Class AutomatonTermsEnum

All Implemented Interfaces:
BytesRefIterator

public class AutomatonTermsEnum extends FilteredTermsEnum
A FilteredTermsEnum that enumerates terms based upon what is accepted by a DFA.

The algorithm is such:

  1. As long as matches are successful, keep reading sequentially.
  2. When a match fails, skip to the next string in lexicographic order that does not enter a reject state.

The algorithm does not attempt to actually skip to the next string that is completely accepted. This is not possible when the language accepted by the FSM is not finite (i.e. * operator).

  • Field Details

    • runAutomaton

      private final ByteRunAutomaton runAutomaton
    • commonSuffixRef

      private final BytesRef commonSuffixRef
    • finite

      private final boolean finite
    • automaton

      private final Automaton automaton
    • visited

      private final short[] visited
    • curGen

      private short curGen
    • seekBytesRef

      private final BytesRefBuilder seekBytesRef
    • linear

      private boolean linear
    • linearUpperBound

      private final BytesRef linearUpperBound
    • transition

      private final Transition transition
    • savedStates

      private final IntsRefBuilder savedStates
  • Constructor Details

    • AutomatonTermsEnum

      public AutomatonTermsEnum(TermsEnum tenum, CompiledAutomaton compiled)
      Construct an enumerator based upon an automaton, enumerating the specified field, working on a supplied TermsEnum
      Parameters:
      compiled - CompiledAutomaton
  • Method Details

    • setVisited

      private void setVisited(int state)
      Records the given state has been visited.
    • isVisited

      private boolean isVisited(int state)
      Indicates whether the given state has been visited.
    • accept

      protected FilteredTermsEnum.AcceptStatus accept(BytesRef term)
      Returns true if the term matches the automaton. Also stashes away the term to assist with smart enumeration.
      Specified by:
      accept in class FilteredTermsEnum
    • nextSeekTerm

      protected BytesRef nextSeekTerm(BytesRef term) throws IOException
      Description copied from class: FilteredTermsEnum
      On the first call to FilteredTermsEnum.next() or if FilteredTermsEnum.accept(org.apache.lucene.util.BytesRef) returns FilteredTermsEnum.AcceptStatus.YES_AND_SEEK or FilteredTermsEnum.AcceptStatus.NO_AND_SEEK, this method will be called to eventually seek the underlying TermsEnum to a new position. On the first call, currentTerm will be null, later calls will provide the term the underlying enum is positioned at. This method returns per default only one time the initial seek term and then null, so no repositioning is ever done.

      Override this method, if you want a more sophisticated TermsEnum, that repositions the iterator during enumeration. If this method always returns null the enum is empty.

      Please note: This method should always provide a greater term than the last enumerated term, else the behaviour of this enum violates the contract for TermsEnums.

      Overrides:
      nextSeekTerm in class FilteredTermsEnum
      Throws:
      IOException
    • setLinear

      private void setLinear(int position)
      Sets the enum to operate in linear fashion, as we have found a looping transition at position: we set an upper bound and act like a TermRangeQuery for this portion of the term space.
    • nextString

      private boolean nextString()
      Increments the byte buffer to the next String in binary order after s that will not put the machine into a reject state. If such a string does not exist, returns false.

      The correctness of this method depends upon the automaton being deterministic, and having no transitions to dead states.

      Returns:
      true if more possible solutions exist for the DFA
    • nextString

      private boolean nextString(int state, int position)
      Returns the next String in lexicographic order that will not put the machine into a reject state.

      This method traverses the DFA from the given position in the String, starting at the given state.

      If this cannot satisfy the machine, returns false. This method will walk the minimal path, in lexicographic order, as long as possible.

      If this method returns false, then there might still be more solutions, it is necessary to backtrack to find out.

      Parameters:
      state - current non-reject state
      position - useful portion of the string
      Returns:
      true if more possible solutions exist for the DFA from this position
    • backtrack

      private int backtrack(int position)
      Attempts to backtrack thru the string after encountering a dead end at some given position. Returns false if no more possible strings can match.
      Parameters:
      position - current position in the input String
      Returns:
      position >= 0 if more possible solutions exist for the DFA