Ejemplo n.º 1
0
  /**
   * Create a new TableStats object, that keeps track of statistics on each column of a table
   *
   * @param tableid The table over which to compute statistics
   * @param ioCostPerPage The cost per page of IO. This doesn't differentiate between
   *     sequential-scan IO and disk seeks.
   */
  public TableStats(int tableid, int ioCostPerPage) {
    // For this function, we use the DbFile for the table in question,
    // then scan through its tuples and calculate the values that you
    // to build the histograms.

    // TODO: Fill out the rest of the constructor.
    // Feel free to change anything already written, it's only a guideline

    this.ioCostPerPage = ioCostPerPage;
    DbFile file = Database.getCatalog().getDbFile(tableid);
    tupleDesc = file.getTupleDesc();
    numPages = ((HeapFile) file).numPages();
    numTuples = 0;

    int numFields = tupleDesc.numFields();

    // TODO: what goes here?
    statistics = new ArrayList<Object>();

    for (int i = 0; i < numFields; i++) {
      if (Type.INT_TYPE.equals(tupleDesc.getFieldType(i))) {
        statistics.add(new IntStatistics(NUM_HIST_BINS));
      } else {
        statistics.add(new StringHistogram(NUM_HIST_BINS));
      }
    }

    final DbFileIterator iter = file.iterator(null);
    try {
      iter.open();

      while (iter.hasNext()) {
        Tuple t = iter.next();
        numTuples++;

        // TODO: and here?
        for (int i = 0; i < numFields; i++) {
          if (Type.INT_TYPE.equals(tupleDesc.getFieldType(i))) {
            ((IntStatistics) statistics.get(i)).addValue(((IntField) t.getField(i)).getValue());
          } else {
            ((StringHistogram) statistics.get(i))
                .addValue(((StringField) t.getField(i)).getValue());
          }
        }
      }
      iter.close();
    } catch (DbException e) {
      e.printStackTrace();
    } catch (TransactionAbortedException e) {
      e.printStackTrace();
    }
  }
Ejemplo n.º 2
0
 /**
  * Estimate the selectivity of predicate <tt>field op constant</tt> on the table.
  *
  * @param field The field over which the predicate ranges
  * @param op The logical operation in the predicate
  * @param constant The value against which the field is compared
  * @return The estimated selectivity (fraction of tuples that satisfy) the predicate
  */
 public double estimateSelectivity(int field, Predicate.Op op, Field constant) {
   // TODO: some code goes here
   if (Type.INT_TYPE.equals(tupleDesc.getFieldType(field))) {
     return ((IntStatistics) statistics.get(field))
         .estimateSelectivity(op, ((IntField) constant).getValue());
   } else {
     return ((StringHistogram) statistics.get(field))
         .estimateSelectivity(op, ((StringField) constant).getValue());
   }
 }