package com.ctc.wstx.dtd;
import java.util.Iterator;
import java.util.TreeSet;
import com.ctc.wstx.util.PrefixedName;
Implementation of PrefixedNameSet
suitable for storing large number of entries; basically anything above trivially small sets (4 or less).
Notes about usage:
- All Strings contained in
PrefixedName
instances are assumed interned, so that equality comparison can be done (both for values stored and keys used)
- It is assumed that sets are never empty, ie. always contain at
least one entry.
- It is assumed that caller has ensured that there are no duplicates
in the set -- this data structure does no further validation.
/**
* Implementation of {@link PrefixedNameSet} suitable for storing large number
* of entries; basically anything above trivially small sets (4 or less).
*<p>
* Notes about usage:
* <ul>
* <li>All Strings contained in {@link PrefixedName} instances are assumed
* interned, so that equality comparison can be done (both for values
* stored and keys used)
* </li>
* <li>It is assumed that sets are never empty, ie. always contain at
* least one entry.
* </li>
* <li>It is assumed that caller has ensured that there are no duplicates
* in the set -- this data structure does no further validation.
* </li>
* </ul>
*/
public final class LargePrefixedNameSet
extends PrefixedNameSet
{
Let's not bother creating tiny hash areas; should seldom be a problem
as smaller sets are usually created using different impl. class.
/**
* Let's not bother creating tiny hash areas; should seldom be a problem
* as smaller sets are usually created using different impl. class.
*/
final static int MIN_HASH_AREA = 8;
final boolean mNsAware;
Primary hash area in which NameKeys are added. Sized to be the smallest
power of two bigger than number of entries; but at least 4 (it doesn't
make sense to create smaller arrays)
/**
* Primary hash area in which NameKeys are added. Sized to be the smallest
* power of two bigger than number of entries; but at least 4 (it doesn't
* make sense to create smaller arrays)
*/
final PrefixedName[] mNames;
Secondary (spill) area, in which keys whose hash values collide
with primary ones are added. Number of buckets is 1/4 of number
of primary entries,
/**
* Secondary (spill) area, in which keys whose hash values collide
* with primary ones are added. Number of buckets is 1/4 of number
* of primary entries,
*/
final Bucket[] mBuckets;
public LargePrefixedNameSet(boolean nsAware, PrefixedName[] names)
{
mNsAware = nsAware;
int len = names.length;
// Let's find the size first... let's except 1/8 slack (88% fill rate)
int minSize = len + ((len + 7) >> 3);
// Let's not create hash areas smaller than certain limit
int tableSize = MIN_HASH_AREA;
while (tableSize < minSize) {
tableSize += tableSize;
}
mNames = new PrefixedName[tableSize];
// and 1/4 of that for spill area... but let's do that lazily
Bucket[] buckets = null;
int mask = (tableSize - 1);
for (int i = 0; i < len; ++i) {
PrefixedName nk = names[i];
int ix = (nk.hashCode() & mask);
if (mNames[ix] == null) { // no collision
mNames[ix] = nk;
} else { // collision, need to add a bucket
ix >>= 2;
Bucket old;
if (buckets == null) {
buckets = new Bucket[tableSize >> 2];
old = null;
} else {
old = buckets[ix];
}
buckets[ix] = new Bucket(nk, old);
}
}
mBuckets = buckets;
}
@Override
public boolean hasMultiple() { return true; }
Returns: True if the set contains specified name; false if not.
/**
* @return True if the set contains specified name; false if not.
*/
@Override
public boolean contains(PrefixedName name)
{
PrefixedName[] hashArea = mNames;
int index = name.hashCode() & (hashArea.length - 1);
PrefixedName res = hashArea[index];
if (res != null && res.equals(name)) {
return true;
}
Bucket[] buckets = mBuckets;
if (buckets != null) {
for (Bucket bucket = buckets[index >> 2]; bucket != null;
bucket = bucket.getNext()) {
res = bucket.getName();
if (res.equals(name)) {
return true;
}
}
}
return false;
}
Method called by debug/error handling code, to get a list of
all names contained.
/**
* Method called by debug/error handling code, to get a list of
* all names contained.
*/
@Override
public void appendNames(StringBuilder sb, String sep)
{
// Let's first get the alphabetized list of all names from main hash
TreeSet<PrefixedName> ts = new TreeSet<PrefixedName>();
for (int i = 0; i < mNames.length; ++i) {
PrefixedName name = mNames[i];
if (name != null) {
ts.add(name);
}
}
// then spill area
if (mBuckets != null) {
for (int i = 0; i < (mNames.length >> 2); ++i) {
Bucket b = mBuckets[i];
while (b != null) {
ts.add(b.getName());
b = b.getNext();
}
}
}
// And then append them:
Iterator<PrefixedName> it = ts.iterator();
boolean first = true;
while (it.hasNext()) {
if (first) {
first = false;
} else {
sb.append(sep);
}
sb.append(it.next().toString());
}
}
/*
///////////////////////////////////////////////////////////
// Helper class(es)
///////////////////////////////////////////////////////////
*/
private final static class Bucket
{
final PrefixedName mName;
final Bucket mNext;
public Bucket(PrefixedName name, Bucket next) {
mName = name;
mNext = next;
}
public PrefixedName getName() { return mName; }
public Bucket getNext() { return mNext; }
/*
public boolean contains(PrefixedName n) {
return mName.equals(n);
}
*/
}
}