// [core#191]
@SuppressWarnings("deprecation")
public void testShortNameCollisionsDirect() throws IOException {
final int COUNT = 600;
// First, char-based
{
CharsToNameCanonicalizer symbols = CharsToNameCanonicalizer.createRoot(1);
for (int i = 0; i < COUNT; ++i) {
String id = String.valueOf((char) i);
char[] ch = id.toCharArray();
symbols.findSymbol(ch, 0, ch.length, symbols.calcHash(id));
}
assertEquals(COUNT, symbols.size());
assertEquals(1024, symbols.bucketCount());
assertEquals(16, symbols.collisionCount());
assertEquals(1, symbols.maxCollisionLength());
}
// then byte-based
{
BytesToNameCanonicalizer symbols =
BytesToNameCanonicalizer.createRoot(1).makeChild(JsonFactory.Feature.collectDefaults());
for (int i = 0; i < COUNT; ++i) {
String id = String.valueOf((char) i);
int[] quads = calcQuads(id.getBytes("UTF-8"));
symbols.addName(id, quads, quads.length);
}
assertEquals(COUNT, symbols.size());
assertEquals(1024, symbols.bucketCount());
assertEquals(209, symbols.collisionCount());
assertEquals(1, symbols.maxCollisionLength());
}
}
// [core#191]
public void testShortQuotedDirectChars() throws IOException {
final int COUNT = 400;
CharsToNameCanonicalizer symbols = CharsToNameCanonicalizer.createRoot(1);
for (int i = 0; i < COUNT; ++i) {
String id = String.format("\\u%04x", i);
char[] ch = id.toCharArray();
symbols.findSymbol(ch, 0, ch.length, symbols.calcHash(id));
}
assertEquals(COUNT, symbols.size());
assertEquals(1024, symbols.bucketCount());
assertEquals(50, symbols.collisionCount());
assertEquals(2, symbols.maxCollisionLength());
}
// [core#187]: unexpectedly high number of collisions for straight numbers
public void testCollisionsWithChars187() throws IOException {
CharsToNameCanonicalizer symbols = CharsToNameCanonicalizer.createRoot(1);
final int COUNT = 30000;
for (int i = 0; i < COUNT; ++i) {
String id = String.valueOf(10000 + i);
char[] ch = id.toCharArray();
symbols.findSymbol(ch, 0, ch.length, symbols.calcHash(id));
}
assertEquals(COUNT, symbols.size());
assertEquals(65536, symbols.bucketCount());
// collision count rather high, but has to do
assertEquals(7127, symbols.collisionCount());
// as well as collision counts
assertEquals(4, symbols.maxCollisionLength());
}
// [Issue#145]
public void testThousandsOfSymbolsWithChars() throws IOException {
final int SEED = 33333;
CharsToNameCanonicalizer symbolsCRoot = CharsToNameCanonicalizer.createRoot(SEED);
int exp = 0;
for (int doc = 0; doc < 100; ++doc) {
CharsToNameCanonicalizer symbolsC =
symbolsCRoot.makeChild(JsonFactory.Feature.collectDefaults());
for (int i = 0; i < 250; ++i) {
String name = "f_" + doc + "_" + i;
char[] ch = name.toCharArray();
String str = symbolsC.findSymbol(ch, 0, ch.length, symbolsC.calcHash(name));
assertNotNull(str);
}
symbolsC.release();
exp += 250;
if (exp > CharsToNameCanonicalizer.MAX_ENTRIES_FOR_REUSE) {
exp = 0;
}
assertEquals(exp, symbolsCRoot.size());
}
}
// Test for verifying stability of hashCode, wrt collisions, using
// synthetic field name generation and character-based input
public void testSyntheticWithChars() {
// pass seed, to keep results consistent:
CharsToNameCanonicalizer symbols = CharsToNameCanonicalizer.createRoot(1);
final int COUNT = 12000;
for (int i = 0; i < COUNT; ++i) {
String id = fieldNameFor(i);
char[] ch = id.toCharArray();
symbols.findSymbol(ch, 0, ch.length, symbols.calcHash(id));
}
assertEquals(16384, symbols.bucketCount());
assertEquals(COUNT, symbols.size());
// System.out.printf("Char stuff: collisions %d, max-coll %d\n", symbols.collisionCount(),
// symbols.maxCollisionLength());
// holy guacamoley... there are way too many. 31 gives 3567 (!), 33 gives 2747
// ... at least before shuffling. Shuffling helps quite a lot, so:
assertEquals(3431, symbols.collisionCount());
assertEquals(6, symbols.maxCollisionLength());
}