org.apache.pig.newplan.logical.relational
Class LogicalSchema

java.lang.Object
  org.apache.pig.newplan.logical.relational.LogicalSchema

public class LogicalSchema
extends Object

Schema, from a logical perspective.

Nested Class Summary

static class	LogicalSchema.LogicalFieldSchema
static class	LogicalSchema.MergeMode

Constructor Summary

LogicalSchema()

Method Summary

void	addField(LogicalSchema.LogicalFieldSchema field) Add a field to this schema.
static boolean	castable(LogicalSchema inSch, LogicalSchema outSch) Recursively compare two schemas to check if the input schema can be cast to the cast schema
LogicalSchema	deepCopy()
static boolean	equals(LogicalSchema schema, LogicalSchema other, boolean relaxInner, boolean relaxAlias) Recursively compare two schemas for equality
int	findField(long uid) Look for the index of the field that contains the specified uid
LogicalSchema.LogicalFieldSchema	getField(int fieldNum) Fetch a field by field number
LogicalSchema.LogicalFieldSchema	getField(String alias) Fetch a field by alias
int	getFieldPosition(String alias)
List<LogicalSchema.LogicalFieldSchema>	getFields() Get all fields
LogicalSchema.LogicalFieldSchema	getFieldSubNameMatch(String alias) Given an alias name, find the associated LogicalFieldSchema.
boolean	isEqual(Object other) Two schemas are equal if they are of equal size and their fields schemas considered in order are equal.
boolean	isEqual(Object other, boolean compareAlias) Two schemas are equal if they are of equal size and their fields schemas considered in order are equal.
static LogicalSchema	merge(LogicalSchema s1, LogicalSchema s2, LogicalSchema.MergeMode mode) Merge two schemas.
static LogicalSchema	mergeSchemaByAlias(LogicalSchema schema1, LogicalSchema schema2) Merges two schemas using their column aliases (unlike mergeSchema(..) functions which merge using positions) Schema will not be merged if types are incompatible, as per DataType.mergeType(..) For Tuples and Bags, SubSchemas have to be equal be considered compatible
static LogicalSchema	mergeSchemasByAlias(List<LogicalSchema> schemas) Merges collection of schemas using their column aliases (unlike mergeSchema(..) functions which merge using positions) Schema will not be merged if types are incompatible, as per DataType.mergeType(..) For Tuples and Bags, SubSchemas have to be equal be considered compatible
LogicalSchema	mergeUid(LogicalSchema uidOnlySchema)
void	normalize() Old Pig schema does not require a tuple schema inside a bag; Now it is required to have that; this method is to fill the gap
void	resetUid() Reset uids of all fieldschema that the schema contains
int	size() Get the size of the schema.
String	toString()
String	toString(boolean verbose)

Methods inherited from class java.lang.Object

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

Constructor Detail

LogicalSchema

public LogicalSchema()

Method Detail

resetUid

public void resetUid()

Reset uids of all fieldschema that the schema contains

castable

public static boolean castable(LogicalSchema inSch,
                               LogicalSchema outSch)

Recursively compare two schemas to check if the input schema can be cast to the cast schema

Parameters:

inSch - schema of the cast input

outSch - schema of the cast operator

Returns:

true if castable

addField

public void addField(LogicalSchema.LogicalFieldSchema field)

Add a field to this schema.

Parameters:

field - to be added to the schema

getField

public LogicalSchema.LogicalFieldSchema getField(String alias)
                                          throws FrontendException

Fetch a field by alias

Parameters:

alias -

Returns:

field associated with alias, or null if no such field

Throws:

FrontendException

getFieldSubNameMatch

public LogicalSchema.LogicalFieldSchema getFieldSubNameMatch(String alias)
                                                      throws FrontendException

Given an alias name, find the associated LogicalFieldSchema. If exact name is not found see if any field matches the part of the 'namespaced' alias. eg. if given alias is nm::a , and schema is (a,b). It will return FieldSchema of a. if given alias is nm::a and schema is (nm2::a, b), it will return null

Parameters:

alias - Alias to look up.

Returns:

LogicalFieldSchema, or null if no such alias is in this tuple.

Throws:

FrontendException

getFieldPosition

public int getFieldPosition(String alias)

getField

public LogicalSchema.LogicalFieldSchema getField(int fieldNum)

Fetch a field by field number

Parameters:

fieldNum - field number to fetch

Returns:

field

getFields

public List<LogicalSchema.LogicalFieldSchema> getFields()

Get all fields

Returns:

list of all fields

size

public int size()

Get the size of the schema.

Returns:

size

isEqual

public boolean isEqual(Object other)

Two schemas are equal if they are of equal size and their fields schemas considered in order are equal. This function does not compare the alias of the fields.

isEqual

public boolean isEqual(Object other,
                       boolean compareAlias)

Two schemas are equal if they are of equal size and their fields schemas considered in order are equal. If compareAlias argument is set to true, the alias of the fields are also compared.

Parameters:

other -

compareAlias -

Returns:

true if equal

findField

public int findField(long uid)

Look for the index of the field that contains the specified uid

Parameters:

uid - the uid to look for

Returns:

the index of the field, -1 if not found

merge

public static LogicalSchema merge(LogicalSchema s1,
                                  LogicalSchema s2,
                                  LogicalSchema.MergeMode mode)
                           throws FrontendException

Merge two schemas.

Parameters:

s1 - In Load/ForEach, s1 is user declared schema; In Union, s1 is left side.

s2 - In Load/ForEach, s2 is infered schema; In Union, s2 is right side.

mode - We merge schema in Load/Foreach/Union. In Load/Foreach, we always take s1 if compatible (s1 is set to be user defined schema), In union, we take more specific type (between numeric and string, we take string). In the case type mismatch in s1/s2, we expect TypeCheckingVisitor will fill the gap later.

Returns:

a merged schema, or null if the merge fails

Throws:

FrontendException

toString

public String toString(boolean verbose)

toString

public String toString()

Overrides:

toString in class Object

mergeUid

public LogicalSchema mergeUid(LogicalSchema uidOnlySchema)
                       throws FrontendException

Throws:

FrontendException

deepCopy

public LogicalSchema deepCopy()

mergeSchemasByAlias

public static LogicalSchema mergeSchemasByAlias(List<LogicalSchema> schemas)
                                         throws FrontendException

Merges collection of schemas using their column aliases (unlike mergeSchema(..) functions which merge using positions) Schema will not be merged if types are incompatible, as per DataType.mergeType(..) For Tuples and Bags, SubSchemas have to be equal be considered compatible

Parameters:

schemas - - list of schemas to be merged using their column alias

Returns:

merged schema

Throws:

FrontendException

mergeSchemaByAlias

public static LogicalSchema mergeSchemaByAlias(LogicalSchema schema1,
                                               LogicalSchema schema2)
                                        throws FrontendException

Merges two schemas using their column aliases (unlike mergeSchema(..) functions which merge using positions) Schema will not be merged if types are incompatible, as per DataType.mergeType(..) For Tuples and Bags, SubSchemas have to be equal be considered compatible

Throws:

FrontendException

equals

public static boolean equals(LogicalSchema schema,
                             LogicalSchema other,
                             boolean relaxInner,
                             boolean relaxAlias)

Recursively compare two schemas for equality

Parameters:

schema -

other -

relaxInner - if true, inner schemas will not be checked

relaxAlias - if true, aliases will not be checked

Returns:

true if schemas are equal, false otherwise

normalize

public void normalize()

Old Pig schema does not require a tuple schema inside a bag; Now it is required to have that; this method is to fill the gap