Time/HU: extend dates (#462)

Summary:
Extend Hungarian dates with new cases

Pull Request resolved: https://github.com/facebook/duckling/pull/462

Differential Revision: D25573636

Pulled By: chessai

fbshipit-source-id: 251698cf9f5126162ad4fbf1489dcbc4c12541ed

Duckling/Time/HU/Corpus.hs

@@ -16,7 +16,6 @@ import Prelude
 import Duckling.Locale
 import Duckling.Resolve
 import Duckling.Time.Corpus
-import Duckling.Time.Types hiding (Month)
 import Duckling.TimeGrain.Types hiding (add)
 import Duckling.Testing.Types hiding (examples)
 
@@ -28,62 +27,258 @@ allExamples = concat
   [ examples (datetime (2013, 2, 12, 4, 30, 0) Second)
              [ "most"
              , "épp most"
+             , "azonnal"
+             , "mostani"
              ]
   , examples (datetime (2013, 2, 12, 0, 0, 0) Day)
              [ "ma"
+             , "máma"
+             , "Mai nap"
+             , "Mai napot"
+             , "mai napon"
+             , "mai"
+             , "mait"
+             , "mai napi"
+             , "mai napit"
              ]
   , examples (datetime (2013, 2, 13, 0, 0, 0) Day)
              [ "holnap"
+             , "holnapi"
+             , "holnapit"
              ]
   , examples (datetime (2013, 2, 14, 0, 0, 0) Day)
              [ "holnapután"
+             , "holnaputáni"
+             , "holnaputánit"
              ]
   , examples (datetime (2013, 2, 11, 0, 0, 0) Day)
              [ "tegnap"
+             , "tegnapi"
+             , "tegnapit"
              ]
   , examples (datetime (2013, 2, 10, 0, 0, 0) Day)
              [ "tegnapelőtt"
+             , "tegnapelőtti"
+             , "tegnapelőttit"
              ]
   , examples (datetime (2013, 3, 1, 0, 0, 0) Month)
              [ "hónap vége"
              , "a hónap vége"
+             , "hó vége"
+             , "hó végi"
+             , "hó végit"
+             , "hó végén"
              ]
   , examples (datetime (2014, 1, 1, 0, 0, 0) Year)
              [ "év vége"
              , "az év vége"
+             , "év végi"
+             , "év végit"
+             , "év végén"
              ]
   , examples (datetime (2013, 2, 18, 0, 0, 0) Day)
              [ "hétfő"
-             , "hét"
+             , "hétfőn"
+             , "hétfőt"
+             , "hétfői"
+             , "hétfőit"
              , "hét."
+             , "hét"
              ]
   , examples (datetime (2013, 2, 19, 0, 0, 0) Day)
              [ "kedd"
+             , "kedden"
+             , "keddet"
+             , "keddi"
+             , "keddit"
+             ]
+  , examples (datetime (2013, 2, 13, 0, 0, 0) Day)
+             [ "szerda"
+             , "szerdán"
+             , "szerdát"
+             , "szerdai"
+             , "szerdait"
+             , "szer"
+             , "szer."
+             ]
+  , examples (datetime (2013, 2, 14, 0, 0, 0) Day)
+             [ "csütörtök"
+             , "csütörtökön"
+             , "csütörtököt"
+             , "csütörtöki"
+             , "csütörtökit"
+             , "csüt"
+             , "csüt."
+             ]
+  , examples (datetime (2013, 2, 15, 0, 0, 0) Day)
+             [ "péntek"
+             , "pénteken"
+             , "pénteket"
+             , "pénteki"
+             , "péntekit"
+             , "pén"
+             , "pén."
+             ]
+  , examples (datetime (2013, 2, 16, 0, 0, 0) Day)
+             [ "szombat"
+             , "szombaton"
+             , "szombatot"
+             , "szombati"
+             , "szombatit"
+             , "szom"
+             , "szom."
              ]
   , examples (datetime (2013, 2, 17, 0, 0, 0) Day)
              [ "vasárnap"
+             , "vasárnapot"
+             , "vasárnapi"
+             , "vasárnapit"
              , "vas"
              , "vas."
              ]
   , examples (datetime (2014, 1, 1, 0, 0, 0) Month)
              [ "január"
+             , "januárban"
+             , "januári"
+             , "januárit"
              , "jan"
              , "jan."
              ]
+  , examples (datetime (2013, 2, 1, 0, 0, 0) Month)
+             [ "február"
+             , "februárban"
+             , "februári"
+             , "februárit"
+             , "feb"
+             , "feb."
+             ]
   , examples (datetime (2013, 3, 1, 0, 0, 0) Month)
              [ "március"
+             , "márciusban"
+             , "márciusi"
+             , "márciusit"
              , "már"
              , "már."
              , "márc"
-             , "márc"
+             , "márc."
+             ]
+  , examples (datetime (2013, 4, 1, 0, 0, 0) Month)
+             [ "április"
+             , "áprilisban"
+             , "áprilisi"
+             , "áprilisit"
+             , "ápr"
+             , "ápr."
+             ]
+  , examples (datetime (2013, 5, 1, 0, 0, 0) Month)
+             [ "május"
+             , "májusban"
+             , "májusi"
+             , "májusit"
+             , "máj"
+             , "máj."
+             ]
+  , examples (datetime (2013, 6, 1, 0, 0, 0) Month)
+             [ "június"
+             , "júniusban"
+             , "júniusi"
+             , "júniusit"
+             , "jún"
