Support malayalam ordinals

Summary: Add support for malayalam ordinals.

Reviewed By: patapizza

Differential Revision: D10097075

fbshipit-source-id: 90be5f05d1f9cf4cc6558a4583c8c72518fe0ab0

Duckling/Dimensions/ML.hs

@@ -14,5 +14,6 @@ import Duckling.Dimensions.Types
 
 allDimensions :: [Some Dimension]
 allDimensions =
-  [ This Numeral
+  [ This Numeral,
+    This Ordinal
   ]

Duckling/Ordinal/ML/Corpus.hs

@@ -0,0 +1,119 @@
+-- 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.Ordinal.ML.Corpus
+  ( corpus
+  ) where
+
+import Data.String
+import Prelude
+
+import Duckling.Locale
+import Duckling.Ordinal.Types
+import Duckling.Resolve
+import Duckling.Testing.Types
+
+context :: Context
+context = testContext {locale = makeLocale ML Nothing}
+
+corpus :: Corpus
+corpus = (context, testOptions, allExamples)
+
+allExamples :: [Example]
+allExamples = concat
+  [ examples (OrdinalData 1)
+             [ "ഒന്നാം"
+             , "1."
+             ]
+  , examples (OrdinalData 2)
+             [ "രണ്ടാം"
+             , "2."
+             ]
+  , examples (OrdinalData 3)
+             [ "മൂന്നാം"
+             , "3."
+             ]
+  , examples (OrdinalData 4)
+             [ "നാലാം"
+             , "4."
+             ]
+  , examples (OrdinalData 5)
+             [ "അഞ്ചാം"
+             , "5."
+             ]
+  , examples (OrdinalData 6)
+             [ "ആറാം"
+             , "6."
+             ]
+  , examples (OrdinalData 7)
+             [ "ഏഴാം"
+             , "7."
+             ]
+  , examples (OrdinalData 8)
+             [ "എട്ടാം"
+             , "8."
+             ]
+  , examples (OrdinalData 9)
+             [ "ഒമ്പതാം"
+             , "9."
+             ]
+  , examples (OrdinalData 10)
+             [ "പത്താം"
+             , "10."
+             ]
+  , examples (OrdinalData 11)
+             [ "പതിനൊന്നാം"
+             , "11."
+             ]
+  , examples (OrdinalData 12)
+             [ "പന്ത്രണ്ടാം"
+             , "12."
+             ]
+  , examples (OrdinalData 20)
+             [ "ഇരുപതാം"
+             , "20."
+             ]
+  , examples (OrdinalData 22)
+             [ "ഇരുപത്തിരണ്ടാം"
+             , "22."
+             ]
+  , examples (OrdinalData 26)
+             [ "ഇരുപത്തിആറാം"
+             , "26."
+             ]
+  , examples (OrdinalData 30)
+             [ "മുപ്പത്തഞ്ചാം"
+             , "30."
+             ]
+   , examples (OrdinalData 33)
+             [ "മുപ്പത്തിമൂന്നാം"
+             , "33."
+             ]
+  , examples (OrdinalData 50)
+             [ "അമ്പതാം"
+             , "50."
+             ]
+   , examples (OrdinalData 54)
+              [ "അമ്പത്തിനാലാം"
+              , "54."
+              ]
+   , examples (OrdinalData 65)
+              [ "അറുപത്തിഅഞ്ചാം"
+              , "65."
+              ]
+   , examples (OrdinalData 76)
+              [ "എഴുപത്തിആറാം"
+              , "76."
+              ]
+   , examples (OrdinalData 87)
+              [ "എൺപത്തിഏഴാം"
+              , "87."
+              ]
+  ]

