Package mondrian.calc

Interface Calc

All Known Subinterfaces:
BooleanCalc, DateTimeCalc, DimensionCalc, DoubleCalc, HierarchyCalc, IntegerCalc, IterCalc, LevelCalc, ListCalc, MemberCalc, StringCalc, TupleCalc, VoidCalc
All Known Implementing Classes:
AbstractBooleanCalc, AbstractCalc, AbstractDateTimeCalc, AbstractDimensionCalc, AbstractDoubleCalc, AbstractHierarchyCalc, AbstractIntegerCalc, AbstractIterCalc, AbstractLevelCalc, AbstractListCalc, AbstractMemberCalc, AbstractStringCalc, AbstractTupleCalc, AbstractVoidCalc, AggregateFunDef.AggregateCalc, CacheCalc, ConstantCalc, GenericCalc, GenericIterCalc, HierarchyCurrentMemberFunDef.CalcImpl, HierarchyCurrentMemberFunDef.FixedCalcImpl, HierarchyDimensionFunDef.CalcImpl, IterableListCalc, LevelHierarchyFunDef.CalcImpl, MemberArrayValueCalc, MemberHierarchyFunDef.CalcImpl, MemberLevelFunDef.CalcImpl, MemberOrderKeyFunDef.CalcImpl, MemberValueCalc, NativizeSetFunDef.NativeListCalc, SetFunDef.ExprIterCalc, SetFunDef.SetListCalc, TupleFunDef.CalcImpl, TupleValueCalc, ValueCalc
public interface Calc
Calc is the base class for all calculable expressions.

Logical and physical expression languages

Mondrian has two expression languages:
  • The logical language of parsed MDX fragments (Exp).
  • The phyiscal language of compiled expressions (Calc).

The two languages allow us to separate logical (how an MDX expression was specified) from physical (how it is to be evaluated). The physical language is more strongly typed, and certain constructs which are implicit in the logical language (such as the addition of calls to the <Member>.CurrentMember function) are made explicit in the physical language.

Compilation

Expressions are generally created from using an expression compiler (ExpCompiler). There are often more than one evaluation strategy for a given expression, and compilation process gives us an opportunity to choose the optimal one.

Implementing expressions

The Calc interface has sub-interfaces for various types: IntegerCalc, BooleanCalc, DoubleCalc, StringCalc are scalar expressions; MemberCalc, LevelCalc, HierarchyCalc, DimensionCalc yield elements of the OLAP model.

Each of these sub-interfaces has an abstract implementation: AbstractIntegerCalc, AbstractBooleanCalc, AbstractDoubleCalc, AbstractStringCalc, AbstractMemberCalc, AbstractLevelCalc, AbstractHierarchyCalc, AbstractDimensionCalc.

GenericCalc is an adapter which implements all of these interfaces and will try to convert any given result to the correct type. Use it sparingly: if you know the expected result type, it is better to write a class which implements a specific TypeCalc interface.

Since:
Sep 26, 2005
Author:
jhyde

  • Method Details

    • evaluate

      Object evaluate(Evaluator evaluator)
      Evaluates this expression.
      Parameters:
      evaluator - Provides dimensional context in which to evaluate this expression
      Returns:
      Result of expression evaluation

    • dependsOn

      boolean dependsOn(Hierarchy hierarchy)
      Returns whether this expression depends upon a given hierarchy.

      If it does not depend on the hierarchy, then re-evaluating the expression with a different member of this context must produce the same answer.

      Some examples:

      • The expression
        [Measures].[Unit Sales]
        depends on all dimensions except [Measures].
      • The boolean expression
        ([Measures].[Unit Sales], [Time].[1997]) > 1000
        depends on all hierarchies except [Measures] and [Time].
      • The list expression
        Filter([Store].[USA].Children, [Measures].[Unit Sales] < 50)
        depends upon all hierarchies except [Store] and [Measures]. How so? Normally the scalar expression would depend upon all hierarchies except [Measures], but the Filter function sets the [Store] context before evaluating the scalar expression, so it is not inherited from the surrounding context.

      Parameters:
      hierarchy - Hierarchy
      Returns:
      Whether this expression's result depends upon the current member of the hierarchy

    • getType

      Type getType()
      Returns the type of this expression.

    • accept

      void accept(CalcWriter calcWriter)
      Prints this expression, by accepting a visiting CalcWriter.
      Parameters:
      calcWriter - Writer

    • getResultStyle

      ResultStyle getResultStyle()
      Returns style in which the result of evaluating this expression is returned.

      One application of this method is for the compiler to figure out whether the compiled expression is returning a mutable list. If a mutable list is required, the compiler can create a mutable copy.

      See Also:

    • isWrapperFor

      boolean isWrapperFor(Class<?> iface)

    • unwrap

      <T> T unwrap(Class<T> iface)