add Kannada support (for Numeral)

Summary: Pull Request resolved: https://github.com/facebook/duckling/pull/322

Reviewed By: patapizza

Differential Revision: D13888983

Pulled By: haoxuany

fbshipit-source-id: b7a3b5c72dae0ccc637cf6faa14670ebf868bb50
This commit is contained in:
akki2825 2019-02-08 11:13:12 -08:00 committed by Facebook Github Bot
parent 301f74ea2e
commit a2494caf61
14 changed files with 286 additions and 2 deletions

View File

@ -45,6 +45,7 @@ import qualified Duckling.Dimensions.IT as ITDimensions
import qualified Duckling.Dimensions.JA as JADimensions
import qualified Duckling.Dimensions.KA as KADimensions
import qualified Duckling.Dimensions.KM as KMDimensions
import qualified Duckling.Dimensions.KN as KNDimensions
import qualified Duckling.Dimensions.KO as KODimensions
import qualified Duckling.Dimensions.LO as LODimensions
import qualified Duckling.Dimensions.ML as MLDimensions
@ -119,6 +120,7 @@ langDimensions IT = ITDimensions.allDimensions
langDimensions JA = JADimensions.allDimensions
langDimensions KA = KADimensions.allDimensions
langDimensions KM = KMDimensions.allDimensions
langDimensions KN = KNDimensions.allDimensions
langDimensions KO = KODimensions.allDimensions
langDimensions LO = LODimensions.allDimensions
langDimensions ML = MLDimensions.allDimensions

18
Duckling/Dimensions/KN.hs Normal file
View File

@ -0,0 +1,18 @@
-- 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.Dimensions.KN
( allDimensions
) where
import Duckling.Dimensions.Types
allDimensions :: [Some Dimension]
allDimensions =
[ This Numeral
]

View File

@ -56,6 +56,7 @@ data Lang
| IT
| JA
| KA
| KN
| KM
| KO
| LO

View File

@ -0,0 +1,68 @@
-- 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.Numeral.KN.Corpus
( corpus
) where
import Data.String
import Prelude
import Duckling.Locale
import Duckling.Numeral.Types
import Duckling.Resolve
import Duckling.Testing.Types
corpus :: Corpus
corpus = (testContext {locale = makeLocale KN Nothing}, testOptions, allExamples)
allExamples :: [Example]
allExamples = concat
[ examples (NumeralValue 0)
[ "ಸೊನ್ನೆ"
, ""
]
, examples (NumeralValue 1)
[ "ಒಂದು"
, ""
]
, examples (NumeralValue 2)
[ "ಎರಡು"
, ""
]
, examples (NumeralValue 3)
[ "ಮೂರು"
, ""
]
, examples (NumeralValue 4)
[ "ನಾಲ್ಕು"
, ""
]
, examples (NumeralValue 5)
[ "ಐದು"
, ""
]
, examples (NumeralValue 6)
[ "ಆರು"
, ""
]
, examples (NumeralValue 7)
[ "ಏಳು"
, ""
]
, examples (NumeralValue 8)
[ "ಎಂಟು"
, ""
]
, examples (NumeralValue 9)
[ "ಒಂಬತ್ತು"
, ""
]
]

View File

@ -0,0 +1,91 @@
-- 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 LambdaCase #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE NoRebindableSyntax #-}
module Duckling.Numeral.KN.Rules
( rules
) where
import Data.HashMap.Strict (HashMap)
import Data.Maybe
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
import Duckling.Regex.Types
import Duckling.Types
ankiMap :: HashMap Char Char
ankiMap = HashMap.fromList
[ ('', '0')
, ('೧', '1')
, ('೨', '2')
, ('೩', '3')
, ('೪', '4')
, ('೫', '5')
, ('೬', '6')
, ('೭', '7')
, ('೮', '8')
, ('೯', '9')
]
ankiToArab :: Char -> Char
ankiToArab c = HashMap.lookupDefault c c ankiMap
ruleAnki :: Rule
ruleAnki = Rule
{ name = "anki forms"
, pattern =
[ regex "([೦೧೨೩೪೫೬೭೮೯]{1,10})"
]
, prod = \case
(Token RegexMatch (GroupMatch (match:_)):_) ->
toInteger <$> parseInt (Text.map ankiToArab match) >>= integer
_ -> Nothing
}
ruleNumeralMap :: [(Text, Integer)]
ruleNumeralMap =
[ ("ಸೊನ್ನೆ", 0)
, ("ಒಂದು", 1)
, ("ಎರಡು", 2)
, ("ಮೂರು", 3)
, ("ನಾಲ್ಕು", 4)
, ("ಐದು", 5)
, ("ಆರು", 6)
, ("ಏಳು", 7)
, ("ಎಂಟು", 8)
, ("ಒಂಬತ್ತು", 9)
]
ruleNumerals :: [Rule]
ruleNumerals =
map constructRule ruleNumeralMap
where
constructRule :: (Text, Integer) -> Rule
constructRule (s, i) = Rule
{ name = "number: " `mappend` s
, pattern =
[ regex $ Text.unpack s
]
, prod = const $ integer i
}
rules :: [Rule]
rules =
[ ruleAnki
]
++ ruleNumerals

