Class InsertSearch


  • public final class InsertSearch
    extends CollatedBaseSearch
    This is a SelectableScheme similar in some ways to the binary tree. When a new row is added, it is inserted into a sorted list of rows. We can then use this list to select out the sorted list of elements.

    This requires less memory than the BinaryTree, however it is not as fast. Even though, it should still perform fairly well on medium size data sets. On large size data sets, insert and remove performance may suffer.

    This object retains knowledge of all set elements unlike BlindSearch which has no memory overhead.

    Performance should be very comparable to BinaryTree for sets that aren't altered much.

    • Field Detail

      • set_list

        private IntegerListInterface set_list
        The sorted list of rows in this set. This is sorted from min to max (not sorted by row number - sorted by entity row value).
      • RECORD_UID

        boolean RECORD_UID
        If this is true, then this SelectableScheme records additional rid information that can be used to very quickly identify whether a value is greater, equal or less.
      • set_comparator

        private IndexComparator set_comparator
        The IndexComparator that we use to refer elements in the set to actual data objects.
      • DEBUG_immutable_set_size

        private int DEBUG_immutable_set_size
        If this is immutable, this stores the number of entries in 'set_list' when this object was made.
    • Constructor Detail

      • InsertSearch

        public InsertSearch​(TableDataSource table,
                            int column)
        The Constructor.
      • InsertSearch

        public InsertSearch​(TableDataSource table,
                            int column,
                            IntegerVector vec)
        Constructor sets the scheme with a pre-sorted list. The Vector 'vec' should not be used again after this is called. 'vec' must be sorted from low key to high key.
      • InsertSearch

        InsertSearch​(TableDataSource table,
                     int column,
                     IntegerListInterface list)
        Constructor sets the scheme with a pre-sorted list. The list 'list' should not be used again after this is called. 'list' must be sorted from low key to high key.
      • InsertSearch

        private InsertSearch​(TableDataSource table,
                             InsertSearch from,
                             boolean immutable)
        Constructs this as a copy of the given, either mutable or immutable copy.
    • Method Detail

      • setupComparator

        private void setupComparator()
        Sets the internal comparator that enables us to sort and lookup on the data in this column.
      • insert

        public void insert​(int row)
        Inserts a row into the list. This will always be thread safe, table changes cause a write lock which prevents reads while we are writing to the table.
        Overrides:
        insert in class CollatedBaseSearch
      • remove

        public void remove​(int row)
        Removes a row from the list. This will always be thread safe, table changes cause a write lock which prevents reads while we are writing to the table.
        Overrides:
        remove in class CollatedBaseSearch
      • safeSetComparator

        private final IndexComparator safeSetComparator()
        This needs to be called to access 'set_comparator' in thread busy methods. Because creating a UID cache will modify set_comparator, we need to make sure we access this variable safely.

        NOTE: This is a throwback method for an idea I had to speed up the 'select*' methods, but it proved unworkable. The reason being that the UID only contains knowledge of relations between rows, and the 'select*' methods find the relationship of a TObject in the column set.

      • readFrom

        public void readFrom​(java.io.InputStream in)
                      throws java.io.IOException
        Reads the entire state of the scheme from the input stream. Throws an exception if the scheme is not empty.
        Overrides:
        readFrom in class CollatedBaseSearch
        Throws:
        java.io.IOException
      • writeTo

        public void writeTo​(java.io.OutputStream out)
                     throws java.io.IOException
        Writes the entire state of the scheme to the output stream.
        Overrides:
        writeTo in class CollatedBaseSearch
        Throws:
        java.io.IOException
      • copy

        public SelectableScheme copy​(TableDataSource table,
                                     boolean immutable)
        Returns an exact copy of this scheme including any optimization information. The copied scheme is identical to the original but does not share any parts. Modifying any part of the copied scheme will have no effect on the original and vice versa.
        Specified by:
        copy in class SelectableScheme
      • checkSchemeSorted

        private void checkSchemeSorted()
        Checks that the scheme is in sorted order. This is a debug check to ensure we maintain a sorted index. NOTE: This *MUST* be removed in a release version because it uses up many cycles for each check.
      • searchFirst

        protected int searchFirst​(TObject val)
        Description copied from class: CollatedBaseSearch
        Finds the position in the collated set of the first value in the column equal to the given value. If the value is not to be found in the column, it returns -(insert_position + 1).
        Specified by:
        searchFirst in class CollatedBaseSearch
      • searchLast

        protected int searchLast​(TObject val)
        Description copied from class: CollatedBaseSearch
        Finds the position in the collated set of the last value in the column equal to the given value. If the value is not to be found in the column, it returns -(insert_position + 1).
        Specified by:
        searchLast in class CollatedBaseSearch