Remove ?nrg_redirect in URL canonicalization (#1277)

energy.gov has apparently started tacking on `?nrg_redirect=123456` to redirects. The value of the query param seems to indicate what URL you were redirected from. This removes that from canonical URLs.

Author: Mr0grog

app/lib/surt/canonicalize.rb

@@ -55,6 +55,7 @@ module Surt::Canonicalize
   #   {param_to_remove: /regex or string to match value/},
   #   {param_to_remove1: /regex for value/, param_to_remove2: /regex for value/}
   # ]
+  # TODO: should rename this -- it's not really just session IDs anymore.
   QUERY_SESSION_IDS = [
     /^(.*)(?:jsessionid=[0-9a-zA-Z]{32})(?:&(.*))?$/i,
     /^(.*)(?:phpsessid=[0-9a-zA-Z]{32})(?:&(.*))?$/i,
@@ -68,7 +69,10 @@ module Surt::Canonicalize
     /^(.*)(?:utm_campaign=[^&]+)(?:&(.*))?$/i,
     /^(.*)(?:sms_ss=[^&]+)(?:&(.*))?$/i,
     /^(.*)(?:awesm=[^&]+)(?:&(.*))?$/i,
-    /^(.*)(?:xtor=[^&]+)(?:&(.*))?$/i
+    /^(.*)(?:xtor=[^&]+)(?:&(.*))?$/i,
+    # TODO: At the moment, we only know `nrg_redirect` to exist on energy.gov.
+    #  it would be nice to have a way to scope this by domain or hostname.
+    /^(.*)(?:nrg_redirect=\d+)(?:&(.*))?$/i
   ].freeze
 
   OCTAL_IP = /^(0[0-7]*)(\.[0-7]+)?(\.[0-7]+)?(\.[0-7]+)?$/

test/lib/surt/surt_test.rb

@@ -309,7 +309,7 @@ def assert_canonicalized(expected, url, message = nil, options = {})
     )
     assert_canonicalized(
       'http://example.com/x',
-      'http://example.com/x?sms_ss=abc&awesm=def&xtor=hij',
+      'http://example.com/x?sms_ss=abc&awesm=def&xtor=hij&nrg_redirect=267439',
       'It failed to remove assorted tracking query params'
     )
   end