Duckling/Ordinal/ML/Rules.hs

@@ -0,0 +1,137 @@
+-- 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 #-}
+{-# LANGUAGE LambdaCase #-}
+{-# LANGUAGE NoRebindableSyntax #-}
+
+module Duckling.Ordinal.ML.Rules
+  ( rules ) where
+
+import Data.HashMap.Strict (HashMap)
+import Data.String
+import Data.Text (Text)
+import Prelude
+import qualified Data.HashMap.Strict as HashMap
+import qualified Data.Text as Text
+
+import Duckling.Dimensions.Types
+import Duckling.Numeral.Helpers (parseInt)
+import Duckling.Ordinal.Helpers
+import Duckling.Regex.Types
+import Duckling.Types
+
+oneToNineteenMap :: HashMap Text Int
+oneToNineteenMap = HashMap.fromList
+  [ ( "ഒന്നാം", 1 )
+  , ( "രണ്ടാം", 2 )
+  , ( "മൂന്നാം", 3 )
+  , ( "നാലാം", 4 )
+  , ( "അഞ്ചാം", 5 )
+  , ( "ആറാം", 6 )
+  , ( "ഏഴാം", 7 )
+  , ( "എട്ടാം", 8 )
+  , ( "ഒമ്പതാം", 9 )
+  , ( "പത്താം", 10 )
+  , ( "പതിനൊന്നാം", 11 )
+  , ( "പന്ത്രണ്ടാം", 12 )
+  , ( "പതിമൂന്നാം", 13 )
+  , ( "പതിനാലാം", 14 )
+  , ( "പതിനഞ്ചാം", 15 )
+  , ( "പതിനാറാം", 16 )
+  , ( "പതിനേഴാം", 17 )
+  , ( "പതിനെട്ടാം", 18 )
+  , ( "പത്തൊൻപതാം", 19 )
+  ]
+
+ruleOneToNineteen :: Rule
+ruleOneToNineteen = Rule
+  { name = "integer (1..19)"
+  , pattern =
+    [ regex "(ഒന്നാം|രണ്ടാം|മൂന്നാം|നാലാം|അഞ്ചാം|ആറാം|ഏഴാം|എട്ടാം|ഒമ്പതാം|പത്താം|പതിനൊന്നാം|പന്ത്രണ്ടാം|പതിമൂന്നാം|പതിനാലാം|പതിനഞ്ചാം|പതിനാറാം|പതിനേഴാം|പതിനെട്ടാം|പത്തൊൻപതാം)"
+    ]
+  , prod = \case
+      (Token RegexMatch (GroupMatch (match:_)):_) ->
+        ordinal <$> HashMap.lookup (Text.toLower match) oneToNineteenMap
+      _ -> Nothing
+  }
+
+tensMap :: HashMap Text Int
+tensMap = HashMap.fromList
+  [ ( "ഇരുപതാം", 20 )
+  , ( "മുപ്പത്തഞ്ചാം", 30 )
+  , ( "നാല്പതാം", 40 )
+  , ( "അമ്പതാം", 50 )
+  , ( "അറുപതാം", 60 )
+  , ( "എഴുപത്താം", 70 )
+  , ( "എൺപത്താം", 80 )
+  , ( "തൊണ്ണൂറാം", 90 )
+  ]
+
+ruleTens :: Rule
+ruleTens = Rule
+  { name = "integer (20..90)"
+  , pattern =
+    [ regex "(ഇരുപതാം|മുപ്പത്തഞ്ചാം|നാല്പതാം|അമ്പതാം|അറുപതാം|എഴുപത്താം|എൺപത്താം|തൊണ്ണൂറാം)"
+    ]
+  , prod = \case
+      (Token RegexMatch (GroupMatch (match:_)):_) ->
+        ordinal <$> HashMap.lookup (Text.toLower match) tensMap
+      _ -> Nothing
+  }
+
+tensOrdinalMap :: HashMap Text Int
+tensOrdinalMap = HashMap.fromList
+  [ ( "ഇരുപത്തി", 20 )
+  , ( "മുപ്പത്തി", 30 )
+  , ( "നാല്പത്തി", 40 )
+  , ( "അമ്പത്തി", 50 )
+  , ( "അറുപത്തി", 60 )
+  , ( "എഴുപത്തി", 70 )
+  , ( "എൺപത്തി", 80 )
+  , ( "തൊണ്ണൂറ്റി", 90 )
+  ]
+
+oneToNineMap :: HashMap Text Int
+oneToNineMap = HashMap.filterWithKey (\_ v -> v <= 9) oneToNineteenMap
+
+
+ruleCompositeTens :: Rule
+ruleCompositeTens = Rule
+  { name = "integer ([2-9][1-9])"
+  , pattern =
+    [ regex "(ഇരുപത്തി|മുപ്പത്തി|നാല്പത്തി|അമ്പത്തി|അറുപത്തി|എഴുപത്തി|എൺപത്തി|തൊണ്ണൂറ്റി)(ആദ്യം|രണ്ടാം|മൂന്നാം|നാലാം|അഞ്ചാം|ആറാം|ഏഴാം|എട്ടാം|ഒമ്പത)"
+    ]
+  , prod = \case
+      (Token RegexMatch (GroupMatch (m1:m2:_)):_) -> do
+        v1 <- HashMap.lookup (Text.toLower m1) tensOrdinalMap
+        v2 <- HashMap.lookup (Text.toLower m2) oneToNineMap
+        Just $ ordinal (v1 + v2)
+      _ -> Nothing
+  }
+
+ruleOrdinalDigits :: Rule
+ruleOrdinalDigits = Rule
+  { name = "ordinal (digits)"
+  , pattern =
+    [ regex "0*(\\d+)\\."
+    ]
+  , prod = \case
+    (   Token RegexMatch (GroupMatch (match :_)) : _)
+          -> ordinal <$> parseInt match
+    _ -> Nothing
+  }
+
+rules :: [Rule]
+rules =
+  [ ruleOrdinalDigits
+  , ruleOneToNineteen
+  , ruleTens
+  , ruleCompositeTens
+  ]

