Added German+Icelandic spelled out Email

Summary: Pull Request resolved: https://github.com/facebook/duckling/pull/272

Reviewed By: chinmay87

Differential Revision: D12955290

Pulled By: patapizza

fbshipit-source-id: 671ee50c92c031762aa147dc6bd98f374693ba6b

Duckling/Email/DE/Corpus.hs

@@ -0,0 +1,65 @@
+-- Copyright (c) 2016-present, Facebook, Inc.
+-- All rights reserved.
+--
+-- This source code is licensed under the BSD-style license found in the
+-- LICENSE file in the root directory of this source tree. An additional grant
+-- of patent rights can be found in the PATENTS file in the same directory.
+
+
+{-# LANGUAGE OverloadedStrings #-}
+
+module Duckling.Email.DE.Corpus
+  ( corpus
+  , negativeCorpus
+  ) where
+
+import Data.String
+import Prelude
+import qualified Data.Text as Text
+
+import Duckling.Email.Types
+import Duckling.Locale
+import Duckling.Resolve
+import Duckling.Testing.Types
+
+context :: Context
+context = testContext {locale = makeLocale DE Nothing}
+
+negativeCorpus :: NegativeCorpus
+negativeCorpus = (context, testOptions, examples)
+  where
+    examples =
+      [ "fitness at 6.40"
+      , "class at 12.00"
+      , "tonight at 9.15"
+      , " dot 2@abci"
+      , "x@ dot x"
+      , "x@ x dot "
+      , "abc@x dot "
+      , Text.replicate 10000 "a at a"
+      ]
+
+corpus :: Corpus
+corpus = (context, testOptions, allExamples)
+
+allExamples :: [Example]
+allExamples = concat
+  [ examples (EmailData "alice@exAmple.io")
+             [ "alice at exAmple.io"
+             ]
+  , examples (EmailData "yo+yo@blah.org")
+             [ "yo+yo at blah.org"
+             ]
+  , examples (EmailData "1234+abc@x.net")
+             [ "1234+abc at x.net"
+             ]
+  , examples (EmailData "jean-jacques@stuff.co.uk")
+             [ "jean-jacques at stuff.co.uk"
+             ]
+  , examples (EmailData "asdf+ab.c@gmail.com")
+             [ "asdf+ab punkt c at gmail punkt com"
+             ]
+  , examples (EmailData "asdf.k@fb.com")
+             [ "asdf punkt k@fb punkt com"
+             ]
+  ]

Duckling/Email/DE/Rules.hs

@@ -0,0 +1,42 @@
+-- Copyright (c) 2016-present, Facebook, Inc.
+-- All rights reserved.
+--
+-- This source code is licensed under the BSD-style license found in the
+-- LICENSE file in the root directory of this source tree. An additional grant
+-- of patent rights can be found in the PATENTS file in the same directory.
+
+
+{-# LANGUAGE GADTs #-}
+{-# LANGUAGE OverloadedStrings #-}
+
+module Duckling.Email.DE.Rules
+  ( rules ) where
+
+import Data.String
+import Prelude
+import qualified Data.Text as Text
+
+import Duckling.Dimensions.Types
+import Duckling.Email.Types (EmailData (..))
+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_+-]+(?:(?: punkt |\\.)[\\w_+-]+){0,10})(?: at |@)([a-zA-Z]+(?:(?:\\.| punkt )[\\w_-]+){1,10})"
+    ]
+  , prod = \xs -> case xs of
+      (Token RegexMatch (GroupMatch (m1:m2:_)):_) ->
+        Just $ Token Email EmailData
+          { TEmail.value = Text.concat [replaceDots m1, "@", replaceDots m2] }
+        where replaceDots = Text.replace " punkt " "."
+      _ -> Nothing
+  }
+
+rules :: [Rule]
+rules =
+  [ ruleEmailSpelledOut
+  ]

