Value before month in zh can be an integers or a chinese char (facebo…

…ok#301) Summary: May 2019 is more often written like' 2019年5月' than ' 2019年五月' so the corresponding integer are added in the regex's rule ruleMonths. Pull Request resolved: facebook#301 Reviewed By: haoxuany Differential Revision: D13606127 Pulled By: patapizza fbshipit-source-id: 7dd2536e36711b4be42a93419b08dbd66cf31523
reinfer · Jan 14, 2019 · a707be9 · a707be9
1 parent 7d6af43
commit a707be9
Show file tree

Hide file tree

Showing 7 changed files with 514 additions and 648 deletions.
diff --git a/Duckling/Ranking/Classifiers/ZH_CN.hs b/Duckling/Ranking/Classifiers/ZH_CN.hs
diff --git a/Duckling/Ranking/Classifiers/ZH_HK.hs b/Duckling/Ranking/Classifiers/ZH_HK.hs
diff --git a/Duckling/Ranking/Classifiers/ZH_MO.hs b/Duckling/Ranking/Classifiers/ZH_MO.hs
diff --git a/Duckling/Ranking/Classifiers/ZH_TW.hs b/Duckling/Ranking/Classifiers/ZH_TW.hs
diff --git a/Duckling/Ranking/Classifiers/ZH_XX.hs b/Duckling/Ranking/Classifiers/ZH_XX.hs
diff --git a/Duckling/Time/ZH/Corpus.hs b/Duckling/Time/ZH/Corpus.hs
@@ -275,26 +275,29 @@ allExamples = concat
              , "下礼拜"
              , "下禮拜"
              ]
-  , examples (datetime (2013, 1, 0, 0, 0, 0) Month)
+  , examples (datetime (2013, 1, 1, 0, 0, 0) Month)
              [ "上月"
              , "上个月"
              , "上個月"
              ]
-  , examples (datetime (2013, 3, 0, 0, 0, 0) Month)
+  , examples (datetime (2013, 3, 1, 0, 0, 0) Month)
              [ "下月"
              , "下个月"
              , "下個月"
+             , "3月"
+             , "3月份"
+             , "三月"
              ]
-  , examples (datetime (2012, 0, 0, 0, 0, 0) Year)
+  , examples (datetime (2012, 1, 1, 0, 0, 0) Year)
              [ "去年"
              , "上年"
              ]
-  , examples (datetime (2013, 0, 0, 0, 0, 0) Year)
+  , examples (datetime (2013, 1, 1, 0, 0, 0) Year)
              [ "今年"
              , "这一年"
              , "這一年"
              ]
-  , examples (datetime (2014, 0, 0, 0, 0, 0) Year)
+  , examples (datetime (2014, 1, 1, 0, 0, 0) Year)
              [ "明年"
              , "下年"
              ]

diff --git a/Duckling/Time/ZH/Rules.hs b/Duckling/Time/ZH/Rules.hs
@@ -764,12 +764,12 @@ ruleMonthNumericWithMonthSymbol = Rule
   { name = "month (numeric with month symbol)"
   , pattern =
     [ Predicate $ isIntegerBetween 1 12
-    , regex "月"
+    , regex "月(份)?"
     ]
   , prod = \tokens -> case tokens of
       (token:_) -> do
         v <- getIntValue token
-        tt . mkLatent $ month v
+        tt . mkOkForThisNext $ month v
       _ -> Nothing
   }
 
@@ -892,22 +892,6 @@ ruleDaysOfWeek = mkRuleDaysOfWeek
   , ( "Sunday", "星期日|星期天|礼拜天|周日|禮拜天|週日|禮拜日" )
   ]
 
-ruleMonths :: [Rule]
-ruleMonths = mkRuleMonths
-  [ ( "January", "一月(份)?" )
-  , ( "February", "二月(份)?" )
-  , ( "March", "三月(份)?" )
-  , ( "April", "四月(份)?" )
-  , ( "May", "五月(份)?" )
-  , ( "June", "六月(份)?" )
-  , ( "July", "七月(份)?" )
-  , ( "August", "八月(份)?" )
-  , ( "September", "九月(份)?" )
-  , ( "October", "十月(份)?" )
-  , ( "November", "十一月(份)?" )
-  , ( "December", "十二月(份)?" )
-  ]
-
 rulePeriodicHolidays :: [Rule]
 rulePeriodicHolidays = mkRuleHolidays
   -- Fixed dates, year over year
@@ -1039,5 +1023,4 @@ rules =
   , ruleTimezone
   ]
   ++ ruleDaysOfWeek
-  ++ ruleMonths
   ++ rulePeriodicHolidays