Duckling/Rules/ML.hs

@@ -19,6 +19,7 @@ import Duckling.Dimensions.Types
 import Duckling.Locale
 import Duckling.Types
 import qualified Duckling.Numeral.ML.Rules as Numeral
+import qualified Duckling.Ordinal.ML.Rules as Ordinal
 
 defaultRules :: Some Dimension -> [Rule]
 defaultRules = langRules
@@ -33,7 +34,7 @@ langRules (This Distance) = []
 langRules (This Duration) = []
 langRules (This Email) = []
 langRules (This Numeral) = Numeral.rules
-langRules (This Ordinal) = []
+langRules (This Ordinal) = Ordinal.rules
 langRules (This PhoneNumber) = []
 langRules (This Quantity) = []
 langRules (This RegexMatch) = []

duckling.cabal

@@ -468,6 +468,8 @@ library
                      , Duckling.Ordinal.KM.Rules
                      , Duckling.Ordinal.KO.Corpus
                      , Duckling.Ordinal.KO.Rules
+                     , Duckling.Ordinal.ML.Corpus
+                     , Duckling.Ordinal.ML.Rules
                      , Duckling.Ordinal.NB.Corpus
                      , Duckling.Ordinal.NB.Rules
                      , Duckling.Ordinal.NL.Corpus
@@ -892,6 +894,7 @@ test-suite duckling-test
                      , Duckling.Ordinal.KA.Tests
                      , Duckling.Ordinal.KM.Tests
                      , Duckling.Ordinal.KO.Tests
+                     , Duckling.Ordinal.ML.Tests
                      , Duckling.Ordinal.NB.Tests
                      , Duckling.Ordinal.NL.Tests
                      , Duckling.Ordinal.PL.Tests

tests/Duckling/Ordinal/ML/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.Ordinal.ML.Tests
+  ( tests
+  ) where
+
+import Data.String
+import Prelude
+import Test.Tasty
+
+import Duckling.Dimensions.Types
+import Duckling.Ordinal.ML.Corpus
+import Duckling.Testing.Asserts
+
+tests :: TestTree
+tests = testGroup "ML Tests"
+  [ makeCorpusTest [This Ordinal] corpus
+  ]

tests/Duckling/Ordinal/Tests.hs

@@ -32,6 +32,7 @@ import qualified Duckling.Ordinal.JA.Tests as JA
 import qualified Duckling.Ordinal.KA.Tests as KA
 import qualified Duckling.Ordinal.KM.Tests as KM
 import qualified Duckling.Ordinal.KO.Tests as KO
+import qualified Duckling.Ordinal.ML.Tests as ML
 import qualified Duckling.Ordinal.NB.Tests as NB
 import qualified Duckling.Ordinal.NL.Tests as NL
 import qualified Duckling.Ordinal.PL.Tests as PL
@@ -67,6 +68,7 @@ tests = testGroup "Ordinal Tests"
   , KA.tests
   , KM.tests
   , KO.tests
+  , ML.tests
   , NB.tests
   , NL.tests
   , PL.tests