duckling/Duckling/Numeral/HI/Rules.hs

-- 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.


{-# LANGUAGE GADTs #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE NoRebindableSyntax #-}

module Duckling.Numeral.HI.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.Numeral.Types (NumeralData (..))
import Duckling.Regex.Types
import Duckling.Types
import qualified Duckling.Numeral.Types as TNumeral

devanagariMap :: HashMap Char Char
devanagariMap = HashMap.fromList
  [ ( '०', '0' )
  , ( '१', '1' )
  , ( '२', '2' )
  , ( '३', '3' )
  , ( '४', '4' )
  , ( '५', '5' )
  , ( '६', '6' )
  , ( '७', '7' )
  , ( '८', '8' )
  , ( '९', '9' )
  ]

devanagariToArab :: Char -> Char
devanagariToArab c = HashMap.lookupDefault c c devanagariMap

ruleDevanagari :: Rule
ruleDevanagari = Rule
  { name = "devanagari forms"
  , pattern =
    [ regex "([०१२३४५६७८९]{1,18})"
    ]
  , prod = \tokens -> case tokens of
      (Token RegexMatch (GroupMatch (match:_)):_) ->
        toInteger <$> parseInt (Text.map devanagariToArab match) >>= integer
      _ -> Nothing
  }

ruleZeroToNinetyNineMap :: HashMap Text Integer
ruleZeroToNinetyNineMap = HashMap.fromList
  [ ( "शून्य", 0 )
  , ( "एक", 1 )
  , ( "दो", 2 )
  , ( "तीन", 3 )
  , ( "चार", 4 )
  , ( "पाँच", 5 )
  , ( "छः", 6 )
  , ( "छह", 6 )
  , ( "सात", 7 )
  , ( "आठ", 8 )
  , ( "नौ", 9 )
  , ( "ग्यारह", 11 )
  , ( "बारह", 12 )
  , ( "तेरह", 13 )
  , ( "चौदह", 14 )
  , ( "पन्द्रह", 15 )
  , ( "सोलह", 16 )
  , ( "सत्रह", 17 )
  , ( "अठारह", 18 )
  , ( "उन्नीस", 19 )
  , ( "इक्कीस", 21 )
  , ( "बाईस", 22 )
  , ( "तेईस", 23 )
  , ( "चौबीस", 24 )
  , ( "पच्चीस", 25 )
  , ( "छब्बीस", 26 )
  , ( "सत्ताईस", 27 )
  , ( "अट्ठाईस", 28 )
  , ( "उनतीस", 29 )
  , ( "इकतीस", 31 )
  , ( "इकत्तीस", 31 )
  , ( "बत्तीस", 32 )
  , ( "तैंतीस", 33 )
  , ( "चौंतीस", 34 )
  , ( "पैंतीस", 35 )
  , ( "छत्तीस", 36 )
  , ( "सैंतीस", 37 )
  , ( "अड़तीस", 38 )
  , ( "उनतालीस", 39 )
  , ( "इकतालीस", 41 )
  , ( "बयालीस", 42 )
  , ( "तैंतालीस", 43 )
  , ( "चौवालीस", 44 )
  , ( "पैंतालीस", 45 )
  , ( "छियालीस", 46 )
  , ( "सैंतालीस", 47 )
  , ( "अड़तालीस", 48 )
  , ( "उनचास", 49 )
  , ( "इक्यावन", 51 )
  , ( "बावन", 52 )
  , ( "तिरेपन", 53 )
  , ( "चौवन", 54 )
  , ( "पचपन", 55 )
  , ( "छप्पन", 56 )
  , ( "सत्तावन", 57 )
  , ( "अट्ठावन", 58 )
  , ( "उनसठ", 59 )
  , ( "इकसठ", 61 )
  , ( "बासठ", 62 )
  , ( "तिरेसठ", 63 )
  , ( "चौंसठ", 64 )
  , ( "पैंसठ", 65 )
  , ( "छियासठ", 66 )
  , ( "सड़सठ", 67 )
  , ( "अड़सठ", 68 )
  , ( "उनहत्तर", 69 )
  , ( "इकहत्तर", 71 )
  , ( "बहत्तर", 72 )
  , ( "तिहत्तर", 73 )
  , ( "चौहत्तर", 74 )
  , ( "पचहत्तर", 75 )
  , ( "छिहत्तर", 76 )
  , ( "सतहत्तर", 77 )
  , ( "अठहत्तर", 78 )
  , ( "उनासी", 79 )
  , ( "इक्यासी", 81 )
  , ( "बयासी", 82 )
  , ( "तिरासी", 83 )
  , ( "चौरासी", 84 )
  , ( "पचासी", 85 )
  , ( "छियासी", 86 )
  , ( "सतासी", 87 )
  , ( "अट्ठासी", 88 )
  , ( "नवासी", 89 )
  , ( "इक्यानवे", 91 )
  , ( "बानवे", 92 )
  , ( "तिरानवे", 93 )
  , ( "चौरानवे", 94 )
  , ( "पचानवे", 95 )
  , ( "छियानवे", 96 )
  , ( "सत्तानवे", 97 )
  , ( "अट्ठानवे", 98 )
  , ( "निन्यानवे", 99 )
  ]

