Module fts 

Full-text indexing and retrieval. Full-text indexing and retrieval.

The lexical half of the engine. Task 12 builds the versioned analysis pipeline whose identity is index metadata; tasks 13 through 15 build the postings, the scorer, dynamic pruning, and the precision surface on top of it.

Modules

bm25

The BM25 scorer and corpus statistics. The BM25 scorer.

dict

The sorted, front-coded term dictionary. The sorted, front-coded term dictionary.

fuzzy

Bounded fuzzy matching. Bounded fuzzy matching by dictionary-guided Levenshtein pruning.

index

The lexical index across segments. The lexical index: per-segment postings and store-wide statistics.

norms

Byte-quantized document length norms. Byte-quantized document length norms.

phonetic

Double Metaphone phonetic encoding. Double Metaphone encoding for the opt-in phonetic side-field.

phrase

Phrase matching over stored positions. Phrase matching over the positions task 13 stored.

postings

Persisted posting blocks with positions. The persisted posting format: bit-packed FOR blocks with positions.

prefix

Prefix queries over the sorted dictionary. Prefix queries as a sorted-dictionary range scan.

prune

Dynamic pruning: block-max MAXSCORE and WAND. Dynamic pruning: block-max MAXSCORE and WAND, counter-gated.

query

Structured query descriptions and deterministic expansion provenance. Structured lexical queries and their deterministic expansion plan.

sealed

The sealed segment: the persisted posting format on the query path. The sealed lexical segment: the persisted format, on the query path.

search

The exhaustive-OR scorer and task 14 oracle. The exhaustive-OR scorer: task 13’s deliverable and task 14’s oracle.

snippet

Snippets and match offsets. Snippets and match offsets: how a hit explains itself.

tokenizer

The versioned analysis pipeline and its epoch identity.