org.codehaus.jackson.sym

Class BytesToNameCanonicalizer

java.lang.Object

org.codehaus.jackson.sym.BytesToNameCanonicalizer

public final class BytesToNameCanonicalizer
extends Object

A caching symbol table implementation used for canonicalizing JSON field names (as Names which are constructed directly from a byte-based input source). Complications arise from trying to do efficient reuse and merging of symbol tables, to be able to make use of usually shared vocabulary of subsequent parsing runs.

Author:

Tatu Saloranta

Field Summary

Fields

Modifier and Type	Field and Description
protected int	_collCount Total number of Names in collision buckets (included in _count along with primary entries)
protected int	_collEnd Index of the first unused collision bucket entry (== size of the used portion of collision list): less than or equal to 0xFF (255), since max number of entries is 255 (8-bit, minus 0 used as 'empty' marker)
protected org.codehaus.jackson.sym.BytesToNameCanonicalizer.Bucket[]	_collList Array of heads of collision bucket chains; size dynamically
protected int	_count Total number of Names in the symbol table; only used for child tables.
protected boolean	_intern Whether canonical symbol Strings are to be intern()ed before added to the table or not
protected int	_longestCollisionList We need to keep track of the longest collision list; this is needed both to indicate problems with attacks and to allow flushing for other cases.
protected int[]	_mainHash Array of 2^N size, which contains combination of 24-bits of hash (0 to indicate 'empty' slot), and 8-bit collision bucket index (0 to indicate empty collision bucket chain; otherwise subtract one from index)
protected int	_mainHashMask Mask used to truncate 32-bit hash value to current hash array size; essentially, hash array size - 1 (since hash array sizes are 2^N).
protected Name[]	_mainNames Array that contains Name instances matching entries in _mainHash.
protected BytesToNameCanonicalizer	_parent Reference to the root symbol table, for child tables, so that they can merge table information back as necessary.
protected AtomicReference<org.codehaus.jackson.sym.BytesToNameCanonicalizer.TableInfo>	_tableInfo Member that is only used by the root table instance: root passes immutable state into child instances, and children may return new state if they add entries to the table.
protected static int	DEFAULT_TABLE_SIZE
protected static int	MAX_TABLE_SIZE Let's not expand symbol tables past some maximum size; this should protected against OOMEs caused by large documents with unique (~= random) names.

Method Summary

Methods

Modifier and Type	Method and Description
Name	addName(String symbolStr, int[] quads, int qlen)
Name	addName(String symbolStr, int q1, int q2)
int	bucketCount()
int	calcHash(int firstQuad)
int	calcHash(int[] quads, int qlen)
int	calcHash(int firstQuad, int secondQuad)
protected static int[]	calcQuads(byte[] wordBytes)
int	collisionCount() Method mostly needed by unit tests; calculates number of entries that are in collision list.
static BytesToNameCanonicalizer	createRoot() Factory method to call to create a symbol table instance with a randomized seed value.
protected static BytesToNameCanonicalizer	createRoot(int hashSeed) Factory method that should only be called from unit tests, where seed value should remain the same.
Name	findName(int firstQuad) Finds and returns name matching the specified symbol, if such name already exists in the table.
Name	findName(int[] quads, int qlen) Finds and returns name matching the specified symbol, if such name already exists in the table; or if not, creates name object, adds to the table, and returns it.
Name	findName(int firstQuad, int secondQuad) Finds and returns name matching the specified symbol, if such name already exists in the table.
static Name	getEmptyName()
int	hashSeed()
BytesToNameCanonicalizer	makeChild(boolean canonicalize, boolean intern) Factory method used to create actual symbol table instance to use for parsing.
int	maxCollisionLength() Method mostly needed by unit tests; calculates length of the longest collision chain.
boolean	maybeDirty() Method called to check to quickly see if a child symbol table may have gotten additional entries.
void	release() Method called by the using code to indicate it is done with this instance.
protected void	reportTooManyCollisions(int maxLen)
int	size()

Methods inherited from class java.lang.Object

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

Field Detail

DEFAULT_TABLE_SIZE

protected static final int DEFAULT_TABLE_SIZE

See Also:

Constant Field Values

MAX_TABLE_SIZE

protected static final int MAX_TABLE_SIZE

Let's not expand symbol tables past some maximum size; this should protected against OOMEs caused by large documents with unique (~= random) names.

See Also:

Constant Field Values

_parent

protected final BytesToNameCanonicalizer _parent

Reference to the root symbol table, for child tables, so that they can merge table information back as necessary.

_tableInfo

protected final AtomicReference<org.codehaus.jackson.sym.BytesToNameCanonicalizer.TableInfo> _tableInfo

Member that is only used by the root table instance: root passes immutable state into child instances, and children may return new state if they add entries to the table. Child tables do NOT use the reference.

