Lucene90BlockTreeTermsWriter

Block-based terms index and dictionary writer.

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

Files:

• .tim: #Termdictionary

• .tmd: #Termmetadata

• .tip: #Termindex

Term Dictionary

The .tim file contains the list of terms in each field along with per-term statistics (such as docfreq) and per-term metadata (typically pointers to the postings list for that term in the inverted index).

The .tim is arranged in blocks: with blocks containing a variable number of entries (by default 25-48), where each entry is either a term or a reference to a sub-block.

NOTE: The term dictionary can plug into different postings implementations: the postings writer/reader are actually responsible for encoding and decoding the Postings Metadata and Term Metadata sections.

• TermsDict (.tim) --> Header, FieldDict^NumFields, Footer

• FieldDict -->PostingsHeader, NodeBlock^NumBlocks

• NodeBlock --> (OuterNode | InnerNode)

• OuterNode --> EntryCount, SuffixLength, Byte^SuffixLength, StatsLength, < TermStats >^EntryCount, MetaLength, <TermMetadata>^EntryCount

• InnerNode --> EntryCount, SuffixLength,Sub, Byte^SuffixLength, StatsLength, < TermStats >^EntryCount, MetaLength, <*TermMetadata

• >^EntryCount

• TermStats --> DocFreq, TotalTermFreq

• Header -->CodecHeader

• EntryCount,SuffixLength,StatsLength,DocFreq,MetaLength --> VInt

• TotalTermFreq -->VLong

• Footer -->CodecFooter

Notes:

• Header is a CodecHeader storing the version information for the BlockTree implementation.

• DocFreq is the count of documents which contain the term.

• TotalTermFreq is the total number of occurrences of the term. This is encoded as the difference between the total number of occurrences and the DocFreq.

• PostingsHeader and TermMetadata are plugged into by the specific postings implementation: these contain arbitrary per-file data (such as parameters or versioning information) and per-term data (such as pointers to inverted files).

• For inner nodes of the tree, every entry will steal one bit to mark whether it points to child nodes(sub-block). If so, the corresponding TermStats and TermMetaData are omitted.

Term Metadata

The .tmd file contains the list of term metadata (such as FST index metadata) and field level statistics (such as sum of total term freq).

• TermsMeta (.tmd) --> Header, NumFields, ^NumFields, TermIndexLength, TermDictLength, Footer

• FieldStats --> FieldNumber, NumTerms, RootCodeLength, Byte^{RootCodeLength}, SumTotalTermFreq, SumDocFreq, DocCount, MinTerm, MaxTerm, IndexStartFP, FSTHeader, FSTMetadata

• Header,FSTHeader -->CodecHeader

• TermIndexLength, TermDictLength -->Uint64

• MinTerm,MaxTerm -->VInt length followed by the byte[]

• NumFields,FieldNumber,RootCodeLength,DocCount -->VInt

• NumTerms,SumTotalTermFreq,SumDocFreq,IndexStartFP --> VLong

• Footer -->CodecFooter

Notes:

• FieldNumber is the fields number from FieldInfos. (.fnm)

• NumTerms is the number of unique terms for the field.

• RootCode points to the root block for the field.

• SumDocFreq is the total number of postings, the number of term-document pairs across the entire field.

• DocCount is the number of documents that have at least one posting for this field.

• MinTerm, MaxTerm are the lowest and highest term in this field.

Term Index

The .tip file contains an index into the term dictionary, so that it can be accessed randomly. The index is also used to determine when a given term cannot exist on disk (in the .tim file), saving a disk seek.

• TermsIndex (.tip) --> Header, FSTIndex^NumFieldsFooter

• Header -->CodecHeader

• FSTIndex -->FST<byte[]>

• Footer -->CodecFooter

Notes:

• The .tip file contains a separate FST for each field. The FST maps a term prefix to the on-disk block that holds all terms starting with that prefix. Each field's IndexStartFP points to its FST.

• It's possible that an on-disk block would contain too many terms (more than the allowed maximum (default: 48)). When this happens, the block is sub-divided into new blocks (called "floor blocks"), and then the output in the FST for the block's prefix encodes the leading byte of each sub-block, and its file pointer.

See also