Merge pull request #25 from qdrant/update-1.1.0

Update to 1.1.0

Author: generall

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "qdrant-client"
-version = "1.0.0"
+version = "1.1.0"
 edition = "2021"
 authors = ["Qdrant Team <team@qdrant.com>"]
 description = "Rust client for Qdrant Vector Search Engine"

proto/collections.proto

@@ -61,6 +61,11 @@ enum PayloadSchemaType {
   Text = 5;
 }
 
+enum QuantizationType {
+  UnknownQuantization = 0;
+  Int8 = 1;
+}
+
 message OptimizerStatus {
   bool ok = 1;
   string error = 2;
@@ -154,6 +159,18 @@ message OptimizersConfigDiff {
   optional uint64 max_optimization_threads = 8;
 }
 
+message ScalarQuantization {
+  QuantizationType type = 1; // Type of quantization
+  optional float quantile = 2; // Number of bits to use for quantization
+  optional bool always_ram = 3; // If true - quantized vectors always will be stored in RAM, ignoring the config of main storage
+}
+
+message QuantizationConfig {
+  oneof quantization {
+    ScalarQuantization scalar = 1;
+  }
+}
+
 message CreateCollection {
   string collection_name = 1; // Name of the collection
   reserved 2; // Deprecated
@@ -168,6 +185,7 @@ message CreateCollection {
   optional uint32 replication_factor = 11; // Number of replicas of each shard that network tries to maintain, default = 1
   optional uint32 write_consistency_factor = 12; // How many replicas should apply the operation for us to consider it successful, default = 1
   optional string init_from_collection = 13; // Specify name of the other collection to copy data from
+  optional QuantizationConfig quantization_config = 14;
 }
 
 message UpdateCollection {
@@ -207,6 +225,7 @@ message CollectionConfig {
   HnswConfigDiff hnsw_config = 2; // Configuration of vector index
   OptimizersConfigDiff optimizer_config = 3; // Configuration of the optimizers
   WalConfigDiff wal_config = 4; // Configuration of the Write-Ahead-Log
+  optional QuantizationConfig quantization_config = 5; // Configuration of the vector quantization
 }
 
 enum TokenizerType {

proto/points.proto

@@ -158,6 +158,18 @@ message WithVectorsSelector {
   }
 }
 
+message QuantizationSearchParams {
+  /*
+  If set to true, search will ignore quantized vector data
+   */
+  optional bool ignore = 1;
+
+  /*
+  If true, use original vectors to re-score top-k results. Default is true.
+   */
+  optional bool rescore = 2;
+}
+
 message SearchParams {
   /*
   Params relevant to HNSW index. Size of the beam in a beam-search.
@@ -169,6 +181,11 @@ message SearchParams {
   Search without approximation. If set to true, search may run long but with exact results.
   */
   optional bool exact = 2;
+
+  /*
+  If set to true, search will ignore quantized vector data 
+  */
+  optional QuantizationSearchParams quantization = 3;
 }
 
 message SearchPoints {
@@ -359,9 +376,19 @@ message Match {
     int64 integer = 2; // Match integer
     bool boolean = 3; // Match boolean
     string text = 4; // Match text
+    RepeatedStrings keywords = 5; // Match multiple keywords
+    RepeatedIntegers integers = 6; // Match multiple integers
   }
 }
 
+message RepeatedStrings {
+  repeated string strings = 1;
+}
+
+message RepeatedIntegers {
+  repeated int64 integers = 1;
+}
+
 message Range {
   optional double lt = 1;
   optional double gt = 2;

