public void TestBothExactAndPrefix()
{
Analyzer a = new MockAnalyzer(Random(), MockTokenizer.WHITESPACE, false);
AnalyzingInfixSuggester suggester = new AnalyzingInfixSuggester(TEST_VERSION_CURRENT, NewDirectory(), a, a, 3);
suggester.Build(new InputArrayIterator(new Input[0]));
suggester.Add(new BytesRef("the pen is pretty"), null, 10, new BytesRef("foobaz"));
suggester.Refresh();
IList <Lookup.LookupResult> results = suggester.DoLookup(TestUtil.StringToCharSequence("pen p", Random()).ToString(), 10, true, true);
assertEquals(1, results.size());
assertEquals("the <b>pen</b> is <b>p</b>retty", results[0].key);
assertEquals(10, results[0].value);
assertEquals(new BytesRef("foobaz"), results[0].payload);
suggester.Dispose();
}
public void TestEmptyAtStart()
{
Analyzer a = new MockAnalyzer(Random(), MockTokenizer.WHITESPACE, false);
AnalyzingInfixSuggester suggester = new AnalyzingInfixSuggester(TEST_VERSION_CURRENT, NewDirectory(), a, a, 3);
suggester.Build(new InputArrayIterator(new Input[0]));
suggester.Add(new BytesRef("a penny saved is a penny earned"), null, 10, new BytesRef("foobaz"));
suggester.Add(new BytesRef("lend me your ear"), null, 8, new BytesRef("foobar"));
suggester.Refresh();
IList <Lookup.LookupResult> results = suggester.DoLookup(TestUtil.StringToCharSequence("ear", Random()).ToString(), 10, true, true);
assertEquals(2, results.size());
assertEquals("a penny saved is a penny <b>ear</b>ned", results[0].key);
assertEquals(10, results[0].value);
assertEquals(new BytesRef("foobaz"), results[0].payload);
assertEquals("lend me your <b>ear</b>", results[1].key);
assertEquals(8, results[1].value);
assertEquals(new BytesRef("foobar"), results[1].payload);
results = suggester.DoLookup(TestUtil.StringToCharSequence("ear ", Random()).ToString(), 10, true, true);
assertEquals(1, results.size());
assertEquals("lend me your <b>ear</b>", results[0].key);
assertEquals(8, results[0].value);
assertEquals(new BytesRef("foobar"), results[0].payload);
results = suggester.DoLookup(TestUtil.StringToCharSequence("pen", Random()).ToString(), 10, true, true);
assertEquals(1, results.size());
assertEquals("a <b>pen</b>ny saved is a <b>pen</b>ny earned", results[0].key);
assertEquals(10, results[0].value);
assertEquals(new BytesRef("foobaz"), results[0].payload);
results = suggester.DoLookup(TestUtil.StringToCharSequence("p", Random()).ToString(), 10, true, true);
assertEquals(1, results.size());
assertEquals("a <b>p</b>enny saved is a <b>p</b>enny earned", results[0].key);
assertEquals(10, results[0].value);
assertEquals(new BytesRef("foobaz"), results[0].payload);
suggester.Dispose();
}
public void TestBasicNRT()
{
Input[] keys = new Input[] {
new Input("lend me your ear", 8, new BytesRef("foobar")),
};
Analyzer a = new MockAnalyzer(Random(), MockTokenizer.WHITESPACE, false);
AnalyzingInfixSuggester suggester = new AnalyzingInfixSuggester(TEST_VERSION_CURRENT, NewDirectory(), a, a, 3);
suggester.Build(new InputArrayIterator(keys));
IList <Lookup.LookupResult> results = suggester.DoLookup(TestUtil.StringToCharSequence("ear", Random()).ToString(), 10, true, true);
assertEquals(1, results.size());
assertEquals("lend me your <b>ear</b>", results.ElementAt(0).key);
assertEquals(8, results.ElementAt(0).value);
assertEquals(new BytesRef("foobar"), results.ElementAt(0).payload);
// Add a new suggestion:
suggester.Add(new BytesRef("a penny saved is a penny earned"), null, 10, new BytesRef("foobaz"));
// Must refresh to see any newly added suggestions:
suggester.Refresh();
results = suggester.DoLookup(TestUtil.StringToCharSequence("ear", Random()).ToString(), 10, true, true);
assertEquals(2, results.size());
assertEquals("a penny saved is a penny <b>ear</b>ned", results.ElementAt(0).key);
assertEquals(10, results.ElementAt(0).value);
assertEquals(new BytesRef("foobaz"), results.ElementAt(0).payload);