_intern

protected final boolean _intern

Whether canonical symbol Strings are to be intern()ed before added to the table or not

_count

protected int _count

Total number of Names in the symbol table; only used for child tables.

_longestCollisionList

protected int _longestCollisionList

We need to keep track of the longest collision list; this is needed both to indicate problems with attacks and to allow flushing for other cases.

Since:

1.9.9

_mainHashMask

protected int _mainHashMask

Mask used to truncate 32-bit hash value to current hash array size; essentially, hash array size - 1 (since hash array sizes are 2^N).

_mainHash

protected int[] _mainHash

Array of 2^N size, which contains combination of 24-bits of hash (0 to indicate 'empty' slot), and 8-bit collision bucket index (0 to indicate empty collision bucket chain; otherwise subtract one from index)

_mainNames

protected Name[] _mainNames

Array that contains Name instances matching entries in _mainHash. Contains nulls for unused entries.

_collList

protected org.codehaus.jackson.sym.BytesToNameCanonicalizer.Bucket[] _collList

Array of heads of collision bucket chains; size dynamically

_collCount

protected int _collCount

Total number of Names in collision buckets (included in _count along with primary entries)

_collEnd

protected int _collEnd

Index of the first unused collision bucket entry (== size of the used portion of collision list): less than or equal to 0xFF (255), since max number of entries is 255 (8-bit, minus 0 used as 'empty' marker)

Method Detail

createRoot

public static BytesToNameCanonicalizer createRoot()

Factory method to call to create a symbol table instance with a randomized seed value.

createRoot

protected static BytesToNameCanonicalizer createRoot(int hashSeed)

Factory method that should only be called from unit tests, where seed value should remain the same.

makeChild

public BytesToNameCanonicalizer makeChild(boolean canonicalize,
                                 boolean intern)

Factory method used to create actual symbol table instance to use for parsing.

Parameters:

intern - Whether canonical symbol Strings should be interned or not

release

public void release()

Method called by the using code to indicate it is done with this instance. This lets instance merge accumulated changes into parent (if need be), safely and efficiently, and without calling code having to know about parent information

size

public int size()

bucketCount

public int bucketCount()

Since:

1.9.9

maybeDirty

public boolean maybeDirty()

Method called to check to quickly see if a child symbol table may have gotten additional entries. Used for checking to see if a child table should be merged into shared table.

hashSeed

public int hashSeed()

Since:

1.9.9

collisionCount

public int collisionCount()

Method mostly needed by unit tests; calculates number of entries that are in collision list. Value can be at most (size() - 1), but should usually be much lower, ideally 0.

Since:

1.9.9

maxCollisionLength

public int maxCollisionLength()

Method mostly needed by unit tests; calculates length of the longest collision chain. This should typically be a low number, but may be up to size() - 1 in the pathological case

Since:

1.9.9

getEmptyName

public static Name getEmptyName()

findName

public Name findName(int firstQuad)

Finds and returns name matching the specified symbol, if such name already exists in the table. If not, will return null.

Note: separate methods to optimize common case of short element/attribute names (4 or less ascii characters)

Parameters:

firstQuad - int32 containing first 4 bytes of the name; if the whole name less than 4 bytes, padded with zero bytes in front (zero MSBs, ie. right aligned)

Returns:

Name matching the symbol passed (or constructed for it)

findName

public Name findName(int firstQuad,
            int secondQuad)

Finds and returns name matching the specified symbol, if such name already exists in the table. If not, will return null.

Note: separate methods to optimize common case of relatively short element/attribute names (8 or less ascii characters)

Parameters:

firstQuad - int32 containing first 4 bytes of the name.

secondQuad - int32 containing bytes 5 through 8 of the name; if less than 8 bytes, padded with up to 3 zero bytes in front (zero MSBs, ie. right aligned)

Returns:

Name matching the symbol passed (or constructed for it)

findName

public Name findName(int[] quads,
            int qlen)

Finds and returns name matching the specified symbol, if such name already exists in the table; or if not, creates name object, adds to the table, and returns it.

Note: this is the general purpose method that can be called for names of any length. However, if name is less than 9 bytes long, it is preferable to call the version optimized for short names.

Parameters:

quads - Array of int32s, each of which contain 4 bytes of encoded name

qlen - Number of int32s, starting from index 0, in quads parameter

Returns:

Name matching the symbol passed (or constructed for it)

addName

public Name addName(String symbolStr,
           int q1,
           int q2)

addName

public Name addName(String symbolStr,
           int[] quads,
           int qlen)

calcHash

public final int calcHash(int firstQuad)

calcHash

public final int calcHash(int firstQuad,
           int secondQuad)

calcHash

public final int calcHash(int[] quads,
           int qlen)

calcQuads

protected static int[] calcQuads(byte[] wordBytes)

reportTooManyCollisions

protected void reportTooManyCollisions(int maxLen)

Since:

1.9.9