public void TestNoDupsAcrossGrams()
{
IEnumerable <Input> keys = AnalyzingSuggesterTest.Shuffle(
new Input("foo bar bar bar bar", 50)
);
Analyzer a = new MockAnalyzer(Random);
FreeTextSuggester sug = new FreeTextSuggester(a, a, 2, (byte)0x20);
sug.Build(new InputArrayIterator(keys));
assertEquals("foo bar/1.00",
ToString(sug.DoLookup("foo b", 10)));
}
public void TestTwoEndingHoles()
{
// Just deletes "of"
Analyzer a = new TestEndingHoleAnalyzer();
IEnumerable <Input> keys = AnalyzingSuggesterTest.Shuffle(
new Input("wizard of of oz", 50)
);
FreeTextSuggester sug = new FreeTextSuggester(a, a, 3, (byte)0x20);
sug.Build(new InputArrayIterator(keys));
assertEquals("",
ToString(sug.DoLookup("wizard of of", 10)));
}
public void TestUnigrams()
{
IEnumerable <Input> keys = AnalyzingSuggesterTest.Shuffle(
new Input("foo bar baz blah boo foo bar foo bee", 50)
);
Analyzer a = new MockAnalyzer(Random);
FreeTextSuggester sug = new FreeTextSuggester(a, a, 1, (byte)0x20);
sug.Build(new InputArrayIterator(keys));
// Sorts first by count, descending, second by term, ascending
assertEquals("bar/0.22 baz/0.11 bee/0.11 blah/0.11 boo/0.11",
ToString(sug.DoLookup("b", 10)));
}
public void TestBasic()
{
IEnumerable <Input> keys = AnalyzingSuggesterTest.Shuffle(
new Input("foo bar baz blah", 50),
new Input("boo foo bar foo bee", 20)
);
Analyzer a = new MockAnalyzer(Random);
FreeTextSuggester sug = new FreeTextSuggester(a, a, 2, (byte)0x20);
sug.Build(new InputArrayIterator(keys));
assertEquals(2, sug.Count);
for (int i = 0; i < 2; i++)
{
// Uses bigram model and unigram backoff:
assertEquals("foo bar/0.67 foo bee/0.33 baz/0.04 blah/0.04 boo/0.04",
ToString(sug.DoLookup("foo b", 10)));
// Uses only bigram model:
assertEquals("foo bar/0.67 foo bee/0.33",
ToString(sug.DoLookup("foo ", 10)));
// Uses only unigram model:
assertEquals("foo/0.33",
ToString(sug.DoLookup("foo", 10)));
// Uses only unigram model:
assertEquals("bar/0.22 baz/0.11 bee/0.11 blah/0.11 boo/0.11",
ToString(sug.DoLookup("b", 10)));
// Try again after save/load:
DirectoryInfo tmpDir = CreateTempDir("FreeTextSuggesterTest");
//tmpDir.Create();
FileInfo path = new FileInfo(Path.Combine(tmpDir.FullName, "suggester"));
using (Stream os = new FileStream(path.FullName, FileMode.Create, FileAccess.Write))
sug.Store(os);
using (Stream @is = new FileStream(path.FullName, FileMode.Open, FileAccess.Read))
{
sug = new FreeTextSuggester(a, a, 2, (byte)0x20);
sug.Load(@is);
}
assertEquals(2, sug.Count);
}
}
public void TestIllegalByteDuringBuild()
{
// Default separator is INFORMATION SEPARATOR TWO
// (0x1e), so no input token is allowed to contain it
IEnumerable <Input> keys = AnalyzingSuggesterTest.Shuffle(
new Input("foo\u001ebar baz", 50)
);
FreeTextSuggester sug = new FreeTextSuggester(new MockAnalyzer(Random));
try
{
sug.Build(new InputArrayIterator(keys));
fail("did not hit expected exception");
}
catch (ArgumentException /*iae*/)
{
// expected
}
}
public void TestEndingHole()
{
// Just deletes "of"
Analyzer a = new TestEndingHoleAnalyzer();
IEnumerable <Input> keys = AnalyzingSuggesterTest.Shuffle(
new Input("wizard of oz", 50)
);
FreeTextSuggester sug = new FreeTextSuggester(a, a, 3, (byte)0x20);
sug.Build(new InputArrayIterator(keys));
assertEquals("wizard _ oz/1.00",
ToString(sug.DoLookup("wizard of", 10)));
// Falls back to unigram model, with backoff 0.4 times
// prop 0.5:
assertEquals("oz/0.20",
ToString(sug.DoLookup("wizard o", 10)));
}
public void TestEmptyString()
{
IEnumerable <Input> keys = AnalyzingSuggesterTest.Shuffle(
new Input("foo bar bar bar bar", 50)
);
Analyzer a = new MockAnalyzer(Random);
FreeTextSuggester sug = new FreeTextSuggester(a, a, 2, (byte)0x20);
sug.Build(new InputArrayIterator(keys));
try
{
sug.DoLookup("", 10);
fail("did not hit exception");
}
catch (ArgumentException /*iae*/)
{
// expected
}
}
