static TestTaxonomyFacetCounts2()
{
CATEGORIES_A = new FacetField[NUM_CHILDREN_CP_A];
for (int i = 0; i < NUM_CHILDREN_CP_A; i++)
{
CATEGORIES_A[i] = new FacetField(CP_A, Convert.ToString(i));
}
CATEGORIES_B = new FacetField[NUM_CHILDREN_CP_B];
for (int i = 0; i < NUM_CHILDREN_CP_B; i++)
{
CATEGORIES_B[i] = new FacetField(CP_B, Convert.ToString(i));
}
// NO_PARENTS categories
CATEGORIES_C = new FacetField[NUM_CHILDREN_CP_C];
for (int i = 0; i < NUM_CHILDREN_CP_C; i++)
{
CATEGORIES_C[i] = new FacetField(CP_C, Convert.ToString(i));
}
// Multi-level categories
CATEGORIES_D = new FacetField[NUM_CHILDREN_CP_D];
for (int i = 0; i < NUM_CHILDREN_CP_D; i++)
{
string val = Convert.ToString(i);
CATEGORIES_D[i] = new FacetField(CP_D, val, val + val); // e.g. D/1/11, D/2/22...
}
}
public override void Run()
{
Random random = Random;
while (numDocs.DecrementAndGet() > 0)
{
try
{
Document doc = new Document();
int numCats = random.Next(3) + 1; // 1-3
while (numCats-- > 0)
{
FacetField ff = NewCategory();
doc.Add(ff);
FacetLabel label = new FacetLabel(ff.Dim, ff.Path);
// add all prefixes to values
int level = label.Length;
while (level > 0)
{
string s = FacetsConfig.PathToString(label.Components, level);
values[s] = s;
--level;
}
}
iw.AddDocument(config.Build(tw, doc));
}
catch (IOException e)
{
throw new Exception(e.ToString(), e);
}
}
}
/// <summary>
/// Adds a new <see cref="FacetField"/> with the specified <paramref name="dim"/> and
/// <paramref name="path"/>.
/// </summary>
/// <param name="document">This <see cref="Document"/>.</param>
/// <param name="dim">Dimension for this field.</param>
/// <param name="path">Facet path for this field.</param>
/// <returns>The field that was added to this <see cref="Document"/>.</returns>
public static FacetField AddFacetField(this Document document, string dim, params string[] path)
{
var field = new FacetField(dim, path);
document.Add(field);
return(field);
}
static TestTaxonomyFacetCounts2()
{
CATEGORIES_A = new FacetField[NUM_CHILDREN_CP_A];
for (int i = 0; i < NUM_CHILDREN_CP_A; i++)
{
CATEGORIES_A[i] = new FacetField(CP_A, Convert.ToString(i));
}
CATEGORIES_B = new FacetField[NUM_CHILDREN_CP_B];
for (int i = 0; i < NUM_CHILDREN_CP_B; i++)
{
CATEGORIES_B[i] = new FacetField(CP_B, Convert.ToString(i));
}
// NO_PARENTS categories
CATEGORIES_C = new FacetField[NUM_CHILDREN_CP_C];
for (int i = 0; i < NUM_CHILDREN_CP_C; i++)
{
CATEGORIES_C[i] = new FacetField(CP_C, Convert.ToString(i));
}
// Multi-level categories
CATEGORIES_D = new FacetField[NUM_CHILDREN_CP_D];
for (int i = 0; i < NUM_CHILDREN_CP_D; i++)
{
string val = Convert.ToString(i);
CATEGORIES_D[i] = new FacetField(CP_D, val, val + val); // e.g. D/1/11, D/2/22...
