public virtual void TestSimple()
{
Directory dest = NewDirectory();
var tw1 = new DirectoryTaxonomyWriter(dest);
tw1.AddCategory(new FacetLabel("Author", "Mark Twain"));
tw1.AddCategory(new FacetLabel("Animals", "Dog"));
tw1.AddCategory(new FacetLabel("Author", "Rob Pike"));
Directory src = NewDirectory();
var tw2 = new DirectoryTaxonomyWriter(src);
tw2.AddCategory(new FacetLabel("Author", "Rob Pike"));
tw2.AddCategory(new FacetLabel("Aardvarks", "Bob"));
tw2.Dispose();
IOrdinalMap map = randomOrdinalMap();
tw1.AddTaxonomy(src, map);
tw1.Dispose();
validate(dest, src, map);
IOUtils.Dispose(dest, src);
}
private void validate(Directory dest, Directory src, IOrdinalMap ordMap)
{
using (var destTr = new DirectoryTaxonomyReader(dest))
{
int destSize = destTr.Count;
using (var srcTR = new DirectoryTaxonomyReader(src))
{
var map = ordMap.GetMap();
// validate taxo sizes
int srcSize = srcTR.Count;
Assert.True(destSize >= srcSize, "destination taxonomy expected to be larger than source; dest=" + destSize + " src=" + srcSize);
// validate that all source categories exist in destination, and their
// ordinals are as expected.
for (int j = 1; j < srcSize; j++)
{
FacetLabel cp = srcTR.GetPath(j);
int destOrdinal = destTr.GetOrdinal(cp);
Assert.True(destOrdinal > 0, cp + " not found in destination");
Assert.AreEqual(destOrdinal, map[j]);
}
}
}
}
/// <summary>
/// Takes the categories from the given taxonomy directory, and adds the
/// missing ones to this taxonomy. Additionally, it fills the given
/// <see cref="IOrdinalMap"/> with a mapping from the original ordinal to the new
/// ordinal.
/// </summary>
public virtual void AddTaxonomy(Directory taxoDir, IOrdinalMap map)
{
EnsureOpen();
DirectoryReader r = DirectoryReader.Open(taxoDir);
try
{
int size = r.NumDocs;
IOrdinalMap ordinalMap = map;
ordinalMap.SetSize(size);
int @base = 0;
TermsEnum te = null;
DocsEnum docs = null;
foreach (AtomicReaderContext ctx in r.Leaves)
{
AtomicReader ar = ctx.AtomicReader;
Terms terms = ar.GetTerms(Consts.FULL);
te = terms.GetIterator(te);
while (te.Next() != null)
{
FacetLabel cp = new FacetLabel(FacetsConfig.StringToPath(te.Term.Utf8ToString()));
int ordinal = AddCategory(cp);
docs = te.Docs(null, docs, DocsFlags.NONE);
ordinalMap.AddMapping(docs.NextDoc() + @base, ordinal);
}
@base += ar.MaxDoc; // no deletions, so we're ok
}
ordinalMap.AddDone();
}
finally
{
r.Dispose();
}
}
public virtual void TestAddToEmpty()
{
Directory dest = NewDirectory();
Directory src = NewDirectory();
DirectoryTaxonomyWriter srcTW = new DirectoryTaxonomyWriter(src);
srcTW.AddCategory(new FacetLabel("Author", "Rob Pike"));
srcTW.AddCategory(new FacetLabel("Aardvarks", "Bob"));
srcTW.Dispose();
DirectoryTaxonomyWriter destTW = new DirectoryTaxonomyWriter(dest);
IOrdinalMap map = randomOrdinalMap();
destTW.AddTaxonomy(src, map);
destTW.Dispose();
validate(dest, src, map);
IOUtils.Dispose(dest, src);
}
private void Dotest(int ncats, int range)
{
AtomicInt32 numCats = new AtomicInt32(ncats);
Directory[] dirs = new Directory[2];
for (int i = 0; i < dirs.Length; i++)
{
dirs[i] = NewDirectory();
var tw = new DirectoryTaxonomyWriter(dirs[i]);
ThreadJob[] addThreads = new ThreadJob[4];
for (int j = 0; j < addThreads.Length; j++)
{
addThreads[j] = new ThreadAnonymousInnerClassHelper(this, range, numCats, tw);
}
foreach (ThreadJob t in addThreads)
{
t.Start();
}
foreach (ThreadJob t in addThreads)
{
t.Join();
}
tw.Dispose();
}
var tw1 = new DirectoryTaxonomyWriter(dirs[0]);
IOrdinalMap map = randomOrdinalMap();
tw1.AddTaxonomy(dirs[1], map);
tw1.Dispose();
validate(dirs[0], dirs[1], map);
IOUtils.Dispose(dirs);
}
private void validate(Directory dest, Directory src, IOrdinalMap ordMap)
{
var destTr = new DirectoryTaxonomyReader(dest);
try
{
int destSize = destTr.Count;
var srcTR = new DirectoryTaxonomyReader(src);
try
{
var map = ordMap.Map;
// validate taxo sizes
int srcSize = srcTR.Count;
Assert.True(destSize >= srcSize, "destination taxonomy expected to be larger than source; dest=" + destSize + " src=" + srcSize);
// validate that all source categories exist in destination, and their
// ordinals are as expected.
for (int j = 1; j < srcSize; j++)
{
FacetLabel cp = srcTR.GetPath(j);
int destOrdinal = destTr.GetOrdinal(cp);
Assert.True(destOrdinal > 0, cp + " not found in destination");
Assert.AreEqual(destOrdinal, map[j]);
}
}
finally
{
((TaxonomyReader)srcTR).Dispose(true);
}
}
finally
{
((TaxonomyReader)destTr).Dispose(true);
}
}