simpledb

Class LogicalPlan

java.lang.Object

simpledb.LogicalPlan

public class LogicalPlan
extends java.lang.Object

LogicalPlan represents a logical query plan that has been through the parser and is ready to be processed by the optimizer.

A LogicalPlan consits of a collection of table scan nodes, join nodes, filter nodes, a select list, and a group by field. LogicalPlans can only represent queries with one aggregation field and one group by field.

LogicalPlans can be converted to physical (optimized) plans using the physicalPlan(simpledb.TransactionId, java.util.Map<java.lang.String, simpledb.TableStats>, boolean) method, which uses the JoinOptimizer to order joins optimally and to select the best implementations for joins.

Field Summary

Fields

Modifier and Type	Field and Description
private java.lang.String	aggField
private java.lang.String	aggOp
private java.util.Vector<LogicalFilterNode>	filters
private java.lang.String	groupByField
private boolean	hasAgg
private boolean	hasOrderBy
private java.util.Vector<LogicalJoinNode>	joins
private boolean	oByAsc
private java.lang.String	oByField
private java.lang.String	query
private java.util.Vector<LogicalSelectListNode>	selectList
private java.util.HashMap<java.lang.String,DbIterator>	subplanMap
private java.util.HashMap<java.lang.String,java.lang.Integer>	tableMap
private java.util.Vector<LogicalScanNode>	tables

Constructor Summary

Constructors

Constructor and Description
LogicalPlan() Constructor -- generate an empty logical plan

Method Summary

Modifier and Type	Method and Description
void	addAggregate(java.lang.String op, java.lang.String afield, java.lang.String gfield) Add an aggregate over the field with the specified grouping to the query.
void	addFilter(java.lang.String field, Predicate.Op p, java.lang.String constantValue) Add a new filter to the logical plan
void	addJoin(java.lang.String joinField1, DbIterator joinField2, Predicate.Op pred) Add a join between a field and a subquery.
void	addJoin(java.lang.String joinField1, java.lang.String joinField2, Predicate.Op pred) Add a join between two fields of two different tables.
void	addOrderBy(java.lang.String field, boolean asc) Add an ORDER BY expression in the specified order on the specified field.
void	addProjectField(java.lang.String fname, java.lang.String aggOp) Add a specified field/aggregate combination to the select list of the query.
void	addScan(int table, java.lang.String name) Add a scan to the plan.
(package private) java.lang.String	disambiguateName(java.lang.String name) Given a name of a field, try to figure out what table it belongs to by looking through all of the tables added via addScan(int, java.lang.String).
(package private) static Aggregator.Op	getAggOp(java.lang.String s) Convert the aggregate operator name s into an Aggregator.op operation.
java.lang.String	getQuery() Get the query text associated with this plan via setQuery(java.lang.String).
java.util.HashMap<java.lang.String,java.lang.Integer>	getTableAliasToIdMapping()
java.lang.Integer	getTableId(java.lang.String alias) Given a table alias, return id of the table object (this id can be supplied to Catalog.getDbFile(int)).
static void	main(java.lang.String[] argv)
DbIterator	physicalPlan(TransactionId t, java.util.Map<java.lang.String,TableStats> baseTableStats, boolean explain) Convert this LogicalPlan into a physicalPlan represented by a DbIterator.
void	setQuery(java.lang.String query) Set the text of the query representing this logical plan.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

joins

private java.util.Vector<LogicalJoinNode> joins

tables

private java.util.Vector<LogicalScanNode> tables

filters

private java.util.Vector<LogicalFilterNode> filters

subplanMap

private java.util.HashMap<java.lang.String,DbIterator> subplanMap

tableMap

private java.util.HashMap<java.lang.String,java.lang.Integer> tableMap

selectList

private java.util.Vector<LogicalSelectListNode> selectList

groupByField

private java.lang.String groupByField

hasAgg

private boolean hasAgg

aggOp

private java.lang.String aggOp

aggField

private java.lang.String aggField

oByAsc

private boolean oByAsc

hasOrderBy

private boolean hasOrderBy

oByField

private java.lang.String oByField

query

private java.lang.String query

Constructor Detail

LogicalPlan

public LogicalPlan()

Constructor -- generate an empty logical plan

Method Detail

setQuery

public void setQuery(java.lang.String query)

Set the text of the query representing this logical plan. Does NOT parse the specified query -- this method is just used so that the object can print the SQL it represents.

Parameters:

query - the text of the query associated with this plan

getQuery

public java.lang.String getQuery()

Get the query text associated with this plan via setQuery(java.lang.String).

getTableId

public java.lang.Integer getTableId(java.lang.String alias)

Given a table alias, return id of the table object (this id can be supplied to Catalog.getDbFile(int)). Aliases are added as base tables are added via addScan(int, java.lang.String).