src/qdrant.rs

@@ -171,6 +171,34 @@ pub struct OptimizersConfigDiff {
 }
 #[allow(clippy::derive_partial_eq_without_eq)]
 #[derive(Clone, PartialEq, ::prost::Message)]
+pub struct ScalarQuantization {
+    /// Type of quantization
+    #[prost(enumeration = "QuantizationType", tag = "1")]
+    pub r#type: i32,
+    /// Number of bits to use for quantization
+    #[prost(float, optional, tag = "2")]
+    pub quantile: ::core::option::Option<f32>,
+    /// If true - quantized vectors always will be stored in RAM, ignoring the config of main storage
+    #[prost(bool, optional, tag = "3")]
+    pub always_ram: ::core::option::Option<bool>,
+}
+#[allow(clippy::derive_partial_eq_without_eq)]
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct QuantizationConfig {
+    #[prost(oneof = "quantization_config::Quantization", tags = "1")]
+    pub quantization: ::core::option::Option<quantization_config::Quantization>,
+}
+/// Nested message and enum types in `QuantizationConfig`.
+pub mod quantization_config {
+    #[allow(clippy::derive_partial_eq_without_eq)]
+    #[derive(Clone, PartialEq, ::prost::Oneof)]
+    pub enum Quantization {
+        #[prost(message, tag = "1")]
+        Scalar(super::ScalarQuantization),
+    }
+}
+#[allow(clippy::derive_partial_eq_without_eq)]
+#[derive(Clone, PartialEq, ::prost::Message)]
 pub struct CreateCollection {
     /// Name of the collection
     #[prost(string, tag = "1")]
@@ -205,6 +233,8 @@ pub struct CreateCollection {
     /// Specify name of the other collection to copy data from
     #[prost(string, optional, tag = "13")]
     pub init_from_collection: ::core::option::Option<::prost::alloc::string::String>,
+    #[prost(message, optional, tag = "14")]
+    pub quantization_config: ::core::option::Option<QuantizationConfig>,
 }
 #[allow(clippy::derive_partial_eq_without_eq)]
 #[derive(Clone, PartialEq, ::prost::Message)]
@@ -286,6 +316,9 @@ pub struct CollectionConfig {
     /// Configuration of the Write-Ahead-Log
     #[prost(message, optional, tag = "4")]
     pub wal_config: ::core::option::Option<WalConfigDiff>,
+    /// Configuration of the vector quantization
+    #[prost(message, optional, tag = "5")]
+    pub quantization_config: ::core::option::Option<QuantizationConfig>,
 }
 #[allow(clippy::derive_partial_eq_without_eq)]
 #[derive(Clone, PartialEq, ::prost::Message)]
@@ -555,6 +588,32 @@ impl PayloadSchemaType {
 }
 #[derive(Clone, Copy, Debug, PartialEq, Eq, Hash, PartialOrd, Ord, ::prost::Enumeration)]
 #[repr(i32)]
+pub enum QuantizationType {
+    UnknownQuantization = 0,
+    Int8 = 1,
+}
+impl QuantizationType {
+    /// String value of the enum field names used in the ProtoBuf definition.
+    ///
+    /// The values are not transformed in any way and thus are considered stable
+    /// (if the ProtoBuf definition does not change) and safe for programmatic use.
+    pub fn as_str_name(&self) -> &'static str {
+        match self {
+            QuantizationType::UnknownQuantization => "UnknownQuantization",
+            QuantizationType::Int8 => "Int8",
+        }
+    }
+    /// Creates an enum from field names used in the ProtoBuf definition.
+    pub fn from_str_name(value: &str) -> ::core::option::Option<Self> {
+        match value {
+            "UnknownQuantization" => Some(Self::UnknownQuantization),
+            "Int8" => Some(Self::Int8),
+            _ => None,
+        }
+    }
+}
+#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash, PartialOrd, Ord, ::prost::Enumeration)]
+#[repr(i32)]
 pub enum TokenizerType {
     Unknown = 0,
     Prefix = 1,
@@ -1657,6 +1716,18 @@ pub mod with_vectors_selector {
 }
 #[allow(clippy::derive_partial_eq_without_eq)]
 #[derive(Clone, PartialEq, ::prost::Message)]
+pub struct QuantizationSearchParams {
+    ///
+    /// If set to true, search will ignore quantized vector data
+    #[prost(bool, optional, tag = "1")]
+    pub ignore: ::core::option::Option<bool>,
+    ///
+    /// If true, use original vectors to re-score top-k results. Default is true.
+    #[prost(bool, optional, tag = "2")]
+    pub rescore: ::core::option::Option<bool>,
+}
+#[allow(clippy::derive_partial_eq_without_eq)]
+#[derive(Clone, PartialEq, ::prost::Message)]
 pub struct SearchParams {
     ///
     /// Params relevant to HNSW index. Size of the beam in a beam-search.
@@ -1667,6 +1738,10 @@ pub struct SearchParams {
     /// Search without approximation. If set to true, search may run long but with exact results.
     #[prost(bool, optional, tag = "2")]
     pub exact: ::core::option::Option<bool>,
+    ///
+    /// If set to true, search will ignore quantized vector data
+    #[prost(message, optional, tag = "3")]
+    pub quantization: ::core::option::Option<QuantizationSearchParams>,
 }
 #[allow(clippy::derive_partial_eq_without_eq)]
 #[derive(Clone, PartialEq, ::prost::Message)]
@@ -2014,7 +2089,7 @@ pub struct FieldCondition {
 #[allow(clippy::derive_partial_eq_without_eq)]
 #[derive(Clone, PartialEq, ::prost::Message)]
 pub struct Match {
-    #[prost(oneof = "r#match::MatchValue", tags = "1, 2, 3, 4")]
+    #[prost(oneof = "r#match::MatchValue", tags = "1, 2, 3, 4, 5, 6")]
     pub match_value: ::core::option::Option<r#match::MatchValue>,
 }
 /// Nested message and enum types in `Match`.
@@ -2034,10 +2109,28 @@ pub mod r#match {
         /// Match text
         #[prost(string, tag = "4")]
         Text(::prost::alloc::string::String),
+        /// Match multiple keywords
+        #[prost(message, tag = "5")]
+        Keywords(super::RepeatedStrings),
+        /// Match multiple integers
+        #[prost(message, tag = "6")]
+        Integers(super::RepeatedIntegers),
     }
 }
 #[allow(clippy::derive_partial_eq_without_eq)]
 #[derive(Clone, PartialEq, ::prost::Message)]
+pub struct RepeatedStrings {
+    #[prost(string, repeated, tag = "1")]
+    pub strings: ::prost::alloc::vec::Vec<::prost::alloc::string::String>,
+}
+#[allow(clippy::derive_partial_eq_without_eq)]
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct RepeatedIntegers {
+    #[prost(int64, repeated, tag = "1")]
+    pub integers: ::prost::alloc::vec::Vec<i64>,
+}
+#[allow(clippy::derive_partial_eq_without_eq)]
+#[derive(Clone, PartialEq, ::prost::Message)]
 pub struct Range {
     #[prost(double, optional, tag = "1")]
     pub lt: ::core::option::Option<f64>,

tests/integration-tests.sh

@@ -11,7 +11,7 @@ function stop_docker()
 # Ensure current path is project root
 cd "$(dirname "$0")/../"
 
-QDRANT_VERSION='v1.0.3'
+QDRANT_VERSION='v1.1.0'
 
 QDRANT_HOST='localhost:6333'