Struct TextIndex 

pub struct TextIndex { /* private fields */ }

A parsed text index: the token table (always resident) plus a posting source.

Implementations

impl TextIndex

pub fn from_section(section: &[u8], codec: u8) -> Result<TextIndex, FileError>

Parse a whole section (local): token table + the full postings blob resident.

pub fn from_token_table( prefix: &[u8], codec: u8, loader: Box<dyn Fn(u64, u64) -> Option<Vec<u8>> + Send + Sync>, ) -> Result<TextIndex, FileError>

Build from a section prefix holding the token table, plus a loader that fetches a posting (offset_within_postings_blob, len) on demand (remote).

pub fn postings_base(section_prefix: &[u8]) -> Option<usize>

Byte offset of the postings blob within the section (token-table end). The remote opener uses this to base its posting-range loader.

pub fn token_count(&self) -> usize

pub fn lookup(&self, token: &str) -> Vec<u32>

Subjects whose literals contain the exact token (case-insensitive — the caller passes a tokenized word), sorted ascending.

pub fn substring(&self, piece: &str) -> Option<Vec<u32>>

Subjects whose literals contain a word that CONTAINS piece as a substring — the union over every matching table token. The token table is always resident so the scan is in-memory; each matching token costs one posting read (a range fetch on the remote path), so an unselective piece matching more than [SUBSTRING_TOKEN_CAP] tokens returns None and the caller falls back to its non-indexed path.

pub fn prefix(&self, prefix: &str) -> Vec<u32>

Subjects whose literals contain a word starting with prefix — the union over every token in the prefix… range. Sorted, deduped.