Duckling/Email/IS/Corpus.hs

@@ -0,0 +1,40 @@
+-- Copyright (c) 2016-present, Facebook, Inc.
+-- All rights reserved.
+--
+-- This source code is licensed under the BSD-style license found in the
+-- LICENSE file in the root directory of this source tree. An additional grant
+-- of patent rights can be found in the PATENTS file in the same directory.
+
+
+{-# LANGUAGE OverloadedStrings #-}
+
+module Duckling.Email.IS.Corpus
+  ( corpus
+  ) where
+
+import Data.String
+import Prelude
+
+import Duckling.Email.Types
+import Duckling.Locale
+import Duckling.Resolve
+import Duckling.Testing.Types
+
+corpus :: Corpus
+corpus = (testContext {locale = makeLocale IS Nothing}, testOptions, allExamples)
+
+allExamples :: [Example]
+allExamples = concat
+  [ examples (EmailData "alice@exAmple.io")
+             [ "alice ingi exAmple.io"
+             ]
+  , examples (EmailData "yo+yo@blah.org")
+             [ "yo+yo ingi blah.org"
+             ]
+  , examples (EmailData "1234+abc@x.net")
+             [ "1234+abc ingi x.net"
+             ]
+  , examples (EmailData "jean-jacques@stuff.co.uk")
+             [ "jean-jacques ingi stuff.co.uk"
+             ]
+  ]

Duckling/Email/IS/Rules.hs

@@ -0,0 +1,40 @@
+-- Copyright (c) 2016-present, Facebook, Inc.
+-- All rights reserved.
+--
+-- This source code is licensed under the BSD-style license found in the
+-- LICENSE file in the root directory of this source tree. An additional grant
+-- of patent rights can be found in the PATENTS file in the same directory.
+
+
+{-# LANGUAGE GADTs #-}
+{-# LANGUAGE OverloadedStrings #-}
+
+module Duckling.Email.IS.Rules
+  ( rules ) where
+
+import Data.String
+import qualified Data.Text as Text
+import Prelude
+
+import Duckling.Dimensions.Types
+import Duckling.Email.Types (EmailData (..))
+import qualified Duckling.Email.Types as TEmail
+import Duckling.Regex.Types
+import Duckling.Types
+
+ruleEmailSpelledOut :: Rule
+ruleEmailSpelledOut = Rule
+  { name = "email spelled out"
+  , pattern =
+    [ regex "([\\w\\._+-]+) ingi ([\\w_-]+(\\.[\\w_-]+)+)"
+    ]
+  , prod = \xs -> case xs of
+      (Token RegexMatch (GroupMatch (m1:m2:_)):_) ->
+        Just $ Token Email EmailData {TEmail.value = Text.concat [m1, "@", m2]}
+      _ -> Nothing
+  }
+
+rules :: [Rule]
+rules =
+  [ ruleEmailSpelledOut
+  ]

Duckling/Rules/DE.hs

@@ -19,6 +19,7 @@ import Duckling.Dimensions.Types
 import Duckling.Locale
 import Duckling.Types
 import qualified Duckling.Duration.DE.Rules as Duration
+import qualified Duckling.Email.DE.Rules as Email
 import qualified Duckling.Ordinal.DE.Rules as Ordinal
 import qualified Duckling.Numeral.DE.Rules as Numeral
 import qualified Duckling.Time.DE.Rules as Time
@@ -35,7 +36,7 @@ langRules :: Some Dimension -> [Rule]
 langRules (This AmountOfMoney) = []
 langRules (This Distance) = []
 langRules (This Duration) = Duration.rules
-langRules (This Email) = []
+langRules (This Email) = Email.rules
 langRules (This Numeral) = Numeral.rules
 langRules (This Ordinal) = Ordinal.rules
 langRules (This PhoneNumber) = []

Duckling/Rules/IS.hs

@@ -17,6 +17,7 @@ module Duckling.Rules.IS
 import Duckling.Dimensions.Types
 import Duckling.Locale
 import Duckling.Types
+import qualified Duckling.Email.IS.Rules as Email
 import qualified Duckling.Numeral.IS.Rules as Numeral
 
 defaultRules :: Some Dimension -> [Rule]
@@ -30,7 +31,7 @@ langRules :: Some Dimension -> [Rule]
 langRules (This Distance) = []
 langRules (This Duration) = []
 langRules (This Numeral) = Numeral.rules
-langRules (This Email) = []
+langRules (This Email) = Email.rules
 langRules (This AmountOfMoney) = []
 langRules (This Ordinal) = []
 langRules (This PhoneNumber) = []

