private void ProcessFacetFields(TaxonomyWriter taxoWriter, IDictionary<string, IList<FacetField>> byField, Document doc)
{
foreach (KeyValuePair<string, IList<FacetField>> ent in byField)
{
string indexFieldName = ent.Key;
//System.out.println(" indexFieldName=" + indexFieldName + " fields=" + ent.getValue());
IntsRef ordinals = new IntsRef(32);
foreach (FacetField facetField in ent.Value)
{
FacetsConfig.DimConfig ft = GetDimConfig(facetField.dim);
if (facetField.path.Length > 1 && ft.Hierarchical == false)
{
throw new System.ArgumentException("dimension \"" + facetField.dim + "\" is not hierarchical yet has " + facetField.path.Length + " components");
}
FacetLabel cp = new FacetLabel(facetField.dim, facetField.path);
checkTaxoWriter(taxoWriter);
int ordinal = taxoWriter.AddCategory(cp);
if (ordinals.Length == ordinals.Ints.Length)
{
ordinals.Grow(ordinals.Length + 1);
}
ordinals.Ints[ordinals.Length++] = ordinal;
//System.out.println("ords[" + (ordinals.length-1) + "]=" + ordinal);
//System.out.println(" add cp=" + cp);
if (ft.MultiValued && (ft.Hierarchical || ft.RequireDimCount))
{
//System.out.println(" add parents");
// Add all parents too:
int parent = taxoWriter.GetParent(ordinal);
while (parent > 0)
{
if (ordinals.Ints.Length == ordinals.Length)
{
ordinals.Grow(ordinals.Length + 1);
}
ordinals.Ints[ordinals.Length++] = parent;
parent = taxoWriter.GetParent(parent);
}
if (ft.RequireDimCount == false)
{
// Remove last (dimension) ord:
ordinals.Length--;
}
}
// Drill down:
for (int i = 1; i <= cp.Length; i++)
{
doc.Add(new StringField(indexFieldName, PathToString(cp.Components, i), Field.Store.NO));
}
}
// Facet counts:
// DocValues are considered stored fields:
doc.Add(new BinaryDocValuesField(indexFieldName, DedupAndEncode(ordinals)));
}
}
private void checkTaxoWriter(TaxonomyWriter taxoWriter)
{
if (taxoWriter == null)
{
throw new ThreadStateException("a non-null TaxonomyWriter must be provided when indexing FacetField or AssociationFacetField");
}
}
private void ProcessAssocFacetFields(TaxonomyWriter taxoWriter, IDictionary<string, IList<AssociationFacetField>> byField, Document doc)
{
foreach (KeyValuePair<string, IList<AssociationFacetField>> ent in byField)
{
byte[] bytes = new byte[16];
int upto = 0;
string indexFieldName = ent.Key;
foreach (AssociationFacetField field in ent.Value)
{
// NOTE: we don't add parents for associations
checkTaxoWriter(taxoWriter);
FacetLabel label = new FacetLabel(field.dim, field.path);
int ordinal = taxoWriter.AddCategory(label);
if (upto + 4 > bytes.Length)
{
bytes = ArrayUtil.Grow(bytes, upto + 4);
}
// big-endian:
bytes[upto++] = (byte)(ordinal >> 24);
bytes[upto++] = (byte)(ordinal >> 16);
bytes[upto++] = (byte)(ordinal >> 8);
bytes[upto++] = (byte)ordinal;
if (upto + field.assoc.Length > bytes.Length)
{
bytes = ArrayUtil.Grow(bytes, upto + field.assoc.Length);
}
Array.Copy(field.assoc.Bytes, field.assoc.Offset, bytes, upto, field.assoc.Length);
upto += field.assoc.Length;
// Drill down:
for (int i = 1; i <= label.Length; i++)
{
doc.Add(new StringField(indexFieldName, PathToString(label.Components, i), Field.Store.NO));
}
}
doc.Add(new BinaryDocValuesField(indexFieldName, new BytesRef(bytes, 0, upto)));
}
}
/// <summary>
/// Translates any added <seealso cref="FacetField"/>s into normal fields for indexing.
///
/// <para>
/// <b>NOTE:</b> you should add the returned document to IndexWriter, not the
/// input one!
