mirror of
https://github.com/facebook/duckling.git
synced 2024-11-28 16:54:59 +03:00
83ea150d94
Summary: We noticed that using UTF-8 characters directly in regexes work. Hence converting back the escaped characters for readability and maintenance. Reviewed By: blandinw Differential Revision: D5787146 fbshipit-source-id: e5a4b9a
282 lines
7.2 KiB
Haskell
282 lines
7.2 KiB
Haskell
-- 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.Numeral.ID.Rules
|
|
( rules ) where
|
|
|
|
import qualified Data.HashMap.Strict as HashMap
|
|
import Data.HashMap.Strict (HashMap)
|
|
import Data.Maybe
|
|
import qualified Data.Text as Text
|
|
import Data.Text (Text)
|
|
import Prelude
|
|
import Data.String
|
|
|
|
import Duckling.Dimensions.Types
|
|
import Duckling.Numeral.Helpers
|
|
import Duckling.Numeral.Types (NumeralData (..))
|
|
import qualified Duckling.Numeral.Types as TNumeral
|
|
import Duckling.Regex.Types
|
|
import Duckling.Types
|
|
|
|
ruleIntegerMap :: HashMap Text Integer
|
|
ruleIntegerMap = HashMap.fromList
|
|
[ ( "kosong" , 0 )
|
|
, ( "nol" , 0 )
|
|
, ( "satu" , 1 )
|
|
, ( "dua" , 2 )
|
|
, ( "tiga" , 3 )
|
|
, ( "empat" , 4 )
|
|
, ( "lima" , 5 )
|
|
, ( "enam" , 6 )
|
|
, ( "tujuh" , 7 )
|
|
, ( "delapan" , 8 )
|
|
, ( "sembilan", 9 )
|
|
, ( "sebelas" , 11 )
|
|
]
|
|
|
|
ruleTeen :: Rule
|
|
ruleTeen = Rule
|
|
{ name = "teen"
|
|
, pattern =
|
|
[ numberBetween 2 10
|
|
, regex "belas"
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token Numeral (NumeralData {TNumeral.value = v}):_) -> double $ v + 10
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleNumeralCommaNumeral :: Rule
|
|
ruleNumeralCommaNumeral = Rule
|
|
{ name = "number comma number"
|
|
, pattern =
|
|
[ dimension Numeral
|
|
, regex "koma"
|
|
, numberWith TNumeral.grain isNothing
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token Numeral nd1:_:Token Numeral nd2:_) ->
|
|
double $ TNumeral.value nd1 + decimalsToDouble (TNumeral.value nd2)
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleNumeralsPrefixWithNegativeOrMinus :: Rule
|
|
ruleNumeralsPrefixWithNegativeOrMinus = Rule
|
|
{ name = "numbers prefix with -, negative or minus"
|
|
, pattern =
|
|
[ regex "-|minus\\s?|negatif\\s?"
|
|
, dimension Numeral
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(_:Token Numeral nd:_) -> double (TNumeral.value nd * (-1))
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleIntegerNumeric :: Rule
|
|
ruleIntegerNumeric = Rule
|
|
{ name = "integer (numeric)"
|
|
, pattern =
|
|
[ regex "(\\d{1,18})"
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token RegexMatch (GroupMatch (match:_)):_) -> do
|
|
v <- toInteger <$> parseInt match
|
|
integer v
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleTen :: Rule
|
|
ruleTen = Rule
|
|
{ name = "ten"
|
|
, pattern =
|
|
[ regex "(se)?puluh"
|
|
]
|
|
, prod = \_ -> integer 10 >>= withGrain 1
|
|
}
|
|
|
|
ruleDecimalWithThousandsSeparator :: Rule
|
|
ruleDecimalWithThousandsSeparator = Rule
|
|
{ name = "decimal with thousands separator"
|
|
, pattern =
|
|
[ regex "(\\d+(\\.\\d\\d\\d)+,\\d+)"
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token RegexMatch (GroupMatch (match:_)):
|
|
_) -> let dot = Text.singleton '.'
