Package org.apache.druid.java.util.common.granularity

Class Granularity

java.lang.Object
org.apache.druid.java.util.common.granularity.Granularity
All Implemented Interfaces:
Cacheable
Direct Known Subclasses:
AllGranularity, DurationGranularity, NoneGranularity, PeriodGranularity
public abstract class Granularity extends Object implements Cacheable

  • Field Details

  • Constructor Details

    • Granularity

      public Granularity()

  • Method Details

    • fromString

      public static Granularity fromString(String str)

    • mergeGranularities

      public static Granularity mergeGranularities(List<Granularity> toMerge)
      simple merge strategy on query granularity that checks if all are equal or else returns null. this can be improved in future but is good enough for most use-cases.

    • granularitiesFinerThan

      public static List<Granularity> granularitiesFinerThan(Granularity gran0)
      Returns a list of standard granularities that are equal to, or finer than, a provided granularity. ALL will not be returned unless the provided granularity is ALL. NONE will never be returned, even if the provided granularity is NONE. This is because the main usage of this function in production is segment allocation, and we do not wish to generate NONE-granular segments. The list of granularities returned contains WEEK only if the requested granularity is WEEK.

    • getFormatter

      public abstract org.joda.time.format.DateTimeFormatter getFormatter(Granularity.Formatter type)

    • increment

      public abstract long increment(long time)

    • increment

      public abstract org.joda.time.DateTime increment(org.joda.time.DateTime time)

    • bucketStart

      public abstract long bucketStart(long time)

    • bucketStart

      public abstract org.joda.time.DateTime bucketStart(org.joda.time.DateTime time)

    • toDate

      public abstract org.joda.time.DateTime toDate(String filePath, Granularity.Formatter formatter)

    • isAligned

      public abstract boolean isAligned(org.joda.time.Interval interval)
      Return true only if the time chunks populated by this granularity includes the given interval time chunk. The interval must fit exactly into the scheme of the granularity for this to return true

    • getTimeZone

      public org.joda.time.DateTimeZone getTimeZone()

    • bucketEnd

      public org.joda.time.DateTime bucketEnd(org.joda.time.DateTime time)

    • toDateTime

      public org.joda.time.DateTime toDateTime(long offset)

    • toDate

      public org.joda.time.DateTime toDate(String filePath)

    • toPath

      public final String toPath(org.joda.time.DateTime time)

    • bucket

      public final org.joda.time.Interval bucket(org.joda.time.DateTime t)
      Return a granularity-sized Interval containing a particular DateTime.

    • isFinerThan

      public boolean isFinerThan(Granularity g)
      Decides whether this granularity is finer than the other granularity
      Returns:
      true if this Granularity is finer than the passed one

    • getIterable

      public Iterable<org.joda.time.Interval> getIterable(org.joda.time.Interval input)
      Return an iterable of granular buckets that overlap a particular interval. In cases where the number of granular buckets is very large, the Iterable returned by this method will take an excessive amount of time to compute, and materializing it into a collection will take an excessive amount of memory. For example, this happens in the extreme case of an input interval of Intervals.ETERNITY and any granularity other than Granularities.ALL, as well as cases like an input interval of ten years with Granularities.SECOND. To avoid issues stemming from large numbers of buckets, this method should be avoided, and code that uses this method should be rewritten to use some other approach. For example: rather than computing all possible buckets in a wide time range, only process buckets related to actual data points that appear.