/// <summary>
/// Command-line tool. </summary>
public static void Main(string[] args)
{
bool printTree = false;
string path = null;
for (int i = 0; i < args.Length; i++)
{
if (args[i].Equals("-printTree"))
{
printTree = true;
}
else
{
path = args[i];
}
}
if (args.Length != (printTree ? 2 : 1))
{
Console.WriteLine("\nUsage: java -classpath ... org.apache.lucene.facet.util.PrintTaxonomyStats [-printTree] /path/to/taxononmy/index\n");
Environment.Exit(1);
}
Store.Directory dir = FSDirectory.Open(new DirectoryInfo(path));
var r = new DirectoryTaxonomyReader(dir);
PrintStats(r, System.Console.Out, printTree);
r.Dispose();
//dir.close();
}
public virtual void TestWriterParent2()
{
var indexDir = NewDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
FillTaxonomy(tw);
tw.Commit();
var tr = new DirectoryTaxonomyReader(indexDir);
CheckWriterParent(tr, tw);
tw.Dispose();
tr.Dispose();
indexDir.Dispose();
}
public virtual void TestWriterLock()
{
// native fslock impl gets angry if we use it, so use RAMDirectory explicitly.
var indexDir = new RAMDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
tw.AddCategory(new FacetLabel("hi", "there"));
tw.Commit();
// we deliberately not close the write now, and keep it open and
// locked.
// Verify that the writer worked:
var tr = new DirectoryTaxonomyReader(indexDir);
Assert.AreEqual(2, tr.GetOrdinal(new FacetLabel("hi", "there")));
// Try to open a second writer, with the first one locking the directory.
// We expect to get a LockObtainFailedException.
try
{
Assert.Null(new DirectoryTaxonomyWriter(indexDir));
Fail("should have failed to write in locked directory");
}
catch (LockObtainFailedException)
{
// this is what we expect to happen.
}
// Remove the lock, and now the open should succeed, and we can
// write to the new writer.
DirectoryTaxonomyWriter.Unlock(indexDir);
var tw2 = new DirectoryTaxonomyWriter(indexDir);
tw2.AddCategory(new FacetLabel("hey"));
tw2.Dispose();
// See that the writer indeed wrote:
var newtr = TaxonomyReader.OpenIfChanged(tr);
Assert.NotNull(newtr);
tr.Dispose();
tr = newtr;
Assert.AreEqual(3, tr.GetOrdinal(new FacetLabel("hey")));
tr.Dispose();
tw.Dispose();
indexDir.Dispose();
}
public virtual void TestTaxonomyReaderRefreshRaces()
{
// compute base child arrays - after first chunk, and after the other
var indexDirBase = NewDirectory();
var twBase = new DirectoryTaxonomyWriter(indexDirBase);
twBase.AddCategory(new FacetLabel("a", "0"));
FacetLabel abPath = new FacetLabel("a", "b");
twBase.AddCategory(abPath);
twBase.Commit();
var trBase = new DirectoryTaxonomyReader(indexDirBase);
ParallelTaxonomyArrays ca1 = trBase.ParallelTaxonomyArrays;
int abOrd = trBase.GetOrdinal(abPath);
int abYoungChildBase1 = ca1.Children()[abOrd];
int numCategories = AtLeast(800);
for (int i = 0; i < numCategories; i++)
{
twBase.AddCategory(new FacetLabel("a", "b", Convert.ToString(i)));
}
twBase.Dispose();
var newTaxoReader = TaxonomyReader.OpenIfChanged(trBase);
Assert.NotNull(newTaxoReader);
trBase.Dispose();
trBase = newTaxoReader;
ParallelTaxonomyArrays ca2 = trBase.ParallelTaxonomyArrays;
int abYoungChildBase2 = ca2.Children()[abOrd];
int numRetries = AtLeast(50);
for (int retry = 0; retry < numRetries; retry++)
{
AssertConsistentYoungestChild(abPath, abOrd, abYoungChildBase1, abYoungChildBase2, retry, numCategories);
}
trBase.Dispose();
indexDirBase.Dispose();
}
public virtual void TestSeparateReaderAndWriter2()
{
var indexDir = NewDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
tw.Commit();
var tr = new DirectoryTaxonomyReader(indexDir);
// Test getOrdinal():
FacetLabel author = new FacetLabel("Author");
Assert.AreEqual(1, tr.Size); // the empty taxonomy has size 1 (the root)
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, tr.GetOrdinal(author));
tw.AddCategory(author);
// before commit and refresh, no change:
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, tr.GetOrdinal(author));
Assert.AreEqual(1, tr.Size); // still root only...
