Unbalanced `^` and `$` anchors in Oracle regexp

Question

Why in some cases number of start-of-string anchors is not equal to number of end-of-string anchors?

regexp_count(chr(10)||'A'||chr(10)||'B', '^') = 2
regexp_count(chr(10)||'A'||chr(10)||'B', '$') = 1

---

UPD :
In multi-line mode disbalance has opposite direction:

regexp_count(chr(10)||'A'||chr(10)||'B', '^', 1, 'mn') = 1 
regexp_count(chr(10)||'A'||chr(10)||'B', '$', 1, 'mn') = 3

fiddle

Answer 1

After investigation Oracle Support concluded that this issue is a bug.

Bug 16707134 : REGEXP_COUNT PATTERN '^' AND NEW LINE UNEXPECTED BEHAVIOR