def __init__(self, tx, idxname, leafLayout):
super(BTreeIndex, self).__init__()
self.tx = tx
# deal with the leaves
self.leaftbl = idxname + "leaf"
self.leafLayout = leafLayout
if tx.size(self.leaftbl) == 0:
blk = tx.append(self.leaftbl)
node = BTPage(tx, blk, leafLayout)
node.format(blk, -1)
# deal with the directory
dirsch = Schema()
dirsch.add("block", leafLayout.schema())
dirsch.add("dataval", leafLayout.schema())
dirtbl = idxname + "dir"
self.dirLayout = Layout(dirsch)
self.rootblk = BlockId(dirtbl, 0)
if tx.size(dirtbl) == 0:
# create new root block
tx.append(dirtbl)
node = BTPage(tx, self.rootblk, self.dirLayout)
node.format(self.rootblk, 0)
# insert initial directory entry
fldtype = dirsch.type("dataval")
minval = Constant(
Integer.MIN_VALUE) if fldtype == INTEGER else Constant("")
node.insertDir(0, minval, 0)
node.close()
class ProjectPlan(Plan):
#
# * Creates a new project node in the query tree,
# * having the specified subquery and field list.
# * @param p the subquery
# * @param fieldlist the list of fields
#
def __init__(self, p, fieldlist):
super(ProjectPlan, self).__init__()
self.p = p
self._schema = Schema()
for fldname in fieldlist:
self._schema.add(fldname, p.schema())
#
# * Creates a project scan for this query.
# * @see Plan#open()
#
def open(self):
s = self.p.open()
return ProjectScan(s, self._schema.fields())
#
# * Estimates the number of block accesses in the projection,
# * which is the same as in the underlying query.
# * @see Plan#blocksAccessed()
#
def blocksAccessed(self):
return self.p.blocksAccessed()
#
# * Estimates the number of output records in the projection,
# * which is the same as in the underlying query.
# * @see Plan#recordsOutput()
#
def recordsOutput(self):
return self.p.recordsOutput()
#
# * Estimates the number of distinct field values
# * in the projection,
# * which is the same as in the underlying query.
# * @see Plan#distinctValues(String)
#
def distinctValues(self, fldname):
return self.p.distinctValues(fldname)
#
# * Returns the schema of the projection,
# * which is taken from the field list.
# * @see Plan#schema()
#
def schema(self):
return self._schema
class GroupByPlan(Plan):
#
# * Create a groupby plan for the underlying query.
# * The grouping is determined by the specified
# * collection of group fields,
# * and the aggregation is computed by the
# * specified collection of aggregation functions.
# * @param p a plan for the underlying query
# * @param groupfields the group fields
# * @param aggfns the aggregation functions
# * @param tx the calling transaction
#
def __init__(self, tx, p, groupfields, aggfns):
super(GroupByPlan, self).__init__()
self.p = SortPlan(tx, p, groupfields)
self.groupfields = groupfields
self.aggfns = aggfns
self.sch = Schema()
for fldname in groupfields:
self.sch.add(fldname, p.schema())
for fn in aggfns:
self.sch.addIntField(fn.fieldName())
#
# * This method opens a sort plan for the specified plan.
# * The sort plan ensures that the underlying records
# * will be appropriately grouped.
# * @see Plan#open()
#
def open(self):
s = self.p.open()
return GroupByScan(s, self.groupfields, self.aggfns)
#
# * Return the number of blocks required to
# * compute the aggregation,
# * which is one pass through the sorted table.
# * It does <i>not</i> include the one-time cost
# * of materializing and sorting the records.
# * @see Plan#blocksAccessed()
#
def blocksAccessed(self):
return self.p.blocksAccessed()
#
# * Return the number of groups. Assuming equal distribution,
# * this is the product of the distinct values
# * for each grouping field.
# * @see Plan#recordsOutput()
#
def recordsOutput(self):
numgroups = 1
for fldname in self.groupfields:
numgroups *= self.p.distinctValues(fldname)
return numgroups
#
# * Return the number of distinct values for the
# * specified field. If the field is a grouping field,
# * then the number of distinct values is the same
# * as in the underlying query.
# * If the field is an aggregate field, then we
# * assume that all values are distinct.
# * @see Plan#distinctValues(String)
#
def distinctValues(self, fldname):
if self.p.schema().hasField(fldname):
return self.p.distinctValues(fldname)
else:
return self.recordsOutput()
#
# * Returns the schema of the output table.
# * The schema consists of the group fields,
# * plus one field for each aggregation function.
# * @see Plan#schema()
#
def schema(self):
return self.sch