// Since 2.6
public void testThousandsOfSymbolsWithNew() throws IOException {
final int SEED = 33333;
ByteQuadsCanonicalizer symbolsBRoot = ByteQuadsCanonicalizer.createRoot(SEED);
final Charset utf8 = Charset.forName("UTF-8");
int exp = 0;
ByteQuadsCanonicalizer symbolsB = null;
// loop to get
for (int doc = 0; doc < 100; ++doc) {
symbolsB = symbolsBRoot.makeChild(JsonFactory.Feature.collectDefaults());
for (int i = 0; i < 250; ++i) {
String name = "f_" + doc + "_" + i;
int[] quads = calcQuads(name.getBytes(utf8));
symbolsB.addName(name, quads, quads.length);
String n = symbolsB.findName(quads, quads.length);
assertEquals(name, n);
}
symbolsB.release();
exp += 250;
if (exp > ByteQuadsCanonicalizer.MAX_ENTRIES_FOR_REUSE) {
exp = 0;
}
assertEquals(exp, symbolsBRoot.size());
}
/* 05-Feb-2015, tatu: Fragile, but it is important to ensure that collision
* rates are not accidentally increased...
*/
assertEquals(6250, symbolsB.size());
assertEquals(4761, symbolsB.primaryCount()); // 80% primary hit rate
assertEquals(1190, symbolsB.secondaryCount()); // 13% secondary
assertEquals(299, symbolsB.tertiaryCount()); // 7% tertiary
assertEquals(0, symbolsB.spilloverCount()); // and couple of leftovers
}
