlist_alloc(&s->tnext, re->op.n, a);
@@ -872,6 +872,9 @@ void re_search_last_chunk(ReSearch *s) {
 	s->flags |= RE_SEARCH_LAST_CHUNK;
 }
 
+void re_search_first_chunk(ReSearch *s) {
+	s->flags |= RE_SEARCH_FIRST_CHUNK;
+}
 
 /* searching */
 
@@ -879,6 +882,18 @@ void re_search_last_chunk(ReSearch *s) {
  * check s->flags to find if the match is done yet
  */
 
+static inline void re_search_chunk_fin(ReSearch *s) {
+	u32 n = s->buf_len;
+	s->buf_idx = n;
+	if (n > 0) {
+		if (s->buf[n-1] == '\n') {
+			s->flags |= RE_SEARCH_WAS_NEWLINE;
+		} else {
+			s->flags &= ~RE_SEARCH_WAS_NEWLINE;
+		}
+	}
+}
+
 int re_search_match_at_start(ReSearch *s, ReMatch *m) {
 	size_t i = s->buf_idx;
 	size_t n = s->buf_len;
@@ -901,7 +916,7 @@ int re_search_match_at_start(ReSearch *s, ReMatch *m) {
 			}
 		}
 	}
-	s->buf_idx = s->buf_len;
+	re_search_chunk_fin(s);
 	return 0;
 }
 
@@ -917,7 +932,7 @@ int re_search_match(ReSearch *s, ReMatch *m) {
 		s->buf_idx = i++;
 		if (re_search_match_at_start(s, m)) return 1;
 	}
-	s->buf_idx = n;
+	re_search_chunk_fin(s);
 	return 0;
 }
 
@@ -929,6 +944,7 @@ ReMatchList re_match_all(RegEx *re, Str s, Arena *a) {
 	ReMatch m;
 	re_search_start(&sr, re, a);
 	re_search_chunk(&sr, s.s, s.n);
+	re_search_first_chunk(&sr);
 	re_search_last_chunk(&sr);
 	while (re_search_match(&sr, &m)) {
 		ReSpan *grp = new_arr(a, ReSpan, re->groups);
@@ -943,6 +959,7 @@ int re_match_full(RegEx *re, Str s, Arena *a) {
 	ReMatch m;
 	re_search_start(&sr, re, a);
 	re_search_chunk(&sr, s.s, s.n);
+	re_search_first_chunk(&sr);
 	re_search_last_chunk(&sr);
 	return re_search_match(&sr, &m) && m.extent.start == 0 && m.extent.len == s.n;
 }
@@ -951,6 +968,7 @@ int re_match(RegEx *re, Str s, ReMatch *out, Arena *a) {
 	ReSearch sr = { 0 };
 	re_search_start(&sr, re, a);
 	re_search_chunk(&sr, s.s, s.n);
+	re_search_first_chunk(&sr);
 	re_search_last_chunk(&sr);
 	return re_search_match(&sr, out);
 }
diff --git a/regex.h b/regex.h
index c21c708..8f95319 100644
--- a/regex.h
+++ b/regex.h
@@ -92,6 +92,7 @@ typedef enum {
 	RE_SEARCH_FIRST_CHUNK = 1,
 	RE_SEARCH_LAST_CHUNK  = 2,
 	RE_SEARCH_MID_MATCH   = 4,
+	RE_SEARCH_WAS_NEWLINE = 8,
 } ReSearchFlags;
 
 typedef struct {
@@ -124,6 +125,7 @@ const char *re_comp_strerror(ReCompErr err);
 
 void re_search_start(ReSearch *s, RegEx *re, Arena *a);
 void re_search_chunk(ReSearch *s, const char *buf, size_t n);
+void re_search_first_chunk(ReSearch *s);
 void re_search_last_chunk(ReSearch *s);
 int re_search_match(ReSearch *s, ReMatch *m);
 int re_search_match_at_start(ReSearch *s, ReMatch *m);