summaryrefslogtreecommitdiff
path: root/parse.y
diff options
context:
space:
mode:
Diffstat (limited to 'parse.y')
-rw-r--r--parse.y14
1 files changed, 14 insertions, 0 deletions
diff --git a/parse.y b/parse.y
index 1d4598c3d7..8284603c0f 100644
--- a/parse.y
+++ b/parse.y
@@ -4842,9 +4842,23 @@ parser_str_new(const char *p, long n, rb_encoding *enc, int coderange)
static VALUE
parser_str_new2(const char *p, long n, rb_encoding *enc, int has8bit,int hasmb)
{
+ /*
+ * Set coderange bit flags based on the presence of 8-bit and
+ * multi-byte characters in the string
+ */
int coderange = ENC_CODERANGE_SINGLE;
if (hasmb) coderange = ENC_CODERANGE_MULTI;
else if (has8bit) coderange = ENC_CODERANGE_UNKNOWN;
+
+ /*
+ * If it is all single byte characters with the 8th bit clear,
+ * and if the specified encoding is ASCII-compatible, then this
+ * string is in the ASCII subset, and we just use the ASCII encoding
+ * instead.
+ */
+ if ((coderange == ENC_CODERANGE_SINGLE) && rb_enc_asciicompat(enc))
+ enc = rb_enc_default();
+
return parser_str_new(p, n, enc, coderange);
}