View File

@ -39,6 +39,7 @@ import qualified Duckling.Ranking.Classifiers.IT_XX as IT_XXClassifiers
import qualified Duckling.Ranking.Classifiers.JA_XX as JA_XXClassifiers
import qualified Duckling.Ranking.Classifiers.KA_XX as KA_XXClassifiers
import qualified Duckling.Ranking.Classifiers.KM_XX as KM_XXClassifiers
import qualified Duckling.Ranking.Classifiers.KN_XX as KN_XXClassifiers
import qualified Duckling.Ranking.Classifiers.KO_XX as KO_XXClassifiers
import qualified Duckling.Ranking.Classifiers.LO_XX as LO_XXClassifiers
import qualified Duckling.Ranking.Classifiers.ML_XX as ML_XXClassifiers
@ -85,6 +86,7 @@ classifiers (Locale IT _) = IT_XXClassifiers.classifiers
classifiers (Locale JA _) = JA_XXClassifiers.classifiers
classifiers (Locale KA _) = KA_XXClassifiers.classifiers
classifiers (Locale KM _) = KM_XXClassifiers.classifiers
classifiers (Locale KN _) = KN_XXClassifiers.classifiers
classifiers (Locale KO _) = KO_XXClassifiers.classifiers
classifiers (Locale LO _) = LO_XXClassifiers.classifiers
classifiers (Locale ML _) = ML_XXClassifiers.classifiers

View File

@ -320,10 +320,10 @@ classifiers
unseen = -4.31748811353631,
likelihoods =
HashMap.fromList
[("<integer> (latent time-of-day)", -0.9718605830289657),
[("<integer> (latent time-of-day)", -0.9718605830289658),
("intersect by \"di\", \"della\", \"del\"", -3.20545280453606),
("day", -2.3581549441488563), ("Lunedi", -3.6109179126442243),
("hour", -0.9718605830289657),
("hour", -0.9718605830289658),
("two time tokens separated by `di`", -3.20545280453606),
("Domenica", -3.6109179126442243)],
n = 33}}),

View File

@ -0,0 +1,22 @@
-- 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.
-----------------------------------------------------------------
-- Auto-generated by regenClassifiers
--
-- DO NOT EDIT UNLESS YOU ARE SURE THAT YOU KNOW WHAT YOU ARE DOING
-- @generated
-----------------------------------------------------------------
{-# LANGUAGE OverloadedStrings #-}
module Duckling.Ranking.Classifiers.KN_XX (classifiers) where
import Data.String
import Prelude
import qualified Data.HashMap.Strict as HashMap
import Duckling.Ranking.Types
classifiers :: Classifiers
classifiers = HashMap.fromList []

View File

@ -46,6 +46,7 @@ import qualified Duckling.Rules.IT as ITRules
import qualified Duckling.Rules.JA as JARules
import qualified Duckling.Rules.KA as KARules
import qualified Duckling.Rules.KM as KMRules
import qualified Duckling.Rules.KN as KNRules
import qualified Duckling.Rules.KO as KORules
import qualified Duckling.Rules.LO as LORules
import qualified Duckling.Rules.ML as MLRules
@ -110,6 +111,7 @@ defaultRules IT = ITRules.defaultRules
defaultRules JA = JARules.defaultRules
defaultRules KA = KARules.defaultRules
defaultRules KM = KMRules.defaultRules
defaultRules KN = KNRules.defaultRules
defaultRules KO = KORules.defaultRules
defaultRules LO = LORules.defaultRules
defaultRules ML = MLRules.defaultRules
@ -154,6 +156,7 @@ localeRules IT = ITRules.localeRules
localeRules JA = JARules.localeRules
localeRules KA = KARules.localeRules
localeRules KM = KMRules.localeRules
localeRules KN = KNRules.localeRules
localeRules KO = KORules.localeRules
localeRules LO = LORules.localeRules
localeRules ML = MLRules.localeRules
@ -190,6 +193,7 @@ langRules FR = FRRules.langRules
langRules GA = GARules.langRules
langRules HE = HERules.langRules
langRules HI = HIRules.langRules
langRules KN = KNRules.langRules
langRules HR = HRRules.langRules
langRules HU = HURules.langRules
langRules ID = IDRules.langRules

