Namespace Lucene.Net.Codecs.BlockTerms

Handles a terms dict, but decouples all details of doc/freqs/positions reading to an instance of PostingsReaderBase. This class is reusable for codecs that use a different format for docs/freqs/positions (though codecs are also free to make their own terms dict impl).

This class also interacts with an instance of TermsIndexReaderBase, to abstract away the specific implementation of the terms dict index.

@lucene.experimental

Writes terms dict, block-encoding (column stride) each term's metadata for each set of terms between two index terms.

@lucene.experimental

TermsIndexReaderBase for simple every Nth terms indexes.

@lucene.experimental

Selects every Nth term as and index term, and hold term bytes (mostly) fully expanded in memory. This terms index supports seeking by ord. See VariableGapTermsIndexWriter for a more memory efficient terms index that does not support seeking by ord.

@lucene.experimental

BlockTermsReader interacts with an instance of this class to manage its terms index. The writer must accept indexed terms (many pairs of BytesRef text + long fileOffset), and then this reader must be able to retrieve the nearest index term to a provided term text.

@lucene.experimental

Similar to TermsEnum, except, the only "metadata" it reports for a given indexed term is the long fileOffset into the main terms dictionary file.

Base class for terms index implementations to plug into BlockTermsWriter.

@lucene.experimental

Terms index API for a single field.

See VariableGapTermsIndexWriter.

@lucene.experimental

Selects index terms according to provided pluggable VariableGapTermsIndexWriter.IndexTermSelector, and stores them in a prefix trie that's loaded entirely in RAM stored as an FST<T>. This terms index only supports unsigned byte term sort order (unicode codepoint order when the bytes are UTF8).

@lucene.experimental

Sets an index term when docFreq >= docFreqThresh, or every interval terms. This should reduce seek time to high docFreq terms.

Hook for selecting which terms should be placed in the terms index.

NewField(FieldInfo) is called at the start of each new field, and IsIndexTerm(BytesRef, TermStats) for each term in that field.

@lucene.experimental