Assert.Null(TaxonomyReader.OpenIfChanged(tr)); // this is not enough, because tw.Commit() hasn't been done yet
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, tr.GetOrdinal(author));
Assert.AreEqual(1, tr.Size); // still root only...
tw.Commit();
// still not enough before refresh:
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, tr.GetOrdinal(author));
Assert.AreEqual(1, tr.Size); // still root only...
var newTaxoReader = TaxonomyReader.OpenIfChanged(tr);
Assert.NotNull(newTaxoReader);
tr.Dispose();
tr = newTaxoReader;
Assert.AreEqual(1, tr.GetOrdinal(author));
Assert.AreEqual(2, tr.Size);
tw.Dispose();
tr.Dispose();
indexDir.Dispose();
}
public virtual void TestSeparateReaderAndWriter()
{
var indexDir = NewDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
tw.Commit();
var tr = new DirectoryTaxonomyReader(indexDir);
Assert.AreEqual(1, tr.Size); // the empty taxonomy has size 1 (the root)
tw.AddCategory(new FacetLabel("Author"));
Assert.AreEqual(1, tr.Size); // still root only...
Assert.Null(TaxonomyReader.OpenIfChanged(tr)); // this is not enough, because tw.Commit() hasn't been done yet
Assert.AreEqual(1, tr.Size); // still root only...
tw.Commit();
Assert.AreEqual(1, tr.Size); // still root only...
var newTaxoReader = TaxonomyReader.OpenIfChanged(tr);
Assert.NotNull(newTaxoReader);
tr.Dispose();
tr = newTaxoReader;
int author = 1;
try
{
Assert.AreEqual(TaxonomyReader.ROOT_ORDINAL, tr.ParallelTaxonomyArrays.Parents()[author]);
// ok
}
catch (System.IndexOutOfRangeException)
{
Fail("After category addition, commit() and refresh(), getParent for " + author + " should NOT throw exception");
}
Assert.AreEqual(2, tr.Size); // finally, see there are two categories
// now, add another category, and verify that after commit and refresh
// the parent of this category is correct (this requires the reader
// to correctly update its prefetched parent vector), and that the
// old information also wasn't ruined:
tw.AddCategory(new FacetLabel("Author", "Richard Dawkins"));
int dawkins = 2;
tw.Commit();
newTaxoReader = TaxonomyReader.OpenIfChanged(tr);
Assert.NotNull(newTaxoReader);
tr.Dispose();
tr = newTaxoReader;
int[] parents = tr.ParallelTaxonomyArrays.Parents();
Assert.AreEqual(author, parents[dawkins]);
Assert.AreEqual(TaxonomyReader.ROOT_ORDINAL, parents[author]);
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, parents[TaxonomyReader.ROOT_ORDINAL]);
Assert.AreEqual(3, tr.Size);
tw.Dispose();
tr.Dispose();
indexDir.Dispose();
}
public virtual void TestRootOnly2()
{
var indexDir = NewDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
tw.Commit();
var tr = new DirectoryTaxonomyReader(indexDir);
Assert.AreEqual(1, tr.Size);
Assert.AreEqual(0, tr.GetPath(0).Length);
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, tr.ParallelTaxonomyArrays.Parents()[0]);
Assert.AreEqual(0, tr.GetOrdinal(new FacetLabel()));
tw.Dispose();
tr.Dispose(true);
indexDir.Dispose();
}
public virtual void TestRootOnly()
{
var indexDir = NewDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
// right after opening the index, it should already contain the
// root, so have size 1:
Assert.AreEqual(1, tw.Size);
tw.Dispose();
var tr = new DirectoryTaxonomyReader(indexDir);
Assert.AreEqual(1, tr.Size);
Assert.AreEqual(0, tr.GetPath(0).Length);
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, tr.ParallelTaxonomyArrays.Parents()[0]);
Assert.AreEqual(0, tr.GetOrdinal(new FacetLabel()));
tr.Dispose(true);
indexDir.Dispose();
}
public virtual void TestReaderParent()
{
var indexDir = NewDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
FillTaxonomy(tw);
tw.Dispose();
var tr = new DirectoryTaxonomyReader(indexDir);
// check that the parent of the root ordinal is the invalid ordinal:
int[] parents = tr.ParallelTaxonomyArrays.Parents();
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, parents[0]);
// check parent of non-root ordinals:
for (int ordinal = 1; ordinal < tr.Size; ordinal++)
{