assertEquals("lend me your <b>ear</b>", results.ElementAt(1).key);
assertEquals(8, results.ElementAt(1).value);
assertEquals(new BytesRef("foobar"), results.ElementAt(1).payload);
results = suggester.DoLookup(TestUtil.StringToCharSequence("ear ", Random()).ToString(), 10, true, true);
assertEquals(1, results.size());
assertEquals("lend me your <b>ear</b>", results.ElementAt(0).key);
assertEquals(8, results.ElementAt(0).value);
assertEquals(new BytesRef("foobar"), results.ElementAt(0).payload);
results = suggester.DoLookup(TestUtil.StringToCharSequence("pen", Random()).ToString(), 10, true, true);
assertEquals(1, results.size());
assertEquals("a <b>pen</b>ny saved is a <b>pen</b>ny earned", results.ElementAt(0).key);
assertEquals(10, results.ElementAt(0).value);
assertEquals(new BytesRef("foobaz"), results.ElementAt(0).payload);
results = suggester.DoLookup(TestUtil.StringToCharSequence("p", Random()).ToString(), 10, true, true);
assertEquals(1, results.size());
assertEquals("a <b>p</b>enny saved is a <b>p</b>enny earned", results.ElementAt(0).key);
assertEquals(10, results.ElementAt(0).value);
assertEquals(new BytesRef("foobaz"), results.ElementAt(0).payload);
// Change the weight:
suggester.Update(new BytesRef("lend me your ear"), null, 12, new BytesRef("foobox"));
// Must refresh to see any newly added suggestions:
suggester.Refresh();
results = suggester.DoLookup(TestUtil.StringToCharSequence("ear", Random()).ToString(), 10, true, true);
assertEquals(2, results.size());
assertEquals("lend me your <b>ear</b>", results.ElementAt(0).key);
assertEquals(12, results.ElementAt(0).value);
assertEquals(new BytesRef("foobox"), results.ElementAt(0).payload);
assertEquals("a penny saved is a penny <b>ear</b>ned", results.ElementAt(1).key);
assertEquals(10, results.ElementAt(1).value);
assertEquals(new BytesRef("foobaz"), results.ElementAt(1).payload);
suggester.Dispose();
}
public void TestRandomNRT()
{
DirectoryInfo tempDir = CreateTempDir("AnalyzingInfixSuggesterTest");
Analyzer a = new MockAnalyzer(Random(), MockTokenizer.WHITESPACE, false);
int minPrefixChars = Random().nextInt(7);
if (VERBOSE)
{
Console.WriteLine(" minPrefixChars=" + minPrefixChars);
}
AnalyzingInfixSuggester suggester = new AnalyzingInfixSuggester(TEST_VERSION_CURRENT, NewFSDirectory(tempDir), a, a, minPrefixChars);
// Initial suggester built with nothing:
suggester.Build(new InputArrayIterator(new Input[0]));
var stop = new AtomicBoolean(false);
Exception[] error = new Exception[] { null };
LookupThread lookupThread = new LookupThread(this, suggester, stop, error);
lookupThread.Start();
int iters = AtLeast(1000);
int visibleUpto = 0;
ISet <long> usedWeights = new HashSet <long>();
ISet <string> usedKeys = new HashSet <string>();
List <Input> inputs = new List <Input>();
List <Update> pendingUpdates = new List <Update>();
for (int iter = 0; iter < iters; iter++)
{
string text;
while (true)
{
text = RandomText();
if (usedKeys.contains(text) == false)
{
usedKeys.add(text);
break;
}
}
// Carefully pick a weight we never used, to sidestep
// tie-break problems:
long weight;
while (true)
{
weight = Random().nextInt(10 * iters);
if (usedWeights.contains(weight) == false)
{
usedWeights.add(weight);
break;
}
}
if (inputs.size() > 0 && Random().nextInt(4) == 1)
{
// Update an existing suggestion
Update update = new Update();
update.index = Random().nextInt(inputs.size());
update.weight = weight;
Input input = inputs.ElementAt(update.index);
pendingUpdates.Add(update);
if (VERBOSE)
{
Console.WriteLine("TEST: iter=" + iter + " update input=" + input.term.Utf8ToString() + "/" + weight);
}
suggester.Update(input.term, null, weight, input.term);
}
else
{
// Add a new suggestion
inputs.Add(new Input(text, weight, new BytesRef(text)));
if (VERBOSE)
{
Console.WriteLine("TEST: iter=" + iter + " add input=" + text + "/" + weight);
