mirror of
https://github.com/facebook/duckling.git
synced 2025-01-08 15:00:59 +03:00
Numeral/GA: HashMap lookups for large regexes
Summary: Use hashMap oneToTenMap instead of guard. Reviewed By: patapizza Differential Revision: D6414352 fbshipit-source-id: a1788df0d3521a3a870e453294708d9fe2c10908
This commit is contained in:
Jana Šefčíková
Facebook Github Bot
parent
8fbd6a0414
commit
7fe748ffec
@ -13,9 +13,12 @@ module Duckling.Numeral.GA.Rules
|
||||
( rules
|
||||
) where
|
||||
|
||||
import Data.HashMap.Strict (HashMap)
|
||||
import qualified Data.HashMap.Strict as HashMap
|
||||
import Data.String
|
||||
import Prelude
|
||||
import Data.Text (Text)
|
||||
import qualified Data.Text as Text
|
||||
import Prelude
|
||||
|
||||
import Duckling.Dimensions.Types
|
||||
import Duckling.Numeral.Helpers
|
||||
@ -51,6 +54,24 @@ ruleIntegerNumeric = Rule
|
||||
_ -> Nothing
|
||||
}
|
||||
|
||||
oneToTenMap :: HashMap Text Integer
|
||||
oneToTenMap = HashMap.fromList
|
||||
[ ("aon", 1)
|
||||
, ("dha", 2)
|
||||
, ("dhá", 2)
|
||||
, ("trí", 3)
|
||||
, ("tri", 3)
|
||||
, ("ceithre", 4)
|
||||
, ("cuig", 5)
|
||||
, ("cúig", 5)
|
||||
, ("sé", 6)
|
||||
, ("se", 6)
|
||||
, ("seacht", 7)
|
||||
, ("ocht", 8)
|
||||
, ("naoi", 9)
|
||||
, ("deich", 10)
|
||||
]
|
||||
|
||||
ruleNumerals2 :: Rule
|
||||
ruleNumerals2 = Rule
|
||||
{ name = "numbers, 1-10"
|
||||
@ -58,22 +79,8 @@ ruleNumerals2 = Rule
|
||||
[ regex "(aon|dh(á|a)|tr(í|i)|ceithre|c(ú|u)ig|seacht|s(é|e)|ocht|naoi|deich)"
|
||||
]
|
||||
, prod = \tokens -> case tokens of
|
||||
(Token RegexMatch (GroupMatch (match:_)):_) -> case Text.toLower match of
|
||||
"aon" -> integer 1
|
||||
"dha" -> integer 2
|
||||
"dhá" -> integer 2
|
||||
"trí" -> integer 3
|
||||
"tri" -> integer 3
|
||||
"ceithre" -> integer 4
|
||||
"cuig" -> integer 5
|
||||
"cúig" -> integer 5
|
||||
"sé" -> integer 6
|
||||
"se" -> integer 6
|
||||
"seacht" -> integer 7
|
||||
"ocht" -> integer 8
|
||||
"naoi" -> integer 9
|
||||
"deich" -> integer 10
|
||||
_ -> Nothing
|
||||
(Token RegexMatch (GroupMatch (match:_)):_) ->
|
||||
HashMap.lookup (Text.toLower match) oneToTenMap >>= integer
|
||||
_ -> Nothing
|
||||
}
|
||||
|
||||
@ -127,6 +134,17 @@ ruleNumeralsSuffixesKMG = Rule
|
||||
_ -> Nothing
|
||||
}
|
||||
|
||||
oldVigNumeralsSMap :: HashMap Text Integer
|
||||
oldVigNumeralsSMap = HashMap.fromList
|
||||
[ ("dá fhichead",40)
|
||||
, ("da fhichead", 40)
|
||||
, ("dhá fhichead", 40)
|
||||
, ("dha fhichead", 40)
|
||||
, ("trí fichid", 60)
|
