[DUCK] repeat 增加 重复区间+时间点/区间 的能力

duckling-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/Helpers.scala

@@ -16,13 +16,16 @@
 
 package com.xiaomi.duckling.dimension.time
 
+import java.time.temporal.ChronoUnit
+
 import com.xiaomi.duckling.Types._
 import com.xiaomi.duckling.dimension.implicits._
 import com.xiaomi.duckling.dimension.time.enums.Grain
 import com.xiaomi.duckling.dimension.time.enums.Grain._
 import com.xiaomi.duckling.dimension.time.form.{TimeOfDay, Month => _}
 import com.xiaomi.duckling.dimension.time.helper.TimeDataHelpers.hour
 import com.xiaomi.duckling.dimension.time.predicates.{TimeDatePredicate, TimePredicate}
+import com.xiaomi.duckling.dimension.time.Types.InstantValue
 
 object Helpers {
 
@@ -66,4 +69,20 @@ object Helpers {
       } else td2
   }
 
+  def countGrains(start: InstantValue, end: InstantValue): Int = {
+    val a = start.datetime.toLocalDatetime
+    val b = end.datetime.toLocalDatetime
+    val n = start.grain match {
+      case NoGrain | Second => ChronoUnit.SECONDS.between(a, b)
+      case Minute => ChronoUnit.MINUTES.between(a, b)
+      case Hour => ChronoUnit.HOURS.between(a, b)
+      case Day => ChronoUnit.DAYS.between(a, b)
+      case Week => ChronoUnit.WEEKS.between(a, b)
+      case Month => ChronoUnit.MONTHS.between(a, b)
+      case Quarter => ChronoUnit.MONTHS.between(a, b) / 3
+      case Year => ChronoUnit.YEARS.between(a, b)
+    }
+    n.intValue()
+  }
+
 }

duckling-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/Rules.scala

@@ -373,8 +373,7 @@ trait Rules extends DimRules {
                 case TimeOfDay(_, _) => Day
                 case _               => NoGrain
               }
-            } else if (g == Year && td.timeGrain == Month) Year
-            else NoGrain
+            } else g
           val coarseDate = cycleNth(g, sign * v, roundGrain)
           tt(intersect(coarseDate, td))
         } else None

duckling-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.Buffer[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)
+      }
+    }
+  )
 }

duckling-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/predicates.scala

@@ -187,6 +187,11 @@ object predicates {
     case _ => false
   }
 
+  val isInterval: Predicate = {
+    case Token(Time, td: TimeData) => td.timePred.isInstanceOf[TimeIntervalsPredicate]
+    case _ => false
+  }
+
   val isIntervalOfDay: Predicate = {
     case Token(Time, td: TimeData) =>
       td.form match {

duckling-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/repeat/Repeat.scala

@@ -25,17 +25,20 @@ import com.xiaomi.duckling.dimension.time.grain.TimeGrain
 import com.xiaomi.duckling.dimension.time.Types._
 import com.xiaomi.duckling.dimension.time.enums.Grain
 import com.xiaomi.duckling.dimension.time.form.Form
+import com.xiaomi.duckling.dimension.time.Helpers.countGrains
 
 case object Repeat extends Dimension with Rules {
   override val name: String = "Repeat"
 
   override val dimDependents: List[Dimension] = List(TimeGrain, Duration, Time)
 }
 
-case class RepeatData(interval: Option[DurationData] = None,
+case class RepeatData(interval: Option[DurationData] = None, // 间隔，如果与其它的配合，表示外层间隔
                       n: Option[Int] = None,
                       start: Option[TimeData] = None,
-                      workdayType: Option[WorkdayType] = None)
+                      workdayType: Option[WorkdayType] = None,
+                      repeatGrain: Option[Grain] = None,
+                      repeatNFromInterval: Option[TimeData] = None)
     extends Resolvable {
 
   override def resolve(context: Context,
@@ -48,7 +51,19 @@ case class RepeatData(interval: Option[DurationData] = None,
         }
       case None => (None, true)
     }
-    if (success) Some(RepeatValue(interval, n, instant, workdayType), false)
+    val repeatN = repeatNFromInterval match {
+      case Some(intervalTimeData) =>
+        intervalTimeData.resolve(context, options) match {
+          case Some((tv: TimeValue, _)) =>
+            tv.timeValue match {
+              case IntervalValue(start, end) => Some(countGrains(start, end))
+              case _ => None
+            }
+          case _ => None
+        }
+      case _ => None
+    }
+    if (success) Some(RepeatValue(interval, n.orElse(repeatN), instant, repeatGrain = repeatNFromInterval.map(_.timeGrain), workdayType), false)
     else None
   }
 }
@@ -63,6 +78,7 @@ case class RepeatData(interval: Option[DurationData] = None,
 case class RepeatValue(interval: Option[DurationData] = None,
                        n: Option[Int] = None,
                        start:  Option[(TimeValue, Option[Form])] = None,
+                       repeatGrain: Option[Grain] = None,
                        workdayType: Option[WorkdayType] = None)
     extends ResolvedValue {
 

duckling-fork-chinese/core/src/main/scala/com/xiaomi/duckling/dimension/time/repeat/Rules.scala