45
Duckling/Rules/KN.hs Normal file
View File

@ -0,0 +1,45 @@
---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 #-}
module Duckling.Rules.KN
( defaultRules
, langRules
, localeRules
) where
import Duckling.Dimensions.Types
import Duckling.Locale
import Duckling.Types
import qualified Duckling.Numeral.KN.Rules as Numeral
defaultRules :: Some Dimension -> [Rule]
defaultRules = langRules
localeRules :: Region -> Some Dimension -> [Rule]
localeRules region (This (CustomDimension dim)) = dimLocaleRules region dim
localeRules _ _ = []
langRules :: Some Dimension -> [Rule]
langRules (This AmountOfMoney) = []
langRules (This CreditCardNumber) = []
langRules (This Distance) = []
langRules (This Duration) = []
langRules (This Numeral) = Numeral.rules
langRules (This Email) = []
langRules (This Ordinal) = []
langRules (This PhoneNumber) = []
langRules (This Quantity) = []
langRules (This RegexMatch) = []
langRules (This Temperature) = []
langRules (This Time) = []
langRules (This TimeGrain) = []
langRules (This Url) = []
langRules (This Volume) = []
langRules (This (CustomDimension dim)) = dimLangRules KN dim

View File

@ -72,6 +72,7 @@ library
, Duckling.Rules.JA
, Duckling.Rules.KA
, Duckling.Rules.KM
, Duckling.Rules.KN
, Duckling.Rules.KO
, Duckling.Rules.LO
, Duckling.Rules.ML
@ -178,6 +179,7 @@ library
, Duckling.Dimensions.JA
, Duckling.Dimensions.KA
, Duckling.Dimensions.KM
, Duckling.Dimensions.KN
, Duckling.Dimensions.KO
, Duckling.Dimensions.LO
, Duckling.Dimensions.ML
@ -425,6 +427,8 @@ library
, Duckling.Numeral.KA.Rules
, Duckling.Numeral.KM.Corpus
, Duckling.Numeral.KM.Rules
, Duckling.Numeral.KN.Corpus
, Duckling.Numeral.KN.Rules
, Duckling.Numeral.KO.Corpus
, Duckling.Numeral.KO.Rules
, Duckling.Numeral.LO.Corpus
@ -928,6 +932,7 @@ test-suite duckling-test
, Duckling.Numeral.JA.Tests
, Duckling.Numeral.KA.Tests
, Duckling.Numeral.KM.Tests
, Duckling.Numeral.KN.Tests
, Duckling.Numeral.KO.Tests
, Duckling.Numeral.LO.Tests
, Duckling.Numeral.ML.Tests

View File

@ -193,6 +193,7 @@ getCorpusForLang IT = ITTime.corpus
getCorpusForLang JA = (testContext, testOptions, [])
getCorpusForLang KA = (testContext, testOptions, [])
getCorpusForLang KM = (testContext, testOptions, [])
getCorpusForLang KN = (testContext, testOptions, [])
getCorpusForLang KO = KOTime.corpus
getCorpusForLang LO = (testContext, testOptions, [])
getCorpusForLang ML = (testContext, testOptions, [])

View File

@ -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.Numeral.KN.Tests
( tests
) where
import Data.String
import Prelude
import Test.Tasty
import Duckling.Dimensions.Types
import Duckling.Numeral.KN.Corpus
import Duckling.Testing.Asserts
tests :: TestTree
tests = testGroup "KN Tests"
[ makeCorpusTest [This Numeral] corpus
]

View File

@ -35,6 +35,7 @@ import qualified Duckling.Numeral.IT.Tests as IT
import qualified Duckling.Numeral.JA.Tests as JA
import qualified Duckling.Numeral.KA.Tests as KA
import qualified Duckling.Numeral.KM.Tests as KM
import qualified Duckling.Numeral.KN.Tests as KN
import qualified Duckling.Numeral.KO.Tests as KO
import qualified Duckling.Numeral.LO.Tests as LO
import qualified Duckling.Numeral.ML.Tests as ML
@ -80,6 +81,7 @@ tests = testGroup "Numeral Tests"
, JA.tests
, KA.tests
, KM.tests
, KN.tests
, KO.tests
, LO.tests
, ML.tests