+             , "jún."
+             ]
+  , examples (datetime (2013, 7, 1, 0, 0, 0) Month)
+             [ "július"
+             , "júliusban"
+             , "júliusi"
+             , "júliusit"
+             , "júl"
+             , "júl."
+             ]
+  , examples (datetime (2013, 8, 1, 0, 0, 0) Month)
+             [ "augusztus"
+             , "augusztusban"
+             , "augusztusi"
+             , "augusztusit"
+             , "aug"
+             , "aug."
+             ]
+  , examples (datetime (2013, 9, 1, 0, 0, 0) Month)
+             [ "szeptember"
+             , "szeptemberben"
+             , "szeptemberi"
+             , "szeptemberit"
+             , "szep"
+             , "szep."
+             , "szept"
+             , "szept."
+             ]
+  , examples (datetime (2013, 10, 1, 0, 0, 0) Month)
+             [ "október"
+             , "októberben"
+             , "októberi"
+             , "októberit"
+             , "okt"
+             , "okt."
+             ]
+  , examples (datetime (2013, 11, 1, 0, 0, 0) Month)
+             [ "november"
+             , "novemberben"
+             , "novemberi"
+             , "novemberit"
+             , "nov"
+             , "nov."
+             ]
+  , examples (datetime (2013, 12, 1, 0, 0, 0) Month)
+             [ "december"
+             , "decemberben"
+             , "decemberi"
+             , "decemberit"
+             , "dec"
+             , "dec."
              ]
   , examples (datetime (2013, 3, 15, 0, 0, 0) Day)
              [ "március 15"
+             , "március 15."
              , "már 15"
              , "már. 15"
              , "márc 15"
              , "márc. 15"
              ]
+  , examples (datetime (2013, 3, 1, 0, 0, 0) Day)
+            [ "március 1-én"
+            , "március 1-jén"
+            , "március 1-je"
+            , "március 1-jei"
+            , "március 1-i"
+            , "március 1én"
+            , "március 1je"
+            , "március 1jei"
+            , "március 1i"
+            ]
+  , examples (datetime (2013, 3, 2, 0, 0, 0) Day)
+            [ "március 2-án"
+            , "március 2-ai"
+            , "március 2-a"
+            , "március 2án"
+            , "március 2ai"
+            , "március 2a"
+            ]
+  , examples (datetime (2013, 3, 4, 0, 0, 0) Day)
+            [ "március 4-e"
+            , "március 4e"
+            ]
+  , examples (datetime (2013, 3, 7, 0, 0, 0) Day)
+            [ "március 7-ei"
+            , "március 7ei"
+            ]
+  , examples (datetime (2013, 3, 15, 0, 0, 0) Day)
+            [ "március 15-én"
+            , "március 15-ei"
+            , "március 15-i"
+            , "március 15én"
+            , "március 15e"
+            , "március 15ei"
+            ]
   , examples (datetime (2013, 3, 1, 0, 0, 0) Month)
              [ "következő hónap"
              , "jövő hónap"
@@ -150,6 +345,20 @@ allExamples = concat
   , examples (datetimeInterval ((2013, 2, 12, 6, 0, 0), (2013, 2, 12, 10, 0, 0)) Hour)
              [ "ma reggel"
              , "reggel"
+             , "ma reggeli"
+             , "reggeli"
+             , "ma reggelit"
+             , "reggelit"
+             ]
+  , examples (datetimeInterval ((2013, 2, 12, 8, 0, 0), (2013, 2, 12, 12, 0, 0)) Hour)
+             [ "délelőtt"
+             , "délelőtti"
+             , "délelőttit"
+             ]
+  , examples (datetimeInterval ((2013, 2, 12, 12, 0, 0), (2013, 2, 12, 13, 0, 0)) Hour)
+             [ "délben"
+             , "déli"
+             , "délit"
              ]
   , examples (datetimeInterval ((2013, 2, 13, 8, 0, 0), (2013, 2, 13, 12, 0, 0)) Hour)
              [ "holnap délelőtt"
@@ -160,16 +369,43 @@ allExamples = concat
   , examples (datetimeInterval ((2013, 2, 12, 12, 0, 0), (2013, 2, 12, 18, 0, 0)) Hour)
              [ "ma délután"
              , "délután"
+             , "délutáni"
+             , "délutánit"
              ]
   , examples (datetimeInterval ((2013, 2, 12, 16, 0, 0), (2013, 2, 12, 20, 0, 0)) Hour)
              [ "ma este"
              , "este"
+             , "esti"
+             , "estit"
              ]
   , examples (datetimeInterval ((2013, 2, 12, 20, 0, 0), (2013, 2, 12, 23, 0, 0)) Hour)
              [ "ma éjszaka"
              , "éjszaka"
+             , "éjszakai"
+             , "éjszakait"
              ]
   , examples (datetimeInterval ((2013, 6, 21, 0, 0, 0), (2013, 9, 24, 0, 0, 0)) Day)
              [ "nyár"
+             , "nyári"
+             , "nyárit"
+             , "nyáron"
+             ]
+   , examples (datetimeInterval ((2013, 9, 23, 0, 0, 0), (2013, 12, 22, 0, 0, 0)) Day)
+             [ "ősz"
+             , "őszi"
+             , "őszit"
+             , "ősszel"
+             ]
+  , examples (datetimeInterval ((2012, 12, 21, 0, 0, 0), (2013, 3, 21, 0, 0, 0)) Day)
+             [ "tél"
+             , "téli"
+             , "télit"
+             , "télen"
+             ]
+    , examples (datetimeInterval ((2013, 3, 20, 0, 0, 0), (2013, 6, 22, 0, 0, 0)) Day)
+             [ "tavasz"
+             , "tavaszi"
+             , "tavaszit"
+             , "tavasszal"
              ]
   ]