}
BytesRef br = new BytesRef(text);
suggester.Add(br, null, weight, br);
}
if (Random().nextInt(15) == 7)
{
if (VERBOSE)
{
Console.WriteLine("TEST: now refresh suggester");
}
suggester.Refresh();
visibleUpto = inputs.size();
foreach (Update update in pendingUpdates)
{
Input oldInput = inputs.ElementAt(update.index);
Input newInput = new Input(oldInput.term, update.weight, oldInput.payload);
inputs[update.index] = newInput;
}
pendingUpdates.Clear();
}
if (Random().nextInt(50) == 7)
{
if (VERBOSE)
{
Console.WriteLine("TEST: now close/reopen suggester");
}
//lookupThread.Finish();
stop.Set(true);
lookupThread.Join();
Assert.Null(error[0], "Unexpcted exception at retry : \n" + stackTraceStr(error[0]));
suggester.Dispose();
suggester = new AnalyzingInfixSuggester(TEST_VERSION_CURRENT, NewFSDirectory(tempDir), a, a, minPrefixChars);
lookupThread = new LookupThread(this, suggester, stop, error);
lookupThread.Start();
visibleUpto = inputs.size();
foreach (Update update in pendingUpdates)
{
Input oldInput = inputs.ElementAt(update.index);
Input newInput = new Input(oldInput.term, update.weight, oldInput.payload);
inputs[update.index] = newInput;
}
pendingUpdates.Clear();
}
if (visibleUpto > 0)
{
string query = RandomText();
bool lastPrefix = Random().nextInt(5) != 1;
if (lastPrefix == false)
{
query += " ";
}
string[] queryTerms = Regex.Split(query, "\\s", RegexOptions.Compiled);
bool allTermsRequired = Random().nextInt(10) == 7;
bool doHilite = Random().nextBoolean();
if (VERBOSE)
{
Console.WriteLine("TEST: lookup \"" + query + "\" allTermsRequired=" + allTermsRequired + " doHilite=" + doHilite);
}
// Stupid slow but hopefully correct matching:
List <Input> expected = new List <Input>();
for (int i = 0; i < visibleUpto; i++)
{
Input input = inputs.ElementAt(i);
string[] inputTerms = Regex.Split(input.term.Utf8ToString(), "\\s");
bool match = false;
for (int j = 0; j < queryTerms.Length; j++)
{
if (j < queryTerms.Length - 1 || lastPrefix == false)
{
// Exact match
for (int k = 0; k < inputTerms.Length; k++)
{
if (inputTerms[k].equals(queryTerms[j]))
{
match = true;
break;
}
}
}
else
{
// Prefix match
for (int k = 0; k < inputTerms.Length; k++)
{
if (inputTerms[k].StartsWith(queryTerms[j], StringComparison.InvariantCulture))
{
match = true;
break;
}
}
}
if (match)
{
if (allTermsRequired == false)
{
// At least one query term does match:
break;
}
match = false;
}
else if (allTermsRequired)
{
// At least one query term does not match:
break;
}
}
if (match)
{
if (doHilite)
{
expected.Add(new Input(Hilite(lastPrefix, inputTerms, queryTerms), input.v, input.term));
}
else
{
expected.Add(input);
}
}
}
expected.Sort(new TestRandomNRTComparator());
if (expected.Any())
{
int topN = TestUtil.NextInt(Random(), 1, expected.size());
IList <Lookup.LookupResult> actual = suggester.DoLookup(TestUtil.StringToCharSequence(query, Random()).ToString(), topN, allTermsRequired, doHilite);
int expectedCount = Math.Min(topN, expected.size());
if (VERBOSE)
{
Console.WriteLine(" expected:");
for (int i = 0; i < expectedCount; i++)
{
Input x = expected.ElementAt(i);
Console.WriteLine(" " + x.term.Utf8ToString() + "/" + x.v);
}
Console.WriteLine(" actual:");
foreach (Lookup.LookupResult result in actual)
{
Console.WriteLine(" " + result);
}
}
assertEquals(expectedCount, actual.size());
for (int i = 0; i < expectedCount; i++)
{
assertEquals(expected.ElementAt(i).term.Utf8ToString(), actual.ElementAt(i).key.toString());
assertEquals(expected.ElementAt(i).v, actual.ElementAt(i).value);
assertEquals(expected.ElementAt(i).payload, actual.ElementAt(i).payload);
}
}
else
{
if (VERBOSE)
{
Console.WriteLine(" no expected matches");
}
}
}
}
//lookupThread.finish();
stop.Set(true);
lookupThread.Join();
Assert.Null(error[0], "Unexpcted exception at retry : \n" + stackTraceStr(error[0]));
suggester.Dispose();
}