public void Search_Result_Is_Same_For_InvertedIndex_And_SimpleSearching()
{
var tokenizer = new DefaultTokenizer();
var ticks = Environment.TickCount;
Console.WriteLine($"Test ticks: {ticks}");
var lines = TestDataGenerator.GetRandomLines(ticks);
var invertedIndex = new InvertedIndex(tokenizer);
var phrase = TestDataGenerator.GetSearchPhrase(ticks);
BuildIndex(invertedIndex, lines);
var inmemoryResult = InmemorySimpleSearch.Find(lines, phrase);
invertedIndex.Find(phrase)
.Select(x => new { x.RowNumber, x.ColNumber })
.Should()
.BeEquivalentTo(inmemoryResult.Select(x => new { x.RowNumber, x.ColNumber }));
}
public void InvertedIndex_Should_Be_Faster_Than_Simple_Searching()
{
const int phrasesCount = 50;
var phrases = new string[phrasesCount];
var tickCount = Environment.TickCount;
Console.WriteLine($"TickCount: {tickCount}");
for (var i = 0; i < phrasesCount; i++)
{
phrases[i] = TestDataGenerator.GetSearchPhrase(tickCount + i);
}
var tokenizer = new DefaultTokenizer();
var stopWatch = new Stopwatch();
var lines = TestDataGenerator.GetRandomLines(tickCount, 50000);
var invertedIndex = new InvertedIndex(tokenizer);
BuildIndex(invertedIndex, lines);
stopWatch.Start();
for (var i = 0; i < phrasesCount; i++)
{
var elapsedBefore = stopWatch.Elapsed;
invertedIndex.Find(phrases[i]);
var elapsed = stopWatch.Elapsed - elapsedBefore;
Console.WriteLine($"Elapsed for phrase: {phrases[i]} {elapsed}");
}
var indexSearchingTime = stopWatch.Elapsed;
stopWatch.Restart();
for (var i = 0; i < phrasesCount; i++)
{
InmemorySimpleSearch.Find(lines, phrases[i]);
}
var simpleSearchingTime = stopWatch.Elapsed;
Console.WriteLine($"InvertedIndex searching time: {indexSearchingTime}");
Console.WriteLine($"Simple searching time: {simpleSearchingTime}");
indexSearchingTime.Should().BeLessThan(simpleSearchingTime);
}
public void Add_And_Read_Simultaneously()
{
var exceptionsCount = 0;
var index = this.GetNewIndex();
var ticks = 100;
Action addAction = () =>
{
try
{
for (var i = 0; i < 5; i++)
{
var lines = TestDataGenerator.GetRandomLines(ticks + i, 10000);
index.Add(lines, "some doc" + i);
}
}
catch (Exception e)
{
Console.WriteLine(e);
Interlocked.Increment(ref exceptionsCount);
throw;
}
};
var stopWatch = Stopwatch.StartNew();
var addTasks = new Task[2];
for (var i = 0; i < addTasks.Length; i++)
{
addTasks[i] = new Task(addAction);
addTasks[i].Start();
}
Action readAction = () =>
{
try
{
for (var i = 0; i < 1000; i++)
{
var phrase = TestDataGenerator.GetSearchPhrase(ticks + i);
index.Find(phrase);
}
}
catch (Exception e)
{
Console.WriteLine(e);
Interlocked.Increment(ref exceptionsCount);
throw;
}
};
var readTasks = new Task[2];
for (var i = 0; i < readTasks.Length; i++)
{
readTasks[i] = new Task(readAction);
readTasks[i].Start();
}
Task.WaitAll(addTasks);
Console.WriteLine($"Build elapsed: {stopWatch.Elapsed}");
Task.WaitAll(readTasks);
Console.WriteLine($"Total elapsed: {stopWatch.Elapsed}");
exceptionsCount.Should().Be(0);
}