|
| 1 | +-- Copyright (c) 2016-present, Facebook, Inc. |
| 2 | +-- All rights reserved. |
| 3 | +-- |
| 4 | +-- This source code is licensed under the BSD-style license found in the |
| 5 | +-- LICENSE file in the root directory of this source tree. An additional grant |
| 6 | +-- of patent rights can be found in the PATENTS file in the same directory. |
| 7 | + |
| 8 | + |
| 9 | +{-# LANGUAGE GADTs #-} |
| 10 | +{-# LANGUAGE OverloadedStrings #-} |
| 11 | +{-# LANGUAGE NoRebindableSyntax #-} |
| 12 | + |
| 13 | +module Duckling.Numeral.HI.Rules |
| 14 | + ( rules |
| 15 | + ) where |
| 16 | + |
| 17 | +import Data.HashMap.Strict (HashMap) |
| 18 | +import Data.Maybe |
| 19 | +import Data.String |
| 20 | +import Data.Text (Text) |
| 21 | +import Prelude |
| 22 | +import qualified Data.HashMap.Strict as HashMap |
| 23 | +import qualified Data.Text as Text |
| 24 | + |
| 25 | +import Duckling.Dimensions.Types |
| 26 | +import Duckling.Numeral.Helpers |
| 27 | +import Duckling.Numeral.Types (NumeralData (..)) |
| 28 | +import Duckling.Regex.Types |
| 29 | +import Duckling.Types |
| 30 | +import qualified Duckling.Numeral.Types as TNumeral |
| 31 | + |
| 32 | +devanagariMap :: HashMap Char Char |
| 33 | +devanagariMap = HashMap.fromList |
| 34 | + [ ( '०', '0' ) |
| 35 | + , ( '१', '1' ) |
| 36 | + , ( '२', '2' ) |
| 37 | + , ( '३', '3' ) |
| 38 | + , ( '४', '4' ) |
| 39 | + , ( '५', '5' ) |
| 40 | + , ( '६', '6' ) |
| 41 | + , ( '७', '7' ) |
| 42 | + , ( '८', '8' ) |
| 43 | + , ( '९', '9' ) |
| 44 | + ] |
| 45 | + |
| 46 | +devanagariToArab :: Char -> Char |
| 47 | +devanagariToArab c = HashMap.lookupDefault c c devanagariMap |
| 48 | + |
| 49 | +ruleDevanagari :: Rule |
| 50 | +ruleDevanagari = Rule |
| 51 | + { name = "devanagari forms" |
| 52 | + , pattern = |
| 53 | + [ regex "([०१२३४५६७८९]{1,18})" |
| 54 | + ] |
| 55 | + , prod = \tokens -> case tokens of |
| 56 | + (Token RegexMatch (GroupMatch (match:_)):_) -> |
| 57 | + toInteger <$> parseInt (Text.map devanagariToArab match) >>= integer |
| 58 | + _ -> Nothing |
| 59 | + } |
| 60 | + |
| 61 | +ruleNumeralMap :: HashMap Text Integer |
| 62 | +ruleNumeralMap = HashMap.fromList |
| 63 | + [ ( "शून्य", 0 ) |
| 64 | + , ( "एक", 1 ) |
| 65 | + , ( "दो" , 2 ) |
| 66 | + , ( "तीन", 3 ) |
| 67 | + , ( "चार", 4 ) |
| 68 | + , ( "पाँच", 5 ) |
| 69 | + , ( "छह", 6 ) |
| 70 | + , ( "सात", 7 ) |
| 71 | + , ( "आठ", 8 ) |
| 72 | + , ( "नौ" , 9 ) |
| 73 | + , ( "दस", 10 ) |
| 74 | + ] |
| 75 | + |
| 76 | +ruleNumeral :: Rule |
| 77 | +ruleNumeral = Rule |
| 78 | + { name = "number (0..10)" |
| 79 | + , pattern = |
| 80 | + [ regex "(शून्य|एक|दो|तीन|चार|पाँच|छह|सात|आठ|नौ|दस)" |
| 81 | + ] |
| 82 | + , prod = \tokens -> case tokens of |
| 83 | + (Token RegexMatch (GroupMatch (match:_)):_) -> |