}
}
private int maxValue; // = maxDocFacets * maxFacetDepth;
public override void GetNextFacets(IList <FacetField> facets)
{
facets.Clear();
int numFacets = 1 + random.Next(maxDocFacets); // at least one facet to each doc
for (int i = 0; i < numFacets; i++)
{
int depth;
if (maxFacetDepth == 2)
{
depth = 2;
}
else
{
depth = 2 + random.Next(maxFacetDepth - 2); // depth < 2 is not useful
}
string dim = random.Next(maxDims).ToString(CultureInfo.InvariantCulture);
string[] components = new string[depth - 1];
for (int k = 0; k < depth - 1; k++)
{
components[k] = random.Next(maxValue).ToString(CultureInfo.InvariantCulture);
AddItem();
}
FacetField ff = new FacetField(dim, components);
facets.Add(ff);
AddBytes(ff.ToString().Length); // very rough approximation
}
}
/// <summary>
/// Register <see cref="IFacetField"/> for use within query
/// </summary>
protected virtual IFacetQueryField FacetInternal(string field, string[] values = null)
{
var facet = new FacetField(field, values);
Fields.Add(facet);
return(new FacetQueryField(this, facet));
}
/// <summary>
/// Sole constructor. </summary>
public SortedSetDocValuesFacetField(string dim, string label)
: base("dummy", TYPE)
{
FacetField.VerifyLabel(label);
FacetField.VerifyLabel(dim);
this.Dim = dim;
this.Label = label;
}
public void TestAddFacetField()
{
FacetField field = null;
string[] path = new[] { "thePath0", "thePath1", "thePath2" };
AssertDocumentExtensionAddsToDocument(document => field = document.AddFacetField("theDim", path));
Assert.AreEqual("theDim", field.Dim);
Assert.AreEqual(path, field.Path);
}
///<inheritdoc/>
public IFacetQueryField Facet(string field, string[] values)
{
var facet = new FacetField(field)
{
Values = values
};
FacetInternal(facet);
return(new FacetQueryField(this, facet));
}
/// <summary>
/// Adds a new <see cref="FacetField"/> with the specified <paramref name="dim"/> and
/// <paramref name="path"/>.
/// </summary>
/// <param name="document">This <see cref="Document"/>.</param>
/// <param name="dim">Dimension for this field.</param>
/// <param name="path">Facet path for this field.</param>
/// <returns>The field that was added to this <see cref="Document"/>.</returns>
/// <exception cref="ArgumentNullException">This <paramref name="document"/> is <c>null</c>. </exception>
public static FacetField AddFacetField(this Document document, string dim, params string[] path)
{
if (document is null)
{
throw new ArgumentNullException(nameof(document));
}
var field = new FacetField(dim, path);
document.Add(field);
return(field);
}
public virtual void TestHugeLabel()
{
Directory indexDir = NewDirectory(), taxoDir = NewDirectory();
IndexWriter indexWriter = new IndexWriter(indexDir, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random())));
DirectoryTaxonomyWriter taxoWriter = new DirectoryTaxonomyWriter(taxoDir, OpenMode.CREATE, new Cl2oTaxonomyWriterCache(2, 1f, 1));
FacetsConfig config = new FacetsConfig();
// Add one huge label:
string bigs = null;
int ordinal = -1;
int len = FacetLabel.MAX_CATEGORY_PATH_LENGTH - 4; // for the dimension and separator
bigs = TestUtil.RandomSimpleString(Random(), len, len);
FacetField ff = new FacetField("dim", bigs);
FacetLabel cp = new FacetLabel("dim", bigs);
ordinal = taxoWriter.AddCategory(cp);
Document doc = new Document();
doc.Add(ff);
indexWriter.AddDocument(config.Build(taxoWriter, doc));
// Add tiny ones to cause a re-hash
for (int i = 0; i < 3; i++)
{
string s = TestUtil.RandomSimpleString(Random(), 1, 10);
taxoWriter.AddCategory(new FacetLabel("dim", s));
doc = new Document();
doc.Add(new FacetField("dim", s));
indexWriter.AddDocument(config.Build(taxoWriter, doc));
}
// when too large components were allowed to be added, this resulted in a new added category
Assert.AreEqual(ordinal, taxoWriter.AddCategory(cp));
IOUtils.Close(indexWriter, taxoWriter);
DirectoryReader indexReader = DirectoryReader.Open(indexDir);
var taxoReader = new DirectoryTaxonomyReader(taxoDir);
IndexSearcher searcher = new IndexSearcher(indexReader);
DrillDownQuery ddq = new DrillDownQuery(new FacetsConfig());
ddq.Add("dim", bigs);
Assert.AreEqual(1, searcher.Search(ddq, 10).TotalHits);
IOUtils.Close(indexReader, taxoReader, indexDir, taxoDir);
}
/// <summary>
/// Creates this from <paramref name="dim"/> and <paramref name="path"/> and an
/// association
/// </summary>
public AssociationFacetField(BytesRef assoc, string dim, params string[] path)
: base("dummy", TYPE)
{
FacetField.VerifyLabel(dim);
foreach (string label in path)
{
FacetField.VerifyLabel(label);
}
this.Dim = dim;
this.Assoc = assoc;
if (path.Length == 0)
{
throw new System.ArgumentException("path must have at least one element");
}
this.Path = path;
}
/// <summary>
/// Checks if the specified FacetField is configured; if not, FacetField is configured as hierarchical and multivalued.
