Email: restrict domain extensions to letters when spelling out

Summary: We would parse things like "tonight at 6.40".

Reviewed By: blandinw

Differential Revision: D6066926

fbshipit-source-id: d18a8c6

Duckling/Email/Corpus.hs

@@ -13,8 +13,8 @@ module Duckling.Email.Corpus
   , negativeCorpus
   ) where
 
-import Prelude
 import Data.String
+import Prelude
 
 import Duckling.Email.Types
 import Duckling.Testing.Types

Duckling/Email/EN/Corpus.hs

@@ -10,6 +10,7 @@
 
 module Duckling.Email.EN.Corpus
   ( corpus
+  , negativeCorpus
   ) where
 
 import Data.String
@@ -18,6 +19,15 @@ import Prelude
 import Duckling.Email.Types
 import Duckling.Testing.Types
 
+negativeCorpus :: NegativeCorpus
+negativeCorpus = (testContext, examples)
+  where
+    examples =
+      [ "fitness at 6.40"
+      , "class at 12.00"
+      , "tonight at 9.15"
+      ]
+
 corpus :: Corpus
 corpus = (testContext, allExamples)
 

Duckling/Email/EN/Rules.hs

@@ -13,20 +13,20 @@ module Duckling.Email.EN.Rules
   ( rules ) where
 
 import Data.String
-import qualified Data.Text as Text
 import Prelude
+import qualified Data.Text as Text
 
 import Duckling.Dimensions.Types
 import Duckling.Email.Types (EmailData (..))
-import qualified Duckling.Email.Types as TEmail
 import Duckling.Regex.Types
 import Duckling.Types
+import qualified Duckling.Email.Types as TEmail
 
 ruleEmailSpelledOut :: Rule
 ruleEmailSpelledOut = Rule
   { name = "email spelled out"
   , pattern =
-    [ regex "([\\w\\._+-]+) at ([\\w_-]+(\\.[\\w_-]+)+)"
+    [ regex "([\\w\\._+-]+) at ([\\w_-]+(\\.[a-zA-Z]+)+)"
     ]
   , prod = \xs -> case xs of
       (Token RegexMatch (GroupMatch (m1:m2:_)):_) ->

tests/Duckling/Email/EN/Tests.hs

@@ -20,4 +20,5 @@ import Duckling.Testing.Asserts
 tests :: TestTree
 tests = testGroup "Email Tests"
   [ makeCorpusTest [This Email] corpus
+  , makeNegativeCorpusTest [This Email] negativeCorpus
   ]