||||
, ("tri fichid", 60)
|
||||
, ("ceithre fichid", 80)
|
||||
]
|
||||
|
||||
ruleOldVigesimalNumeralsS :: Rule
|
||||
ruleOldVigesimalNumeralsS = Rule
|
||||
{ name = "old vigesimal numbers, 20s"
|
||||
@ -134,18 +152,22 @@ ruleOldVigesimalNumeralsS = Rule
|
||||
[ regex "is (dh?(á|a) fhichead|tr(í|i) fichid|ceithre fichid)"
|
||||
]
|
||||
, prod = \tokens -> case tokens of
|
||||
(Token RegexMatch (GroupMatch (match:_)):_) -> case Text.toLower match of
|
||||
"dá fhichead" -> integer 40
|
||||
"da fhichead" -> integer 40
|
||||
"dhá fhichead" -> integer 40
|
||||
"dha fhichead" -> integer 40
|
||||
"trí fichid" -> integer 60
|
||||
"tri fichid" -> integer 60
|
||||
"ceithre fichid" -> integer 80
|
||||
_ -> Nothing
|
||||
(Token RegexMatch (GroupMatch (match:_)):_) ->
|
||||
HashMap.lookup (Text.toLower match) oldVigNumeralsSMap >>= integer
|
||||
_ -> Nothing
|
||||
}
|
||||
|
||||
oldVigNumeralsS2Map :: HashMap Text Integer
|
||||
oldVigNumeralsS2Map = HashMap.fromList
|
||||
[ ("fiche", 30)
|
||||
, ("dá fhichead", 50)
|
||||
, ("da fhichead", 50)
|
||||
, ("dha fhichead", 50)
|
||||
, ("trí fichid", 70)
|
||||
, ("tri fichid", 70)
|
||||
, ("ceithre fichid", 90)
|
||||
]
|
||||
|
||||
ruleOldVigesimalNumeralsS2 :: Rule
|
||||
ruleOldVigesimalNumeralsS2 = Rule
|
||||
{ name = "old vigesimal numbers, 20s + 10"
|
||||
@ -153,16 +175,8 @@ ruleOldVigesimalNumeralsS2 = Rule
|
||||
[ regex "d(é|e)ag is (fiche|dh?(á|a) fhichead|tr(í|i) fichid|ceithre fichid)"
|
||||
]
|
||||
, prod = \tokens -> case tokens of
|
||||
(Token RegexMatch (GroupMatch (match:_)):_) -> case Text.toLower match of
|
||||
"fiche" -> integer 30
|
||||
"dá fhichead" -> integer 50
|
||||
"da fhichead" -> integer 50
|
||||
"dhá fhichead" -> integer 50
|
||||
"dha fhichead" -> integer 50
|
||||
"trí fichid" -> integer 70
|
||||
"tri fichid" -> integer 70
|
||||
"ceithre fichid" -> integer 90
|
||||
_ -> Nothing
|
||||
(Token RegexMatch (GroupMatch (match:_)):_) ->
|
||||
HashMap.lookup (Text.toLower match) oldVigNumeralsS2Map >>= integer
|
||||
_ -> Nothing
|
||||
}
|
||||
|
||||
@ -175,6 +189,22 @@ ruleAmhin = Rule
|
||||
, prod = \_ -> integer 1
|
||||
}
|
||||
|
||||
twentyToNinetyMap :: HashMap Text Integer
|
||||
twentyToNinetyMap = HashMap.fromList
|
||||
[ ("fiche", 20)
|
||||
, ("triocha", 30)
|
||||
, ("tríocha", 30)
|
||||
, ("daichead", 40)
|
||||
, ("caoga", 50)
|
||||
, ("seasca", 60)
|
||||
, ("seachto", 70)
|
||||
, ("seachtó", 70)
|
||||
, ("ochto", 80)
|
||||
, ("ochtó", 80)
|
||||
, ("nócha", 90)
|