FacetLabel me = tr.GetPath(ordinal);
int parentOrdinal = parents[ordinal];
FacetLabel parent = tr.GetPath(parentOrdinal);
if (parent == null)
{
Fail("Parent of " + ordinal + " is " + parentOrdinal + ", but this is not a valid category.");
}
// verify that the parent is indeed my parent, according to the strings
if (!me.Subpath(me.Length - 1).Equals(parent))
{
Fail("Got parent " + parentOrdinal + " for ordinal " + ordinal + " but categories are " + Showcat(parent) + " and " + Showcat(me) + " respectively.");
}
}
tr.Dispose();
indexDir.Dispose();
}
public virtual void TestReaderBasic()
{
var indexDir = NewDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
FillTaxonomy(tw);
tw.Dispose();
var tr = new DirectoryTaxonomyReader(indexDir);
// test TaxonomyReader.getSize():
Assert.AreEqual(ExpectedCategories.Length, tr.Size);
// test round trips of ordinal => category => ordinal
for (int i = 0; i < tr.Size; i++)
{
Assert.AreEqual(i, tr.GetOrdinal(tr.GetPath(i)));
}
// test TaxonomyReader.getCategory():
for (int i = 1; i < tr.Size; i++)
{
FacetLabel expectedCategory = new FacetLabel(ExpectedCategories[i]);
FacetLabel category = tr.GetPath(i);
if (!expectedCategory.Equals(category))
{
Fail("For ordinal " + i + " expected category " + Showcat(expectedCategory) + ", but got " + Showcat(category));
}
}
// (also test invalid ordinals:)
Assert.Null(tr.GetPath(-1));
Assert.Null(tr.GetPath(tr.Size));
Assert.Null(tr.GetPath(TaxonomyReader.INVALID_ORDINAL));
// test TaxonomyReader.GetOrdinal():
for (int i = 1; i < ExpectedCategories.Length; i++)
{
int expectedOrdinal = i;
int ordinal = tr.GetOrdinal(new FacetLabel(ExpectedCategories[i]));
if (expectedOrdinal != ordinal)
{
Fail("For category " + Showcat(ExpectedCategories[i]) + " expected ordinal " + expectedOrdinal + ", but got " + ordinal);
}
}
// (also test invalid categories:)
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, tr.GetOrdinal(new FacetLabel("non-existant")));
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, tr.GetOrdinal(new FacetLabel("Author", "Jules Verne")));
tr.Dispose();
indexDir.Dispose();
}
public virtual void TestNrt()
{
var dir = NewDirectory();
var writer = new DirectoryTaxonomyWriter(dir);
var reader = new DirectoryTaxonomyReader(writer);
FacetLabel cp = new FacetLabel("a");
writer.AddCategory(cp);
var newReader = TaxonomyReader.OpenIfChanged(reader);
Assert.NotNull(newReader, "expected a new instance");
Assert.AreEqual(2, newReader.Size);
Assert.AreNotSame(TaxonomyReader.INVALID_ORDINAL, newReader.GetOrdinal(cp));
reader.Dispose();
reader = newReader;
writer.Dispose();
reader.Dispose();
dir.Dispose();
}
public virtual void TestChildrenArraysInvariants()
{
var indexDir = NewDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
FillTaxonomy(tw);
tw.Dispose();
var tr = new DirectoryTaxonomyReader(indexDir);
ParallelTaxonomyArrays ca = tr.ParallelTaxonomyArrays;
int[] children = ca.Children();
Assert.AreEqual(tr.Size, children.Length);
int[] olderSiblingArray = ca.Siblings();
Assert.AreEqual(tr.Size, olderSiblingArray.Length);
// test that the "youngest child" of every category is indeed a child:
int[] parents = tr.ParallelTaxonomyArrays.Parents();
for (int i = 0; i < tr.Size; i++)
{
int youngestChild = children[i];
if (youngestChild != TaxonomyReader.INVALID_ORDINAL)
{
Assert.AreEqual(i, parents[youngestChild]);
}
}
// test that the "older sibling" of every category is indeed older (lower)
// (it can also be INVALID_ORDINAL, which is lower than any ordinal)
for (int i = 0; i < tr.Size; i++)
{
Assert.True(olderSiblingArray[i] < i, "olderSiblingArray[" + i + "] should be <" + i);
}
// test that the "older sibling" of every category is indeed a sibling
// (they share the same parent)
for (int i = 0; i < tr.Size; i++)
{
int sibling = olderSiblingArray[i];
if (sibling == TaxonomyReader.INVALID_ORDINAL)
{
continue;
}
Assert.AreEqual(parents[i], parents[sibling]);
}
// And now for slightly more complex (and less "invariant-like"...)