/// </summary>
/// <param name="facetsConfig">The facets configuration.</param>
/// <param name="facetField">The facet field.</param>
public static void EnsureConfig(this FacetsConfig facetsConfig, FacetField facetField)
{
if (facetsConfig == null)
throw new ArgumentNullException(nameof(facetsConfig));
if (facetField == null)
throw new ArgumentNullException(nameof(facetField));
var facetName = facetField.Dim;
if (!facetsConfig.Contains(facetName))
{
// Configure the FacetField if not already configured.
// By default set it to hierarchical, and multi-valued.
lock (facetsConfig)
{
if (!facetsConfig.Contains(facetName))
{
facetsConfig.SetHierarchical(facetName, true);
facetsConfig.SetMultiValued(facetName, true);
}
}
}
}
public virtual void TestHugeLabel()
{
Directory indexDir = NewDirectory(), taxoDir = NewDirectory();
IndexWriter indexWriter = new IndexWriter(indexDir, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random())));
DirectoryTaxonomyWriter taxoWriter = new DirectoryTaxonomyWriter(taxoDir, OpenMode.CREATE, new Cl2oTaxonomyWriterCache(2, 1f, 1));
FacetsConfig config = new FacetsConfig();
// Add one huge label:
string bigs = null;
int ordinal = -1;
int len = FacetLabel.MAX_CATEGORY_PATH_LENGTH - 4; // for the dimension and separator
bigs = TestUtil.RandomSimpleString(Random(), len, len);
FacetField ff = new FacetField("dim", bigs);
FacetLabel cp = new FacetLabel("dim", bigs);
ordinal = taxoWriter.AddCategory(cp);
Document doc = new Document();
doc.Add(ff);
indexWriter.AddDocument(config.Build(taxoWriter, doc));
// Add tiny ones to cause a re-hash
for (int i = 0; i < 3; i++)
{
string s = TestUtil.RandomSimpleString(Random(), 1, 10);
taxoWriter.AddCategory(new FacetLabel("dim", s));
doc = new Document();
doc.Add(new FacetField("dim", s));
indexWriter.AddDocument(config.Build(taxoWriter, doc));
}
// when too large components were allowed to be added, this resulted in a new added category
Assert.AreEqual(ordinal, taxoWriter.AddCategory(cp));
IOUtils.Close(indexWriter, taxoWriter);
DirectoryReader indexReader = DirectoryReader.Open(indexDir);
var taxoReader = new DirectoryTaxonomyReader(taxoDir);
IndexSearcher searcher = new IndexSearcher(indexReader);
DrillDownQuery ddq = new DrillDownQuery(new FacetsConfig());
ddq.Add("dim", bigs);
Assert.AreEqual(1, searcher.Search(ddq, 10).TotalHits);
IOUtils.Close(indexReader, taxoReader, indexDir, taxoDir);
}
/// <summary>
/// Converts the specified category string to a Lucene FacetField.