Duckling/Time/HU/Rules.hs

@@ -13,8 +13,6 @@ module Duckling.Time.HU.Rules
   ( rules ) where
 
 import Data.Maybe
-import Data.String
-import Data.Text (Text)
 import Prelude
 import qualified Data.Text as Text
 
@@ -41,41 +39,41 @@ ruleIntersect = Rule
 
 ruleInstants :: [Rule]
 ruleInstants = mkRuleInstants
-  [ ("right now",            TG.Second, 0 , "((\x00E9pp )?most)|azonnal" )
+  [ ("right now",            TG.Second, 0 , "mostani(t)?|((\x00E9pp )?most)|azonnal"                      )
-  , ("today",                TG.Day,    0 , "ma"                         )
+  , ("today",                TG.Day,    0 , "m(á|a)(i( nap(o(t|n)|it?)?|t)?|ma)?"                         )
-  , ("tomorrow",             TG.Day,    1 , "holnap"                     )
+  , ("tomorrow",             TG.Day,    1 , "holnap(i(t)?)?"                                              )
-  , ("day after tomorrow",   TG.Day,    2 , "holnaput\x00E1n"            )
+  , ("day after tomorrow",   TG.Day,    2 , "holnaput\x00E1n(i(t)?)?"                                     )
-  , ("yesterday",            TG.Day,    -1, "tegnap"                     )
+  , ("yesterday",            TG.Day,    -1, "tegnap(i(t)?)?"                                              )
-  , ("day before yesterday", TG.Day,    -2, "tegnapel\x0151tt"           )
+  , ("day before yesterday", TG.Day,    -2, "tegnapel\x0151tt(i(t)?)?"                                    )
-  , ("end of month",         TG.Month,  1 , "(a )?h\x00F3nap v\x00E9ge"  )
+  , ("end of month",         TG.Month,  1 , "(a )?h\x00F3nap v\x00E9ge|h\x00F3 v\x00E9g(e|i(t)?|\x00E9n)" )
-  , ("end of year",          TG.Year,   1 , "(az )?\x00E9v v\x00E9ge"    )
+  , ("end of year",          TG.Year,   1 , "(az )?\x00E9v v\x00E9g(e|i(t)?|\x00E9n)"                     )
   ]
 