// tests:
// test that the "youngest child" is indeed the youngest (so we don't
// miss the first children in the chain)
for (int i = 0; i < tr.Size; i++)
{
// Find the really youngest child:
int j;
for (j = tr.Size - 1; j > i; j--)
{
if (parents[j] == i)
{
break; // found youngest child
}
}
if (j == i) // no child found
{
j = TaxonomyReader.INVALID_ORDINAL;
}
Assert.AreEqual(j, children[i]);
}
// test that the "older sibling" is indeed the least oldest one - and
// not a too old one or -1 (so we didn't miss some children in the
// middle or the end of the chain).
for (int i = 0; i < tr.Size; i++)
{
// Find the youngest older sibling:
int j;
for (j = i - 1; j >= 0; j--)
{
if (parents[j] == parents[i])
{
break; // found youngest older sibling
}
}
if (j < 0) // no sibling found
{
j = TaxonomyReader.INVALID_ORDINAL;
}
Assert.AreEqual(j, olderSiblingArray[i]);
}
tr.Dispose();
indexDir.Dispose();
}
public virtual void TestChildrenArraysGrowth()
{
var indexDir = NewDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
tw.AddCategory(new FacetLabel("hi", "there"));
tw.Commit();
var tr = new DirectoryTaxonomyReader(indexDir);
ParallelTaxonomyArrays ca = tr.ParallelTaxonomyArrays;
Assert.AreEqual(3, tr.Size);
Assert.AreEqual(3, ca.Siblings().Length);
Assert.AreEqual(3, ca.Children().Length);
Assert.True(Arrays.Equals(new int[] { 1, 2, -1 }, ca.Children()));
Assert.True(Arrays.Equals(new int[] { -1, -1, -1 }, ca.Siblings()));
tw.AddCategory(new FacetLabel("hi", "ho"));
tw.AddCategory(new FacetLabel("hello"));
tw.Commit();
// Before refresh, nothing changed..
ParallelTaxonomyArrays newca = tr.ParallelTaxonomyArrays;
Assert.AreSame(newca, ca); // we got exactly the same object
Assert.AreEqual(3, tr.Size);
Assert.AreEqual(3, ca.Siblings().Length);
Assert.AreEqual(3, ca.Children().Length);
// After the refresh, things change:
var newtr = TaxonomyReader.OpenIfChanged(tr);
Assert.NotNull(newtr);
tr.Dispose();
tr = newtr;
ca = tr.ParallelTaxonomyArrays;
Assert.AreEqual(5, tr.Size);
Assert.AreEqual(5, ca.Siblings().Length);
Assert.AreEqual(5, ca.Children().Length);
Assert.True(Arrays.Equals(new int[] { 4, 3, -1, -1, -1 }, ca.Children()));
Assert.True(Arrays.Equals(new int[] { -1, -1, -1, 2, 1 }, ca.Siblings()));
tw.Dispose();
tr.Dispose();
indexDir.Dispose();
}
public virtual void TestChildrenArrays()
{
var indexDir = NewDirectory();
var tw = new DirectoryTaxonomyWriter(indexDir);
FillTaxonomy(tw);
tw.Dispose();
var tr = new DirectoryTaxonomyReader(indexDir);
ParallelTaxonomyArrays ca = tr.ParallelTaxonomyArrays;
int[] youngestChildArray = ca.Children();
Assert.AreEqual(tr.Size, youngestChildArray.Length);
int[] olderSiblingArray = ca.Siblings();
Assert.AreEqual(tr.Size, olderSiblingArray.Length);
for (int i = 0; i < ExpectedCategories.Length; i++)
{
// find expected children by looking at all expectedCategories
// for children
List<int?> expectedChildren = new List<int?>();
for (int j = ExpectedCategories.Length - 1; j >= 0; j--)
{
if (ExpectedCategories[j].Length != ExpectedCategories[i].Length + 1)
{
continue; // not longer by 1, so can't be a child
}
bool ischild = true;
for (int k = 0; k < ExpectedCategories[i].Length; k++)
{
if (!ExpectedCategories[j][k].Equals(ExpectedCategories[i][k]))
{
ischild = false;
break;
}
}
if (ischild)
{
expectedChildren.Add(j);
}
}
// check that children and expectedChildren are the same, with the
// correct reverse (youngest to oldest) order:
if (expectedChildren.Count == 0)
{
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, youngestChildArray[i]);
}
else
{
int child = youngestChildArray[i];
Assert.AreEqual((int)expectedChildren[0], child);
for (int j = 1; j < expectedChildren.Count; j++)
{
child = olderSiblingArray[child];
Assert.AreEqual((int)expectedChildren[j], child);
// if child is INVALID_ORDINAL we should stop, but
// AssertEquals would fail in this case anyway.
