Fix Regexp#to_s for GC compaction

peterzhu2118 · peterzhu2118 · commit fadda88903d9 · 2023-12-23T16:52:05.000-05:00
The test fails when RGENGC_CHECK_MODE is turned on:

    TestRegexp#test_to_s_under_gc_compact_stress = 13.46 s
    1) Failure:
    TestRegexp#test_to_s_under_gc_compact_stress [test/ruby/test_regexp.rb:81]:
    &lt;"(?-mix:abcd\u3042)"&gt; expected but was
    &lt;"(?-mix:\u5C78\u3030\u5C78\u3030\u5C78\u3030\u5C78\u3030\u5C78\u3030)"&gt;.
diff --git a/re.c b/re.c
@@ -565,8 +565,6 @@ rb_reg_str_with_term(VALUE re, int term)
 {
     int options, opt;
     const int embeddable = ONIG_OPTION_MULTILINE|ONIG_OPTION_IGNORECASE|ONIG_OPTION_EXTEND;
-    long len;
-    const UChar* ptr;
     VALUE str = rb_str_buf_new2("(?");
     char optbuf[OPTBUF_SIZE + 1]; /* for '-' */
     rb_encoding *enc = rb_enc_get(re);
@@ -575,8 +573,9 @@ rb_reg_str_with_term(VALUE re, int term)
 
     rb_enc_copy(str, re);
     options = RREGEXP_PTR(re)->options;
-    ptr = (UChar*)RREGEXP_SRC_PTR(re);
-    len = RREGEXP_SRC_LEN(re);
+    VALUE src_str = RREGEXP_SRC(re);
+    const UChar *ptr = (UChar *)RSTRING_PTR(src_str);
+    long len = RSTRING_LEN(src_str);
   again:
     if (len >= 4 && ptr[0] == '(' && ptr[1] == '?') {
         int err = 1;
@@ -666,6 +665,8 @@ rb_reg_str_with_term(VALUE re, int term)
     }
     rb_enc_copy(str, re);
 
+    RB_GC_GUARD(src_str);
+
     return str;
 }
 
diff --git a/test/ruby/test_regexp.rb b/test/ruby/test_regexp.rb
@@ -72,6 +72,18 @@ def test_to_s
     end
   end
 
+  def test_to_s_under_gc_compact_stress
+    EnvUtil.under_gc_compact_stress do
+      str = "abcd\u3042"
+      [:UTF_16BE, :UTF_16LE, :UTF_32BE, :UTF_32LE].each do |es|
+        enc = Encoding.const_get(es)
+        rs = Regexp.new(str.encode(enc)).to_s
+        assert_equal("(?-mix:abcd\u3042)".encode(enc), rs)
+        assert_equal(enc, rs.encoding)
+      end
+    end
+  end
+
   def test_to_s_extended_subexp
     re = /#\g#{"\n"}/x
     re = /#{re}/
