Support for intervals

Summary:
- Supports singular (like English "a")
- Supports intervals (e.g. EUR 100-200)
- Supports open intervals (like English "more than", "less than")

Reviewed By: haoxuany

Differential Revision: D16069296

fbshipit-source-id: 74808ad28f5673c2ca82f07dcc0934f82910b566

Duckling/AmountOfMoney/HR/Corpus.hs

@@ -23,11 +23,18 @@ corpus = (testContext {locale = makeLocale HR Nothing}, testOptions, allExamples
 
 allExamples :: [Example]
 allExamples = concat
-  [ examples (simple Dollar 10)
+  [ examples (simple Dollar 1)
+             [ "$1"
+             , "jedan dolar"
+             ]
+  , examples (simple Dollar 10)
              [ "$10"
              , "10$"
              , "deset dolara"
              ]
+  , examples (simple Cent 1)
+             [ "jedan cent"
+             ]
   , examples (simple Cent 10)
              [ "deset centa"
              ]
@@ -45,6 +52,9 @@ allExamples = concat
              , "2 dolara 23"
              , "dva dolara i 23"
              ]
+  , examples (simple HRK 1)
+             [ "jedna kuna"
+             ]
   , examples (simple HRK 2.23)
              [ "2 kune i 23 lipe"
              , "dvije kune 23 lipe"
@@ -61,6 +71,9 @@ allExamples = concat
              , "dvjesto kuna"
              , "dvjesta kuna"
              ]
+  , examples (simple EUR 1)
+             [ "jedan euro"
+             ]
   , examples (simple EUR 20)
              [ "20€"
              , "20 euros"
@@ -93,4 +106,28 @@ allExamples = concat
              [ "GBP3,01"
              , "GBP 3,01"
              ]
-  ]
+ , examples (between Dollar (100, 200))
+            [ "od 100 do 200 dolara"
+            , "od 100 dolara do 200 dolara"
+            , "otprilike 100 do 200 dolara"
+            , "otprilike 100 dolara do 200 dolara"
+            , "100 dolara - 200 dolara"
+            , "100-200 dolara"
+            , "oko 100-200 dolara"
+            , "u blizini 100-200 dolara"
+            , "skoro 100-200 dolara"
+            , "približno 100 do 200 dolara"
+            , "izmedju 100 i 200 dolara"
+            , "izmedju 100 dolara i 200 dolara"
+            ]
+  , examples (under EUR 7)
+             [ "manje od EUR 7"
+             , "ispod EUR 7"
+             ]
+  , examples (above Dollar 1.42)
+             [ "više od jednog dolara i četrdeset dva centa"
+             , "najmanje 1,42 dolara"
+             , "preko 1,42 dolara"
+             , "iznad 1,42 dolara"
+             ]
+ ]

Duckling/AmountOfMoney/HR/Rules.hs

@@ -28,6 +28,20 @@ import Duckling.Types
 import qualified Duckling.AmountOfMoney.Types as TAmountOfMoney
 import qualified Duckling.Numeral.Types as TNumeral
 
+ruleACurrency :: Rule
+ruleACurrency = Rule
+  { name = "a <currency>"
+  , pattern =
+    [ regex "jed(an|na|no)"
+    , Predicate isCurrencyOnly
+    ]
+  , prod = \tokens -> case tokens of
+      (_:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.currency = c}:
+       _) -> Just . Token AmountOfMoney $ withValue 1 $ currencyOnly c
+      _ -> Nothing
+  }
+
 ruleUnitAmount :: Rule
 ruleUnitAmount = Rule
   { name = "<unit> <amount>"
@@ -197,9 +211,172 @@ ruleAed = Rule
   , prod = \_ -> Just . Token AmountOfMoney $ currencyOnly AED
   }
 