ruleZeroToNinetyNine :: Rule
ruleZeroToNinetyNine = Rule
  { name = "number (0..99)"
  , pattern =
    [ regex "(शून्य|एक|दो|तीन|चार|पाँच|छे|छह|सात|आठ|नौ|ग्यारह|बारह|तेरह|चौदह|पन्द्रह|सोलह|सत्रह|अठारह|उन्नीस|इक्कीस|बाईस|तेईस|चौबीस|पच्चीस|छब्बीस|सत्ताईस|अट्ठाईस|उनतीस|इकतीस|बत्तीस|तैंतीस|चौंतीस|पैंतीस|छत्तीस|सैंतीस|अड़तीस|उनतालीस|इकतालीस|बयालीस|तैंतालीस|चौवालीस|पैंतालीस|छियालीस|सैंतालीस|अड़तालीस|उनचास|इक्यावन|बावन|तिरेपन|चौवन|पचपन|छप्पन|सत्तावन|अट्ठावन|उनसठ|इकसठ|बासठ|तिरेसठ|चौंसठ|पैंसठ|छियासठ|सड़सठ|अड़सठ|उनहत्तर|इकहत्तर|बहत्तर|तिहत्तर|चौहत्तर|पचहत्तर|छिहत्तर|सतहत्तर|अठहत्तर|उनासी|इक्यासी|बयासी|तिरासी|चौरासी|पचासी|छियासी|सतासी|अट्ठासी|नवासी|इक्यानवे|बानवे|तिरानवे|चौरानवे|पचानवे|छियानवे|सत्तानवे|अट्ठानवे|निन्यानवे)"
    ]
  , prod = \tokens -> case tokens of
      (Token RegexMatch (GroupMatch (match:_)):_) ->
        HashMap.lookup match ruleZeroToNinetyNineMap >>= integer
      _ -> Nothing
  }

tensMap :: HashMap Text Integer
tensMap = HashMap.fromList
  [ ("दस", 10)
  , ( "बीस", 20 )
  , ( "तीस", 30 )
  , ( "चालीस", 40 )
  , ( "पचास", 50 )
  , ( "साठ", 60 )
  , ( "सत्तर", 70 )
  , ( "अस्सी", 80 )
  , ( "नब्बे", 90 )
  ]

ruleTens :: Rule
ruleTens = Rule
  { name = "integer (20,30..90)"
  , pattern =
    [ regex "(दस|बीस|तीस|चालीस|पचास|साठ|सत्तर|अस्सी|नब्बे)"
    ]
  , prod = \tokens -> case tokens of
      (Token RegexMatch (GroupMatch (match:_)):_) ->
        HashMap.lookup match tensMap >>= integer
      _ -> Nothing
  }

rulePowersOfTen :: Rule
rulePowersOfTen = Rule
  { name = "powers of tens"
  , pattern = [regex "(सौ|हज़ार|हज़ार)"]
  , prod = \tokens -> case tokens of
      (Token RegexMatch (GroupMatch (match : _)) : _) ->
        case Text.toLower match of
          "सौ" -> double 1e2 >>= withGrain 2 >>= withMultipliable
          "हज़ार" -> double 1e3 >>= withGrain 3 >>= withMultipliable
          "हज़ार" -> double 1e3 >>= withGrain 3 >>= withMultipliable
          _ -> Nothing
      _ -> Nothing
  }

ruleMultiply :: Rule
ruleMultiply = Rule
  { name = "compose by multiplication"
  , pattern =
    [ Predicate isPositive
    , Predicate isMultipliable
    ]
  , prod = \tokens -> case tokens of
      (token1:token2:_) -> multiply token1 token2
      _ -> Nothing
  }

ruleCompositeHundreds :: Rule
ruleCompositeHundreds = Rule
  { name = "integer 100s.."
  , pattern =
    [ oneOf [100,200..5000]
    , regex "[\\s\\-]+"
    , Predicate $ numberBetween 1 99
    ]
  , prod = \tokens -> case tokens of
      (Token Numeral NumeralData{TNumeral.value = hundreds}:
       _:
       Token Numeral NumeralData{TNumeral.value = remaining}:
       _) -> double $ hundreds + remaining
      _ -> Nothing
  }


ruleCompositeThousands :: Rule
ruleCompositeThousands = Rule
  { name = "integer 100s.."
  , pattern =
    [ oneOf [1000,2000..50000]
    , regex "[\\s\\-]+"
    , Predicate $ numberBetween 1 999
    ]
  , prod = \tokens -> case tokens of
      (Token Numeral NumeralData{TNumeral.value = thousands}:
       _:
       Token Numeral NumeralData{TNumeral.value = remaining}:
       _) -> double $ thousands + remaining
      _ -> Nothing
  }

rules :: [Rule]
rules =
  [ ruleDevanagari
  , ruleZeroToNinetyNine
  , ruleTens
  , rulePowersOfTen
  , ruleMultiply
  , ruleCompositeHundreds
  , ruleCompositeThousands
  ]