/// </para>
/// </summary>
public virtual Document Build(TaxonomyWriter taxoWriter, Document doc)
{
// Find all FacetFields, collated by the actual field:
IDictionary<string, IList<FacetField>> byField = new Dictionary<string, IList<FacetField>>();
// ... and also all SortedSetDocValuesFacetFields:
IDictionary<string, IList<SortedSetDocValuesFacetField>> dvByField = new Dictionary<string, IList<SortedSetDocValuesFacetField>>();
// ... and also all AssociationFacetFields
IDictionary<string, IList<AssociationFacetField>> assocByField = new Dictionary<string, IList<AssociationFacetField>>();
var seenDims = new HashSet<string>();
foreach (IndexableField field in doc.Fields)
{
if (field.FieldType() == FacetField.TYPE)
{
FacetField facetField = (FacetField)field;
FacetsConfig.DimConfig dimConfig = GetDimConfig(facetField.dim);
if (dimConfig.MultiValued == false)
{
CheckSeen(seenDims, facetField.dim);
}
string indexFieldName = dimConfig.IndexFieldName;
IList<FacetField> fields;
if (!byField.TryGetValue(indexFieldName, out fields))
{
fields = new List<FacetField>();
byField[indexFieldName] = fields;
}
fields.Add(facetField);
}
if (field.FieldType() == SortedSetDocValuesFacetField.TYPE)
{
var facetField = (SortedSetDocValuesFacetField)field;
FacetsConfig.DimConfig dimConfig = GetDimConfig(facetField.Dim);
if (dimConfig.MultiValued == false)
{
CheckSeen(seenDims, facetField.Dim);
}
string indexFieldName = dimConfig.IndexFieldName;
IList<SortedSetDocValuesFacetField> fields;
if (!dvByField.TryGetValue(indexFieldName, out fields))
{
fields = new List<SortedSetDocValuesFacetField>();
dvByField[indexFieldName] = fields;
}
fields.Add(facetField);
}
if (field.FieldType() == AssociationFacetField.TYPE)
{
AssociationFacetField facetField = (AssociationFacetField)field;
FacetsConfig.DimConfig dimConfig = GetDimConfig(facetField.dim);
if (dimConfig.MultiValued == false)
{
CheckSeen(seenDims, facetField.dim);
}
if (dimConfig.Hierarchical)
{
throw new System.ArgumentException("AssociationFacetField cannot be hierarchical (dim=\"" + facetField.dim + "\")");
}
if (dimConfig.RequireDimCount)
{
throw new System.ArgumentException("AssociationFacetField cannot requireDimCount (dim=\"" + facetField.dim + "\")");
}
string indexFieldName = dimConfig.IndexFieldName;
IList<AssociationFacetField> fields;
if (!assocByField.TryGetValue(indexFieldName, out fields))
{
fields = new List<AssociationFacetField>();
assocByField[indexFieldName] = fields;
}
fields.Add(facetField);
// Best effort: detect mis-matched types in same
// indexed field:
string type;
if (facetField is IntAssociationFacetField)
{
type = "int";
}
else if (facetField is FloatAssociationFacetField)
{
type = "float";
}
else
{
type = "bytes";
}
// NOTE: not thread safe, but this is just best effort:
string curType;
if (!assocDimTypes.TryGetValue(indexFieldName, out curType))
{
assocDimTypes[indexFieldName] = type;
}
else if (!curType.Equals(type))
{
throw new System.ArgumentException("mixing incompatible types of AssocationFacetField (" + curType + " and " + type + ") in indexed field \"" + indexFieldName + "\"; use FacetsConfig to change the indexFieldName for each dimension");
}
}
}
Document result = new Document();
ProcessFacetFields(taxoWriter, byField, result);
processSSDVFacetFields(dvByField, result);
ProcessAssocFacetFields(taxoWriter, assocByField, result);
//System.out.println("add stored: " + addedStoredFields);
foreach (IndexableField field in doc.Fields)
{
IndexableFieldType ft = field.FieldType();
if (ft != FacetField.TYPE && ft != SortedSetDocValuesFacetField.TYPE && ft != AssociationFacetField.TYPE)
{
result.Add(field);
}
}
return result;
}