|
|
comma = Text.singleton ','
|
|
fmt = Text.replace comma dot $ Text.replace dot Text.empty match
|
|
in parseDouble fmt >>= double
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleMultiply :: Rule
|
|
ruleMultiply = Rule
|
|
{ name = "compose by multiplication"
|
|
, pattern =
|
|
[ dimension Numeral
|
|
, numberWith TNumeral.multipliable id
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(token1:token2:_) -> multiply token1 token2
|
|
_ -> Nothing
|
|
}
|
|
|
|
|
|
ruleDecimalNumeral :: Rule
|
|
ruleDecimalNumeral = Rule
|
|
{ name = "decimal number"
|
|
, pattern =
|
|
[ regex "(\\d*,\\d+)"
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token RegexMatch (GroupMatch (match:_)):_) -> parseDecimal False match
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleInteger3 :: Rule
|
|
ruleInteger3 = Rule
|
|
{ name = "integer 21..99"
|
|
, pattern =
|
|
[ oneOf [70, 20, 60, 50, 40, 90, 30, 80]
|
|
, numberBetween 1 10
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token Numeral (NumeralData {TNumeral.value = v1}):
|
|
Token Numeral (NumeralData {TNumeral.value = v2}):
|
|
_) -> double $ v1 + v2
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleIntersect :: Rule
|
|
ruleIntersect = Rule
|
|
{ name = "intersect"
|
|
, pattern =
|
|
[ numberWith (fromMaybe 0 . TNumeral.grain) (>1)
|
|
, numberWith TNumeral.multipliable not
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token Numeral (NumeralData {TNumeral.value = val1, TNumeral.grain = Just g}):
|
|
Token Numeral (NumeralData {TNumeral.value = val2}):
|
|
_) | (10 ** fromIntegral g) > val2 -> double $ val1 + val2
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleSomefewcouple :: Rule
|
|
ruleSomefewcouple = Rule
|
|
{ name = "some/few/couple"
|
|
, pattern =
|
|
[ regex "beberapa"
|
|
]
|
|
, prod = \_ -> integer 3
|
|
}
|
|
|
|
ruleNumeralsSuffixesKMG :: Rule
|
|
ruleNumeralsSuffixesKMG = Rule
|
|
{ name = "numbers suffixes (K, M, G)"
|
|
, pattern =
|
|
[ dimension Numeral
|
|
, regex "([kmg])(?=[\\W\\$€]|$)"
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token Numeral (NumeralData {TNumeral.value = v}):
|
|
Token RegexMatch (GroupMatch (match:_)):
|
|
_) -> case Text.toLower match of
|
|
"k" -> double $ v * 1e3
|
|
"m" -> double $ v * 1e6
|
|
"g" -> double $ v * 1e9
|
|
_ -> Nothing
|
|
_ -> Nothing
|
|
}
|
|
|
|
rulePowersOfTen :: Rule
|
|
rulePowersOfTen = Rule
|
|
{ name = "powers of tens"
|
|
, pattern =
|
|
[ regex "(se)?(ratus|ribu|juta)"
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token RegexMatch (GroupMatch (_:match:_)):_) -> case Text.toLower match of
|
|
"ratus"-> double 1e2 >>= withGrain 2 >>= withMultipliable
|
|
"ribu" -> double 1e3 >>= withGrain 3 >>= withMultipliable
|
|
"juta" -> double 1e6 >>= withGrain 6 >>= withMultipliable
|
|
_ -> Nothing
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleDozen :: Rule
|
|
ruleDozen = Rule
|
|
{ name = "dozen"
|
|
, pattern =
|
|
[ regex "(se)?lusin"
|
|
]
|
|
, prod = \_ -> integer 12 >>= withGrain 1 >>= withMultipliable
|
|
}
|
|
|
|
ruleInteger :: Rule
|
|
ruleInteger = Rule
|
|
{ name = "integer (0..9 11)"
|
|
, pattern =
|
|
[ regex "(kosong|nol|satu|dua|tiga|empat|lima|enam|tujuh|delapan|sembilan|sebelas)"
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token RegexMatch (GroupMatch (match:_)):_) ->
|
|
HashMap.lookup (Text.toLower match) ruleIntegerMap >>= integer
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleInteger2 :: Rule
|
|
ruleInteger2 = Rule
|
|
{ name = "integer 20..90"
|
|
, pattern =
|
|
[ numberBetween 2 10
|
|
, numberWith TNumeral.value (== 10)
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token Numeral (NumeralData {TNumeral.value = v1}):
|
|
Token Numeral (NumeralData {TNumeral.value = v2, TNumeral.grain = Just g}):
|
|
_) -> double (v1 * v2) >>= withGrain g
|
|
_ -> Nothing
|
|
}
|
|
|
|
ruleIntegerWithThousandsSeparator :: Rule
|
|
ruleIntegerWithThousandsSeparator = Rule
|
|
{ name = "integer with thousands separator ."
|
|
, pattern =
|
|
[ regex "(\\d{1,3}(\\.\\d\\d\\d){1,5})"
|
|
]
|
|
, prod = \tokens -> case tokens of
|
|
(Token RegexMatch (GroupMatch (match:_)):_) ->
|
|
parseDouble (Text.replace (Text.singleton '.') Text.empty match) >>= double
|
|
_ -> Nothing
|
|
}
|
|
|
|
rules :: [Rule]
|
|
rules =
|
|
[ ruleDecimalNumeral
|
|
, ruleDecimalWithThousandsSeparator
|
|
, ruleDozen
|
|
, ruleInteger
|
|
, ruleInteger2
|
|
, ruleInteger3
|
|
, ruleIntegerNumeric
|
|
, ruleIntegerWithThousandsSeparator
|
|
, ruleIntersect
|
|
, ruleMultiply
|
|
, ruleNumeralCommaNumeral
|
|
, ruleNumeralsPrefixWithNegativeOrMinus
|
|
, ruleNumeralsSuffixesKMG
|
|
, rulePowersOfTen
|
|
, ruleSomefewcouple
|
|
, ruleTeen
|
|
, ruleTen
|
|
]
|