Regex: Fix handling of all unicode codepoint as start chars

1 files changed, 7 insertions, 0 deletions

diff --git a/src/regex_impl.cc b/src/regex_impl.cc
index 8b6626fb..993119b9 100644
--- a/src/regex_impl.cc
+++ b/src/regex_impl.cc
@@ -858,6 +858,8 @@ private:
                     else
                         start_chars.map[node->value] = true;
                 }
+                else
+                    start_chars.accept_other = true;
                 return node->quantifier.allows_none();
             case ParsedRegex::AnyChar:
                 for (auto& b : start_chars.map)
@@ -1333,6 +1335,11 @@ auto test_regex = UnitTest{[]{
     }
 
     {
+        TestVM<> vm{R"(д)"};
+        kak_assert(vm.exec("д", RegexExecFlags::Search));
+    }
+
+    {
         TestVM<> vm{R"(\0\x0A\u260e\u260F)"};
         const char str[] = "\0\n☎☏"; // work around the null byte in the literal
         kak_assert(vm.exec({str, str + sizeof(str)-1}));