||||
, ("nocha", 90)
|
||||
]
|
||||
|
||||
ruleNumerals :: Rule
|
||||
ruleNumerals = Rule
|
||||
{ name = "numbers, 20-90"
|
||||
@ -182,20 +212,8 @@ ruleNumerals = Rule
|
||||
[ regex "(fiche|tr(í|i)ocha|daichead|caoga|seasca|seacht(ó|o)|ocht(ó|o)|n(ó|o)cha)"
|
||||
]
|
||||
, prod = \tokens -> case tokens of
|
||||
(Token RegexMatch (GroupMatch (match:_)):_) -> case Text.toLower match of
|
||||
"fiche" -> integer 20
|
||||
"triocha" -> integer 30
|
||||
"tríocha" -> integer 30
|
||||
"daichead" -> integer 40
|
||||
"caoga" -> integer 50
|
||||
"seasca" -> integer 60
|
||||
"seachto" -> integer 70
|
||||
"seachtó" -> integer 70
|
||||
"ochto" -> integer 80
|
||||
"ochtó" -> integer 80
|
||||
"nócha" -> integer 90
|
||||
"nocha" -> integer 90
|
||||
_ -> Nothing
|
||||
(Token RegexMatch (GroupMatch (match:_)):_) ->
|
||||
HashMap.lookup (Text.toLower match) twentyToNinetyMap >>= integer
|
||||
_ -> Nothing
|
||||
}
|
||||
|
||||
@ -211,6 +229,26 @@ ruleIntegerWithThousandsSeparator = Rule
|
||||
_ -> Nothing
|
||||
}
|
||||
|
||||
countNumeralsMap :: HashMap Text Integer
|
||||
countNumeralsMap = HashMap.fromList
|
||||
[ ("naid", 0)
|
||||
, ("náid", 0)
|
||||
, ("haon", 1)
|
||||
, ("dó", 2)
|
||||
, ("do", 2)
|
||||
, ("trí", 3)
|
||||
, ("tri", 3)
|
||||
, ("ceathair", 4)
|
||||
, ("cuig", 5)
|
||||
, ("cúig", 5)
|
||||
, ("sé", 6)
|
||||
, ("se", 6)
|
||||
, ("seacht", 7)
|
||||
, ("hocht", 8)
|
||||
, ("naoi", 9)
|
||||
, ("deich", 10)
|
||||
]
|
||||
|
||||
ruleCountNumerals :: Rule
|
||||
ruleCountNumerals = Rule
|
||||
{ name = "count numbers"
|
||||
@ -218,24 +256,8 @@ ruleCountNumerals = Rule
|
||||
[ regex "a (n(á|a)id|haon|d(ó|o)|tr(í|i)|ceathair|c(ú|u)ig|s(é|e)|seacht|hocht|naoi|deich)"
|
||||
]
|
||||
, prod = \tokens -> case tokens of
|
||||
(Token RegexMatch (GroupMatch (match:_)):_) -> case Text.toLower match of
|
||||
"naid" -> integer 0
|
||||
"náid" -> integer 0
|
||||
"haon" -> integer 1
|
||||
"dó" -> integer 2
|
||||
"do" -> integer 2
|
||||
"trí" -> integer 3
|
||||
"tri" -> integer 3
|
||||
"ceathair" -> integer 4
|
||||
"cuig" -> integer 5
|
||||
"cúig" -> integer 5
|
||||
"sé" -> integer 6
|
||||
"se" -> integer 6
|
||||
"seacht" -> integer 7
|
||||
"hocht" -> integer 8
|
||||
"naoi" -> integer 9
|
||||
"deich" -> integer 10
|
||||
_ -> Nothing
|
||||
(Token RegexMatch (GroupMatch (match:_)):_) ->
|
||||
HashMap.lookup (Text.toLower match) countNumeralsMap >>= integer
|
||||
_ -> Nothing
|
||||
}
|
||||
|
||||
|
||||
Loading…
Reference in New Issue
Block a user