 ruleDaysOfWeek :: [Rule]
 ruleDaysOfWeek = mkRuleDaysOfWeek
-  [ ( "Monday"   , "h\x00E9tf\x0151|h\x00E9t\\.?"           )
+  [ ( "Monday"   , "h\x00E9tf\x0151(n|t|i(t)?)?|h\x00E9t\\.?"                   )
-  , ( "Tuesday"  , "kedd"                                   )
+  , ( "Tuesday"  , "kedd(en|et|i(t)?)?"                                         )
-  , ( "Wednesday", "szerda|szer\\.?"                        )
+  , ( "Wednesday", "szerda(i(t)?)?|szerdá(n|t)|szer\\.?"                        )
-  , ( "Thursday" , "cs\x00FCt\x00F6rt\x00F6k|cs\x00FCt\\.?" )
+  , ( "Thursday" , "cs\x00FCt\x00F6rt\x00F6k(\x00F6(n|t)|i(t)?)?|cs\x00FCt\\.?" )
-  , ( "Friday"   , "p\x00E9ntek|p\x00E9n\\.?"               )
+  , ( "Friday"   , "p\x00E9ntek(e(n|t)|i(t)?)?|p\x00E9n\\.?"                    )
-  , ( "Saturday" , "szombat|szom\\.?"                       )
+  , ( "Saturday" , "szombat(o(n|t)|i(t)?)?|szom\\.?"                            )
-  , ( "Sunday"   , "vas\x00E1rnap|vas\\.?"                  )
+  , ( "Sunday"   , "vas\x00E1rnap(ot|i(t)?)?|vas\\.?"                           )
   ]
 
 ruleMonths :: [Rule]
 ruleMonths = mkRuleMonths
-  [ ( "January"  , "janu\x00E1r|jan\\.?"         )
+  [ ( "January"  , "janu\x00E1r(ban|i(t)?)?|jan\\.?"         )
-  , ( "February" , "febru\x00E1r|febr?\\.?"      )
+  , ( "February" , "febru\x00E1r(ban|i(t)?)?|febr?\\.?"      )
-  , ( "March"    , "m\x00E1rcius|m\x00E1rc?\\.?" )
+  , ( "March"    , "m\x00E1rcius(ban|i(t)?)?|m\x00E1rc?\\.?" )
-  , ( "April"    , "\x00E1prilis|\x00E1pr\\.?"   )
+  , ( "April"    , "\x00E1prilis(ban|i(t)?)?|\x00E1pr\\.?"   )
-  , ( "May"      , "m\x00E1jus|m\x00E1j\\.?"     )
+  , ( "May"      , "m\x00E1jus(ban|i(t)?)?|m\x00E1j\\.?"     )
-  , ( "June"     , "j\x00FAnius|j\x00FAn\\.?"    )
+  , ( "June"     , "j\x00FAnius(ban|i(t)?)?|j\x00FAn\\.?"    )
-  , ( "July"     , "j\x00FAlius|j\x00FAl\\.?"    )
+  , ( "July"     , "j\x00FAlius(ban|i(t)?)?|j\x00FAl\\.?"    )
-  , ( "August"   , "augusztus|aug\\.?"           )
+  , ( "August"   , "augusztus(ban|i(t)?)?|aug\\.?"           )
-  , ( "September", "szeptember|szept?\\.?"       )
+  , ( "September", "szeptember(ben|i(t)?)?|szept?\\.?"       )
-  , ( "October"  , "okt\x00F3ber|okt\\.?"        )
+  , ( "October"  , "október(ben|i(t)?)?|okt\\.?"             )
-  , ( "November" , "november|nov\\.?"            )
+  , ( "November" , "november(ben|i(t)?)?|nov\\.?"            )
-  , ( "December" , "december|dec\\.?"            )
+  , ( "December" , "december(ben|i(t)?)?|dec\\.?"            )
   ]
 