/// </summary>
/// <param name="categoryString">The category.</param>
/// <returns></returns>
public static FacetField ToLuceneFacetField(this string categoryString)
{
const string COLON = ":";
const string ESCAPED_COLON = @"\:";
const string ESCAPED_COLON_TEMP_TOKEN = @"<<€€€€€€>>";
const string SLASH = @"/";
const string ESCAPED_SLASH = @"\/";
const string ESCAPED_SLASH_TEMP_TOKEN = @"<<$$$$$$>>";
// Sample categoryStrings:
// "Author:Arthur Dent" => facet name is "Author", value is "Arthur Dent"
// "Publish Date:2013/Mar/12" => facet name is "Publish Date", value (hierarchical) is "2013/Mar/12"
// A category string must have a name and value => {name}:{value}
if (String.IsNullOrWhiteSpace(categoryString))
throw new ArgumentNullException(nameof(categoryString));
// Convert \/ to <<$$$$$$>> and \: to <<€€€€€€>>
var categoryStringCopy = categoryString.Trim()
.Replace(ESCAPED_SLASH, ESCAPED_SLASH_TEMP_TOKEN)
.Replace(ESCAPED_COLON, ESCAPED_COLON_TEMP_TOKEN);
// Split the string using ":" to get the facet name
var facetName = categoryStringCopy.Contains(COLON) ?
categoryStringCopy.Split(new[] { COLON }, StringSplitOptions.None).FirstOrDefault() :
null;
if (String.IsNullOrWhiteSpace(facetName))
throw new SchemaException($"Invalid category string: '{categoryString}'");
facetName = facetName.Trim()
.Replace(ESCAPED_SLASH_TEMP_TOKEN, SLASH)
.Replace(ESCAPED_COLON, COLON);
var facetValueStartIndex = categoryStringCopy.IndexOf(COLON, StringComparison.InvariantCulture) + 1;
categoryStringCopy = categoryStringCopy.Substring(facetValueStartIndex);
// Split the string using / as separateor to get the facet values
var categoryParts = categoryStringCopy.Split(new[] { SLASH }, StringSplitOptions.None);
if (categoryParts.Count() >= 1)
{
var facetValues = categoryParts.Select(s => s.Trim()
.Replace(ESCAPED_SLASH_TEMP_TOKEN, SLASH)
.Replace(ESCAPED_COLON_TEMP_TOKEN, COLON))
.ToArray();
if (facetValues.Any(s => String.IsNullOrWhiteSpace(s)))
throw new SchemaException($"Invalid category string: '{categoryString}'");
var facetField = new FacetField(facetName, facetValues);
return facetField;
}
return null;
}
/// <summary>
/// Add documents.
/// </summary>
/// <param name="writer">The index writer.</param>
/// <param name="facetWriter">The facet index writer.</param>
/// <param name="directoryInfo">The directory information where all the files that are to be added are located.</param>
/// <param name="files">The list of files that are to be added.</param>
/// <param name="documents">The supported documents search filter, used to indicate what files are to be added.</param>
/// <param name="facetField">The facet field information.</param>
/// <param name="config">The facet configuration information.</param>
public void AddDocuments(Lucene.Net.Index.IndexWriter writer, DirectoryTaxonomyWriter facetWriter,
DirectoryInfo directoryInfo, string[] files, SupportedDocumentExtension documents, FacetField facetField, FacetsConfig config)
{
FieldType pathFieldType = new Lucene.Net.Documents.FieldType()
{
Indexed = true,
Tokenized = false,
Stored = true,
IndexOptions = FieldInfo.IndexOptions.DOCS_AND_FREQS_AND_POSITIONS,
};
FieldType contentFieldType = new Lucene.Net.Documents.FieldType()
{
Indexed = true,
Tokenized = documents.TokenizeContent,
Stored = documents.StoreContent,
IndexOptions = FieldInfo.IndexOptions.DOCS_AND_FREQS_AND_POSITIONS,
};
// For each file.