@@ -22,11 +22,12 @@ import com.xiaomi.duckling.Types._
 import com.xiaomi.duckling.dimension.DimRules
 import com.xiaomi.duckling.dimension.implicits._
 import com.xiaomi.duckling.dimension.matcher.GroupMatch
-import com.xiaomi.duckling.dimension.matcher.Prods.{regexMatch, singleRegexMatch}
-import com.xiaomi.duckling.dimension.time.{form, Time, TimeData}
+import com.xiaomi.duckling.dimension.matcher.Prods.regexMatch
+import com.xiaomi.duckling.dimension.time.{form, GrainWrapper, Time, TimeData}
 import com.xiaomi.duckling.dimension.time.duration.{Duration, DurationData}
 import com.xiaomi.duckling.dimension.time.enums.{Grain, Hint}
-import com.xiaomi.duckling.dimension.time.predicates.{isAPartOfDay, isATimeOfDay, isHint, isNotLatent, isTimeDatePredicate, IntersectTimePredicate, TimeDatePredicate, TimeIntervalsPredicate}
+import com.xiaomi.duckling.dimension.time.helper.TimeDataHelpers.intersect
+import com.xiaomi.duckling.dimension.time.predicates._
 
 trait Rules extends DimRules with LazyLogging {
   /**
@@ -90,10 +91,12 @@ trait Rules extends DimRules with LazyLogging {
     }
   )
 
+  private val predicateEveryGrain = "每(一个?|个)?(年度?|月|周|星期|天|小时|分钟)的?".regex
+
   val ruleEveryGrainDatetime = Rule(
     name = "<every> <grain> <datetime>",
     pattern = List(
-      "每(一个?|个)?(年度?|月|周|星期|天|小时|分钟)的?".regex,
+      predicateEveryGrain,
       and(isDimension(Time), isNotLatent).predicate),
     prod = tokens {
       case Token(_, GroupMatch(_ :: _ :: grainToken :: _)) :: Token(_, td: TimeData) :: _
@@ -155,4 +158,54 @@ trait Rules extends DimRules with LazyLogging {
       workdaysTime(rd, td)
     }
   )
+
+  // 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(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))
+    }
+  )
+
+  val ruleEveryRepeat = Rule(
+    name = "每 x <repeat>",
+    pattern = List(predicateEveryGrain, isDimension(Repeat).predicate),
+    prod = tokens { case Token(_, GroupMatch(_ :: _ :: grainToken :: _)) :: Token(_, repeat: RepeatData):: _ =>
+      val grainHint: Option[Grain] = toGrain(grainToken)
+      (grainHint, repeat.repeatNFromInterval) match {
+        case (Some(everyGrain), Some(td)) if everyGrain >= td.timeGrain =>
+          val interval = DurationData(1, grainHint.getOrElse(everyGrain))
+          Token(Repeat, repeat.copy(interval = interval))
+        case _ => None
+      }
+    }
+  )
+
+  val ruleEveryRepeat1 = Rule(
+    name = "每 <repeat>",
+    pattern = List("每个?".regex, isDimension(Repeat).predicate),
+    prod = tokens { case _ :: Token(_, repeat: RepeatData):: _ =>
+      repeat.repeatNFromInterval match {
+        case Some(td) if td.timePred.maxGrain.isDefined =>
+          val interval = DurationData(1, td.timePred.maxGrain.get)
+          Token(Repeat, repeat.copy(interval = interval))
+        case _ => None
+      }
+    }
+  )
 }

duckling-fork-chinese/learning/src/main/scala/com/xiaomi/duckling/dimension/time/repeat/Examples.scala

@@ -34,6 +34,14 @@ object Examples extends DimExamples {
   implicit def _toTuple(tv: TimeValue) = Option(tv, None: Option[Form])
 
   override def pairs: List[(ResolvedValue, List[String])] = List(
+    (RepeatValue(interval = DurationData(1, Week), start = (datetimeInterval(
+      new DuckDateTime(LocalDateTime.of(2013, 2, 18, 12, 0, 0)),
+      new DuckDateTime(LocalDateTime.of(2013, 2, 18, 18, 0, 0)),
+      Hour), None), repeatGrain = Day, n = 3), List("每个周一到周三下午", "每周一到周三的下午")),
+    (RepeatValue(start = (datetimeInterval(
+      new DuckDateTime(LocalDateTime.of(2013, 2, 18, 12, 0, 0)),
+      new DuckDateTime(LocalDateTime.of(2013, 2, 18, 18, 0, 0)),
+      Hour), None), repeatGrain = Day, n = 3), List("周一到周三下午", "周一到周三的下午")),
     (RepeatValue(DurationData(1, Day, schema = "P1D")), List("每天")),
     (RepeatValue(DurationData(1, Week, schema = "P1W")), List("每周")),
     (RepeatValue(DurationData(15, Minute, schema = "PT15M")), List("每隔15分钟", "隔15分钟")),

duckling-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()
+      }
     }
   }
 

duckling-fork-chinese/project/build.properties

@@ -1 +1 @@
-sbt.version=1.8.0
+sbt.version=1.10.1