`Date._iso8601` regression following the 3.2.1 security fix

[casperisfine]

Previous versions

>> Date._iso8601(nil)
=> {}

3.2.1

>> Date._iso8601(nil)
(irb):2:in `_iso8601': no implicit conversion of nil into String (TypeError)

[hsbt]

And I also faced the regression of CVE-2021-41817 with ruby/spec.

ruby/ruby@17e64cc

[casperisfine]

Oh, good one, I didn't check symbols in my fix.

[casperisfine]

Yep:

Error: test__iso8601(TestDateParse): TypeError: no implicit conversion of Symbol into String
/Users/byroot/src/github.com/casperisfine/date/test/date/test_date_parse.rb:855:in `_iso8601'
/Users/byroot/src/github.com/casperisfine/date/test/date/test_date_parse.rb:855:in `test__iso8601'
     852:     h = Date._iso8601(nil)
     853:     assert_equal({}, h)
     854: 
  => 855:     h = Date._iso8601('01-02-03T04:05:06Z'.to_sym)
     856:     assert_equal([2001, 2, 3, 4, 5, 6, 0],
     857: 		 h.values_at(:year, :mon, :mday, :hour, :min, :sec, :offset))
     858:   end

@hsbt maybe it's best to report this as a separate issue?

[hsbt]

@casperisfine Thanks for your confirmation. It's good to separate from this issue or #40 .

[casperisfine]

I pushed the fix for symbols in #40

[Vasfed]

Looks like this also affectsDateTime.rfc2822(nil), previously it raised ArgumentError, but in 2.0.1 it raises no implicit conversion of nil into String (TypeError), this breaks some applications and gems like faraday.

Looking forward to a new release with #40 merged