Struct Search 

pub struct Search { /* private fields */ }

A whole search request.

Implementations

impl Search

pub fn new() -> Search

Every document, ten at a time — what the cluster does with an empty body.

pub fn query_only(query: impl Into<Query>) -> Search

Start from a query, for the common case where nothing else is set.

pub fn query(self, query: impl Into<Query>) -> Search

pub fn from(self, from: u64) -> Search

How many hits to skip.

from + size may not exceed 10,000 without raising index.max_result_window, and paging deep with it gets slower on every page because every shard has to sort everything up to the offset. search_after is the answer past a few pages; this is right for the first few.

pub fn size(self, size: u64) -> Search

How many hits to return. Ten if not set.

Zero is legitimate and useful: an aggregation-only request pays for no hits at all.

pub fn sort_by(self, field: impl Into<String>, order: Order) -> Search

pub fn sort_asc(self, field: impl Into<String>) -> Search

pub fn sort_desc(self, field: impl Into<String>) -> Search

pub fn sort_by_score(self) -> Search

Sort by relevance, best first — the default when nothing is sorted.

Worth naming because adding any other sort silently replaces scoring rather than tie-breaking on it, so a sorted search is no longer ranked unless _score is put back in the list explicitly.

pub fn source<S: Into<String>>( self, fields: impl IntoIterator<Item = S>, ) -> Search

Return only these fields of the document.

pub fn source_excluding<S: Into<String>>( self, fields: impl IntoIterator<Item = S>, ) -> Search

Return everything except these fields.

pub fn without_source(self) -> Search

Return the ids and scores but no document bodies.

pub fn aggregate( self, name: impl Into<String>, aggregation: impl Into<Aggregation>, ) -> Search

pub fn track_total_hits(self, exact: bool) -> Search

Count every match rather than stopping at 10,000.

This is the one setting behind crate::TotalHits being a relation and not a number. Since Elasticsearch 7 the total stops being counted once it passes index.max_result_window, and the response then says “at least 10000” — which a caller that reads only value will render as “10000 results” on a page listing millions. Turning this on makes the number exact and makes the query slower; that trade is the caller’s to make, so it is never made here.

pub fn track_total_hits_up_to(self, limit: u64) -> Search

Count exactly up to a limit, then stop.

The middle ground: enough to say “more than 1,000” honestly without paying to count a million.

pub fn min_score(self, score: f64) -> Search

Drop hits scoring below this.

pub fn body(&self) -> Json

The JSON body this search sends.

pub fn count_body(&self) -> Json

The body _count accepts, which is the query and nothing else.

_count answers 400 for size, from, sort and aggs, so sending the search body would turn a perfectly good search into a failed count.

Trait Implementations

impl Clone for Search

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Search

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for Search

fn default() -> Self

Returns the “default value” for a type.

impl PartialEq for Search

fn eq(&self, other: &Self) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for Search