- 修复“一九五六年的昨天”抽取错误的问题 (#209)

- 精简相关规则
XiaoMi · Jan 26, 2024 · 05ec5a5 · 05ec5a5
1 parent 8e91e92
commit 05ec5a5
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 34 deletions.
diff --git a/...ling-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/date/Rules.scala b/...ling-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/date/Rules.scala
@@ -146,9 +146,12 @@ trait Rules extends DimRules {
 
   val ruleYearNumericWithYearSymbol = Rule(
     name = "date - year (numeric with year symbol)",
-    pattern = List(seqYearOf1000to9999.predicate, "(年?版|年)".regex),
+    pattern = List(seqYearOf1000to9999or00to99.predicate, "(年?版|年)".regex),
     prod = tokens {
-      case token :: _ => getIntValue(token).map(i => Token(Date, year(i.toInt)))
+      case token :: _ => getIntValue(token).map(i => {
+        val y = if (i < 30) i + 2000 else if (i < 100) i + 1900 else i
+        Token(Date, year(y.toInt))
+      })
     }
   )
 
@@ -174,34 +177,6 @@ trait Rules extends DimRules {
 
   val singleNumberPredicate = singleNumeber.predicate
 
-  val ruleTwoDigitYear = Rule(
-    name = "date - year (like 九八年)",
-    pattern = List(singleNumberPredicate, singleNumberPredicate, "(年?版|年)".regex),
-    prod = tokens {
-      case t1 :: t2 :: _ =>
-        val y = for {
-          thirdDigit <- getIntValue(t1)
-          fourthDigit <- getIntValue(t2)
-        } yield {
-          val lastTwo = thirdDigit * 10 + fourthDigit
-          if (thirdDigit >= 3) 1900 + lastTwo
-          else 2000 + lastTwo
-        }
-        y.map(i => Token(Date, year(i.toInt)))
-    }
-  )
-
-  val ruleTowDigitYear06 = Rule(
-    name = "date - year (like 06年)",
-    pattern = List(raw"\d{2}".regex, "(年?版|年)".regex),
-    prod = singleRegexMatch {
-      case s =>
-        val lastTwo = Integer.parseInt(s)
-        val y = adjustYear(lastTwo, s).get
-        Token(Date, year(y))
-    }
-  )
-
   val ruleMonthNumericWithMonthSymbol = Rule(
     name = "date: month (numeric with month symbol)",
     pattern = List(isIntegerBetween(1, 12).predicate, "月(份)?".regex),

diff --git a/...ling-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/predicates.scala b/...ling-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/predicates.scala
@@ -114,10 +114,15 @@ object predicates {
 
   val singleNumeber: Predicate = isIntegerBetween(0, 9)
 
-  val seqYearOf1000to9999: Predicate = {
-    case Token(DigitSequence, DigitSequenceData(seq, zh, raw)) if seq.length == 4 =>
-      val v = seq.toDouble
-      v > 1000 && v < 9999
+  val seqYearOf1000to9999or00to99: Predicate = {
+    case Token(DigitSequence, DigitSequenceData(seq, zh, raw)) =>
+      if (seq.length == 4) {
+        val v = seq.toDouble
+        v > 1000 && v < 9999
+      } else if (seq.length == 2) {
+        val v = seq.toDouble
+        v >= -0.1 && v < 99.1
+      } else false
   }
 
   val arabicSeqOf1950to2050: Predicate = {

diff --git a/...ng-fork-chinese/learning/src/main/scala/com/xiaomi/duckling/dimension/time/Examples.scala b/...ng-fork-chinese/learning/src/main/scala/com/xiaomi/duckling/dimension/time/Examples.scala
@@ -41,6 +41,7 @@ object Examples extends DimExamples {
     (ymd(2013, 2, 14), List("后天", "后日")),
     (ymd(2013, 2, 10), List("前天", "前日")),
     (ymd(2012, 2, 12), List("去年的今天")),
+    (ymd(1956, 2, 11), List("一九五六年的昨天")),
     (ymd(2010, 2, 12), List("去年的前年的今天")),
     (ymd(2013, 2, 15), List("明天的后天")),
     (ymd(2013, 2, 28), List("这个月月底")),