duckling.cabal

@@ -342,10 +342,14 @@ library
                      , Duckling.Duration.Types
 
                      -- Email
+                     , Duckling.Email.DE.Corpus
+                     , Duckling.Email.DE.Rules
                      , Duckling.Email.EN.Corpus
                      , Duckling.Email.EN.Rules
                      , Duckling.Email.FR.Corpus
                      , Duckling.Email.FR.Rules
+                     , Duckling.Email.IS.Corpus
+                     , Duckling.Email.IS.Rules
                      , Duckling.Email.IT.Corpus
                      , Duckling.Email.IT.Rules
                      , Duckling.Email.Corpus

tests/Duckling/Email/DE/Tests.hs

@@ -0,0 +1,24 @@
+-- Copyright (c) 2016-present, Facebook, Inc.
+-- All rights reserved.
+--
+-- This source code is licensed under the BSD-style license found in the
+-- LICENSE file in the root directory of this source tree. An additional grant
+-- of patent rights can be found in the PATENTS file in the same directory.
+
+
+module Duckling.Email.DE.Tests
+  ( tests
+  ) where
+
+import Data.String
+import Test.Tasty
+
+import Duckling.Dimensions.Types
+import Duckling.Email.DE.Corpus
+import Duckling.Testing.Asserts
+
+tests :: TestTree
+tests = testGroup "DE Tests"
+  [ makeCorpusTest [This Email] corpus
+  , makeNegativeCorpusTest [This Email] negativeCorpus
+  ]

tests/Duckling/Email/EN/Tests.hs

@@ -18,7 +18,7 @@ import Duckling.Email.EN.Corpus
 import Duckling.Testing.Asserts
 
 tests :: TestTree
-tests = testGroup "Email Tests"
+tests = testGroup "EN Tests"
   [ makeCorpusTest [This Email] corpus
   , makeNegativeCorpusTest [This Email] negativeCorpus
   ]

tests/Duckling/Email/FR/Tests.hs

@@ -18,6 +18,6 @@ import Duckling.Email.FR.Corpus
 import Duckling.Testing.Asserts
 
 tests :: TestTree
-tests = testGroup "Email Tests"
+tests = testGroup "FR Tests"
   [ makeCorpusTest [This Email] corpus
   ]

tests/Duckling/Email/IS/Tests.hs

@@ -0,0 +1,23 @@
+-- Copyright (c) 2016-present, Facebook, Inc.
+-- All rights reserved.
+--
+-- This source code is licensed under the BSD-style license found in the
+-- LICENSE file in the root directory of this source tree. An additional grant
+-- of patent rights can be found in the PATENTS file in the same directory.
+
+
+module Duckling.Email.IS.Tests
+  ( tests
+  ) where
+
+import Data.String
+import Test.Tasty
+
+import Duckling.Dimensions.Types
+import Duckling.Email.IS.Corpus
+import Duckling.Testing.Asserts
+
+tests :: TestTree
+tests = testGroup "IS Tests"
+  [ makeCorpusTest [This Email] corpus
+  ]

tests/Duckling/Email/IT/Tests.hs

@@ -18,6 +18,6 @@ import Duckling.Email.IT.Corpus
 import Duckling.Testing.Asserts
 
 tests :: TestTree
-tests = testGroup "Email Tests"
+tests = testGroup "IT Tests"
   [ makeCorpusTest [This Email] corpus
   ]

tests/Duckling/Email/Tests.hs

@@ -6,24 +6,30 @@
 -- of patent rights can be found in the PATENTS file in the same directory.
 
 
-module Duckling.Email.Tests (tests) where
+module Duckling.Email.Tests
+  ( tests
+  ) where
 
-import Prelude
 import Data.String
+import Prelude
 import Test.Tasty
 
 import Duckling.Dimensions.Types
 import Duckling.Email.Corpus
+import Duckling.Testing.Asserts
+import qualified Duckling.Email.DE.Tests as DE
 import qualified Duckling.Email.EN.Tests as EN
 import qualified Duckling.Email.FR.Tests as FR
+import qualified Duckling.Email.IS.Tests as IS
 import qualified Duckling.Email.IT.Tests as IT
-import Duckling.Testing.Asserts
 
 tests :: TestTree
 tests = testGroup "Email Tests"
   [ makeCorpusTest [This Email] corpus
   , makeNegativeCorpusTest [This Email] negativeCorpus
+  , DE.tests
   , EN.tests
   , FR.tests
+  , IS.tests
   , IT.tests
   ]