[WIP] repeat 增加重复区间+时间点/区间的能力

XiaoMi · Jul 31, 2024 · ef11b77 · ef11b77
1 parent 195850e
commit ef11b77
Show file tree

Hide file tree

Showing 3 changed files with 58 additions and 8 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
@@ -18,7 +18,9 @@ package com.xiaomi.duckling.dimension.time.date
 
 import scalaz.std.string.parseInt
 import java.time.LocalDate
+import java.util.regex.Pattern
 
+import scala.collection.mutable
 import scala.util.Try
 
 import com.xiaomi.duckling.Types._
@@ -33,7 +35,7 @@ import com.xiaomi.duckling.dimension.time.duration.{isADecade, Duration, Duratio
 import com.xiaomi.duckling.dimension.time.enums.Grain._
 import com.xiaomi.duckling.dimension.time.enums.Hint.{NoHint, RecentNominal, YearMonth}
 import com.xiaomi.duckling.dimension.time.enums.IntervalType.{Closed, Open}
-import com.xiaomi.duckling.dimension.time.enums.{Grain, Hint}
+import com.xiaomi.duckling.dimension.time.enums.{Grain, Hint, IntervalType}
 import com.xiaomi.duckling.dimension.time.helper.TimeDataHelpers._
 import com.xiaomi.duckling.dimension.time.predicates.{EndOfGrainPredicate, SequencePredicate, TimeDatePredicate, _}
 import com.xiaomi.duckling.dimension.time.{GrainWrapper, TimeData}
@@ -398,4 +400,38 @@ trait Rules extends DimRules {
         }
     }
   )
+
+  private val WeekX = Pattern.compile("(周|星期)([一二三四五六日天])")
+  private val WeekDayStr = "12345671234567"
+
+  val ruleWeekXyz = Rule(
+    name = "周一周二",
+    pattern = List("((周|星期)[一二三四五六日天]){2,}".regex),
+    prod = regexMatch {case text :: _ =>
+      val m = WeekX.matcher(text)
+      var pos = 0
+      val days = mutable.MutableList[Int]()
+      while (m.find(pos)) {
+        val d = m.group(2) match {
+          case "一" => 1
+          case "二" => 2
+          case "三" => 3
+          case "四" => 4
+          case "五" => 5
+          case "六" => 6
+          case "日" | "天" => 7
+        }
+        days += d
+        pos = m.end()
+      }
+      WeekDayStr.indexOf(days.mkString("")) match {
+        case -1 => None
+        case _ =>
+          val td = TimeData(
+          TimeIntervalsPredicate(IntervalType.Closed, TimeDatePredicate(dayOfWeek = days.head), TimeDatePredicate(dayOfWeek = days.last), beforeEndOfInterval = true),
+          timeGrain = Day)
+          Token(Date, td)
+      }
+    }
+  )
 }
diff --git a/...ng-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/repeat/Rules.scala b/...ng-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/repeat/Rules.scala
@@ -157,14 +157,25 @@ trait Rules extends DimRules with LazyLogging {
     }
   )
 
-  // 周一到周五的早上八点
+  // 周一到周五早上八点
   val ruleIntervalTime = Rule(
     name = "<interval> <time/interval>",
     pattern = List(isInterval.predicate, isDimension(Time).predicate),
     prod = tokens { case Token(Time, outer: TimeData) :: Token(Time, inner: TimeData):: _ if outer.timeGrain > inner.timeGrain =>
       val oInterval = outer.timePred.asInstanceOf[TimeIntervalsPredicate]
       // start
-      val start = intersect(TimeData(oInterval.p1, timeGrain=outer.timeGrain), inner)
+      val start = intersect(inner, TimeData(oInterval.p1, timeGrain=outer.timeGrain))
+      Token(Repeat, RepeatData(start = start, repeatNFromInterval = outer))
+    }
+  )
+
+  // 周一到周五早上的八点
+  val ruleIntervalTime1 = Rule(
+    name = "<interval> 的 <time/interval>",
+    pattern = List(isInterval.predicate, "的".regex, isDimension(Time).predicate),
+    prod = tokens { case Token(Time, outer: TimeData) :: _ :: Token(Time, inner: TimeData):: _ if outer.timeGrain > inner.timeGrain =>
+      val oInterval = outer.timePred.asInstanceOf[TimeIntervalsPredicate]
+      val start = intersect(inner.copy(hint = Hint.NoHint), TimeData(oInterval.p1, timeGrain=outer.timeGrain).copy(hint = Hint.NoHint))
       Token(Repeat, RepeatData(start = start, repeatNFromInterval = outer))
     }
   )

diff --git a/...ing-fork-chinese/learning/src/main/scala/com/xiaomi/duckling/task/NaiveBayesConsole.scala b/...ing-fork-chinese/learning/src/main/scala/com/xiaomi/duckling/task/NaiveBayesConsole.scala
@@ -19,6 +19,8 @@ package com.xiaomi.duckling.task
 import java.nio.charset.StandardCharsets
 import java.time.ZonedDateTime
 
+import scala.util.control.Breaks.break
+
 import org.fusesource.jansi.Ansi
 import org.jline.reader.{EndOfFileException, LineReader, LineReaderBuilder}
 import org.jline.reader.impl.completer._
@@ -143,7 +145,7 @@ object NaiveBayesConsole extends LazyLogging {
     } else (false, options)
   }
 
-  def round(reader: LineReader, options: Options): Options = {
+  def round(reader: LineReader, options: Options): (Boolean, Options) = {
     reader.getTerminal.flush()
 
     val line =
@@ -152,7 +154,6 @@ object NaiveBayesConsole extends LazyLogging {
       } catch {
         case ex: EndOfFileException =>
           logger.info("bye!")
-          System.exit(0)
           ""
       }
 
@@ -175,11 +176,11 @@ object NaiveBayesConsole extends LazyLogging {
       }
     }
 
-    _options
+    (line == "", _options)
   }
 
   def run(): Unit = {
-    var options = Options(targets = Set(), withLatent = false)
+    var options: Options = Options(targets = Set(), withLatent = false)
     options.rankOptions.setRanker(Ranker.NaiveBayes)
     options.rankOptions.setWinnerOnly(true)
     options.rankOptions.setCombinationRank(false)
@@ -197,7 +198,9 @@ object NaiveBayesConsole extends LazyLogging {
 
     val reader = buildLineReader()
     while (true) {
-      options = round(reader, options)
+      round(reader, options) match {
+        case (true, _) => break()
+      }
     }
   }