 ruleMonthDOMNumeral :: Rule
@@ -90,6 +88,31 @@ ruleMonthDOMNumeral = Rule
       _ -> Nothing
   }
 
+ruleMonthDOMNumeralSuffix :: Rule
+ruleMonthDOMNumeralSuffix = Rule
+  { name = "<named-month> <day-of-month> (non ordinal)"
+  , pattern =
+    [ Predicate isAMonth
+    , Predicate isDOMInteger
+    , regex "(-?(jei|jén|án|én|ai|ei|je|ji|a|e|i))"
+    ]
+  , prod = \tokens -> case tokens of
+      (Token Time td:token:_) -> Token Time <$> intersectDOM td token
+      _ -> Nothing
+  }
+
+ruleMonthDOMOrdinal :: Rule
+ruleMonthDOMOrdinal = Rule
+  { name = "<named-month> <day-of-month> (ordinal)"
+  , pattern =
+    [ Predicate isAMonth
+    , Predicate isDOMOrdinal
+    ]
+  , prod = \tokens -> case tokens of
+      (Token Time td:token:_) -> Token Time <$> intersectDOM td token
+      _ -> Nothing
+  }
+
 ruleCycleThisLastNext :: Rule
 ruleCycleThisLastNext = Rule
   { name = "this|last|next <cycle>"
@@ -220,17 +243,27 @@ rulePartOfDays :: Rule
 rulePartOfDays = Rule
   { name = "part of days"
   , pattern =
-    [ regex "(reggel|d\x00E9lel\x0151tt|d\x00E9lben|d\x00E9lut\x00E1n|este|\x00E9jszaka)"
+    [ regex "(reggel(i(t)?)?|d\x00E9lel\x0151tt(i(t)?)?|d\x00E9lben|d\x00E9li(t)?|d\x00E9lut\x00E1n(i(t)?)?|est(e|i(t)?)|\x00E9jszaka(i(t)?)?)"
     ]
   , prod = \tokens -> case tokens of
       (Token RegexMatch (GroupMatch (match:_)):_) -> do
         let (start, end) = case Text.toLower match of
-              "reggel"             -> (hour False 6, hour False 10)
+              "reggel"               -> (hour False 6, hour False 10)
-              "d\x00E9lel\x0151tt" -> (hour False 08, hour False 12)
+              "reggeli"              -> (hour False 6, hour False 10)
-              "d\x00E9lben"        -> (hour False 12, hour False 13)
+              "reggelit"             -> (hour False 6, hour False 10)
-              "d\x00E9lut\x00E1n"  -> (hour False 12, hour False 18)
+              "d\x00E9lel\x0151tt"   -> (hour False 08, hour False 12)
-              "este"               -> (hour False 16, hour False 20)
+              "d\x00E9lel\x0151tti"  -> (hour False 08, hour False 12)
-              _                    -> (hour False 20, hour False 23)
+              "d\x00E9lel\x0151ttit" -> (hour False 08, hour False 12)
+              "d\x00E9lben"          -> (hour False 12, hour False 13)
+              "d\x00E9li"            -> (hour False 12, hour False 13)
+              "d\x00E9lit"           -> (hour False 12, hour False 13)
+              "d\x00E9lut\x00E1n"    -> (hour False 12, hour False 18)
+              "d\x00E9lut\x00E1ni"   -> (hour False 12, hour False 18)
+              "d\x00E9lut\x00E1nit"  -> (hour False 12, hour False 18)
+              "este"                 -> (hour False 16, hour False 20)
+              "esti"                 -> (hour False 16, hour False 20)
+              "estit"                -> (hour False 16, hour False 20)
+              _                      -> (hour False 20, hour False 23)
         td <- interval TTime.Open start end
         tt . partOfDay $ td
