[Bug #19902] Update the coderange regarding the changed region

3 files changed, 66 insertions, 0 deletions

diff --git a/ext/-test-/string/set_len.c b/ext/-test-/string/set_len.c
index 219cea404c..049da2cdb5 100644
--- a/ext/-test-/string/set_len.c
+++ b/ext/-test-/string/set_len.c
@@ -7,8 +7,18 @@ bug_str_set_len(VALUE str, VALUE len)
     return str;
 }
 
+static VALUE
+bug_str_append(VALUE str, VALUE addendum)
+{
+    StringValue(addendum);
+    rb_str_modify_expand(str, RSTRING_LEN(addendum));
+    memcpy(RSTRING_END(str), RSTRING_PTR(addendum), RSTRING_LEN(addendum));
+    return str;
+}
+
 void
 Init_string_set_len(VALUE klass)
 {
     rb_define_method(klass, "set_len", bug_str_set_len, 1);
+    rb_define_method(klass, "append", bug_str_append, 1);
 }
diff --git a/string.c b/string.c
index deeed4a12a..18d03dc72e 100644
--- a/string.c
+++ b/string.c
@@ -2985,6 +2985,33 @@ rb_str_set_len(VALUE str, long len)
     if (len > (capa = (long)str_capacity(str, termlen)) || len < 0) {
         rb_bug("probable buffer overflow: %ld for %ld", len, capa);
     }
+
+    int cr = ENC_CODERANGE(str);
+    if (cr == ENC_CODERANGE_UNKNOWN) {
+        /* Leave unknown. */
+    }
+    else if (len > RSTRING_LEN(str)) {
+        if (ENC_CODERANGE_CLEAN_P(cr)) {
+            /* Update the coderange regarding the extended part. */
+            const char *const prev_end = RSTRING_END(str);
+            const char *const new_end = RSTRING_PTR(str) + len;
+            rb_encoding *enc = rb_enc_get(str);
+            rb_str_coderange_scan_restartable(prev_end, new_end, enc, &cr);
+            ENC_CODERANGE_SET(str, cr);
+        }
+        else if (cr == ENC_CODERANGE_BROKEN) {
+            /* May be valid now, by appended part. */
+            ENC_CODERANGE_SET(str, ENC_CODERANGE_UNKNOWN);
+        }
+    }
+    else if (len < RSTRING_LEN(str)) {
+        if (cr != ENC_CODERANGE_7BIT) {
+            /* ASCII-only string is keeping after truncated.  Valid
+             * and broken may be invalid or valid, leave unknown. */
+            ENC_CODERANGE_SET(str, ENC_CODERANGE_UNKNOWN);
+        }
+    }
+
     STR_SET_LEN(str, len);
     TERM_FILL(&RSTRING_PTR(str)[len], termlen);
 }
diff --git a/test/-ext-/string/test_set_len.rb b/test/-ext-/string/test_set_len.rb
index 67ba961194..e3eff75d9b 100644
--- a/test/-ext-/string/test_set_len.rb
+++ b/test/-ext-/string/test_set_len.rb
@@ -34,4 +34,33 @@ class Test_StrSetLen < Test::Unit::TestCase
     assert_equal 128, Bug::String.capacity(str)
     assert_equal 127, str.set_len(127).bytesize, bug12757
   end
+
+  def test_coderange_after_append
+    u = -"\u3042"
+    str = Bug::String.new(encoding: Encoding::UTF_8)
+    bsize = u.bytesize
+    str.append(u)
+    assert_equal 0, str.bytesize
+    str.set_len(bsize)
+    assert_equal bsize, str.bytesize
+    assert_predicate str, :valid_encoding?
+    assert_not_predicate str, :ascii_only?
+    assert_equal u, str
+  end
+
+  def test_coderange_after_trunc
+    u = -"\u3042"
+    bsize = u.bytesize
+    str = Bug::String.new(u)
+    str.set_len(bsize - 1)
+    assert_equal bsize - 1, str.bytesize
+    assert_not_predicate str, :valid_encoding?
+    assert_not_predicate str, :ascii_only?
+    str.append(u.byteslice(-1))
+    str.set_len(bsize)
+    assert_equal bsize, str.bytesize
+    assert_predicate str, :valid_encoding?
+    assert_not_predicate str, :ascii_only?
+    assert_equal u, str
+  end
 end