Struct MatchIndexer 

pub struct MatchIndexer { /* private fields */ }

Implementations

impl MatchIndexer

pub fn new(_: MatchIndexerOptions) -> Self

pub fn validate_query_input(&self, value: &str) -> Result<(), EncryptionError>

Reject a match QUERY input that would tokenize to nothing.

A zero-token input yields an empty bloom filter, and the EQL match comparison (query bits ⊆ row bits) is vacuously true for an empty query — so the term silently matches every row. That’s never what a query means, so callers minting match query terms must reject the input instead. The most common trigger is an input shorter than an Ngram tokenizer’s token_length (default 3), but this also catches e.g. a Stop token filter swallowing every token.

Storage-side indexing must NOT use this: storing a short string is legal (the value simply isn’t findable via the match index).

The error message never contains the input itself — only its length after the char filter, and the index’s minimum.

pub fn encrypt( &self, plaintext: &Plaintext, index_key: &IndexKey, ) -> Result<IndexTerm, EncryptionError>

Trait Implementations

impl Default for MatchIndexer

fn default() -> Self

Returns the “default value” for a type.

impl IndexerInit for MatchIndexer

type Args = MatchIndexerOptions

type Error = EncryptionError

fn try_init<A>(opts: A) -> Result<Self, Self::Error>

where Self::Args: TryFrom<A, Error = Self::Error>,