       _ -> Nothing
@@ -253,21 +286,45 @@ ruleSeasons :: Rule
 ruleSeasons = Rule
   { name = "seasons"
   , pattern =
-    [ regex "(ny\x00E1r|\x0151sz|t\x00E9l|tavasz)"
+    [ regex "(ny\x00E1r(i(t)?|on)?|\x0151sz(i(t)?)?|\x0151sszel|t\x00E9l(i(t)?|en)?|tavasz(i(t)?)?|tavasszal)"
     ]
   , prod = \tokens -> case tokens of
       (Token RegexMatch (GroupMatch (match:_)):_) -> do
         start <- case Text.toLower match of
-          "ny\x00E1r" -> Just $ monthDay 6 21
+          "ny\x00E1r"      -> Just $ monthDay 6 21
-          "\x0151sz"  -> Just $ monthDay 9 23
+          "ny\x00E1ri"     -> Just $ monthDay 6 21
-          "t\x00E9l"  -> Just $ monthDay 12 21
+          "ny\x00E1rit"    -> Just $ monthDay 6 21
-          "tavasz"    -> Just $ monthDay 3 20
+          "ny\x00E1ron"    -> Just $ monthDay 6 21
+          "\x0151sz"       -> Just $ monthDay 9 23
+          "\x0151szi"      -> Just $ monthDay 9 23
+          "\x0151szit"     -> Just $ monthDay 9 23
+          "\x0151sszel"    -> Just $ monthDay 9 23
+          "t\x00E9l"       -> Just $ monthDay 12 21
+          "t\x00E9li"      -> Just $ monthDay 12 21
+          "t\x00E9lit"     -> Just $ monthDay 12 21
+          "t\x00E9len"     -> Just $ monthDay 12 21
+          "tavasz"         -> Just $ monthDay 3 20
+          "tavaszi"        -> Just $ monthDay 3 20
+          "tavaszit"       -> Just $ monthDay 3 20
+          "tavasszal"      -> Just $ monthDay 3 20
           _ -> Nothing
         end <- case Text.toLower match of
-          "ny\x00E1r" -> Just $ monthDay 9 23
+          "ny\x00E1r"      -> Just $ monthDay 9 23
-          "\x0151sz"  -> Just $ monthDay 12 21
+          "ny\x00E1ri"     -> Just $ monthDay 9 23
-          "t\x00E9l"  -> Just $ monthDay 3 20
+          "ny\x00E1rit"    -> Just $ monthDay 9 23
-          "tavasz"    -> Just $ monthDay 6 21
+          "ny\x00E1ron"    -> Just $ monthDay 9 23
+          "\x0151sz"       -> Just $ monthDay 12 21
+          "\x0151szi"      -> Just $ monthDay 12 21
+          "\x0151szit"     -> Just $ monthDay 12 21
+          "\x0151sszel"    -> Just $ monthDay 12 21
+          "t\x00E9l"       -> Just $ monthDay 3 20
+          "t\x00E9li"      -> Just $ monthDay 3 20
+          "t\x00E9lit"     -> Just $ monthDay 3 20
+          "t\x00E9len"     -> Just $ monthDay 3 20
+          "tavasz"         -> Just $ monthDay 6 21
+          "tavaszi"        -> Just $ monthDay 6 21
+          "tavaszit"       -> Just $ monthDay 6 21
+          "tavasszal"      -> Just $ monthDay 6 21
           _ -> Nothing
         Token Time <$> interval TTime.Open start end
       _ -> Nothing
@@ -278,6 +335,8 @@ rules :: [Rule]
 rules =
   [ ruleIntersect
   , ruleMonthDOMNumeral
+  , ruleMonthDOMNumeralSuffix
+  , ruleMonthDOMOrdinal
   , ruleCycleThisLastNext
   , ruleNextDOW
   , ruleHHMM

tests/Duckling/Time/HU/Tests.hs

@@ -8,7 +8,6 @@ module Duckling.Time.HU.Tests
   ( tests ) where
 
 import Data.String
-import Prelude
 import Test.Tasty
 
 import Duckling.Dimensions.Types