| 84 | + HashMap.lookup match ruleNumeralMap >>= integer |
| 85 | + _ -> Nothing |
| 86 | + } |
| 87 | + |
| 88 | +elevenToNineteenMap :: HashMap Text Integer |
| 89 | +elevenToNineteenMap = HashMap.fromList |
| 90 | + [ ( "ग्यारह", 11 ) |
| 91 | + , ( "बारह", 12 ) |
| 92 | + , ( "तेरह", 13 ) |
| 93 | + , ( "चौदह", 14 ) |
| 94 | + , ( "पन्द्रह", 15 ) |
| 95 | + , ( "सोलह", 16 ) |
| 96 | + , ( "सत्रह", 17 ) |
| 97 | + , ( "अठारह", 18 ) |
| 98 | + , ( "उन्नीस", 19 ) |
| 99 | + ] |
| 100 | + |
| 101 | +ruleElevenToNineteen :: Rule |
| 102 | +ruleElevenToNineteen = Rule |
| 103 | + { name = "number (11..19)" |
| 104 | + , pattern = |
| 105 | + [ regex "(ग्यारह|बारह|तेरह|चौदह|पन्द्रह|सोलह|सत्रह|अठारह|उन्नीस)" |
| 106 | + ] |
| 107 | + , prod = \tokens -> case tokens of |
| 108 | + (Token RegexMatch (GroupMatch (match:_)):_) -> |
| 109 | + HashMap.lookup match elevenToNineteenMap >>= integer |
| 110 | + _ -> Nothing |
| 111 | + } |
| 112 | + |
| 113 | +twentyoneToTwentynineMap :: HashMap Text Integer |
| 114 | +twentyoneToTwentynineMap = HashMap.fromList |
| 115 | + [ ( "इक्कीस", 21 ) |
| 116 | + , ( "बाईस", 22 ) |
| 117 | + , ( "तेईस", 23 ) |
| 118 | + , ( "चौबीस", 24 ) |
| 119 | + , ( "पच्चीस", 25 ) |
| 120 | + , ( "छब्बीस", 26 ) |
| 121 | + , ( "सत्ताईस", 27 ) |
| 122 | + , ( "अट्ठाईस", 28 ) |
| 123 | + , ( "उनतीस", 29 ) |
| 124 | + ] |
| 125 | + |
| 126 | +ruleTwentyoneToTwentynine :: Rule |
| 127 | +ruleTwentyoneToTwentynine = Rule |
| 128 | + { name = "number (21..29)" |
| 129 | + , pattern = |
| 130 | + [ regex "(इक्कीस|बाईस|तेईस|चौबीस|पच्चीस|छब्बीस|सत्ताईस|अट्ठाईस|उनतीस)" |
| 131 | + ] |
| 132 | + , prod = \tokens -> case tokens of |
| 133 | + (Token RegexMatch (GroupMatch (match:_)):_) -> |
| 134 | + HashMap.lookup match twentyoneToTwentynineMap >>= integer |
| 135 | + _ -> Nothing |
| 136 | + } |
| 137 | + |
| 138 | +tensMap :: HashMap Text Integer |
| 139 | +tensMap = HashMap.fromList |
| 140 | + [ ( "बीस", 20 ) |
| 141 | + , ( "तीस", 30 ) |
| 142 | + , ( "चालीस", 40 ) |
| 143 | + , ( "पचास", 50 ) |
| 144 | + , ( "साठ", 60 ) |
| 145 | + , ( "सत्तर", 70 ) |
| 146 | + , ( "अस्सी", 80 ) |
| 147 | + , ( "नब्बे", 90 ) |
| 148 | + ] |
| 149 | + |
| 150 | +ruleTens :: Rule |
| 151 | +ruleTens = Rule |
| 152 | + { name = "integer (20,30..90)" |
| 153 | + , pattern = |
| 154 | + [ regex "(बीस|तीस|चालीस|पचास|साठ|सत्तर|अस्सी|नब्बे)" |
| 155 | + ] |
| 156 | + , prod = \tokens -> case tokens of |
| 157 | + (Token RegexMatch (GroupMatch (match:_)):_) -> |
| 158 | + HashMap.lookup match tensMap >>= integer |
| 159 | + _ -> Nothing |
| 160 | + } |
| 161 | + |
| 162 | + |
| 163 | +rules :: [Rule] |
| 164 | +rules = |
| 165 | + [ ruleDevanagari |
| 166 | + , ruleNumeral |
| 167 | + , ruleElevenToNineteen |
| 168 | + , ruleTwentyoneToTwentynine |
| 169 | + , ruleTens |
| 170 | + ] |
0 commit comments