}
// When we're done comparing, olderSiblingArray should now point
// to INVALID_ORDINAL, saying there are no more children. If it
// doesn't, we found too many children...
Assert.AreEqual(-1, olderSiblingArray[child]);
}
}
tr.Dispose();
indexDir.Dispose();
}
private void AssertConsistentYoungestChild(FacetLabel abPath, int abOrd, int abYoungChildBase1, int abYoungChildBase2, int retry, int numCategories)
{
var indexDir = new SlowRAMDirectory(-1, null); // no slowness for intialization
var tw = new DirectoryTaxonomyWriter(indexDir);
tw.AddCategory(new FacetLabel("a", "0"));
tw.AddCategory(abPath);
tw.Commit();
var tr = new DirectoryTaxonomyReader(indexDir);
for (int i = 0; i < numCategories; i++)
{
var cp = new FacetLabel("a", "b", Convert.ToString(i));
tw.AddCategory(cp);
Assert.AreEqual(TaxonomyReader.INVALID_ORDINAL, tr.GetOrdinal(cp), "Ordinal of " + cp + " must be invalid until Taxonomy Reader was refreshed");
}
tw.Dispose();
var stop = new AtomicBoolean(false);
Exception[] error = new Exception[] { null };
int[] retrieval = new int[] { 0 };
var thread = new ThreadAnonymousInnerClassHelper(this, abPath, abOrd, abYoungChildBase1, abYoungChildBase2, retry, tr, stop, error, retrieval);
thread.Start();
indexDir.SleepMillis = 1; // some delay for refresh
var newTaxoReader = TaxonomyReader.OpenIfChanged(tr);
if (newTaxoReader != null)
{
newTaxoReader.Dispose();
}
stop.Set(true);
thread.Join();
Assert.Null(error[0], "Unexpcted exception at retry " + retry + " retrieval " + retrieval[0] + ": \n" + stackTraceStr(error[0]));
tr.Dispose();
}
public virtual void TestBasic()
{
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
// Writes facet ords to a separate directory from the
// main index:
DirectoryTaxonomyWriter taxoWriter = new DirectoryTaxonomyWriter(taxoDir, IndexWriterConfig.OpenMode_e.CREATE);
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir);
FacetsConfig config = new FacetsConfig();
// Reused across documents, to add the necessary facet
// fields:
Document doc = new Document();
doc.Add(new IntField("num", 10, Field.Store.NO));
doc.Add(new FacetField("Author", "Bob"));
writer.AddDocument(config.Build(taxoWriter, doc));
doc = new Document();
doc.Add(new IntField("num", 20, Field.Store.NO));
doc.Add(new FacetField("Author", "Lisa"));
writer.AddDocument(config.Build(taxoWriter, doc));
doc = new Document();
doc.Add(new IntField("num", 30, Field.Store.NO));
doc.Add(new FacetField("Author", "Lisa"));
writer.AddDocument(config.Build(taxoWriter, doc));
doc = new Document();
doc.Add(new IntField("num", 40, Field.Store.NO));
doc.Add(new FacetField("Author", "Susan"));
writer.AddDocument(config.Build(taxoWriter, doc));
doc = new Document();
doc.Add(new IntField("num", 45, Field.Store.NO));
doc.Add(new FacetField("Author", "Frank"));
writer.AddDocument(config.Build(taxoWriter, doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.Reader);
writer.Dispose();
// NRT open
var taxoReader = new DirectoryTaxonomyReader(taxoWriter);
taxoWriter.Dispose();
// Aggregate the facet counts:
FacetsCollector c = new FacetsCollector();
// MatchAllDocsQuery is for "browsing" (counts facets
// for all non-deleted docs in the index); normally
// you'd use a "normal" query and one of the
// Facets.search utility methods:
searcher.Search(new MatchAllDocsQuery(), c);
TaxonomyFacetSumValueSource facets = new TaxonomyFacetSumValueSource(taxoReader, new FacetsConfig(), c, new IntFieldSource("num"));
// Retrieve & verify results:
Assert.AreEqual("dim=Author path=[] value=145.0 childCount=4\n Lisa (50.0)\n Frank (45.0)\n Susan (40.0)\n Bob (10.0)\n", facets.GetTopChildren(10, "Author").ToString());
taxoReader.Dispose();
searcher.IndexReader.Dispose();
dir.Dispose();
taxoDir.Dispose();
}