for (int i = 0; i < files.Length; i++)
{
// If the file exists
if (File.Exists(files[i]))
{
Lucene.Net.Documents.Document document = new Lucene.Net.Documents.Document();
try
{
FileInfo fileInfo = new FileInfo(files[i]);
string file = files[i].Replace(directoryInfo.Root.FullName, "").ToLower();
Lucene.Net.Documents.Field path = new Field("path", file.ToLower().Replace("\\", "/"), pathFieldType);
Lucene.Net.Documents.Field modified = new Field("modified", fileInfo.LastWriteTime.ToShortDateString() + " " + fileInfo.LastWriteTime.ToShortTimeString(), pathFieldType);
// Add the fields.
document.Add(facetField);
document.Add(path);
document.Add(modified);
// Create the stream reader.
OpenDocument(files[i]);
string content = Nequeo.Xml.Document.ExtractContent(_xDocument);
// If content exists.
if (!String.IsNullOrEmpty(content))
{
// Split the white spaces from the text.
string[] words = content.Words();
// If words exist.
if (words != null && words.Length > 0)
{
// Add the query for each word.
for (int j = 0; j < words.Length; j++)
{
// Format the word.
string word = words[j].ToLower().RemovePunctuationFromStartAndEnd();
// If a word exists.
if (!String.IsNullOrEmpty(word))
{
Lucene.Net.Documents.Field contentField = new Field("facetcontent", word, contentFieldType);
document.Add(contentField);
}
}
}
}
// Add the document.
writer.AddDocument(config.Build(facetWriter, document));
_document.Close();
// Commit after a set number of documents.
documents.TotalDocumentSize += fileInfo.Length;
if (documents.TotalDocumentSize > documents.MaxDocumentSizePerCommit)
{
// Commit the index.
writer.Commit();
facetWriter.Commit();
documents.TotalDocumentSize = 0;
}
}
catch (Exception)
{
throw;
}
finally
{
CloseDocument();
}
}
}
}
public virtual void TestEmptyNullComponents()
{
// LUCENE-4724: CategoryPath should not allow empty or null components
string[][] components_tests = new string[][] {
new string[] { "", "test" }, // empty in the beginning
new string[] { "test", "" }, // empty in the end
new string[] { "test", "", "foo" }, // empty in the middle
new string[] { null, "test" }, // null at the beginning
new string[] { "test", null }, // null in the end
new string[] { "test", null, "foo" } // null in the middle
};
foreach (string[] components in components_tests)
{
try
{
Assert.IsNotNull(new FacetLabel(components));
fail("empty or null components should not be allowed: " + Arrays.ToString(components));
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new FacetField("dim", components);
fail("empty or null components should not be allowed: " + Arrays.ToString(components));
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new AssociationFacetField(new BytesRef(), "dim", components);
fail("empty or null components should not be allowed: " + Arrays.ToString(components));
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new Int32AssociationFacetField(17, "dim", components);
fail("empty or null components should not be allowed: " + Arrays.ToString(components));
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new SingleAssociationFacetField(17.0f, "dim", components);
fail("empty or null components should not be allowed: " + Arrays.ToString(components));
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
}
try
{
_ = new FacetField(null, new string[] { "abc" });
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new FacetField("", new string[] { "abc" });
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new Int32AssociationFacetField(17, null, new string[] { "abc" });
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new Int32AssociationFacetField(17, "", new string[] { "abc" });
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new SingleAssociationFacetField(17.0f, null, new string[] { "abc" });
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new SingleAssociationFacetField(17.0f, "", new string[] { "abc" });
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new AssociationFacetField(new BytesRef(), null, new string[] { "abc" });
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new AssociationFacetField(new BytesRef(), "", new string[] { "abc" });
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new SortedSetDocValuesFacetField(null, "abc");
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new SortedSetDocValuesFacetField("", "abc");
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new SortedSetDocValuesFacetField("dim", null);
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
try
{
_ = new SortedSetDocValuesFacetField("dim", "");
fail("empty or null components should not be allowed");
}
catch (Exception e) when(e.IsIllegalArgumentException())
{
// expected
}
}