Parameters:

alias - the table alias to return a table id for

Returns:

the id of the table corresponding to alias, or null if the alias is unknown

getTableAliasToIdMapping

public java.util.HashMap<java.lang.String,java.lang.Integer> getTableAliasToIdMapping()

addFilter

public void addFilter(java.lang.String field,
                      Predicate.Op p,
                      java.lang.String constantValue)
               throws ParsingException

Add a new filter to the logical plan

Parameters:

field - The name of the over which the filter applies; this can be a fully qualified field (tablename.field or alias.field), or can be a unique field name without a tablename qualifier. If it is an ambiguous name, it will throw a ParsingException

p - The predicate for the filter

constantValue - the constant to compare the predicate against; if field is an integer field, this should be a String representing an integer

Throws:

ParsingException - if field is not in one of the tables added via addScan(int, java.lang.String) or if field is ambiguous (e.g., two tables contain a field named field.)

addJoin

public void addJoin(java.lang.String joinField1,
                    java.lang.String joinField2,
                    Predicate.Op pred)
             throws ParsingException

Add a join between two fields of two different tables.

Parameters:

joinField1 - The name of the first join field; this can be a fully qualified name (e.g., tableName.field or alias.field) or may be an unqualified unique field name. If the name is ambiguous or unknown, a ParsingException will be thrown.

joinField2 - The name of the second join field

pred - The join predicate

Throws:

ParsingException - if either of the fields is ambiguous, or is not in one of the tables added via addScan(int, java.lang.String)

addJoin

public void addJoin(java.lang.String joinField1,
                    DbIterator joinField2,
                    Predicate.Op pred)
             throws ParsingException

Add a join between a field and a subquery.

Parameters:

joinField1 - The name of the first join field; this can be a fully qualified name (e.g., tableName.field or alias.field) or may be an unqualified unique field name. If the name is ambiguous or unknown, a ParsingException will be thrown.

joinField2 - the subquery to join with -- the join field of the subquery is the first field in the result set of the query

pred - The join predicate.

Throws:

ParsingException - if either of the fields is ambiguous, or is not in one of the tables added via addScan(int, java.lang.String)

addScan

public void addScan(int table,
                    java.lang.String name)

Add a scan to the plan. One scan node needs to be added for each alias of a table accessed by the plan.

Parameters:

table - the id of the table accessed by the plan (can be resolved to a DbFile using Catalog.getDbFile(int)

name - the alias of the table in the plan

addProjectField

public void addProjectField(java.lang.String fname,
                            java.lang.String aggOp)
                     throws ParsingException

Add a specified field/aggregate combination to the select list of the query. Fields are output by the query such that the rightmost field is the first added via addProjectField.

Parameters:

fname - the field to add to the output

aggOp - the aggregate operation over the field.

Throws:

ParsingException

addAggregate

public void addAggregate(java.lang.String op,
                         java.lang.String afield,
                         java.lang.String gfield)
                  throws ParsingException

Add an aggregate over the field with the specified grouping to the query. SimpleDb only supports a single aggregate expression and GROUP BY field.

Parameters:

op - the aggregation operator

afield - the field to aggregate over

gfield - the field to group by

Throws:

ParsingException

addOrderBy

public void addOrderBy(java.lang.String field,
                       boolean asc)
                throws ParsingException

Add an ORDER BY expression in the specified order on the specified field. SimpleDb only supports a single ORDER BY field.

Parameters:

field - the field to order by

asc - true if should be ordered in ascending order, false for descending order

Throws:

ParsingException

disambiguateName

java.lang.String disambiguateName(java.lang.String name)
                           throws ParsingException

Given a name of a field, try to figure out what table it belongs to by looking through all of the tables added via addScan(int, java.lang.String).

Returns:

A fully qualified name of the form tableAlias.name. If the name parameter is already qualified with a table name, simply returns name.

Throws:

ParsingException - if the field cannot be found in any of the tables, or if the field is ambiguous (appears in multiple tables)

getAggOp

static Aggregator.Op getAggOp(java.lang.String s)
                       throws ParsingException

Convert the aggregate operator name s into an Aggregator.op operation.

Throws:

ParsingException - if s is not a valid operator name

physicalPlan

public DbIterator physicalPlan(TransactionId t,
                               java.util.Map<java.lang.String,TableStats> baseTableStats,
                               boolean explain)
                        throws ParsingException

Convert this LogicalPlan into a physicalPlan represented by a DbIterator. Attempts to find the optimal plan by using JoinOptimizer.orderJoins(java.util.HashMap<java.lang.String, simpledb.TableStats>, java.util.HashMap<java.lang.String, java.lang.Double>, boolean) to order the joins in the plan.

Parameters:

t - The transaction that the returned DbIterator will run as a part of

baseTableStats - a HashMap providing a TableStats object for each table used in the LogicalPlan. This should have one entry for each table referenced by the plan, not one entry for each table alias (so a table t aliases as t1 and t2 would have just one entry with key 't' in this HashMap).

explain - flag indicating whether output visualizing the physical query plan should be given.

Returns:

A DbIterator representing this plan.

Throws:

ParsingException - if the logical plan is not valid

main

public static void main(java.lang.String[] argv)