Enum QueryExpr 

pub enum QueryExpr {
    Points {
        ids: Vec<PointId>,
    },
    Nearest {
        input: QueryInput,
        using: Option<VectorTarget>,
        prefetch: Vec<Prefetch>,
        mmr: Option<Box<MmrConfig>>,
    },
    Recommend {
        positive: Vec<QueryInput>,
        negative: Vec<QueryInput>,
        strategy: Option<RecommendStrategy>,
        using: Option<VectorTarget>,
        prefetch: Vec<Prefetch>,
    },
    Context {
        pairs: Vec<ContextPair>,
        using: Option<VectorTarget>,
        prefetch: Vec<Prefetch>,
    },
    Discover {
        target: QueryInput,
        context: Vec<ContextPair>,
        using: Option<VectorTarget>,
        prefetch: Vec<Prefetch>,
    },
    OrderBy {
        field: String,
        direction: OrderDirection,
        start_from: Option<Value>,
    },
    SampleRandom,
    Fusion {
        method: FusionMethod,
        prefetch: Vec<Prefetch>,
    },
    Formula {
        expression: Box<FormulaExpr>,
        defaults: Vec<(String, Value)>,
        prefetch: Vec<Prefetch>,
    },
    RelevanceFeedback {
        target: QueryInput,
        feedback: Vec<FeedbackItem>,
        strategy: FeedbackStrategy,
        using: Option<VectorTarget>,
        prefetch: Vec<Prefetch>,
    },
    Hybrid {
        text: String,
        model: Option<String>,
        dense_vector: Option<String>,
        sparse_vector: Option<String>,
        fusion: FusionMethod,
        text_param: Option<String>,
    },
    Rerank {
        input: QueryInput,
        model: String,
        using: Option<VectorTarget>,
        prefetch: Vec<Prefetch>,
    },
    CrossRerank {
        query: String,
        model: String,
        field: Option<String>,
        prefetch: Vec<Prefetch>,
        query_param: Option<String>,
    },
}

QUERY expression body — the retrieval strategy and its inputs.

Variants

Points

QUERY POINTS (ids) — direct retrieval of the listed points.

Fields

ids: Vec<PointId>

Point IDs to fetch.

Nearest

QUERY [NEAREST] <input> FROM <coll> — vector nearest-neighbor search.

Fields

input: QueryInput

Embeddable input, vector, or reference point.

using: Option<VectorTarget>

USING vector target; schema-resolved when None.

prefetch: Vec<Prefetch>

Multi-stage PREFETCH pipeline.

mmr: Option<Box<MmrConfig>>

MMR re-diversification settings.

Recommend

QUERY RECOMMEND POSITIVE … [NEGATIVE …] — recommend from examples.

Fields

positive: Vec<QueryInput>

Examples to move toward.

negative: Vec<QueryInput>

Examples to move away from.

strategy: Option<RecommendStrategy>

Scoring strategy; server default when None.

using: Option<VectorTarget>

USING vector target; schema-resolved when None.

prefetch: Vec<Prefetch>

Multi-stage PREFETCH pipeline.

Context

QUERY CONTEXT (POSITIVE … NEGATIVE …) — search guided by example pairs.

Fields

pairs: Vec<ContextPair>

Positive/negative example pairs.

using: Option<VectorTarget>

USING vector target; schema-resolved when None.

prefetch: Vec<Prefetch>

Multi-stage PREFETCH pipeline.

Discover

QUERY DISCOVER TARGET … CONTEXT (…) — discovery from target plus pairs.

Fields

target: QueryInput

Primary target input.

context: Vec<ContextPair>

Guiding positive/negative pairs.

using: Option<VectorTarget>

USING vector target; schema-resolved when None.

prefetch: Vec<Prefetch>

Multi-stage PREFETCH pipeline.

OrderBy

QUERY ORDER BY field [ASC|DESC] [START FROM <value>] — payload-value ordering.

Fields

field: String

Payload field to sort on.

direction: OrderDirection

Sort direction (ASC default).

start_from: Option<Value>

Optional paging origin: resume ordering from this payload value (OpenAPI OrderBy.start_from: integer, float, or datetime string).

SampleRandom

QUERY SAMPLE RANDOM — random sample of points.

Fusion

QUERY FUSION RRF|DBSF — fuse results of the prefetch stages.

Fields

method: FusionMethod

Fusion algorithm.

prefetch: Vec<Prefetch>

Stages whose results are fused (must be non-empty).

Formula

QUERY FORMULA <expr> [DEFAULTS (…)] — formula-expression rescoring.

Fields

expression: Box<FormulaExpr>

Rescoring expression tree.

defaults: Vec<(String, Value)>

DEFAULTS bindings for formula variables.

prefetch: Vec<Prefetch>

Candidate stages the formula rescoring applies to.

RelevanceFeedback

QUERY RELEVANCE FEEDBACK TARGET … FEEDBACK (…) — naive feedback search.

Fields

target: QueryInput

Base target input.

feedback: Vec<FeedbackItem>

Weighted feedback examples.

strategy: FeedbackStrategy

STRATEGY NAIVE (a = …, b = …, c = …) weights.

using: Option<VectorTarget>

USING vector target; schema-resolved when None.

prefetch: Vec<Prefetch>

Multi-stage PREFETCH pipeline.

Hybrid

QUERY HYBRID TEXT … [DENSE n] [SPARSE n] [FUSION m] — dense+sparse fusion.

Fields

text: String

Query text embedded for both stages.

model: Option<String>

Optional embedding model override.

dense_vector: Option<String>

Named dense vector; schema-resolved when None.

sparse_vector: Option<String>

Named sparse vector; schema-resolved when None.

fusion: FusionMethod

Fusion method for the two stages.

text_param: Option<String>

Parameter placeholder (:name or ?idx) when the text was not provided as a literal string.

Rerank

QUERY RERANK <input> MODEL '…' — late-interaction rerank over prefetch.

Fields

input: QueryInput

Query input embedded via USING.

model: String

ColBERT-style late-interaction model.

using: Option<VectorTarget>

Dense (or multivector) target for document embeddings.

prefetch: Vec<Prefetch>

Multi-stage PREFETCH pipeline.

CrossRerank

Cross-encoder pair rerank: score query against PREFETCH document texts. Not sent to Qdrant as MaxSim — executor scores client-side then reorders.

Fields

query: String

Query string scored against each document.

model: String

Cross-encoder model id (e.g. bge-reranker-base).

field: Option<String>

Payload field holding document text (default "text" at resolve time).

prefetch: Vec<Prefetch>

Candidate stages whose documents are reranked.

query_param: Option<String>

Parameter placeholder (:name or ?idx) when the query was not provided as a literal string.

Trait Implementations

impl Clone for QueryExpr

fn clone(&self) -> QueryExpr

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for QueryExpr

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

Formats the value using the given formatter.

impl PartialEq for QueryExpr

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

Equality operator ==.

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

Inequality operator !=.

impl StructuralPartialEq for QueryExpr