Preserve input encoding when parsing time zones

Pass the input encoding through `date__strptime` so non-ASCII trailing
data is treated as a parse failure instead of raising an encoding error.

Fixes #179.

Author: nobu

ext/date/date_core.c

@@ -4515,7 +4515,8 @@ d_new_by_frags(VALUE klass, VALUE hash, VALUE sg)
 }
 
 VALUE date__strptime(const char *str, size_t slen,
-		     const char *fmt, size_t flen, VALUE hash);
+		     const char *fmt, size_t flen,
+		     VALUE hash, rb_encoding *enc);
 
 static VALUE
 date_s__strptime_internal(int argc, VALUE *argv, VALUE klass,
@@ -4524,6 +4525,7 @@ date_s__strptime_internal(int argc, VALUE *argv, VALUE klass,
     VALUE vstr, vfmt, hash;
     const char *str, *fmt;
     size_t slen, flen;
+    rb_encoding *enc;
 
     rb_scan_args(argc, argv, "11", &vstr, &vfmt);
 
@@ -4537,33 +4539,18 @@ date_s__strptime_internal(int argc, VALUE *argv, VALUE klass,
     if (argc < 2) {
 	fmt = default_fmt;
 	flen = strlen(default_fmt);
+	enc = rb_enc_get(vstr);
     }
     else {
 	if (!rb_enc_str_asciicompat_p(vfmt))
 	    rb_raise(rb_eArgError,
 		     "format should have ASCII compatible encoding");
+	enc = rb_enc_check(vstr, vfmt);
 	fmt = RSTRING_PTR(vfmt);
 	flen = RSTRING_LEN(vfmt);
     }
     hash = rb_hash_new();
-    if (NIL_P(date__strptime(str, slen, fmt, flen, hash)))
-	return Qnil;
-
-    {
-	VALUE zone = ref_hash("zone");
-	VALUE left = ref_hash("leftover");
-
-	if (!NIL_P(zone)) {
-	    rb_enc_copy(zone, vstr);
-	    set_hash("zone", zone);
-	}
-	if (!NIL_P(left)) {
-	    rb_enc_copy(left, vstr);
-	    set_hash("leftover", left);
-	}
-    }
-
-    return hash;
+    return date__strptime(str, slen, fmt, flen, hash, enc);
 }
 
 /*

ext/date/date_strptime.c

@@ -41,7 +41,6 @@ static const int ABBREVIATED_MONTH_NAME_LENGTH = 3;
 
 #define f_match(r,s) rb_funcall(r, rb_intern("match"), 1, s)
 #define f_aref(o,i) rb_funcall(o, rb_intern("[]"), 1, i)
-#define f_end(o,i) rb_funcall(o, rb_intern("end"), 1, i)
 
 #define issign(c) ((c) == '-' || (c) == '+')
 
@@ -144,7 +143,7 @@ valid_range_p(VALUE v, int a, int b)
 do { \
     size_t l; \
     l = date__strptime_internal(&str[si], slen - si, \
-				fmt, sizeof fmt - 1, hash); \
+				fmt, sizeof fmt - 1, hash, enc); \
     if (fail_p()) \
 	return 0; \
     si += l; \
@@ -160,7 +159,8 @@ head_match_p(size_t len, const char *name, const char *str, size_t slen, size_t
 
 static size_t
 date__strptime_internal(const char *str, size_t slen,
-			const char *fmt, size_t flen, VALUE hash)
+			const char *fmt, size_t flen,
+			VALUE hash, rb_encoding *enc)
 {
     size_t si, fi;
     int c;
@@ -597,15 +597,14 @@ date__strptime_internal(const char *str, size_t slen,
 
 		    b = rb_backref_get();
 		    rb_match_busy(b);
-		    m = f_match(pat, rb_usascii_str_new(&str[si], slen - si));
+		    m = f_match(pat, rb_enc_str_new(&str[si], slen - si, enc));
 
 		    if (!NIL_P(m)) {
-			VALUE s, l, o;
+			VALUE s, o;
 
 			s = rb_reg_nth_match(1, m);
-			l = f_end(m, INT2FIX(0));
 			o = date_zone_to_diff(s);
-			si += NUM2LONG(l);
+			si += RSTRING_LEN(s);
 			set_hash("zone", s);
 			set_hash("offset", o);
 			rb_backref_set(b);
@@ -654,20 +653,21 @@ date__strptime_internal(const char *str, size_t slen,
 
 VALUE
 date__strptime(const char *str, size_t slen,
-	       const char *fmt, size_t flen, VALUE hash)
+	       const char *fmt, size_t flen,
+	       VALUE hash, rb_encoding *enc)
 {
     size_t si;
     VALUE cent, merid;
 
-    si = date__strptime_internal(str, slen, fmt, flen, hash);
+    si = date__strptime_internal(str, slen, fmt, flen, hash, enc);
 
     if (fail_p())
 	return Qnil;
 
     if (slen > si) {
 	VALUE s;
 
-	s = rb_usascii_str_new(&str[si], slen - si);
+	s = rb_enc_str_new(&str[si], slen - si, enc);
 	set_hash("leftover", s);
     }
 

test/date/test_date_strptime.rb

@@ -533,4 +533,11 @@ def to_str
     assert_equal(6, d[:mon])
     assert_equal(1, d[:mday])
   end
+
+  def test_nonascii_string
+    nonalpha = "\u{2600 fe0f}"
+    s = "2011-10-05T22:26:12#{nonalpha}"
+
+    assert_nil(DateTime._strptime(s))
+  end
 end