+rulePrecision :: Rule
+rulePrecision = Rule
+  { name = "about|exactly <amount-of-money>"
+  , pattern =
+    [ regex "oko|otprilike|u blizini|skoro|približno"
+    , Predicate isMoneyWithValue
+    ]
+  , prod = \tokens -> case tokens of
+      (_:token:_) -> Just token
+      _ -> Nothing
+  }
+
+ruleIntervalNumeralDash :: Rule
+ruleIntervalNumeralDash = Rule
+  { name = "<numeral> - <amount-of-money>"
+  , pattern =
+    [ Predicate isNatural
+    , regex "-"
+    , Predicate isSimpleAmountOfMoney
+    ]
+  , prod = \tokens -> case tokens of
+      (Token Numeral NumeralData{TNumeral.value = from}:
+       _:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just to,
+                  TAmountOfMoney.currency = c}:
+       _) | from < to->
+         Just . Token AmountOfMoney . withInterval (from, to) $ currencyOnly c
+      _ -> Nothing
+  }
+
+ruleIntervalDash :: Rule
+ruleIntervalDash = Rule
+  { name = "<amount-of-money> - <amount-of-money>"
+  , pattern =
+    [ Predicate isSimpleAmountOfMoney
+    , regex "-"
+    , Predicate isSimpleAmountOfMoney
+    ]
+  , prod = \tokens -> case tokens of
+      (Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just from,
+                  TAmountOfMoney.currency = c1}:
+       _:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just to,
+                  TAmountOfMoney.currency = c2}:
+       _) | from < to && c1 == c2 ->
+        Just . Token AmountOfMoney . withInterval (from, to) $ currencyOnly c1
+      _ -> Nothing
+  }
+
+ruleIntervalBetweenNumeral1 :: Rule
+ruleIntervalBetweenNumeral1 = Rule
+  { name = "between|from <numeral> to|and <amount-of-money>"
+  , pattern =
+    [ regex "od|otprilike|približno"
+    , Predicate isPositive
+    , regex "do"
+    , Predicate isSimpleAmountOfMoney
+    ]
+  , prod = \tokens -> case tokens of
+      (_:
+       Token Numeral NumeralData{TNumeral.value = from}:
+       _:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just to,
+                  TAmountOfMoney.currency = c}:
+       _) | from < to ->
+        Just . Token AmountOfMoney . withInterval (from, to) $ currencyOnly c
+      _ -> Nothing
+  }
+
+ruleIntervalBetweenNumeral2 :: Rule
+ruleIntervalBetweenNumeral2 = Rule
+  { name = "between|from <numeral> to|and <amount-of-money>"
+  , pattern =
+    [ regex "izmedju"
+    , Predicate isPositive
+    , regex "i"
+    , Predicate isSimpleAmountOfMoney
+    ]
+  , prod = \tokens -> case tokens of
+      (_:
+       Token Numeral NumeralData{TNumeral.value = from}:
+       _:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just to,
+                  TAmountOfMoney.currency = c}:
+       _) | from < to ->
+        Just . Token AmountOfMoney . withInterval (from, to) $ currencyOnly c
+      _ -> Nothing
+  }
+
+ruleIntervalBetween1 :: Rule
+ruleIntervalBetween1 = Rule
+  { name = "between|from <amount-of-money> to|and <amount-of-money>"
+  , pattern =
+    [ regex "od|otprilike"
+    , Predicate isSimpleAmountOfMoney
+    , regex "do"
+    , Predicate isSimpleAmountOfMoney
+    ]
+  , prod = \tokens -> case tokens of
+      (_:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just from,
+                  TAmountOfMoney.currency = c1}:
+       _:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just to,
+                  TAmountOfMoney.currency = c2}:
+       _) | from < to && c1 == c2 ->
+        Just . Token AmountOfMoney . withInterval (from, to) $ currencyOnly c1
+      _ -> Nothing
+  }
+
+ruleIntervalBetween2 :: Rule
+ruleIntervalBetween2 = Rule
+  { name = "between|from <amount-of-money> to|and <amount-of-money>"
+  , pattern =
+    [ regex "izmedju"
+    , Predicate isSimpleAmountOfMoney
+    , regex "i"
+    , Predicate isSimpleAmountOfMoney
+    ]
+  , prod = \tokens -> case tokens of
+      (_:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just from,
+                  TAmountOfMoney.currency = c1}:
+       _:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just to,
+                  TAmountOfMoney.currency = c2}:
+       _) | from < to && c1 == c2 ->
+        Just . Token AmountOfMoney . withInterval (from, to) $ currencyOnly c1
+      _ -> Nothing
+  }
+
+ruleIntervalMin :: Rule
+ruleIntervalMin = Rule
+  { name = "over/above/at least/more than <amount-of-money>"
+  , pattern =
+    [ regex "više od|najmanje|preko|iznad"
+    , Predicate isSimpleAmountOfMoney
+    ]
+  , prod = \tokens -> case tokens of
+      (_:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just to,
+                  TAmountOfMoney.currency = c}:
+       _) -> Just . Token AmountOfMoney . withMin to $ currencyOnly c
+      _ -> Nothing
+  }
+
+ruleIntervalMax :: Rule
+ruleIntervalMax = Rule
+  { name = "under/less/lower/no more than <amount-of-money>"
+  , pattern =
+    [ regex "ispod|manje od"
+    , Predicate isSimpleAmountOfMoney
+    ]
+  , prod = \tokens -> case tokens of
+      (_:
+       Token AmountOfMoney AmountOfMoneyData{TAmountOfMoney.value = Just to,
+                  TAmountOfMoney.currency = c}:
+       _) -> Just . Token AmountOfMoney . withMax to $ currencyOnly c
+      _ -> Nothing
+  }
+
+
 rules :: [Rule]
 rules =
-  [ ruleUnitAmount
+  [ ruleACurrency
+  , ruleUnitAmount
   , ruleDollar
   , rulePound
   , ruleOtherPounds
@@ -214,4 +391,13 @@ rules =
   , ruleKwd
   , ruleQar
   , ruleSar
+  , rulePrecision
+  , ruleIntervalNumeralDash
+  , ruleIntervalDash
+  , ruleIntervalBetweenNumeral1
+  , ruleIntervalBetweenNumeral2
+  , ruleIntervalBetween1
+  , ruleIntervalBetween2
+  , ruleIntervalMin
+  , ruleIntervalMax
   ]

Duckling/Ranking/Classifiers/HR_XX.hs

@@ -1944,4 +1944,4 @@ classifiers
                                likelihoods = HashMap.fromList [("", 0.0)], n = 4},
                    koData =
                      ClassData{prior = -infinity, unseen = -0.6931471805599453,
-                               likelihoods = HashMap.fromList [], n = 0}})]
+                               likelihoods = HashMap.fromList [], n = 0}})]