Model attribute equality step definition doesn't handle string values correctly

I'm trying to use the step defined here. However, it's capturing the double quotes that surround the string, leading to a failure. Here's my cucumber output:

  Scenario: Edit a geography with invalid values                   # features/manage_geographies.feature:36
    Given a geography exists with name: "Los Angeles", state: "CA" # features/step_definitions/pickle_steps.rb:4
    And I am logged in as an admin                                 # features/step_definitions/authentication_steps.rb:13
    And I am on the homepage                                       # features/step_definitions/web_steps.rb:18
    When I follow these links: "Admin", "Geographies", "Edit"      # features/step_definitions/additional_web_steps.rb:5
    And I fill in "Name" with ""                                   # features/step_definitions/web_steps.rb:38
    And I press "Save"                                             # features/step_definitions/web_steps.rb:26
    Then I should see the record validation error message          # features/step_definitions/additional_web_steps.rb:13
    And that geography's name should be "Los Angeles"              # features/step_definitions/pickle_steps.rb:75
      
      expected "\"Los Angeles\""
           got "Los Angeles"
      
      (compared using eql?)
       (Spec::Expectations::ExpectationNotMetError)
      ./features/step_definitions/pickle_fixes.rb:16:in `send'
      ./features/step_definitions/pickle_fixes.rb:16:in `/^((?:(?:I|myself|me|my)|(?:(?:a|an|another|the|that) )?(?:(?:(?:(?:first|last|(?:\d+(?:st|nd|rd|th))) )?(?:address|slug|promotion[_ ]view|business|encoded[_ ]email|active[_ ]admin|promotion|customer[_ ]information|merchant[_ ]payment|promotion[_ ]transaction|delayed[_ ]job|admin|active[_ ]user|voucher|business[_ ]link|promotion[_ ]referrer|geography|promotion[_ ]user|merchant|business[_ ]category|user))|(?:(?:address|slug|promotion[_ ]view|business|encoded[_ ]email|active[_ ]admin|promotion|customer[_ ]information|merchant[_ ]payment|promotion[_ ]transaction|delayed[_ ]job|admin|active[_ ]user|voucher|business[_ ]link|promotion[_ ]referrer|geography|promotion[_ ]user|merchant|business[_ ]category|user)(?::? "(?:[^\"]|\.)*")))))'s (\w+) (should(?: not)?) be ((?:"(?:[^\"]|\.)*"|nil|true|false|[+-]?\d+(?:\.\d+)?))$/'
      features/manage_geographies.feature:44:in `And that geography's name should be "Los Angeles"'

I'm willing to take a stab at fixing this and submit my fix, but I wanted to get feedback about the best way to go about this. I see two ways to fix this:

1. By fixing #capture_value so that it doesn't include the quotes in the capture.
   
2. By stripping the quotes from the value before checking equality, using something like expected = expected.gsub(/^\"/, '').gsub(/\"$/, '').

I like the 1st solution better overall, but it's complicated by how pickle implements the #capture_value. The problem is that the basic regex is defined by #match_value, which uses quotes to tell that it's a string. Then this dynamic code surrounds it with the capturing parenthesis when it defines #capture_value. I don't see a way to fix the regular expression to not capture the double quotes unless we define the entire #capture_value method rather than defining it dynamically.

Any thoughts on how this should be fixed?